FAQ | Print Page | Sign In | Register
2019 Abstracts
Share |

Below is a "sneak peek" at the 2019 abstracts. Soon, they'll be published in the Online Journal of Public Health Informatics. Until then, they are offered below. There is no order to the abstracts, so use your browser's search function to look for authors or titles to help locate the abstracts you may wish to view.

 

Lightning Talk

Improving Varicella Investigation Completeness in Pennsylvania

Jonah Long, Wayne Fleming, Janine Strick

Pennsylvania Department of Health, Jackson Center, Pennsylvania, United States


Objective The objective of this study was to evaluate the impact of efforts made to improve the completeness of select varicella (chickenpox) case investigation variables.
Introduction Routine childhood administration of varicella-containing vaccine has resulted in the number of varicella (chickenpox) cases in Pennsylvania falling from nearly 3,000 cases in 2007 to less than 400 cases in 2017. Prior to 2018, the completeness of varicella case investigation data documented in Pennsylvania’s electronic disease surveillance system (PA-NEDSS) was not routinely monitored by Department of Health (DOH) staff. A pilot project was initiated in April 2018 to monitor and improve completeness of select varicella case investigation variables.
Methods Varicella cases reported to PA-NEDSS during MMWR year 2018 (MMWR weeks 1 – 26) in Pennsylvania (excluding Philadelphia County) with a classification status of probable or confirmed were included in the pilot project (n=223). DOH epidemiology staff prioritized 11 key varicella investigation variables and developed a SAS program to identify cases with missing data, which were summarized in weekly reports and provided to DOH immunization staff for follow-up. DOH immunization staff reviewed missing data reports and communicated with case investigators to reconcile missing data. Varicella case data from the project period were compared with a 10-year baseline to evaluate the 11 targeted variables for change in percent completion.
Results Percent completion of all 11 variables improved during the intervention period, with a median relative increase of 10.2% (range: 4.2% — 25.5%) compared to baseline. All but two variables (pregnancy status and number of days hospitalized) exhibited a statistically significant (p<0.05) improvement in percent completion. In addition, among eight variables that include an unknown response option, only one variable (number of varicella vaccine doses received) measured an increase in the percentage of unknown responses during the project period compared with baseline; however, this increase was not statistically significant (p=0.180).
Conclusions Prioritization of key varicella investigation variables for improved completion was successful and did not result in significant increases of unknown responses. As varicella cases become less common, varicella case investigation data become increasingly important. Increased completeness of these data will enhance DOH communication of varicella surveillance findings, particularly for severe cases. Based on the success of this interagency collaboration, similar efforts are being developed for additional reportable conditions.
Acknowledgement
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Jonah Long is an ELC-funded epidemiologist who has worked for the Pennsylvania Department of Health since 2011. In his current role, he works to improve and enhance information systems used by the Bureau of Epidemiology, as well as to improve usage and dissemination of surveillance data.
Brief summary (100 words) of Presentation to be Used in Conference Program As varicella (chickenpox) infections become less common due to routine childhood immunization, varicella case investigation data grow increasingly important. The Pennsylvania Department of Health began an interagency collaboration to monitor and improve the completeness of varicella investigations in 2018. Key variables were prioritized for follow up and completeness rates were compared to varicella investigations from the previous ten years.


Table 1. Varicella variable completeness during 2008-2017 and 2018.


Variable 2008-2017 (n=8,895) 2018 (n=223)  
Mean Percent Complete (standard deviation) Percent Complete Relative Percent Change χ2 p-value
Onset date 92.4 (2.9) 99.6 +7.8 <0.001
Hospitalization 90.8 (5.0) 100.0 +10.2 <0.001
Days hospitalized1 72.4 (11.7) 90.9 +25.5 0.1614
Pregnant2 89.8 (3.2) 97.7 +8.8 0.05904
Rash onset date 94.2 (1.3) 99.6 +5.7 <0.001
Lesion severity 89.9 (1.7) 99.6 +10.7 <0.001
Immunocompromised 81.5 (4.9) 99.6 +22.1 <0.001
Complications 81.2 (7.4) 99.6 +22.6 <0.001
Transmission setting known 82.9 (4.5) 99.6 +20.1 <0.001
Received varicella vaccine 90.9 (1.8) 99.6 +9.5 <0.001
Varicella vaccine doses received3 96.0 (1.7) 100.0 +4.2 0.02044



Evaluation of First Electronic Case Reports received in Illinois

Stacey Hoferka

Illinois Department of Public Health, Chicago, Illinois, United States


Objective Comparison of content in eCR and ELR cases reporting
Review technical challenges and strategies for data management
Introduction Communicable disease reporting from providers can be a time-consuming process that results in delayed or incomplete reporting of infectious diseases, limiting public health's ability to respond quickly to prevent or control disease. The recent development of an HL7 standard for automated Electronic initial case reports (eICR) represents an important advancement for public health surveillance. The Illinois Department of Public Health (IDPH) participated in a pilot with the Public Health Informatics Institute and an Illinois-based provider group to accept eICR reports for Gonorrhea and Chlamydia.
Methods The provider group working with their EHR vendor submitted a batch of CT and GC reports directly to IDPH in September 2017 according to the published eICR standard. A summary of the provider and PHII work has been presented previously in the STI eCR Learning Community. The eICR reports received from the provider were compared to case report data in the communicable disease surveillance system, I-NEDSS. Data was extracted from I-NEDSS that included race and ethnicity, timing of specimen collection, result, ELR submission surveillance action and treatment.
Results IDPH received a batch of 89 files containing 77 unique persons, with 54 chlamydia (CT), 13 Gonorrhea (GC) and 10 co-infected case reports. The communicable disease surveillance system had captured 76 (98.7%) of the persons reported in the pilot. Among those, an Electronic Laboratory Report (ELR) was received for 72 (95%) cases, on average within 1 day of the lab report date. Data in I-NEDSS had a completion of 45% for race and ethnicity compared to 99% for race and 92% for ethnicity in the eICR files. Information on treatment in the surveillance system was reported for 18 (24%) cases compared to 67 (87%) cases.
Conclusions This pilot was the first submission of real patient data submitted using the eICR standard to IDPH. Data was more complete from provider eICR reports for key demographic of race and ethnicity and treatment. A comparison with the current surveillance system showed near complete and timely case capture from ELR data. Integrated reporting of both ELR and eICR can produce a more complete case report through automated submissions and potentially reduce burden of data collection on health department communicable disease investigators. As public health reporting moves in this direction, public health agencies will have some substantial tasks to correctly ingest, map and interpret the increased amounts of information that are contained in the eICR. Further, the advantages of case reporting will be dependent on automated processes within the communicable disease system to merge data and apply business rules to automatically process completed case reports for high volume diseases, such as STIs. This work will continue as providers are ready to submit reports from different vendor products from a near real-time production environment.
Acknowledgement
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Stacey Hoferka is the Epidemiology Informatician at the Illinois Department of Public Health. She has worked on implementing informatic solutions since 2012, including implementation of statewide syndromic surveillance, and program managment of the communicable disease surveillance system, I-NEDSS
Brief summary (100 words) of Presentation to be Used in Conference Program The Illinois Department of Public Health (IDPH) participated in a pilot with the Public Health Informatics Institute and an Illinois-based provider group to accept eICR reports for Gonorrhea and Chlamydia. The eICR reports received from the provider were compared to case report data in the communicable disease surveillance system. A comparison with the current surveillance system showed near complete and timely case capture from ELR data. Integrated reporting of both ELR and eICR can produce a more complete case report through automated submissions and potentially reduce burden of data collection on health department communicable disease investigators.


Automating Antimicrobial Usage Reporting

Andrew Walsh1, Cindy Hou2, Nikunj Vyas2

1Health Monitoring, Pittsburgh, Pennsylvania, United States, 2Jefferson Health, Cherry Hill, New Jersey, United States


Objective To leverage existing healthcare transaction messages to automate the aggregation of antimicrobial usage statistics in a method compatible with submission to the National Healthcare Safety Network (NHSN) Antimicrobial Usage module.
Introduction Antimicrobial stewardship is crucial to the ongoing viability of existing therapies. To facilitate this stewardship, NHSN allows hospitals to submit data on their antimicrobial usage and receive feedback on how their usage compares to other facilities.1 This feedback can be used by hospital personnel to assess whether their antimicrobial policies are consistent with current best practices.
Participation in this program has so far been limited. There are several barriers to participating, including the challenge of mapping local medication information to the NHSN list of antimicrobials, the burden of tabulating the necessary statistics, and the technical requirements of generating appropriate CDA documents for submission. An automated solution that obtained the necessary data from existing HL7 interfaces and generated CDA documents in the correct format could significantly lower some of the barriers to submitting antimicrobial usage information to NHSN.
Methods A continuous feed of HL7 ADT and RAS messages was established between a multi-hospital health system and the EpiCenter syndromic surveillance system. Medication administration data elements included time of administration, patient location, and a facility-specific medication identifier. Patient data included time of admissions, transfers and discharges and all relevant patient locations.
Facility medication codes were reconciled to NHSN antimicrobial identifiers via a multi-step, semi-automated process. A medication formulary was provided by the health system mapping their codes to National Drug Code (NDC) identifiers. The RxNorm API2 was queried to map NDC identifiers to RxNorm identifiers. A second round of RxNorm API queries linked the formulation-specific RxNorm identifiers to related parent RxNorm identifiers for antimicrobials provided by NHSN. The final mapping from facility medication codes to NHSN antimicrobial identifiers was manually reviewed and edited to remove duplicates and to add links by name that were not found automatically.
NHSN requires reporting by administration route, which was provided in most cases by the formulary. In rare cases, that route was not sufficiently specific and had to be refined by looking at the site of administration for individual doses.
Therapy days were calculated as the total number of unique patient identifiers receiving at least one dose of a given antimicrobial; these were totaled per day, per route, and per inpatient location. Days present were calculated as the total number of unique patient identifiers associated with a given inpatient location at any time during each day. Facility-wide inpatient admissions were calculated as the total number of unique patient identifiers associated with an admission to any inpatient location during each day.
Results The RxNorm API yielded mapping between 15,472 NDC identifiers and 847 RxNorm codes, covering 86 (96%) NHSN antimicrobials. An initial merge using the NDC identifiers from the provided formulary yielded 252 matches to NHSN antimicrobials. Manual reconciliation eliminated duplicates to leave 239 unique antimicrobials from the formulary. Since not all NDC identifiers in the formulary could be associated with an RxNorm code, there was the potential for additional antimicrobials to be present but not matched to an NHSN code. The names of the NHSN antimicrobials were used to search the generic and brand names of medications in the formulary, yielding 6 additional antimicrobials with appropriate routes. After these steps of automated and manual reconciliation and excluding formulations administered via nonreportable routes, a total of 216 antimicrobial formulations were identified that can be reported to NHSN. These covered 67 (74%) NHSN antimicrobials.
For July 2018, 206,921 medication administration messages were received, including 11,637 administrations of 48 NHSN antimicrobials at 14 NHSN inpatient locations were observed across all four NHSN routes. These represented 7% of completed administrations. They accounted for 6,909 days of therapy with all antimicrobials via all routes at all locations. Figure 1 shows the time series of days of therapy by NHSN route of administration. A total of 950 (0.6%) administrations had medication code 99999 and could not be identified.
An additional 189,717 ADT messages for 5,420 distinct visits were received. These yielded 13,885 facility-wide days present and a 14,797 location-specific days present summed across all inpatient locations and facilities, from 4,054 facility-wide admissions. Figure 2 shows the time series of facility-wide days present and summed location-specific days present for each facility.
Conclusions Reconciling local facility formularies with a national standardized list of antimicrobials can be a complicated task requiring some amount of human intervention. Once completed, however, HL7 messages from existing interface engines can supply sufficient information for calculating the necessary antimicrobial usage statistics to report to NHSN.
Acknowledgement Health Monitoring would like to thank the New Jersey Department of Health for financial support of this work.
References 1. Centers for Disease Control and Prevention [Internet]. Atlanta: National Health Safety Network; 2017 Dec 29. Antimicrobial Use and Resistance (AUR) Module; 2018 Jan [cited 2018 Sep 10]. Available from: https://www.cdc.gov/nhsn/PDFs/pscManual/11pscAURcurrent.pdf
2. Peters LB, Bodenreider O. RESTful services for accessing RxNorm. AMIA Annu Symp Proc. 2010:983.
Brief bio for lead author/ presenter to be used by session moderators at the conference Andy Walsh is the Chief Science Officer at Health Monitoring where he oversees research for EpiCenter, a syndromic surveillance system for a large portion of the United States. Previously, he developed software for visualizing and analyzing viral genome data, as part of a research program for HIV and influenza vaccines. He has a PhD in molecular microbiology from the Bloomberg School of Public Health.
Brief summary (100 words) of Presentation to be Used in Conference Program Hospitals can submit antimicrobial usage data to the National Healthcare Safety Network (NHSN) and receive feedback regarding antimicrobial stewardship. To submit, hospital medication codes must be linked to the NHSN antimicrobial list. Automated RxNorm API queries, manual reconciliation and excluding nonreportable formulations identified 216 reportable antimicrobial formulations from a large health system’s formulary.
For July 2018, 11,637 administrations of 48 NHSN antimicrobials were observed, accounting for 6,909 days of therapy. ADT messages yielded 13,885 facility-wide days present from 4,054 facility-wide admissions. Calculating these statistics using processed messages demonstrates that levering existing interfaces can lower barriers to NHSN antimicrobial usage submission.



Figure 1: Time Series of Antimicrobial Days of Therapy by Route




Figure 2: Time Series of Days-Present from ADT Data by Facility and Aggregation



Epi Archive: Automated Synthesis of Global Notifiable
Disease Data

Hari S. Khalsa, Sergio R. Cordova, Nicholas Generous, Prabhu S. Khalsa

A-1, Los Alamos National Laboratory, Los Alamos, New Mexico, United States


Objective Automatically collect and synthesize global notifiable disease data and make it available to humans and computers. Provide the data on the web and within the Biosurveillance Ecosystem (BSVE) as a novel data stream. These data have many applications including improving the prediction and early warning of disease events.
Introduction Government reporting of notifiable disease data is common and widespread, though most countries do not report in a machine-readable format. This is despite the WHO International Health Regulations stating that “[e]ach State Party shall notify WHO, by the most efficient means of communication available.” 1
Data are often in the form of a file that contains text, tables and graphs summarizing weekly or monthly disease counts. This presents a problem when information is needed for more data intensive approaches to epidemiology, biosurveillance and public health. While most nations likely store incident data in a machine-readable format, governments can be hesitant to share data openly for a variety of reasons that include technical, political, economic, and motivational2.
A survey conducted by LANL of notifiable disease data reporting in over fifty countries identified only a few websites that report data in a machine-readable format. The majority (>70%) produce reports as PDF files on a regular basis. The bulk of the PDF reports present data in a structured tabular format, while some report in natural language or graphical charts.
The structure and format of PDF reports change often; this adds to the complexity of identifying and parsing the desired data. Not all websites publish in English, and it is common to find typos and clerical errors.
LANL has developed a tool, Epi Archive, to collect global notifiable disease data automatically and continuously and make it uniform and readily accessible.
Methods A survey of the national notifiable disease reporting systems is periodically conducted notating how the data are reported and in what formats. We determined the minimal metadata that is required to contextualize incident counts properly, as well as optional metadata that is commonly found.
The development of software to regularly ingest notifiable disease data and make it available involves three to four main steps: scraping, detecting, parsing and persisting.
Scraping: we examine website design and determine reporting mechanisms for each country/website, as well as what varies across the reporting mechanisms. We then design and write code to automate the downloading of data for each country. We store all artifacts presented as files (PDF, XLSX, etc.) in their original form, along with appropriate metadata for parsing and data provenance.
Detecting: This step is required when parsing structured non-machine-readable data, such as tabular data in PDF files. We combine the Nurminen methodology of PDF table detection with in-house heuristics to find the desired data within PDF reports3.
Parsing: We determine what to extract from each dataset and parse these data into uniform data structures, correctly accommodating the variations in metadata (e.g., time interval definitions) and the various human languages.
Persisting: We store the data in the Epi Archive database and make it available on the internet and through the BSVE. The data is persisted into a structured and normalized SQL database.
Results Epi Archive currently contains national and/or subnational notifiable disease data from thirty-nine nations. When a user accesses the Epi Archive site, they are able to peruse, chart and download data by country, subregion, disease and time interval. Access to a cached version of the original artifacts (e.g. PDF files), a link to the source and additional metadata is also available through the user interface. Finally, to ensure machine-readability, the data from Epi Archive can be reached through a REST API. http://epiarchive.bsvgateway.org/
Conclusions LANL, as part of a currently funded DTRA effort, is automatically and continually collecting global notifiable disease data. While thirty-nine nations are in production, more are being brought online in the near future. These data are already being utilized and have many applications, including improving the prediction and early warning of disease events.
Acknowledgement This project is supported by the Chemical and Biological Technologies Directorate Joint Science and Technology Office (JSTO), Defense Threat Reduction Agency (DTRA).
References [1] WHO International Health Regulations, edition 3. http://apps.who.int/iris/bitstream/10665/246107/1/9789241580496-eng.pdf
[2] van Panhuis WG, Paul P, Emerson C, et al. A systematic review of barriers to data sharing in public health. BMC Public Health. 2014. 14:1144. doi:10.1186/1471-2458-14-1144
[3] Nurminen, Anssi. "Algorithmic extraction of data in tables in PDF documents." (2013).
Brief bio for lead author/ presenter to be used by session moderators at the conference Hari Khalsa is a computer scientist at the Los Alamos National Laboratory. He is the prinicple investigator of Epi Archive as well as the lead artchitect and developer. He has an extensive background in Software Engineering and data science, with much experience in modeling and simulation. He enjoys taking large heterogenous data sets and making sense of it (as in Epi Archive).
Brief summary (100 words) of Presentation to be Used in Conference Program The presentation will provide an overview of the motivation and methods of Epi Archive, software designed to synthesize global notifiable disease data. Gathering epidemiological data from multiple government public health websites can be a difficult journey, from finding the disparate sources, to interpreting and translating (if necessary). Parsing the data for analysis, decision support or simple comparison can be even more difficult, due to various irregular formats most of which is not machine-readable. Epi Archive solves these problems by bringing together notifiable disease data from many countries and presenting the data through a human-friendly website and a REST API.




Methods for combining data from multi-jurisdiction sentinel surveillance projects

Elizabeth Torrone1, Eloisa Llata1, Jaeyoung Hong1, Preeti Pathela2

1DSTDP, CDC, Atlanta, Georgia, United States, 2New York City Department of Health and Mental Hygiene , New York City, New York, United States


Objective To identify best practices for combining public health data for multi-jurisdiction surveillance projects.
Introduction Sentinel surveillance, where selected jurisdictions follow standardized protocols to collect and report enhanced public health data not available through other routine surveillance efforts, is a key part of national surveillance of sexually transmitted diseases (STDs). Although four STDs are nationally notifiable conditions (chlamydia, gonorrhea, syphilis and chancroid), the burden of these conditions (over 2.3 million cases were reported in 2017) limits the amount of detailed clinical and demographic data available for all cases. Sentinel surveillance in clinical settings serving at-risk populations, such as STD clinics, provides an opportunity to collect enhanced data elements on persons seeking STD-related services, such as sex of sex partners and anatomic site of infection. However, there are challenges in combining data across jurisdictions as estimated effect measures may vary by jurisdiction (e.g., some may have higher observed burden of disease among certain populations) and the amount of data contributed by jurisdiction may vary; combined this could lead to biased estimates if heterogeneity is not taken into account.
Methods Using data from the STD Surveillance Network (SSuN), a sentinel surveillance project implemented in 10 jurisdictions, we investigated the effect of using different statistical methods to combine data across jurisdictions. We evaluated 5 methodologies:

● “Fully stratified” where estimates were provided separately for each jurisdiction;
● “Aggregated” where numerators and denominators were summed across jurisdictions without weighting;
● “Mean estimate” where the mean of the jurisdiction-specific estimates was estimated;
● “Random effects” where jurisdiction-specific estimates were combined using an inverse variance weighted random effects model to adjust for heterogeneity between jurisdictions; and
● “Stratified random effects” where a possible effect modifier was identified and used to group jurisdictions prior to calculating the estimate from the random effects model.

Through SSuN, jurisdictions collect visit-level data on patients attending selected STD clinics and report clinical and demographic data. As an illustrative example, we estimated rectal gonorrhea positivity among gay, bisexual, and other men who have sex with men (MSM) attending participating clinics. Jurisdiction-specific positivity was estimated as the # of unique MSM testing positive at least once for rectal gonorrhea divided by all MSM tested 1 or more times for rectal gonorrhea in all of the clinics in the jurisdiction. The stratifying variable for the stratified random effects method was the percent of MSM screened in the jurisdiction’s clinics, as low screening coverage may reflect targeted testing of MSM likely to be infected which may inflate observed positivity. For each of the five methods, we estimated rectal gonorrhea positivity and the corresponding 95% confidence interval (CI).
Results In 2017, 123,210 patients attended 30 STD clinics participating in the 10 SSuN jurisdictions, of which 31,052 (25.2%) were identified as MSM (jurisdiction-specific range: 8.8% to 70.0%). (Table 1) One jurisdiction (I) accounted for 39% of all MSM included in the analysis while one jurisdiction (J) accounted for only 1.6% of MSM included. The proportion of MSM tested for rectal gonorrhea at least once varied by jurisdiction, ranging from 44.3% to 76.9%. The fully stratified method identified differences in rectal gonorrhea positivity across jurisdictions, with jurisdiction-specific positivity ranging from 9.9% to 24.1%. Aggregating across jurisdictions masked this heterogeneity and provided a single summary estimate of 15.2% (95% CI: 14.7, 15.7). Taking the mean across the jurisdiction-specific estimates also provided a summary estimate; however, the uncertainty of the estimate increased (15.8%, 95% CI: 13.3, 18.7). Accounting for the heterogeneity by using a random effects model resulted in an estimate of 15.5% (95% CI: 13.9, 17.2). After stratifying by a likely confounder (% of MSM screened); the random effects estimate among 3 jurisdictions with lower screening coverage (<60%) was 19.7% (95% CI: 14.6, 24.8) and among 7 jurisdictions with higher screening coverage (≥60%) was 14.3% (95% CI: 12.9, 15.7).
Conclusions In a sentinel surveillance project implemented in 10 jurisdictions, there was substantial heterogeneity in the observed proportion of MSM testing positive for rectal gonorrhea in selected STD clinics. Although a stratified analysis captured the heterogeneity across jurisdictions, it may not be feasible to present fully stratified estimates for all analyses (e.g., surveillance reports likely provide metrics for multiple diseases). Additionally, it limits the ability to succinctly communicate key findings. Aggregating numerators and denominators across jurisdictions to calculate a single summary estimate masks this heterogeneity and biases estimates toward high volume jurisdictions. Taking the mean across jurisdictions ensures that high-volume jurisdictions do not bias the overall estimate; however, the mean may be biased by very high or very low positivity estimates in a few jurisdictions. Using a random effects model accounted for both varying sample sizes and differences in observed heterogeneity; although the summary estimate was similar to the aggregate in this example, the wider 95% CI more accurately reflects the uncertainty in the estimate. Finally, stratifying by a likely effect measure modifier (% of MSM screened) prior to estimating the measure from the random effects model captured key differences in jurisdictions while still providing a limited number of summary estimates. Analysts using data from multi-jurisdiction surveillance projects should fully investigate possible biases when combining estimates across jurisdictions. If there is observed heterogeneity across jurisdictions and it is not feasible to provide fully stratified estimates, analysts could consider using methods to account for heterogeneity and minimize bias due to differing sample sizes, such as stratified random effects models.
Acknowledgement The authors thank the members of the SSuN Working Group: Bob Kohn, Robbie Madera, and Roxanne Kerani, as well as the SSuN jurisdictions that contributed data to this analysis.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Elizabeth Torrone is the lead of the Surveillance and Special Studies team in the Division of STD Prevention at the Centers for Disease Control and Prevention.
Brief summary (100 words) of Presentation to be Used in Conference Program Using data from the STD Surveillance Network (SSuN), a sentinel surveillance project implemented in 10 jurisdictions, we investigated the effect of using different statistical methods to combine data across jurisdictions to estimate trends in rectal gonorrhea positivity among men who have sex with men tested in selected STD clinics.


Trends in rectal gonorrhea among MSM tested in STD clinics participating in the STD Surveillance Network (SSuN), 2017


Jurisdiction # of patients % of patients identified as MSM % of MSM patients screened for rectal gonorrhea % MSM testing positive for rectal gonorrhea
Fully stratified Aggregated across jurisdictions Mean of jurisdiction estimates Random effects model Stratified random effects model
A 6,262 14.3% 44.3% 14.4% n/a n/a n/a 19.7%
(95% CI: 14.6, 24.8)
B 8,160 8.8% 46.4% 24.1% n/a n/a n/a
C 13,519 15.7% 58.5% 20.9% n/a n/a n/a
D 6,017 44.6% 65.6% 18.2% n/a n/a n/a 14.3%
(95% CI: 12.9, 15.7)
E 15,083 16.8% 65.8% 14.7% n/a n/a n/a
F 9,188 48.4% 67.8% 15.8% n/a n/a n/a
G 10,657 30.1% 68.1% 13.5% n/a n/a n/a
H 3,665 46.2% 70.5% 12.3% n/a n/a n/a
I 49,792 24.4% 71.7% 14.3% n/a n/a n/a
J 867 70.0% 76.9% 9.9% n/a n/a n/a
TOTAL 123,210 25.2% 67.5% n/a 15.2%
(95% CI: 14.7, 15.7)
15.8%
(95% CI: 13.3, 18.7)
15.5%
(95% CI: 13.9, 17.2)
n/a
 



An Algorithm for Early Outbreak Detection in Multiple Data Streams

Sesha K. Dassanayake1, Joshua French2

1Mathematics and Computer Science, Rhodes College, Memphis, Colorado, United States, 2University of Colorado Denver, Denver, Colorado, United States


Objective To propose a computationally simple, fast, and reliable temporal method for early event detection in multiple data streams
Introduction Current biosurveillance systems run multiple univariate statistical process control (SPC) charts to detect increases in multiple data streams1. The method of using multiple univariate SPC charts is easy to implement and easy to interpret. By examining alarms from each control chart, it is easy to identify which data stream is causing the alarm. However, testing multiple data streams simultaneously can lead to multiple testing problems that inflate the combined false alarm probability. Although methods such as the Bonferroni correction can be applied to address the multiple testing problem by lowering the false alarm probability in each control chart, these approaches can be extremely conservative.

Biosurveillance systems often make use of variations of popular univariate SPC charts such as the Shewart Chart, the cumulative sum chart (CUSUM), and the exponentially weighted moving average chart (EWMA). In these control charts an alarm is signaled when the charting statistic exceeds a pre-defined control limit. With the standard SPC charts, the false alarm rate is specified using the in-control average run length (ARL0). If multiple charts are used, the resulting multiple testing problem is often addressed using family-wise error rate (FWER) based methods – that are known to be conservative - for error control.

A new temporal method is proposed for early event detection in multiple data streams. The proposed method uses p-values instead of the control limits that are commonly used with standard SPC charts. In addition, the proposed method uses false discovery rate (FDR) for error control over the standard ARL0 used with conventional SPC charts. With the use of FDR for error control, the proposed method makes use of more powerful and up-to-date procedures for handling the multiple testing problem than FWER-based methods.
Methods The proposed method can be applied to multiple univariate CUSUM or EWMA control charts. It can also be applied to a variation of the Hotelling T2 chart which is a common multivariate process monitoring method. The Hotelling T2 chart is analogous to the Shewart chart. Montgomery et. al2 proposed a variation of the Hotelling T2 chart where the T2 statistic is decomposed into components that reflect the contribution of each data stream.

First, a tolerable FDR level specified. Then, at each new time step disease counts from each of the m geographic regions Y1t, Y2t, … , Ymt are collected. These disease counts are used to calculate the charting statistics S1t, S2t, … , Smt for each region. Meanwhile by inspecting historical data from each region, a non-outbreak period is identified. Using data from the non-outbreak period, bootstrap samples are drawn with replacement from each region and charting statistics are calculated. Using the charting statistics, empirical non-outbreak distributions are generated for each region. With the empirical non-outbreak distributions and the current charting statistic for each region S1t, S2t, … , Smt , corresponding p-values p1t, p2t, … , pmt are calculated. The multiple testing problem that occurs in comparing multiple p-values simultaneously is handled using the Storey -Tibshirani multiple comparison procedure3 to signal alarms.
Results As an illustration, all three methods – EWMA, CUSUM, and Hotelling T2 (components) - were applied to a data set consisting of weekly disease count data from 16 German federal sates gathered over a 11 year period from 2004-2014. The first two years of data from 2004-2005 were used to calibrate the model. Figure 1 shows the results for the state of Rhineland Palatinate. The three plots in Figure 1 show (a) the weekly disease counts for Rhineland Palatinate (b) the EWMA statistic (shown in red), the CUSUM statistic (shown in dark green) and (c) the component of the Hotelling T2 statistic corresponding to the illustrated state (shown in blue). The actual outbreak occurred on week 306 (shown by the orange line). Notice the two false alarms – alarms that occur before week 306 - with the Hotelling T2 statistic (dark green) on weeks 34 and 292; similarly, the CUSUM statistic signals a false alarm on week 57. However, the EWMA statistic does not signal any false alarms before the outbreak (red). Figure 2 zooms on the alarm statistics for the time period from weeks 280 – 330. The Hotelling T2 statistic misses the onset of actual outbreak on week 306. The CUSUM statistic detects the outbreak on week 307 – one week later. However, the EWMA statistic detects the outbreak right at the onset on week 306.
Conclusions Extensive simulation studies were conducted to compare the performance of the three control charts. Performance was compared in terms of (i) speed of detection and (ii) false alarm rates. Simulation results provide convincing evidence that the EWMA and the CUSUM are considerably speedier in detecting outbreaks compared to Hotelling T2 statistic: compared to the CUSUM, the EWMA is relatively faster. Similarly, the false alarm rates are larger for Hotelling T2 statistic compared to the EWMA and the CUSUM: false alarms are rare with both the EWMA and the CUSUM statistics with EWMA statistic having a slight edge. Overall, EWMA has the best performance out of the three methods with the new algorithm. Thus, the new algorithm applied to the EWMA statistic provides a simple, fast, and a reliable method for early event detection in multiple data streams.
Acknowledgement
References 1. Fricker RD. Introduction to Statistical Methods for Biosurveillance. New York, NY: Cambridge University Press; 2013. 399p.
2. Runger GC, Alt FB, Montgomery DC. Contributors to Multivariate Statistical Process Control Signal. Communications in Statistics – Theory and Methods. 1996; 25(10): 2203-2213.
3. Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proceedings of the National Academy of Sciences USA 2003; 100:9440–9445.

Brief bio for lead author/ presenter to be used by session moderators at the conference Sesha Dassanayake is an Assistant Professor at the Department of Mathematics and Computer Science at Rhodes College, Tennessee. He is a statistician and his current research interests are in developing statistical methods for biosurveillance using statistical process control charts.
Brief summary (100 words) of Presentation to be Used in Conference Program A novel algorithm is presented for detecting outbreaks in multiple data streams using statistical process control charts (SPC). The proposed temporal method is computationally simple, fast and reliable. The new algorithm uses p-values instead of the control limits that are commonly used with standard SPC charts. Furthermore, the proposed method uses false discovery rate (FDR) for error control over the standard ARL0 used with conventional SPC charts. With the use of FDR for error control, the proposed method makes use of more powerful procedures for handling the multiple testing problem than FWER-based methods.



Figure 1




Figure 2



Examining and improving reproducible research practices in public health

Kimberly J. Johnson, Bobbi J. Carothers, Xiaoyan Wang, Todd Combs, Douglas A. Luke, Jenine K. Harris

Public Health, Washington University in St. Louis, St. Louis, Missouri, United States


Objective Our presentation will explain current use, and barriers to use, of reproducible research practices in public health. We will also introduce a set of modules for researchers wishing to increase their use of reproducible research practices.
Introduction An important goal of surveillance is to inform public health interventions that aim to reduce the burden of disease in the population. Ensuring accuracy of results is paramount to achieving this goal. However, science is currently facing a “reproducibility crisis” where researchers have found it difficult or impossible to reproduce study results. Organized and well-documented statistical source code that is publicly available could increase research reproducibility, especially for research relying on publicly available surveillance data like the BRFSS, NHANES, GSS, SEER, and others. As part of our overall goal to improve training around reproducible research practices, we surveyed public health data analysts to determine current practices and barriers to code sharing.
Methods We conducted a cross-sectional web-based survey about code organization, documenting, storage, and sharing. We surveyed public health scientists who reported recently conducting statistical analyses for a report or manuscript. A total of 247 of 278 screened eligible to filled out the survey, and 209 answered every applicable question. We used traditional descriptive statistics and graphs to examine the survey data.
Results Most participants reported using some promising coding practices, with 67% including a prolog to introduce the code and 85% including comments in statistical code to explain operations and analyses. Of 10 common code organization strategies (e.g., naming variables logically, using white space), most (82%) respondents reported employing at least three of the strategies and just under half (47%) reported using five or more. Over half of participants (59%) reported code was developed or checked by two or more people. Many participants also reported promising file management habits for data and code used in publications. Three-quarters (75%) had a variable dictionary to accompany the dataset used, 48% created clean versions of code files, and 64% created clean versions of data files at the time of publication. Forty three percent of participants reported that if they suddenly left their current position, it would not be easy for others to find their statistical code files. Public code sharing was much less common among participants with just 9% reporting sharing code publicly from a recent publication and 20% of those surveyed reported ever having shared code publicly.

The top two barriers to using reproducible research practices were lack of training in reproducible research (n=108) and data privacy issues (n=105). Journals and funders not requiring reproducible practices were barriers selected by 94 and 84 participants, respectively. Few participants identified fear of errors being discovered (n=26) or a lack of workplace incentives (n=32) as barriers.
Conclusions Most participants were using some promising practices for organizing and formatting statistical code but few were sharing statistical code publicly. The second most frequently identified barrier to using reproduciible practices was data privacy, which could prohibit easily sharing a data source. With surveillance data often being publicly available, researchers working with surveillance data have overcome this top barrier without any change to current research practices. Researchers using surveillance data could greatly increase research reproduciblity by adopting promising practices for code formatting, like using logical variable names and limiting line length, and posting code in a public repository like GitHub.

To overcome the top barrier to use of reproducible research practices, lack of training, we developed brief training modules on formatting, documenting, and sharing statistical code and data. As part of our presentation we will introduce and provide access to these online modules. The introduction will focus on the relevant modules for surveillance data users, which include statistical code formatting and statistical code sharing via GitHub.

With fewer barriers to practicing reproducible research, public health researchers using surveillance data have the opportunity to be leaders in improving the adoption of reproducible research practices and subsequently improving the quality of research we rely on to improve public health.
Acknowledgement This project was supported by the Robert Wood Johnson Foundation (RWJF) Increasing
Openness and Transparency in Research program.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Dr. Johnson is an epidemiologist with over a decade of experience conducting cancer research. Her research has primarily focused on the etiology of pediatric cancer in both the general and high risk (Neurofibromatosis Type 1) populations. She has recently become interested in disparities in access to care among those diagnosed with cancer. She has published extensively using surveillance data from the Surveillance, Epidemiology and End Results program. In addition to her research activities, she leads the Brown School epidemiology and biostatistics specialization and teaches two epidemiology courses in the Brown School Masters of Public Health (MPH) program (Foundations in Public Health: Epidemiology and Advanced Data Analysis).
Brief summary (100 words) of Presentation to be Used in Conference Program An important goal of surveillance is to inform interventions that reduce the burden of disease in the population. Ensuring accuracy of results is paramount to achieving this goal. Science currently faces a “reproducibility crisis” where researchers find it difficult or impossible to reproduce study results. We surveyed public health scientists who reported recently conducting statistical analyses for a report or manuscript. Our results suggested opportunities to improve the use of reproducible research practices in public health. With the data often publicly available, researchers working with surveillance data have an opportunity to lead widespread adoption of reproducible research practices.


Comparing spatio-temporal methods of non-communicable disease surveillance.

Mico Hamlyn, Frederic B. Piel

SAHSU, Imperial College London, London, United Kingdom


Objective To determine the merits of different surveillance methods for cluster detection, in particular when used in conjuction with small area data. This will be investigated using a simulated framework. This is with a view to support further surviellance work using real small area data.
Introduction Health surveillance is well established for infectious diseases, but less so for non-communicable diseases. When spatio-temporal methods are used, selection often appears to be driven by arbitrary criteria, rather than optimal detection capabilities. Our aim is to use a theoretical simulation framework with known spatio-temporal clusters to investigate the sensitivity and specificity of several traditional (e.g. SatScan and Cusum) and Bayesian (incl. BaySTDetect and Dcluster) statistical methods for spatio-temporal cluster detection of non-communicable disease.
Methods Count data were generated using various random effects (RE). A subset of areas was randomly given an increased relative risk (RR) to simulate disease clusters. Simulations were conducted in R using a grid of 625 areas. We used 12 times= nteps within a hierarchical Poisson model. Multiple values of model parameters, including REs and the RR within clusters, were then tested. The range of RE (values) was derived from real-world data from England on common and rare diseases. RR ranging between 1.2 and 1.8 were tested to reflect both low and high exposures to pollutants and other risk factors. ROC analysis, based on 50 simulations, was used to assess the performance of each statistical method for each combination of parameter values.
Results Our ROC analysis suggested that SaTScan usually had the highest specificity at low sensitivities (<0.5), although its maximum sensitivity was often lower than when using the Bayesian methods. In scenarios where the RR within clusters was lower, all methods had less sensitivity at a given specificity. Cusum usually performed quite similarly to SatScan, while the two Bayesian methods considered often misidentified a high proportion of disease clusters. P-values generated by SaTScan need to be considered with caution as they did not relate closely with the sensitivity or specificity of the ROC curves from our simulations.
Conclusions Real-world investigations of spatio-temporal signals (e.g. disease clusters) are often complex and time consuming. Identifying the best method to reduce the risks of identifying false positives and of missing real clusters is therefore essential. Despite the inherent constraints of theoretical simulations, such a framework allows to objectively assess the performance of different methods. Overall, our simulation framework suggested that SatScan would usually be the easiest, most user-friendly and best performing space-time methods for non-communicable disease surveillance.
Acknowledgement
References
Brief bio for lead author/ presenter to be used by session moderators at the conference I am a Research Assistant with the Small Area Health Statistics Unit at Imperial College London. As an undergraduate I studied Natural Sciences at the University of Cambridge. My Masters in Public Health is from Imperial College London. My work focuses on surveillance methodologies, and on cancer surviellance. Other areas of interest include causal methods and metaresearch.
Brief summary (100 words) of Presentation to be Used in Conference Program Given the range of methods for performing cluster detection it can be challengin to determine which method to choose. We use a simulation framework to assess the merits of different surviellance methods, particularly when used in conjuction with small area data. This is with a view to support further surviellance work using real small area data. We also suggest that simulations can be useful to help determine detection thresholds and control false detection rates.


Influenza laboratory testing and its application in timely Department of Defense biosurveillance

Jessica F. Deerin, Paul E. Lewis

Armed Forces Health Surveillance Branch, Alexandria, Virginia, United States


Objective To describe influenza laboratory testing and results in the Military Health System and how influenza laboratory results may be used in DoD Electronic Surveillance System for Early Notification of Community-based Epidemics (ESSENCE)
Introduction Timely influenza data can help public health decision-makers identify influenza outbreaks and respond with preventative measures. DoD ESSENCE has the unique advantage of ingesting multiple data sources from the Military Health System (MHS), including outpatient, inpatient, and emergency department (ED) medical encounter diagnosis codes and laboratory-confirmed influenza data, to aid in influenza outbreak monitoring. The Influenza-like Illness (ILI) syndrome definition includes ICD-9 or ICD-10 codes that may increase the number of false positive alerts. Laboratory-confirmed influenza data provides an increased positive predictive value (PPV). The gold standard for influenza testing is molecular assays or viral culture. However, the tests may take 3-10 days to result. Rapid influenza diagnostic tests (RIDTs) have a lower sensitivity, but the timeliness of receiving a result improves to within <15 minutes. We evaluate the utility of RIDTs for routine ILI surveillance.
Methods Administrative medical encounters for ILI and influenza laboratory-confirmed data were analyzed from the MHS from June 2013 – September 2017 (Figure 1). The medical encounters and laboratory data include outpatient, inpatient, and ED data. The ILI syndrome case definition is a medical encounter during the study period with an ICD-9 or ICD-10 codes in any diagnostic position (ICD-9 codes = 79.99, 382.9, 460, 461.9, 465.8, 465.9, 466.0, 486, 487.0, 487.1, 487.8, 488, 490, 780.6, or 786.2; ICD-10 codes = B97.89, H66.9, J00, J01.9, J06.9, J09, J09.X, J10, J10.0, J10.1, J10.2, J10.8, J11, J11.0, J11.1, J11.2, J11.8, J12.89, J12.9, J18, J20.9, J40, R05, R50.9). The ILI dataset was limited to care provided in the MHS as laboratory data is only available for direct care. We describe influenza laboratory testing practices in the MHS. We aggregated the ILI encounters and RIDT positive results into daily counts and generated a weekly Pearson’s correlation.
Results Influenza tests are ordered throughout the year; the mean weekly percentage of ILI encounters in which an influenza laboratory test is ordered is 5.62%, with a range from 0.68% in the off season to 19.2% during peak influenza activity. The mean weekly percentage of positive influenza laboratory results among all ILI encounters is 0.82%, with a range from 0.01% to 5.73% (Figure 2). The percent of ILI encounters in which a test is ordered increases as the influenza season progresses. Influenza laboratory tests conducted in the MHS include RIDTs, PCR, culture, and DFA. Among all influenza tests ordered in the MHS, 66.0% were RIDTs, 22.7% were PCR, and 11.3% were viral culture. Often, a confirmatory test is ordered following a RIDT; 20% of RIDTs have follow-up tests. The mean timeliness of influenza test result data in the MHS was 11.26 days for viral culture, 2.94 days for PCR, and 0.11 days for RIDTs. The RIDT results were moderately correlated with ILI encounters for the entire year (mean weekly Pearson correlation coefficient rho=0.60, 95% CI: 0.55, 0.66, Figure 3). During the influenza season, the mean weekly Pearson correlation coefficient increases to rho=0.75, 95% CI: 0.70, 0.79.
Conclusions The DoD has the unique advantage of access to the electronic health record and laboratory tests and results of all MHS beneficiaries. This analysis provides evidence for increased utilization of positive RIDTs in ESSENCE. The moderate correlation between the ILI syndrome and positive RIDTs may be associated with ICD-10 codes included in the ILI syndrome definition that contribute to false positive influenza cases. Ongoing research is focused on improving this ILI syndrome definition using ICD-10 codes. Rapid influenza diagnostic tests provide more timely results than other influenza test types. In conjunction with ILI medical encounter data, positive RIDT data provides a more complete and timely picture of the true burden of influenza on the MHS population for early warning of influenza outbreaks.
Acknowledgement The views expressed in this article are those of the author and do not reflect the official policy or position of the Department of Defense or the U.S. Government. Support was provided by the Armed Forces Health Surveillance Branch of the Public Health Division at the Defense Health Agency.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Jessica Deerin is an epidemiologist with the Armed Forces Health Surveillance Branch at the Defense Health Agency. She has 10 years of public health experience working in infectious disease epidemiology. She is currently pursuing a doctorate in epidemiology from the George Washington University.
Brief summary (100 words) of Presentation to be Used in Conference Program The DoD has the unique capability of ingesting laboratory results in ESSENCE to provide a comprehenisve picture of the disease burden on the Military Health System population. The rapid influenza diagnostic tests (RIDTs) data is analyzed here to assess its representatitiveness of influenza laboratory orders, timeliness, and correlation with ILI syndrome. In conjunction with ILI medical encounter data, positive RIDT data provides a more complete and timely picture of the true burden of influenza on the MHS population for early warning of influenza outbreaks.








Potential Applications of Emerging Technologies in Disease Surveillance

Jay Huang, Wayne Loschen

Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland, United States


Objective The objective of this presentation is to explore emerging technologies and how they will impact the public health field. New technologies such as blockchain, artificial intelligence (AI), and the Internet of Things (IoT) will likely be incorporated into epidemiological methods and processes. This presentation will provide an overview of these technologies and focus on how they may impact public health surveillance in the future.
Introduction With the increase in the amount of public health data along with the growth of public health informatics, it is important for epidemiologists to understand the current trends in technology and the impact they may have in the field. Because it is unfeasible for public health professionals to be an expert in every emerging technology, this presentation seeks to provide them with a better understanding of how emerging technologies may impact the field and the level of expertise required to realize benefits from the new technologies. Furthermore, understanding the capabilities provided by emerging technologies may guide future training and continuing education for public health professionals.
Methods Analysis of current capabilities and potential advances in emerging technologies such as blockchain, AI, and IoT were performed by reviewing articles and whitepapers. In addition to a literature review, interviews will be performed with public health experts to determine how the emerging technologies align with current practices and the extent to which they may solve existing public health surveillance challenges.
Results The literature review revealed many emerging technologies and potential applications in the public health field, including:

Blockchain
Blockchains can serve as electronic health information exchanges that hold the metadata and access information for patient electronic health records (EHRs).1 These systems can ensure data privacy protections while also facilitate relevant data sharing from EHRs to disease surveillance systems. Furthermore, blockchain technology can be used in food supply chain management systems. During food contamination events, epidemiologists can trace through the blockchain to identify possible sources of the contamination.2

AI
AI can be used to improve the prediction and detection capabilities of disease surveillance systems. Machine learning algorithms can reveal patterns in the data and enable faster anomaly detection. Furthermore, machine learning models can be trained on data to create predictive models.

IoT
Urban IoT systems can monitor environmental indices including water and air quality, energy consumption, waste management, and traffic congestion in smart cities.3 The data collected from such systems can be incorporated into more comprehensive disease surveillance systems and assist epidemiologists in better understanding populations and environmental risk factors.

We will analyze and discuss such prospective applications with public health professionals to determine their potential impact on public health processes and practices in the next one, five, and ten years.
Conclusions Blockchain, AI, IoT and other emerging technologies have applications in public health surveillance and impact the field to varying degrees. In addition to technological advances, there will be barriers to adoption that must be overcome before the value provided by the technologies can be realized. Many new technologies will require significant collaboration between public health departments, healthcare providers, and other partners to successfully incorporate the technologies into epidemiological processes. These collaborations include forming consortiums to exchange data in a blockchain and working with IoT providers for data access. Some technologies will require public health professionals to obtain additional training before they can take full advantage of the capabilities provided, while other technologies may be implemented by external partners allowing epidemiologists to utilize the new capabilities without the need to completely understand the underlying concepts. As emerging technologies are introduced into the public health field, a strong understanding of their capabilities and suitable applications will allow public health professionals to fully capture the benefits provided by the new technologies.
Acknowledgement
References 1. Ekblaw A, Azaria A, Halamka JD, Lippman A. A Case Study for Blockchain in Healthcare:“MedRec” prototype for electronic health records and medical research data. InProceedings of IEEE open & big data conference 2016 Aug 22 (Vol. 13, p. 13).
2. Yiannas F. A New Era of Food Transparency Powered by Blockchain. Innovations: Technology, Governance, Globalization. 2018 Jul;12(1-2):46-56.
3. Zanella A, Bui N, Castellani A, Vangelista L, Zorzi M. Internet of things for smart cities. IEEE Internet of Things Journal. 2014 Feb 14;1(1):22-32.
Brief bio for lead author/ presenter to be used by session moderators at the conference Jay Huang is a software engineer at the Johns Hopkins University Applied Physics Laboratory. He is a member of the ESSENCE software development team. He graduated with a B.S. in Computer Science from the University of Maryland, College Park and is pursuing a M.S. in Computer Science at Johns Hopkins University.
Brief summary (100 words) of Presentation to be Used in Conference Program With the increase in the amount of public health data along with the growth of public health informatics, it is important for epidemiologists to understand the current trends in technology and the impact they may have in the field. Emerging technologies such as blockchain, artificial intelligence, and the Internet of Things will likely be incorporated into epidemiological methods and processes. This presentation will provide an overview of these technologies and focus on how they may impact public health surveillance in the future.


Building Electronic Disease Surveillance Capacity in the Peruvian Navy with SAGES

Shraddha Patel, Miles Stewart, Martina Siwek

Johns Hopkins Applied Physics Laboratory, Laurel, Maryland, United States


Objective To introduce SMS-based data collection into the Peruvian Navy’s public health surveillance system for increased reporting rates and timeliness, particularly from remote areas, as well as improve capabilities for analysis of surveillance data by decision makers.
Introduction In the past 15 years, public health surveillance has undergone a revolution driven by advances in information technology (IT) with vast improvements in the collection, analysis, visualization, and reporting of health data. Mobile technologies and open source software have played a key role in advancing surveillance techniques, particularly in resource-limited settings. Johns Hopkins University Applied Physics Laboratory (JHU/APL) is an internationally recognized leader in the area of electronic disease surveillance. In addition to the Electronic Surveillance System for the Early Notification of Community-based Epidemics (ESSENCE) used by several state and local jurisdictions and the CDC in the U.S., JHU/APL has also developed the Suite for Automated Global Electronic bioSurveillance (SAGES). SAGES is a collection of modular, open-source software tools designed to meet the challenges of electronic disease surveillance in resource-limited settings.

JHU/APL is working with the Peruvian Navy health system to improve their electronic disease surveillance capabilities. The Peruvian Navy currently uses a SAGES-based system called Alerta DISAMAR that was implemented several years ago in an effort supported by the Armed Forces Health Surveillance Branch, and in collaboration with the Naval Medical Research Unit No. 6 (NAMRU-6). The system uses both web-based and IVR-based (interactive voice response) data collection from several Navy health facilities in Peru. For the present effort, JHU/APL is implementing a new SMS-based data collection capability for the Peruvian Navy.
Methods JHU/APL is engaged with the Peruvian Navy Health System to upgrade the existing SAGES-based Alerta DISAMAR surveillance system which relies on remote data collection using IVR (interactive voice recording) technology, with a SAGES-based system that uses SMS (short message service) text messages for remote data collection. Based on Peruvian Navy requirements, JHU/APL created mobile data entry forms for Android smartphones using the SAGES mCollect application. SAGES mCollect is built using Open Data Kit open source tools along with added features such as 128-bit encryption and quality checks.

The JHU/APL team engages closely with end users and other stakeholders to determine system requirements and to deploy the system, as well as to train end users and the system administrators who will need to maintain the system once it is deployed. The JHU/APL team, consisting of both information technology and public health expertise, conduct a country-level capabilities and needs assessment to address design considerations and operational end user requirements. This assessment takes into account the requirements and objectives of the Peruvian Navy, while keeping in mind infrastructure, cost, and personnel constraints. A pilot test of SMS-based data collection is currently underway with 10 health clinics within the Navy.
Results Many challenges exist when implementing electronic disease surveillance tools in resource-limited settings, but using a tailored approach to implementation in which specific needs, constraints, and expectations are identified with stakeholders helps increase the overall adoption and sustainment of the system. JHU/APL believes SMS-based data collection will be more sustainable than IVR-based data collection for the Peruvian Navy.
Conclusions JHU/APL is deploying a SAGES-based electronic disease surveillance system for the Peruvian Navy that has great potential to increase reporting rates from its health facilities as well as improve data quality and timeliness, thus resulting in greater awareness and enhanced public health decision making.
Acknowledgement This project is supported by the Uniformed Services University, Center for Global Health Engagement and is conducted in close coordination with United States Southern Command Surgeon’s Office.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Shraddha Patel, MPH is the project manager for SAGES at the Johns Hopkins Applied Physics Laboratory. For the past five years she has led the implementation of SAGES solutions in several countries around the world. She holds a Bachelors degree in Computer Science and a Master in Public Health from George Washington University.
Brief summary (100 words) of Presentation to be Used in Conference Program The Johns Hopkins University Applied Physics Laboratory is implementing an SMS-based data collection capability for the Peruvian Navy Health System with the goal of improving reporting rates, data quality, and timeliness.


A Novel Method for Rapid Mapping of the Spatial Intensity of Influenza Epidemics

David J. Muscatello1, Robert N. Leong1, Robin M. Turner2, Anthony T. Newall1

1School of Public Health and Community Medicine, UNSW Sydney, UNSW Sydney, New South Wales, Australia, 2University of Otago, Dunedin, New Zealand


Objective Using the epidemic of influenza type A in 2016 in Australia, we demonstrated a simple but statistically sound adaptive method of automatically representing the spatial intensity and evolution of an influenza epidemic that could be applied to a laboratory surveillance count data stream that does not have a denominator.
Introduction Surveillance of influenza epidemics is a priority for risk assessment and pandemic preparedness. Mapping epidemics can be challenging because influenza infections are incompletely ascertained, ascertainment can vary spatially, and often a denominator is not available. Rapid, more refined geographic or spatial intelligence could facilitate better preparedness and response.
Methods Weekly counts of persons with laboratory confirmed influenza type A infections in Australia in 2016 were analysed by 86 sub-state geographical areas. Weekly standardised epidemic intensity was represented by a z-score calculated using the standard deviation of below-median counts in the previous 52 weeks. A geographic information system was used to present the epidemic progression.
Results There were 79,628 notifications of influenza A infections included. Of these, 79,218 (99.5%) were allocated to a geographical area. The maps indicated areas of elevated epidemic intensity across Australia by week and area, that were consistent with the observed start, peak and decline of the epidemic when compared with weekly counts aggregated at the state and territory level. An example is shown in Figure 1.
Conclusions The methods could be automated to rapidly generate spatially varying epidemic intensity maps using a surveillance data stream. This could improve local level epidemic intelligence in a variety of settings and for other diseases. It may also increase our understanding of geographic epidemic dynamics.
Acknowledgement We thank the Australian Department of Health for providing the influenza laboratory data used in the study.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Dr David Muscatello is a Senior Lecturer in infectious diseases epidemiology. He has a PhD in the epidemiology of pandemic and seasonal influenza. He also has many years' experience in government as an epidemiologist specialising in acute disease surveillance using administrative databases, public health intelligence and biostatistics including time series analysis. He played a major surveillance role in the New South Wales government response to pandemic influenza in 2009 and has served on the Australian National Influenza Surveillance Committee. David is also a graduate of the New South Wales Public Health Officer Training Program and has supervised and trained numerous Public Health Officer and Biostatistical trainees. He is particularly interested in the use of time series analysis for estimating mortality and morbidity from infectious and other diseases and for assessing the impact of health policies on populations. He contributes to the World Health Organisation's (WHO) activities for estimating the global burden of deaths and hospitalisations attributable to influenza.
Brief summary (100 words) of Presentation to be Used in Conference Program Influenza laboratory data streams are influenced by spatially and temporally varying ascertainment. If no denominator for the number of tests is available, estimating meaningful changes in incidence over time and space is impossible. This study uses a moving window of z-scores based on the Poisson distribution for count data to provide an adaptive method for representing epidemic intensity across time and space. Using a geographic information system (GIS) to represent the results provides a compelling picture of the progression of the intensity of the epidemic.



Figure 1. Example map as at 9 September 2016, showing intensity by area during the peak week of the season. Insets show capital cities.



PocketAID: The Pocket Atlas of Infectious Diseases Mobile Application

Bonnie Gale, Hamid Mansoor, Chen-Yeou Yu, Lauren E. Charles

Pacific Northwest National Laboratory, Richland, Washington, United States


Objective The Pocket Atlas of Infectious Diseases (PocketAID) mobile application developed at Pacific Northwest National Laboratory (PNNL) provides infectious disease education and decision support offline for an enhanced personal situational risk assessment anywhere in the world. The app integrates a user’s location, demographic information, and infectious disease data to present the user with important information including personalized, calculated risk level. PocketAID features a global disease distribution map and epidemiological curve of country-based case counts by year. Filter options allow users to customize disease lists available to aid in situational awareness. PocketAID, first of its kind, is being developed for offline decision support use by Department of Defense’s Defense Threat Reduction Agency (DTRA).
Introduction There are a wide variety of available web-based apps, such as CDC’s Epidemic Information Exchange, that provide infectious disease information and disease distribution [1]. Publicly available, online data can be used to inform a user of general risks based on disease distribution maps and case count data. Unfortunately, each app contains different aspects of the data, which is often represented in different ways and incompatible formats. This heterogeneity can overwhelm a user with confusing information making it difficult to interpret or gain valuable insight into their own situational risk in a specified location. In addition, online resources do not filter information based on the user’s current location or situational needs and, therefore, reduces the value of information a user may be interpreting.
However, information formatted and represented appropriately in a single app could be used to better understand an individual’s situational infectious disease risk. In addition, this information may further educate a user based on a situation or incident to prevent disease spread, especially in higher risk populations. To accomplish these goals, PNNL has developed an offline, Android app that provides the user with simple, easy to understand filterable global infectious disease information integrated with their location to provide personalized situational health risk and decision support in the field.
Methods This prototype mobile app was a product of PNNL’s Biosurveillance Application Competition, sponsored by DTRA. Our implementation of this prototype consisted of two parallel efforts: data collection and Android app development.
Data. Infectious disease information was collected from CDC, WHO, Biosurveillance Resource Directory, and Analytics for Investigation of Disease Outbreaks websites [1-4]. Visualization feature data for global disease distribution and the case count curves was collected from CDC, WHO, and ECDC websites [1, 2, 5]. Data used for the disease filter and risk level warning features were associated to the collected infectious disease information and user inputted demographic information.
Application. The prototype app was built using Android operating system. Information about diseases, e.g., transmission mode, symptoms, properties, was stored in SQLite database that was imported into the phone at install time to provide offline information access. We used OSMDroid, an open source project, for map and location services. Downloaded map tiles made zoomable, interactive maps available offline.
Results PocketAID biosurveillance Android app was targeted for active duty service members, although deemed useful to a much broader audience. Given the various challenges that service members can face during deployment, such as no connectivity in remote areas, the app provides full functionality offline. The general purpose of PocketAID is to provide a user with infectious disease situational awareness and decision support, not be used as an analytic tool to test, treat, or diagnose disease.
Upon launch, the user is shown their location on a zoomable, interactive map and a list of diseases that are known to be present in their current country (detected automatically using the device’s GPS). The user can change their location by selecting a country from the location dropdown menu, filtering the populated list of diseases. The user can further filter diseases by disease attributes: symptoms, transmission, and properties. Clicking on a disease redirects the user to a page with more details about the disease, an interactive map of global disease distribution, and epidemiological curve displaying case counts by year for selected disease in selected country.
The user can input basic demographic information (i.e., age, gender, occupation, and pregnancy status) in the settings page of the app, which then enables an automated assessment of disease risk. Since specific diseases pose an increased risk to certain groups of people, the app can personalize the user’s risk level. In other words, if a user’s demographic information matches a disease’s risk groups, the user is shown a warning alert.
The app was awarded second prize in the competition by judges from across the government for its perceived benefit to biosurveillance, innovation and originality, quality of user experience, and long-term value and sustainability.
Conclusions The PocketAID provides global disease distribution on a zoomable map, infectious disease background information, disease case counts, offline capabilities, and diseases filtered by the location. This educational app offers a situational health risk assessment for the user through accessing infectious disease information with a disease attribute filter, personalized risk level warning, and user’s GPS or selected location to help improve decision support and reduce situational risk. The app was vetted by domain experts across the US Government, who found it to be useful and valuable.
Acknowledgement This work was funded by the Defense Threat Reduction Agency (project number CB10190).
References 1. Centers for Disease Control and Prevention [Internet]. Atlanta (GA): U.S. Department of Health & Human Services; [cited 2018 Aug 17]. Available from: https://www.cdc.gov/.
2. World Health Organization [Internet]. Geneva (Switzerland): World Health Organization; c2018 [cited 2018 Aug 17]. Available from: http://www.who.int/gho/en/.
3. Margevicius KJ, Generous N, Taylor-McCabe KJ, et. al. Advancing a Framework to Enable Characterization and Evaluation of Data Streams Useful for Biosurveillance. PLOS ONE 2014;9(1): e83730. doi: 10.1371/journal.pone.0083730
4. Analytics for Investigation of Disease Outbreaks [Internet]. Los Alamos (NM): Los Alamos National Security, LLC for the U.S Dept. of Energy's NNSA; c2018 [cited 2018 Aug 17]. Available from: https://aido.bsvgateway.org/.
5. Surveillance Atlas of Infectious Diseases [Internet]. Solna (Sweden): European Centre for Disease Prevention & Control; c2018 [cited 2018 Aug 17]. Available from: http://atlas.ecdc.europa.eu/public/index.aspx.
Brief bio for lead author/ presenter to be used by session moderators at the conference Bonnie Gale is a recent graduate of Emory University Rollins School of Public Health with her Master of Public Health in Epidemiology. Her research interests include infectious and zoonotic disease, mathematical modeling of infectious disease, and outbreak preparedness and response.
Brief summary (100 words) of Presentation to be Used in Conference Program The Pocket Atlas of Infectious Diseases (PocketAID) mobile application (app) developed at Pacific Northwest National Laboratory provides infectious disease education and decision support offline for an enhanced personal situational risk assessment anywhere in the world. The app integrates a user’s location, demographic information, and infectious disease data to present the user with important information including personalized, calculated risk level. PocketAID features a global disease distribution map and epidemiological curve of country-based case counts by year. Filter options allow users to customize disease lists available to aid in situational awareness. PocketAID, first of its kind, is being developed for offline decision support use by Department of Defense’s Defense Threat Reduction Agency (DTRA).



Figure 1. Application prototype main page (left) and disease information page (right). The main page includes filtered diseases by the user’s current location or a selected location from a drop-down menu and further filtering by disease transmission, properties, or symptoms. Disease information pages include a risk level warning dialogue box, global disease distribution on a map, epidemiological curves, and disease information.



Flexibility of ED surveillance system to monitor dengue outbreak in Reunion Island

Pascal Vilain1, Muriel Vincent1, Anne Fouillet2, Katia Mougin-Damour3, Xavier Combes4, Adrien Vague5, Fabien Vaniet5, Laurent Filleul2, Luce Menudier1

1Regional office of French National Public Health Agency in Indian Ocean, Saint-Denis, Réunion, 2French National Public Health Agency, Saint-Maurice, France, 3 Hospital Centre, Saint-Paul, Réunion, 4 University Hospital Centre, Saint-Denis, Réunion, 5Hospital Centre, Saint-Benoît, Réunion


Objective To describe the characteristics of ED vitis related to dengue fever and to show how the syndromic surveillance system can be flexible for the monitoring of this outbreak.
Introduction In Reunion Island, a French overseas territory located in the southwestern of Indian Ocean, the dengue virus circulation is sporadic. Since 2004, between 10 and 221 probable and confirmed autochthonous dengue fever cases have been reported annually. Since January 2018, the island has experienced a large epidemic of DENV serotype 2. As of 4 September 2018, 6,538 confirmed and probable autochthonous cases have been notified1. From the beginning of the epidemic, the regional office of National Public Health Agency (ANSP) in Indian Ocean enhanced the syndromic surveillance system in order to monitor the outbreak and to provide hospital morbidity data to public health authorities.
Methods In Reunion Island, the syndromic surveillance system called OSCOUR® network (Organisation de la Surveillance Coordonnée des Urgences) is based on all emergency departments (ED)2. Anonymous data are collected daily directly from the patients’ computerized medical files completed during medical consultations. Every day, data files are sent to the ANSP via a regional server over the internet using a file transfer protocol. Each file transmitted to ANSP includes all patient visits to the ED logged during the previous 24 hours (midnight to midnight). Finally, data are integrated in a national database (including control of data quality regarding authorized thesauri) and are made available to the regional office through an online application3.
Following the start of dengue outbreak in week 4 of 2018, the regional office organized meetings with physicians in each ED to present the dengue epidemiological update and to recommend the coding of ED visit related to dengue for any suspect case (acute fever disease and two or more of the following signs or symptoms: nausea, vomiting, rash, headache, retro-orbital pain, myalgia). During these meetings, it was found that the version of ICD-10 (International Classification of Diseases) was different from one ED to another. Indeed, some ED used A90, A91 (ICD-10 version: 2015) for visit related to dengue and others used A97 and subdivisions (ICD-10 version: 2016). As the ICD-10 version: 2015 was implemented at the national server, some passages could be excluded. In this context, the thesaurus of medical diagnosis implemented in the national database has been updated so that all codes can be accepted. ED visits related to dengue fever has been then described according to age group, gender and hospitalization.
Results From week 9 of 2018, the syndromic surveillance system was operational to monitor dengue outbreak. The regional office has provided each week, an epidemic curve of ED visits for dengue and a dashboard on descriptive characteristic of these visits. In total, 441 ED visits for dengue were identified from week 9 to week 34 of 2018 (Figure 1). On this period, the weekly number of ED visits for dengue was correlated with the weekly number of probable and confirmed autochthonous cases (rho=0.86, p<0.001). Among these visits, the male/female ratio was 0.92 and median (min-max) age was 44 (2-98) years. The distribution by age group showed that 15-64 year-old (72.1%, n=127) were most affected. Age groups 65 years and more and 0-14 year-old represented respectively 21.8% (n=96) and 6.1% (n=27) of dengue visits. About 30% of dengue visits were hospitalized.
Conclusions According Buehler et al., “the flexibility of a surveillance system refers to the system's ability to change as needs change. The adaptation to changing detection needs or operating conditions should occur with minimal additional time, personnel, or other resources. Flexibility generally improves the more data processing is handled centrally rather than distributed to individual data-providing facilities because fewer system and operator behavior changes are needed...” 4.
During this dengue outbreak, the syndromic surveillance system seems to have met this purpose. In four weeks (from week 5 to week 9 of 2018), the system was able to adapt to the epidemiological situation with minimal additional resources and personnel. Indeed, updates were not made in the IT systems of each EDs’ but at the level of the national ANSP server (by one person). This surveillance system was also flexible thank to the reactivity of ED physicians who timely implemented coding of visits related to dengue fever.
In conclusion, ED surveillance system constitutes an added-value for the dengue outbreak monitoring in Reunion Island. The automated collection and analysis data allowed to provide hospital morbidity (severe dengue) data to public health authorities. Although the epidemic has decreased, this system also allows to continue a routine active surveillance in order to quickly identify a new increase.
Acknowledgement All emergency departments of the Reunion Island.
References 1Santé publique France. Surveillance de la dengue à la Réunion. Point épidémiologique au 4 septembre 2018. http://invs.santepubliquefrance.fr/fr/Publications-et-outils/Points-epidemiologiques/Tous-les-numeros/Ocean-Indien/2018/Surveillance-de-la-dengue-a-la-Reunion.-Point-epidemiologique-au-4-septembre-2018. [Accessed September 8, 2018].
2Vilain P, Filleul F. La surveillance syndromique à la Réunion : un système de surveillance intégré. [Syndromic surveillance in Reunion Island: integrated surveillance system]. Bulletin de Veille Sanitaire. 2013;(21):9-12. http://invs.santepubliquefrance.fr/fr/Publications-et-outils/Bulletin-de-veille-sanitaire/Tous-les-numeros/Ocean-indien-Reunion-Mayotte/Bulletin-de-veille-sanitaire-ocean-Indien.-N-21-Septembre-2013. [Accessed September 4, 2018].
3Fouillet A, Fournet N, Caillère N et al. SurSaUD® Software: A Tool to Support the Data Management, the Analysis and the Dissemination of Results from the French Syndromic Surveillance System. OJPHI. 2013; 5(1): e118.
4Buehler JW, Hopkins RS, Overhage JM, Sosin DM, Tong V; CDC Working Group. Framework for evaluating public health surveillance systems for early detection of outbreaks: recommendations from the CDC Working Group. MMWR Recomm Rep. 2004;53(RR-5):1-11.
Brief bio for lead author/ presenter to be used by session moderators at the conference Vilain Pascal Epidemiologist Since 2009, has been working at the Regional Office of French national public health agency in Indian Ocean on syndromic surveillance on Reunion Island and Mayotte Island.
Brief summary (100 words) of Presentation to be Used in Conference Program Since January 2018, Reunion Island has experienced a large epidemic of dengue fever. In order to monitor this outbreak, the syndromic surveillance system based on emergency departments has been enhanced. Thanks to his flexibility and the reactivity of ED physicians, hospital morbidity data has been provided to public health authorities four weeks after the beginning of the outbreak. Although the epidemic has decreased, this system also allows to continue a routine active surveillance in order to quickly identify a new increase.



Figure1. Weekly number of ED visits related to dengue and weekly number of probable and confirmed autochthonous dengue fever cases, from week 1 to week 34, 2018, Reunion Island.



Use of the OSCOUR network data to describe low back pain attendances in French ED

Cécile Forgeot1, Gilles Viudes2, guilhem Noel3, Anne Fouillet1, Céline Caserio-Schönemann1

1DATA, Sante publique France, Saint-maurice, France, 2Fedoru, Emergency regional observatory Federation, Paris, France, 3PACA Emergency Network Observatory (ORUPACA), Marseille, France


Objective The study describes the characteristics of attendances for low back pain (LBP) in the French emergency departments (ED) network Oscour®, in order to give an overview of this disease before launching a prevention campaign.
Introduction LBP is one of the leading contributors to disease burden worldwide [1]. In France, LBP is a frequent reason of general practice consultations. According to a study published in 2017 and based on 2014 data issued of the National Health Insurance Cross-Schemes Information System (Sniiram) [2], this pathology stands for 30% of thickness leave and 4 of 5 people will suffer of low back pain during their own life. Most often, LBP is a chronic pathology with acute episodes which most often require emergency care.
In order to prevent chronicity, French health care insurance launched into a mainstream national prevention campaign during spring 2018. This campaign was also targeted for health professional to inform them of the best recommendations to provide to their patients. Then the French society of emergency medicine (SFMU) has been asked to relay this campaign to emergency departments (ED) where LBP is a frequent reason of attendance.
Since 2004, the French syndromic surveillance system SurSaUD® [3] coordinated by the French Public Health Agency (Santé publique France) daily collects morbidity data from the emergency departments (ED) network Oscour®. Almost 92% of the French ED attendances were recorded by the system in 2017.
The availability of this large ED dataset on the whole territory since several years gives the opportunity to describe LBP attendances before the potential fallout of the national prevention campaign.
Methods For each ED attendance, the SurSaUD® system daily collects individual data containing demographic (age, gender, zip code), administrative (ED unit, date of attendance, transport…) and medical information (medical diagnosis (ICD10), chief complaint, severity, hospit.). These data are routinely analyzed to detect and follow-up various expected or unusual public health events all over the territory [3] and also constitute a large database to perform in-depth studies on specific public health issues.
ED attendances with a medical diagnosis of LBP have been identified using at least one of the following ICD10 codes “M545”, “M5450”, “M5456”, “M5457”, “M5458”, “M5459”. Those data have been analyzed from 01/01/2014 to 31/12/2017 (504 ED) for the following age groups; less than 18 years old (yo), 18 to 34 yo, 35 to 49 yo, 50 to 64 yo, 65 to 84 yo and 85 yo and over, at national and regional levels. ED attendances have been also described by month, day of week and hour of day. Hospitalizations after discharge, stay duration in ED services, transport and associated diagnoses were also analyzed.
Results From 2014 to 2017, 481,291 ED attendances for LBP were recorded corresponding to 1.12% of the total number of ED attendances with a coded diagnosis. 60% of annual ED attendances for LBP concern 18 to 50 years old adults. The proportion of LBP attendances among the all-cause activity remains stable between 2014 and 2017.
At the regional level, LBP proportion among the all-cause activity is similar to the national value in metropolitan regions (0.8% in Brittany to 1.6% in Corsica) and is lower than the national value in overseas regions (0.4% in Mayotte to 0.8% in Guyane) except for Saint-Barthélémy (1.8%).
At the national level, almost 10% of ED attendances for LBP are hospitalized after discharge. This proportion increases with age to reach 43% for the 85 years old and more. Proportion of hospitalization ranges between 5.6% (in Paris area) and 17.1% (in Brittany) in metropolitan regions and between 2.8% (Guyane) and 9.3% (Reunion island) in overseas regions.
From 2014 to 2017, ED attendances for LBP remain stable by month. However, we observed a slight decrease along the week with more attendances on Monday (17.8% of LBP attendances) than the other days. The attendances are more frequent in the morning (between 6 and 12 AM).
At the national level, mean stay duration for LBP attendances in ED is almost 5 hours whereas median stay duration is 2 hours and 45 minutes. Stay duration is longer for patient arrived during night hours (from midnight to 6 AM) and for those hospitalized after discharge. Stay duration is also increasing with age. At the regional level, mean stay duration varies from 3 to more than 6 hours.
Conclusions The broad coverage of the French ED network on the whole territory since several years enables to give an overview of ED attendances for acute LBP and their characteristics.
One strength of the system is its strong partnership between epidemiologists and the ED physicians. It enables to verify that the results of the study are consistent with their perception on the field.
The results of this study will be used as reference to evaluate potential benefits of this campaign.
Finally, this study is a good illustration of how the syndromic surveillance system in collaboration with ED physicians, can quickly provide valuable data to support political strategies.
Acknowledgement
References [1] Maher et al, Non-specific low back pain Lancet 2017; 389: 736–47 Published Online October 10, 2016, http://dx.doi.org/10.1016/ S0140-6736(16)30970-9
[2] Assurance Maladie, Le patient adulte atteint de lombalgie commune; Livret d’information Octobre 2017 données SNIRAAM 2014, https://www.ameli.fr/sites/default/files/Documents/346618/document/lombalgie-professionnels-de-sante_assurance-maladie.pdf
[3] Caserio-Schönemann C, Bousquet V, Fouillet A, Henry V, pour l’équipe projet SurSaUD®. Le système de surveillance syndromique SurSaUD®. Bull Epidémiol Hebd. 2014;(3-4):38-44.
Brief bio for lead author/ presenter to be used by session moderators at the conference After a master's degree, the author arrived in French National Public Health Agency in april 2011. She is actually in charge of the animation of the french ED network OSCOUR® and analysis of data collected by the French syndromic surveillance system SurSaUD®.
Brief summary (100 words) of Presentation to be Used in Conference Program Low back pain (LBP) is one of the leading contributors to disease burden worldwide. Most often, acute episodes of LBP require emergency care. This study used data of the French ED network Oscour® to give an overview of ED attendances for LBP. At the national level, LBP represent 1.12% of ED all-cause activity and 60% of those attendances concern adults aged 18 to 50 years old. Almost 10% of ED attendances for LBP are hospitalized after discharge. The results of this study constitute a reference to evaluate the benefits of the national prevention campaign launched in spring 2018.


Environmental Surveillance and Vaccine Derived Polio Virus type 2 Isolation, Gombe State, Nigeria.

Raymond S. Dankoli

World Health Organization (W.H.O), Gombe, Gombe, Nigeria


Objective To evaluate Vaccine Derived Polio Virus 2 isolation rate from Environmental Surveillance and its contribution to Polio Eradication Initiative (PEI)
Introduction Nigeria is the only country in Africa yet to be certified free of Wild Polio Virus (WPV). The country consists of 36 States and a Federal Capital Territory. Gombe is one of the 19 Polio high risk States in the North-eastern geo-political zone of the country. The last case of WPV isolated in Gombe State was in 2013.
One of the strategies for Polio eradication is a sensitive Acute Flaccid Paralysis (AFP) surveillance system in which any AFP is promptly detected and timely investigated. The focus of the investigation is to analyze two faecal samples of the patient, and/or sometimes those from contacts for any possible isolation of Polio Virus1 (PV). AFP surveillance is meant to be applicable to any human population at any time; however, there are situations in which there are good reasons to suspect that negative results of AFP surveillance are not reliable. Supplementary information is required in such situations and one approach for that is Environmental Surveillance (ES), in which a search for PV is made in environmental specimens contaminated by human feaces2
ES in the African region started in Nigeria in July 20113,4. Since the introduction of this strategy, it has achieved its objective of complimenting the AFP surveillance system. There has been a gradual increase in the number of ES sites in Nigeria from 2011 to date4. The increase is largely due to the successes recorded in terms of the PV isolation from the sites, PV epidemiology, the large population size and mobility4,5. The last cases of WPV1 and WPV3 from environmental samples had dates of collection in May 2014 (Kaduna) and July 2012 (Kano) respectively4.
ES was initiated in Gombe State in December 2016. Four ES sites were identified and sample collection began soon after training of personnel responsible for collection of the sewage sample. The four identified ES sites are Baba Roba Valley, Unguwauku Railway Bridge, Gadan Bayan Moonshine and Dan Gusau Bridge. Since inception of ES in Gombe State, 2 ambiguous Vaccine Derived Poliovirus type 2 (aVDPV2) were confirmed from sewage samples collected from Baba Roba Valley site on the 30th January 2017 and from Dan Gusau Bridge site on the 6th March 2017. In 2018, a circulating Vaccine Derived Poliovirus type 2 (cVDPV2) was also detected from sewage samples collected on the 9th April 2018 from Baba Roba Valley site. We reviewed the laboratory results from the 2 surveillance methods so as to evaluate the VDPV2 isolation rate.
Methods ES involves collection of one litre of environmental sample (sewage water) via grab sampling method in accordance with World Health Organization’s (WHO) Guidelines for Environmental Surveillance for Polioviruses2. All ES sewage samples were transported in a 1 litre container appropriately packaged in a Giostyle with 8 frozen icepacks to maintain reverse cold chain to a Polio Laboratory where the samples are analyzed as per WHO ES testing standard operating procedures. Poliovirus type 2 isolates are sent to the reference laboratory at the US Centre for Disease Control for sequencing for PV isolation.
We reviewed all the results of the environmental samples (ES) and stool samples from patients with Acute flaccid paralysis (AFP) from January 2017 to June 2018. The environmental samples were from five pre-selected sites that was based on the perceived risks for polio circulation that included poor sanitation, overcrowding, extend of drainage population, availability of sewage system and absence of discharge into the sites. The stool samples were from patients detected with AFP in Gombe local government area.
The results from the two methods of surveillance for PV were evaluated and compared based on yields and isolates (Negative results, VDPV2, Non-polio enterovirus (NPENT).
Results A total of 309 sewage samples from five (5) sites and 142 AFP stool samples from Gombe LGA were collected from January 2017 to June 2018. Three 3(0.97%) of the sewage samples yielded VDPV2, 102(33.01%) had Non-polio enteroviruses (NPENT) and 41 (13.27%) negative samples. On the other hand, no VDPV was isolated from the AFP stool samples, the NPENT detection rate was 13(9.16%) and 121(85.21%) samples were negative. The Non-polio AFP (NPAFP) and stool adequacy rates for Gombe LGA during the reporting period were calculated to be 17.2 and 100% respectively.
Conclusions The polio virus (VDPV) isolation from ES in this review is higher than in AFP surveillance. This has demonstrated amongst others benefit of ES its ability to detect polio virus even in the absence of the virus among AFP cases. ES can thus detect virus that are probably missed by AFP surveillance and hence allow for early response so as to curtail further transmission. The high NPAFP and stool adequacy rates are indication of a sensitive surveillance system nonetheless, the virus isolation from the AFP surveillance was very low. It is important to mention here that other laboratory indicators were not factored into this review. We recommend therefore that both ES and AFP surveillance be done together where facility, resources and personnel are available to implement
Acknowledgement We sincerely acknowledge the Gombe State Ministry of Health, Gombe State Primary Health Care Development Agency, all Sample collectors and State Data Assistance W.H.O Gombe office for their support and information sharing during this review.
References [1] WHO. Field guide for supplementary activities aimed at achieving polio eradication, publication no. WHO/
EPI/GEN/95.1. Geneva: World Health Organization, 1995.

[2] WHO. Guidelines for environmental surveillance of poliovirus circulation. World Health Organization 2003,
Department of Vaccines and Biologicals, 2003. (http://www.who.int/vaccines-documents/DoxGen/H5-Surv.
htm). Accessed 6 October 2010.
[3] Nicksy Gumede et al. Status of environmental surveillance in the African Regional. African health monitor. March 2015 Issue 19: Pg 38-41
[4] Ticha Johnson Muluh et al. Contribution of Environmental Surveillance Toward Interruption of Poliovirus Transmission in Nigeria, 2012–2015. The Journal of Infectious Diseases, Volume 213, Issue suppl_3, 1 May 2016, Pages S131–S135, https://doi.org/10.1093/infdis/jiv767
[5] Humayun Ashgar et al. Environmental Surveillance for Polioviruses in the Global Polio Eradication Initiative. The Journal of Infectious Diseases, Volume 210, Issue suppl_1, 1 November 2014, Pages S294–S303, https://doi.org/10.1017/S095026881000316X
Brief bio for lead author/ presenter to be used by session moderators at the conference Raymond S. Dankoli is a Medical Doctor by training. He holds a Master in Public Health (MPH) degree from the Ahmadu Bello University Zaria, Nigeria. He is a graduate of the Nigerian Field Epidemiology and Laboratory Training Programme (NFELTP) where he received training in Field Epidemiology. He was Head of Department of Medicine, State Specialist Hospital Gombe, Nigeria. He is a Member of the International society for infectious Diseases. He is a member of the WHO AFRO Regional Environmental Surveillance Working Group (RESWG). He is currently a Surveillance Officer with WHO Borno State Field Office, North-East, Nigeria.
Brief summary (100 words) of Presentation to be Used in Conference Program Nigeria is the only endemic country for Polio in Africa. One of the strategies for Polio eradication is a sensitive Acute Flaccid Paralysis (AFP) surveillance system in which any AFP is promptly detected and timely investigated. Environmental Surveillance (ES) is also being employed as a complimentary surveillance system. AFP surveillance examines stool sample from an AFP patient while ES examines sewage sample contaminated with feaces. The laboratory results from the two methods were compared to evaluate the Polio Virus isolation. ES has a higher yield from our review. ES is therefore recommended an essential compliment to AFP surveillance and should be adopted where resources and logistics permit.


Summary of result from five (5) ES sites and AFP Stool Samples in Gombe LGA January 2017 – June 2018


Name of ES Site NO. of AFP cases NO. of sewage/
stool samples collected
VDPV2

Sabin

NPENT

Negative
Sabin + NPENT
Sabin 2

NPAFP Rate % stool Adequacy
Baba Roba Valley - 66 2 14 18 3 16 13 - -
Dan Gusau Bridge - 65 1 12 21 10 9 12 - -
UnguwaUku Railway Bridge - 64 0 4 27 12 10 11 - -
Gadan Bayan Moonshine - 65 0 18 17 5 14 11 - -
Unguwan Bari Bari - 49 0 7 19 11 7 5 - -
ES Total (%) - 309 3 (1.0) 55(17.8) 102(33.0) 41(13.3) 56(18.1) 52(16.8) - -
AFP Stool Samples (%) 71 142 0 (0) 8 (5.6) 13(9.2) 121 (85.2) 0 (0) 0 (0) 17.2 100


Knowledge of Malaria and Antimalarial Drug Dispensing Practices in Buea Community

Marcelus U. Ajonina1, 2

1Health Sciences, Meridian Global University, Buea, CENTRAL REGION, Cameroon, 2Meridian Global Education and Research Foundation, Buea, Southwest Region, Cameroon


Objective This study was aimed at assessing the knowledge of malaria as well as perception and dispensing practices of antimalarials among vendors in Buea community
Introduction Lack of knowledge of rational use of antimalarial drugs among dispensers is a serious problem, especially in areas of intense transmission thus increasing the risk of resistance and adverse drug reactions.
Methods A community-based cross-sectional survey of a random sample of 140 drug vendors living within the Buea community was conducted between March and June 2017. Questionnaire was designed to obtain information from drug vendors on the general knowledge of malaria as well as dispensing practices. Data were analyzed using SPSS Statistics 20.0 and were considered significant at P ≤ 0.05
Results Knowledge of malaria symptoms, transmission, and prevention was reasonable among 55.8% (77) of the respondents. Only 33.6% (47) of the respondents could attribute the cause of malaria to protozoan of genus Plasmodium species. Of the 140 vendors, 115(82.7%) prescribe antimalarial drugs. The knowledge of the national protocol was malaria case management among dispensers was 35.0%. Vendors in hospital/community pharmacies were 2.4 times (OR = 3.14, 95% CI: 4.14 - 8.74, P < 0.001). more knowledgeable about malaria treatment protocol than those of in drugstores. The prevalence of self-prescription of antimalarials was 39.3%. Self-prescription was significantly higher in drugstores than hospital/community pharmacies (P=0.004). In all, 56(40.6%) of vendors showed good practices regarding antimalarial drug dispensing with majority (51.7%) from community pharmacies (OR=2.27,95% CI: 1.13-4.56).
Conclusions Findings reveal moderate knowledge of malaria but poor prescription and dispensing practices of antimalarial drugs among vendors, thus indicating a need for routine monitoring and evaluation to prevent emergence of resistant strains to current efficacious antimalarials
Acknowledgement We thank the all participants from various drug retail outlets from the Buea community who made this study possible by giving their consent. We equally thank Meridian Global Education and Research Foundation (MGERF), Cameroon for financial support.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Marcelus U. Ajonina holds a PhD in Public Health from Charisma University, a Master degree in Biochemistry and a Bachelor degree in Biochemistry from the University of Buea, Cameroon. His research interests are geared towards control of infectious and non-communicable diseases including malaria, onchocerciasis, HIV/AIDS, hypertension, etc. He has served as a lecturer in some universities among which inclue Savannah State University, Saint Monica University. He is currently the President and CEO of Meridian Global Education and Research Foundation (MGERF) and the Vice Chancellor of Meridian Global University, Cameroon. He has over 12 publications in peer-reviewed journals to his credit and has been serving as an editorial board member
Brief summary (100 words) of Presentation to be Used in Conference Program Malaria continues to be a threat to public health in Sub-Saharan African countries due to lack of appropriate vector and parasite control strategies. Irrational dispensing of antimalarial drugs has resulted to increased drug resistance to current drug of choice. To combat drug resistance, policies regarding to antimalaria drug dispensing must be strictly implemented. We herein assessed the knowledge of malaria as well as perception and dispensing practices of antimalarials among vendors in Buea community, Southwestern Cameroon.


Epidemiological Distribution of Reported West Nile cases in Houston, Texas, 2014-2017

Razina Khayat, Alex Nguyen, Sudipa Biswas, Hafeez Rehman, Kirstin Short, Najmus Abdullah

Informatics, City of Houston , Houston, Texas, United States


Objective To demonstrate an overview of the epidemiological and clinical distribution of reportable West Nile cases in Houston, Texas, from 2015-2017.
Introduction West Nile virus (WNV) is considered the leading cause of domestically acquired arboviral disease and is spread through mosquitoes. In general, the majority of the cases are asymptomatic. One in five people infected will display mild symptoms like fever, headache, body ache, nausea, and vomiting. Only about 1 in 150 people infected will develop serious neurologic complications such as encephalitis and meningitis. According to CDC, in 2017, there were 133 confirmed cases including 5 deaths and 14 presumptive blood donors reported in the State of Texas. Out of the confirmed cases, there were 85 neuroinvasive and 48 non-neuroinvasive disease cases.
Methods Data were extracted from Houston’s Electronic Disease Surveillance System (HEDSS) from January 1, 2014, to December 31, 2017. A total of 45 confirmed cases are included in this analysis to examine the epidemiologic characteristics of the WNV cases.
A confirmed case is an illness with onset of acute focal limb weakness and an MRI showing a spinal cord lesion largely restricted to gray matter and spanning one or more spinal segments.

Results Among the confirmed cases, 67% of were males. Age group 60 and above (47%) had the highest proportion of WNV cases. Whites (26%) represented the highest number of confirmed cases followed by Hispanics (24%).
Seventy six percent of the cases were hospitalized. Non-neuroinvasive clinical presentations found among confirmed WNV cases were fever (94%), headache (76%) followed by chills and rigors (68%). Among the neuroinvasive presentations, altered mental status had the highest proportion (24%), followed by stiff necks (18%), Ataxia (12%), and seizure (9%).
Conclusions WNV is mostly prevalent in White male adults over 60 years of age, with majority of cases have common neuroinvasive symptoms like altered mental status, stiff necks, and Ataxia. For non-neuroinvasive cases clinical symptoms were fever, headache, chills and rigors.
WNV infection is a markedly underreported disease as most of the infected people do not seek medical care due to mild or no symptoms. Currently there are no specific treatments available. Thus, continued monitoring and surveillance activities are warranted for prevention and control of WNV complications as well as decreasing the risk of infection.
Acknowledgement We thank the flollowing agencies for providing data for this study:
Texas Department of State Health Services.
Houston Health Department, Diivision of Disease Prevention and Control.
Houston Electronic Disease Surveillance System.
References CDC: West Nile Virus [Internet]. Atlanta: Center for Disease Control and Prevention (CDC), National Center for Emerging and Zoonotic Infectious Diseases (NCEZID), and Division of Vector-Borne Diseases (DVBD); Last reviewed: September 19, 2018.
Available from: https://www.cdc.gov/westnile/index.html
Brief bio for lead author/ presenter to be used by session moderators at the conference I am Razina Khayat, currently working as an Informatic, Biostatician at the City of Houston Health Department. I also worked as an Infectious Disease Epidemiologist at the same department. I'm interested in Emerging Infectious Disease Research and prevention.
Brief summary (100 words) of Presentation to be Used in Conference Program West Nile Virus (WNV) is an emerging infectious disease of high concern. It is transmitted via mosquitoes, thus it is more prominent in tropical areas. Data used in this study were extracted from Houston’s Electronic Disease Surveillance System (HEDSS) from January 1, 2014 to December 31, 2017. Among the confirmed cases, males and age group 60 and above had the highest proportion of cases.Currently there are no specific treatments available. Thus, continued monitoring and surveillance activities are warranted for prevention and control of WNV.


TIME SERIES ANALYSIS OF INFECTIOUS DISEASE MORTALITY IN UKRAINE (1965-2015)

Hennadii Mokhort, Olga Sokolovska

Epidemiology, Bogomolets National Medical University, Kyiv, Ukraine


Objective The aim of our work is to determine the main trends and structure in infectious disease mortality in Ukraine over the last 50 years.
Introduction Monitoring of long-term infectious disease mortality trends is of great value to national public health systems both in estimation of the efficacy of preventive programs, and in development of the new strategies of preventive measures. In the developed countries, there are a number of studies with long-term time series of infectious disease mortality analysis in epidemiological and historical aspects. Our research was based on the work by Armstrong GL, Conn LA and Pinner RW, 1999. Literature review revealed that such analysis has been never carried out in Ukraine up to now.
Methods Our study is designed as a descriptive retrospective epidemiological analysis. We constructed time series of infectious mortality in all oblasts of Ukraine during the period of 1965-2015 years. We used annual statistical forms C-8 “Distribution of deceased by sex, age and death cause” provided by Ministry of Health. The cause of death was accounted in accordance with international statistical classification of diseases, injuries, and causes of death: based on the recommendations of the ninth revision conference, 1975. We analyzed infectious diseases belonging to the class of Infectious and parasitic diseases (45 nosology and nosology groups – codes 001 – 139). We also included into our analysis some other infectious diseases belonging to other classes: Neoplasm (cervix carcinoma – code 180); Heart diseases (Rheumatic fever, rheumatic heart disease – codes 390-398); Diseases of the respiratory system (Acute Respiratory Infection, influenza, viral pneumonia, pneumococcal pneumonia, other acute forms of pneumonia – codes 460-466, 487, 480, 481, 482, 483, 485, 486); diseases of nervous system (non-infectious and non-parasitic meningitis, codes 320-322).
Therefore, all infections that are reported in Ukraine were included to this research. Nosologies were grouped using several disease classifications: in accordance with International classification (belonging or not to Infectious disease class); by transmission method or localization of an infectious agent (respiratory, intestinal or alimentary, blood borne, contact and other infections); by ecological principle (anthroponosis, zoonosis, sapronosis and other).
All time series were divided onto two periods: 1965-1991 (soviet period) and 1992-2015 (period of independent Ukraine). Average mortality (mortality coefficient) of these periods was compared to each other for calculation of percentage decrease/increase of each disease’s mortality rate. Additionally, we determined the proportion (%) of infectious mortality compare to the total mortality of population of Ukraine. Limited scope of this study does not allow us to present data regarding the age distribution, thus we focus on general characteristics. Although the practice of presenting data was changed over the course of 50 years covered by this abstract, the data are comparable and can be used for analysis.
Results Total number of fatal cases caused by infectious diseases in Ukraine during 1965-2015 years is 1,268,560 or 4.05% of all deaths caused by different reasons. 550,329 deaths or 43.38% of all infectious deaths belong to class of infectious and parasitic diseases, other 718,231 or 56.62% belong to infectious diseases of other classes. Percentage of respiratory infections is 80.28%, intestinal infections – 1.72%, blood infections – 16.94% and other infections – 1.05%. Additionally, proportion of anthroponosis is 98.31%, proportion of zoonosis – 0.42%, sapronosis – 0.22%, other – 1.05%. During 1965-2015, percentage of infectious diseases in overall structure was within the range from 10.53% (1965) to 2.99% (2015). Overall mortality rate of infectious diseases decreased from 80.49 per 100,000 population (1965) to 41.77 per 100,000 (2015). This finding demonstrates to reduction of overall infectious mortality in Ukraine. It is important to mention that decrease of overall infectious mortality happened simultaneously with an increase of mortality caused by non-infectious diseases. Non-infectious mortality increased from 683.92 per 100 000 population (1965) to 1354.77 per 100 000 population (2015).
The first 10 causes of death from infectious diseases in Ukraine in 1965-2015 included the following nosological units and infectious groups: 1. Respiratory tuberculosis and other forms of tuberculosis (30.3%); 2. Acute respiratory infections + Influenza + Viral pneumonia + Pneumococcal pneumonia and Other acute pneumonias (28.36%); 3. Acute rheumatic fever + Chronic rheumatic heart disease (15.93%); 4. Malignant neoplasm of cervix uteri (10.42%); 5. AIDS (4.8%); 6. Septicemia (2.86%); 7. Meningococcal infection + Meningitis, excluding infectious and parasitic meningitis (2.64%); 8. Other infectious and parasitic diseases and long-term effects of other infectious and parasitic diseases (1.05%); 9. Acute intestinal infections due to unspecified micro-organisms and ill-defined, including toxical dyspepsia (0.93%) and 10. Viral hepatitis (0.79%).
The average mortality rate declined for the most infectious diseases during 1992-2015 comparing to 1965-1991. For certain diseases or their groups, the range varied from 12.71% (pneumonia) to 92.69% (influenza). The biggest decrease was observed in intestinal infections group (up to 73.99%). All respiratory infections demonstrated a decrease to 18.14%; other infections with unknown path of transmission resulted in 4.6% decrease, but blood borne infections demonstrated an increase of 59.72 % (mostly caused by AIDS). Some other infections also demonstrated increase in mortality rate: foodborne botulism – up to 29.94%, tuberculosis – up to 26.52%, diphtheria – up to 376,53%, erysipelas – to 371,26%, AIDS – to 100%.
Conclusions Infectious diseases are not the main mortality cause among the population of Ukraine during the past 50 years. Over the last half-century, the proportion of infectious diseases in the mortality structure of the population of Ukraine demonstrated a decreasing tendency, while non-infectious disease mortality had an opposite trend, which can be explained by epidemiological transition (Omran AR, 1971). However, there is always a possibility of rapid spreading of infectious diseases and increasing their proportion in the structure of total mortality. Possible growth of mortality rate caused by AIDS, tuberculosis and diphtheria is an issue of concern. International experience demonstrated that these three infections could be successfully controlled. The long-term trends of AIDS, tuberculosis and diphtheria mortality rates in Ukraine require regulatory interventions and show the need for emergency measures by the state services to these and some other infections, including vaccine-controlled. Thus, our study of the long-term trends of infectious mortality can be used to make decisions of public health in Ukraine on the control of infectious morbidity and mortality.
Acknowledgement The author is thankful to the Ministry of Health of Ukraine for giving access and sharing their invaluable data set on the disease mortality.
References Armstrong GL, Conn LA and Pinner RW. Trends in Infectious Disease Mortality in the United States During the 20th Century. JAMA, 1999, Vol. 281 (1): 61-66.
Omran AR. The epidemiologic transition: a theory of the epidemiology of population change. Milbank Q. 1971; 49:509-538.
Brief bio for lead author/ presenter to be used by session moderators at the conference Hennadii Mokhort was born in 1971 in Kyiv, Ukraine. He received a medical degree at Bogomolets National Medical University, Kyiv, Ukraine, in 1995. In 1995-1998, Hennadii worked as an epidemiologist in the Kyiv State Sanitary and Epidemiological Service. Since 1998, Hennadii has been teaching epidemiology at Bogomolets National Medical University. In 2003, he received his PhD, thesis title was Manifestations of the Diphtheria Epidemic Process in Presents Days and Ways of Improving of its Epidemiological Surveillance. The thesis were based on materials on the epidemic process of diphtheria, which took place in the 1990s in Ukraine and other countries of the former Soviet Union. Since 2008, Hennadii holds a position of Associate Professor at the Department of Epidemiology of Bogomolets National Medical University. Hennadii research interests are epidemiology of infectious diseases and vaccine prevention.
Brief summary (100 words) of Presentation to be Used in Conference Program The study contains results of time series analysis of morbidity among Ukrainian population during 1965-2015. All fatal cases due to infectious diseases that are subjects of mandatory reporting in Ukraine according to Ninth Revision of the International Statistical Classification of Diseases, Injuries, and Causes of Death. The majority of infectious diseases showed decrease of mean morbidity rate in 1992-2015 comparing to 1965-1991. For some diseases and their groups, the decrease varied from 12.71% (pneumonia) to 92.69% (influenza); other diseases demonstrated increase of morbidity: food-borne botulism – by 29.94%; tuberculosis – by 26.52%; diphtheria – by 376.53%; erysipelas – by 371.26%, and AIDS – by 100%.


Factors Predicting Retention In Care and Health Outcomes Among HIV Patients

Merlene Ramnon

Florida Department of Health, State, Wellington, Florida, United States


Objective To povide knowledge on the factors that predict retention in care and health outmomes among HIV patients and be able to understand viral load and its relation to retention in care.
Introduction The prevalence of persons living and diagnosed with HIV infection in the United States in 2010 to 2014 increased in number and rate (Center for Disease Control & Prevention (CDC), 2016).
In 2015, persons aged 25–29 years had the highest rate (33.4), followed by persons aged 20–24 years (31.2) (CDC, 2016).
Consistent reduced viral load is associated with reduced morbidity and mortality and a lower likelihood of transmitting HIV to sex partners (CDC, 2011). Retention into HIV care promotes life and decreases the risk for HIV transmission (Yehia et. al. 2015).
Preventing HIV transmission, prevention intervention strategy is critical and should be ongoing to all HIV patients consistently.
New cases of HIV in the United States are increasing by approximately 30, 000 per year and with this increase, more providers are needed (Weiser et al.2016).
Methods Quantitative cross sectional study: 2017 Palm Beach County Needs Assessment Survey was used, The data used was secondary-deidentified data. The sample size consisted of 357 survey participants.The surveys were collected from September 2016 to January 2017. The Florida Department of Health (FDOH) Institutional Review Board ( IRB) approval was granted before data Collection.. The participants were not at risk due to de-indentifieddata. The demographic and clinical data was reviewed. Ethical practices were followed by securing data and only the data needed to conduct study were utilized.
The Independent Variables were: Age, Educational Level, Race, Gender, Condom Use, Unprotected sex, Sexual Orientation, Blood Tests-Viral Load, Medical care type facility. The Dependent Variables were: Medical Care/In Care, Miss HIV Meds and Hospitalization.
Four Research Questions are posed in this study, the results section list the research questions. Statistical Test were computed with the use of SPSS with ANOVA and Linear Regression
Results RQ:Is there a statistical significant association between age of HIV patients, retention in care and health outcomes, in Palm Beach County?
Analysis of variance (ANOVA) was conducted to investigate if there was a statistical significant association between age of HIV patients and retention in care .
Analysis Result: ANOVA, F (9, 0.393.) =2.181, p<0.05 (p=0.023). There was statistically significant association between age and retention in care between groups.
Post Hoc (Dunnett test revealed differences between the 50-54 p =0.006, between 55-59, p=0.009 and 60 ≥ p=0.010
RQ2: Is there a statistically significant association between HIV patients at risk for sexually transmitted diseases and retention in care as evidenced by unprotected sex?
Analysis of variance (ANOVA) was conducted to investigate if there was a statistical significant association between at risk for STD of HIV patients and retention in care as evidenced by unprotected sex.
Analysis Result: ANOVA , F (3, 4.531) =15.975, p<0.001 (p=0.000). There was statistically significant association between at risk for STD and retention in care as evidenced by unprotected sex .
Post Hoc (Dunnett) test revealed differences between retention in care and risk for sexually transmitted diseases as evidenced by unprotected sex, p=0
RQ3: Are MSM HIV patients who attend health department clinics and or other health care facilities, more likely to retain in care than other groups of HIV patients?
Analysis of variance (ANOVA) was conducted to investigate if MSN patients who attend health department clinics and other health care facilities, more likely to retain in care than other groups of HIV patients?
ANOVA , F (4, 0.280) = 1.516, p > 0.05 (p= 0.197). There was no statistically significant association between MSN HIV patients who attend health department clinics and other health care facilities than other groups of HIV patients more likely to remain in care?
RQ4: Do patients knowledge of viral load test predict retention in care?
Logistic Regression was conducted to investigate knowledge of viral load and retention in care.
Retention in care and viral load tests regression model was statistically significant
The regression model showed P < 0.01, p=0.000
Viral Load test significantly predicted retention in care.Coefficients of Viral Load greater than 1000 and Less than 200 were statistically significant:Viral Load >1000 p = 0.010;Viral Load < 200 p = 0.004
Conclusions Lmitations to the study included the time frame to complete the study and the use of secondary data which was available to conduct the study. Low viral load is indicative of better health outcomes. Many studies have attempted to address barriers to retain in care and more work is needed to address the factors that impact retention in care.
Findings are consistent with other research that retention in care are due to social, behavioral and system factors. Some of the reasons the patients gave for their not in care are treatment of staff in clinic and or doctors office, long wait times, transportation, language barrier, child care and the clinic hours. The three most frequent answers were treatment of staff in clinic, long wait times and transportation. The burden o fnew HIV infection transmitting HIV if patients do not remain in care. Findings are consistent with other research that retention in care are due to social, behavioral and system factors. Three most frequent answers were treatment by staff, long wait times and transportation.
Acknowledgement 1. Florida Department of Health- Research Excellence Initiative
2. Ryan White-Palm Beach County
3. Palm Beach County Health Department: Medical Director-Dr. Alina Alonso, Center Administrator-Ms. Lawanta Stewart and HIV Provider-Dr. Samuel Frimpong.
References Center for Disease Control & Prevention. Diagnoses of HIV infectionin the United States and dependent areas, 2015 HIV Surveillance Report, 2016; 27.

Drachler, M.D., Drachler, C. W., Teixeira, L.B., & Leite, J. C. D. The Scale of Self-Efficacy Expectations of Adherence to Antiretroviral Treatment: A Tool for dentifying Risk for Non-Adherence to Treatment for HIV. PLoSONE, 2016; 11(2),e0147443.
.
Kambugu, A., Zhang, Y., Braitstein, P., Christopoulos, K.A…Martin, J.N. (2010). Retention in care among HIV infected patients in resource- limited settings: Emerging insights and new directions.
Current HIV/AIDS Report, 2010.; 7(4), 234-244.

Roscoe, C., & Hachey, D.M. Topic 8: Retention in HIV Care. National HIV Curriculum, 2017.

Thompson, M.A., Mugavero, M.J., Amico, K.R., Cargill, V.A., Chang, L.W., Gross, R…Nachega, J.B. Guidelines for improving entry into and retention in care and antiretroviral adherence for persons with HIV: Evidence- based recommendations from an international association of physicians in AIDS care panel. Annals of Internal Medicine, 2012; 156(11), 817-833.

Weiser, J., Beer, L., West, B.T., Duke, C.C., Gremel, G.W., & Skarbinsky, J. Qualifications, demographics, satisfaction and future capacity of the HIVcare provider workforce in the United States, 2013- 2014.Clinical Infectious Disease, 2016; 63(7), 966-975.
Yehia, B. R., Stewart, L., Momplaisir, F., Mody, A, Holtzman, C.W... Shea, J.A. Barriers and facilitators to patient retention in HIV care. Biomedical Central, 2015; 15, 246.
Brief bio for lead author/ presenter to be used by session moderators at the conference Dr. Merlene Ramnon holds a PhD in Public Health, specializing in Epidemiology. She works as a nursing consultant with the Florida Department of Health, Palm Beach. Merlene has presented findings to the public, government organizations, and private organizations through many forums including epi com, APHA and Vaccinology conferences, and webinars. She has been teaching for more than eight (8) years as a nursing professor in various academic settings.
Merlene likes epidemiology/research and wants to commit more time to research.
Brief summary (100 words) of Presentation to be Used in Conference Program Retention in care has been a major concern for many health care providers and it is very important that HIV patients retain into care therefore, health care providers need to include regular prevention services and be aware of the factors that affect retention in care to prevent HIV transmission.The factors assessed in this study predicted retention in care and health outmomes among HIV patients.


Logistic Regression Result


Model DF Mean Square F Sig
Regression 2 3.005 17.663 0.000b
Residual 353 .170    
Total 355      



Linear Regression


Model R RSquare Adjusted R Square Standard Error of Estimates Durbin Watson
1 .032Bb .091 .086 .41244 1.803
a. Predictor Constant Blood Test, Viral Load < 200 Blood Test Viral Load > 1000        
b. Dependent Variable: Medical Care          



Neonatal tetanus surveillance in Bayelsa state of Nigeria: a five-year review

Abisoye S. Oyeyemi, Hilda C. Afakwu, Esievoadje Akpofure, Luke E. Izibekien

Community Medicine, Niger Delta University, Yenagoa, Bayelsa, Nigeria


Objective To assess the performance of neonatal tetanus surveillance in Bayelsa state of Nigeria.
Introduction Neonatal tetanus (NT) though a preventable disease, remains a disturbing cause of neonatal morbidity and mortality particularly in low income countries where maternal and child care are substandard and antitetanus immunization coverage is still poor. The disease, which is mostly fatal, is particularly common in hard to reach and rural areas where deliveries take place at home or with untrained attendants without adequate sterile procedures and in unclean environment. Since eliminating NT became a global target, significant reductions in NT deaths have been reported. The most recent estimates by WHO (2015) put death of newborns due to NT at 34,019, a 96% reduction from the situation in the late 1980s.

All countries are committed to “elimination” of maternal and neonatal tetanus (MNT), i.e., a reduction of NT incidence to below one case per 1000 live births per year in every district. A strong neonatal tetanus surveillance (NTS) is however required to achieve this. As of March 2018, only 14 countries were yet to eliminate MNT and this includes Nigeria.

The different types of NTS recommended are conducted to varying degrees of efficiency and effectiveness in Nigeria under the major surveillance strategy – the Integrated Disease Surveillance and Response (IDSR). These include routine monthly surveillance, zero reporting, active surveillance and retrospective record review.

Nigeria comprises six geopolitical zones, 36 states and a Federal Capital territory (FCT), and is made up of 774 Local Government areas (LGA) (districts) – an LGA being the lowest administrative level. This study was conducted in Bayelsa state – one of the six states in the south zone. It is made up of eight LGAs, more than half of which are riverine and consists of many hard-to-reach communities, where formal functional health facilities are few and far between. Health workers are in short supply and funding of health care delivery is poor in the state.
Methods This was a retrospective review of all confirmed cases of neonatal tetanus that were managed at the two tertiary hospitals in the state - Niger Delta University Teaching Hospital Okolobiri (NDUTH), and Federal Medical Centre Yenagoa (FMC) - between January 2009 and December 2013. These were the only two public facilities that had the capacity to manage NT cases in the state. Relevant data including sociodemographics, pregnancy and birth history of patients, cord care and tetanus toxoid immunization of mothers were abstracted from the case files. The cases were traced to the office of the State Epidemiologist, where all cases were expected to be documented and investigated in line with the existing neonatal tetanus surveillance. Ethical approval was obtained from the Research and Ethics Committee of NDUTH for the research and permission was given to access case files.
Results A total of 48 cases were managed in both facilities (36/75.0% in NDUTH and 12/25.0% in FMC) in the period under review but only 13 cases (27.1%) were reported to the office of the State Epidemiologist. Figure 1 shows the number of cases per year of review. The cases were resident in seven out of the eight LGAs. The mean age of cases was 8.98 (SD = 5.14) days and 29 (60.4%) were male while 19 (39.6%) were female. Available evidence showed that only 2.1% of the cases were protected at birth (mothers had TT2+); 91.7% of mothers did not have antenatal care and all the mothers were delivered by traditional birth attendants; 70.8% had their umbilical cord cut with new (?sterile) blade; and 43.8% had their cord treated with methylated spirit, others were treated with just water or some herbal preparation. Educational attainment of mothers of cases was primary (54.2%) and secondary (45.8%).
Conclusions There were gaps in Neonatal Tetanus Surveillance in Bayelsa State as only 27.1% of cases were captured at the state level. Many mothers and their newborns were still not protected against tetanus, and delivery and cord care were done in unhygienic conditions. There is an urgent need to strengthen NT surveillance, improve vaccination against tetanus, and encourage skilled birth attendance in the state.
Acknowledgement We thank the staff of the hospitals and that of the office of the State Epidemiologist who facilitated data collection.
References 1. WHO. Immunization, Vaccines and Biologicals: Tetanus. http://www.who.int/immunization/diseases/tetanus/en/. Accessed on 23 Jul 2018
2. WHO. Immunization, Vaccines and Biologicals: Maternal and Neonatal Tetanus Elimination (MNTE): The initiative and challenges.http://www.who.int/immunization/diseases/MNTE_initiative/en/ Accessed on 23 Jul 2018.
3. WHO. WHO-recommended standards for surveillance of selected vaccine-preventable disease. WHO. 2003
4. Bayelsa State Ministry of Health. Health facilities and their distribution across the Local Government Areas of Bayelsa State. 2010.
Brief bio for lead author/ presenter to be used by session moderators at the conference Abisoye Oyeyemi holds an MBBS degree (University of Ibadan, Nigeria), an MPH degree (University of Lagos, Nigeria) and is a Fellow of the National Postgraduate Medical College of Nigeria in the Faculty of Public Health (FMCPH). He has been in medical practice for about 20 years and most of the time has been spent in public health and disease control. His academic and professional interests are infectious disease epidemiology - particularly malaria and HIV/AIDS; disease surveillance; and operations research. He is currently a Senior Lecturer at the Niger Delta University, Bayelsa State of Nigeria and a Consultant Public Health Physician. He provides technical support to Bayelsa State Ministry of Health on disease surveillance, having previously served as the State Epidemiologist.
Brief summary (100 words) of Presentation to be Used in Conference Program Neonatal tetanus (NT) is a preventable disease but remains a disturbing cause of neonatal mortality in resource-poor settings where maternal and child care is still poor. This study assessed performance of neonatal tetanus surveillance (NTS) in Bayelsa State of Nigeria through a retrospective case review over five years. Results show gaps in NTS in the state as only 27.1% of cases were notified to the state. Many mothers and their newborns were still not protected against tetanus, and delivery and cord care were done in unhygienic conditions, a situation that calls for strengthening of NTS and improvement of vaccination coverage.



Number of cases seen at the facilities and reported to the office of the State Epidemiologist



Validation of a surveillance-based definition for hepatitis B treatment eligibility.

Kevin Guerra, Regan Deming, Angelica Bocour, Ann Winters

Disease Control / Communicable Disease, New York City Department of Health, Long Island City, New York, United States


Objective To assess the accuracy of a surveillance-based definition for hepatitis B treatment eligibility among New York City residents with chronic hepatitis B infection.
Introduction Approximately 100,000 New York City (NYC) residents are currently diagnosed with chronic hepatitis B virus (HBV) infection.1 Routine monitoring and treatment, where indicated, are necessary to reduce HBV disease progression. Using the 2017 European Association for the Study of the Liver (EASL) 2 guidelines on HBV infection management, we developed a surveillance-based definition for treatment eligibility. Validation of this definition will support the creation of a population-level HBV care continuum, which will allow us to monitor gaps from HBV diagnosis to viral suppression and to develop public health interventions to address these gaps.
Methods Laboratories everywhere are required to electronically report the following HBV tests to the NYC Department of Health and Mental Hygiene (DOHMH) for all NYC residents: positive and negative (as of April 2018) DNA, positive surface antigen, positive e antigen, positive core IgM, and Alanine aminotransferase (ALT) (when ordered at the same time as another reportable HBV test). Using reportable HBV tests, treatment eligibility was defined as ever having an HBV DNA result >2000 IU/mL and ALT>40 U/L. We assessed the accuracy of the surveillance-based definition by calculating sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) by applying the definition to the test data of people participating in two DOHMH programs that included clinical information on treatment eligibility: the Enhanced Surveillance Project (provider interviews conducted for 300 randomly selected patients with chronic HBV) and the Check Hep B Patient Navigation Program (program providing HBV-related patient navigation at community organizations, health centers, and hospitals). Everyone meeting inclusion criteria in the Enhanced Surveillance Project who were also identified as being in care and being monitored (two or more HBV DNA results reported at any time) were included in our analysis. For Check Hep B, we included everyone enrolled prior December 31, 2017 who also met our criteria of being in care and being monitored. To determine treatment eligibility using surveillance data, we used all HBV DNA and ALT results reported prior to January 31st, 2016 for the Enhanced Surveillance project and prior to December 31st, 2017 for Check Hep B.
Results Treatment eligibility was 62.0% (145/234) among people from the Enhanced Surveillance Project (Table 1A) and 40.0% (161/402) among people enrolled in Check Hep B (Table 1B). Sensitivity of the surveillance-based definition was low using both data sources (Enhanced Surveillance Project: 26.2%; Check Hep B: 24.2%) and specificity high (Enhanced Surveillance Project: 92.1%; Check Hep B: 94.2%). PPV was 84.4% and 73.6% for the Enhanced Surveillance project and Check Hep B, respectively, while NPV was 43.4% and 65.0% for the Enhanced Surveillance project and Check Hep B respectively.
Conclusions Our surveillance-based definition had high specificity, indicating that the great majority of patients who were truly not treatment-eligible were correctly classified. However, sensitivity was low, indicating that the surveillance-based definition was unable to accurately identify those considered treatment-eligible from either data source. Low sensitivity suggests that clinicians are likely using other clinical factors not included in laboratory-based reporting to assess a patient’s eligibility for treatment, such as fibrosis and cirrhosis, and that clinicians might be using guidelines other than EASL (e.g., American Association for the Study of Liver Diseases (AASLD)3) to determine treatment eligibility. We will conduct chart reviews to better understand the variability in criteria being used. These chart reviews will allow us to further refine our surveillance-based definition (e.g., by incorporating different HBV tests or for clinical criteria that are not laboratory-based, including information from external sources such as Regional Health Information Organizations (RHIOs)), eventually supporting the creation of an HBV care continuum for NYC.
Acknowledgement
References 1. France AM, Bornschlegel K, Lazaroff J, Kennedy J, Balter S. Estimating the prevalence of chronic hepatitis B virus infection--New York City, 2008. Journal of urban health: bulletin of the New York Academy of Medicine 2012; 89(2): 373-83.

2. European Association for the Study of the Liver. Electronic address eee, European Association for the Study of the L. EASL 2017 Clinical Practice Guidelines on the management of hepatitis B virus infection. J Hepatol 2017; 67(2): 370-98.

3. Terrault NA, Bzowej NH, Chang KM, Hwang JP, Jonas MM, Murad MH; American Association for the Study of Liver Diseases. AASLD guidelines for treatment of chronic hepatitis B. Hepatology. 2016;63(1):261–83. https://doi.org/10.1002/hep.28156.
Brief bio for lead author/ presenter to be used by session moderators at the conference Kevin Guerra is a data analyst in the Viral Hepatitis program at the New York City Department of Health. He has a a background in infectious disease epidemiology and his current work focuses on the surveillance of Hepatitis B and C.
Brief summary (100 words) of Presentation to be Used in Conference Program Using the 2017 EASL guidelines, we developed a surveillance-based definition for treatment eligibility. The surveillance-based definition was defined as having an HBV DNA result >2000 IU/mL and ALT>40 U/L. The definition was validated using two data sources, an Enhanced Surveillance project and the NYC Check Hep B Patient Navigation Program. For each data source specificity was high while sensitivity was low. These results indicate that NYC clinicians may be relying on different guidelines and other clinical factors such as fibrosis and cirrhosis to determine whether to start treatment in patients with chronic hepatitis B.




Epidemiological trends of Reported Legionnaires’ disease in Houston, Texas, 2014-2017

Razina Khayat, Najmus Abdullah, Sudipa Biswas, Hafeez Rehman, Kirstin Short

Informatics, City of Houston , Houston, Texas, United States


Objective To study trends and patterns in legionnaires’ disease cases in Houston, Texas, from 2014-2017.
Introduction Legionellosis is a respiratory illness that is mostly (80-90%) caused by the bacterium Legionella pneumophila. It is associated with a mild febrile illness, Pontiac fever, or Legionnaires’ disease (1), a source of severe, community-acquired pneumonia. Legionella bacteria mostly affect elderly persons specifically those with underlying debilitating illnesses and with lowered immune systems. Water is the major natural reservoir for Legionella, and the pathogen is found in many different natural and artificial aquatic environments such as cooling towers or water systems in buildings, including hospitals. An abrupt increase in the incidence of Legionnaires’ has been noted since 2003 throughout the nation. According to CDC, about 6,000 cases of Legionnaires’ disease were reported in the United State in 2015 (1). Incidence rates of Legionnaires for the year 2015 were 1.06 and 1.90 (ref) for Texas and the United States respectively (2). Increased number of reported cases might be due to the fact of an older population, more at risk individuals, aging plumbing infrastructure, and increased testing for Legionnaires’ disease by various hospitals and laboratories.
Methods Data were extracted from Houston’s Electronic Disease Surveillance System (HEDSS) from January 1, 2014, to December 07, 2018. Confirmed cases were analyzed to examine the epidemiologic trends across years 2014 to 2018. Demographic characteristics such as age, race, and gender were also analyzed. Incidence rates, case fatality and time lapse from date of diagnosis to date of reporting to the health department were also studied. Data were analyzed using SAS statistical software, version 9.4. Only Houston residents were included in the analysis. To be considered confirmed, a case must be clinically compatible and fulfill at least one of the confirmatory laboratory criteria.
Results There were 218 cases of LD reported to the City of Huston from 2014 to 2018. Only 116 cases (53%) were classified as confirmed. Reported cases may have been not confirmed due to the lack of fulfilling the case criteria for the case. Providers may have ordered a non-confirmative test, or the case may not have satisfied the clinical compatibility due to loss to follow-up or for other reasons.
Most of the confirmed cases were reported from larger for-profit hospitals (500+beds) in the area. The majority of cases were diagnosed by urinary antigen test (95, 82%). There were four deaths due to legionnaires disease during this period giving a case fatality rate of 3.4%. Death rates were inaccurate, though, and could be higher than reported since cases were not followed up after being reported to the state. From 2014 to 2018, legionnaires’ disease incidence rates increased from 0.71 to 1.36 per 100,000, an average annual increase of 17%.
In 2014–2018, the incidence of LD was higher among men compared with women. 67 cases (58%) were male, and 49 (42%) were female. Female cases remained stable throughout the years while male cases increased from 6 to 23, an increase of approximately four folds. The median age was 60 years with a range of 21 to 96 years. LD incidence increased with age; it was highest among residents 65 years and older (42,36%). African Americans had the highest incidence of LD (40, 35%) followed by Hispanics (29, 25%). African Americans cases had more than doubled through years 2014-2018 from 6 to 13. Cases were higher in warmer months specifically in July (14) an August (13).
Conclusions Cases were higher in the warmer months and the highest among the elderly, men, and those of African American race. ELR was the prime source of initial case reporting to the health department. The number of legionnaire’s cases observed were increasing with each passing year. The ratio of confirmed cases to those reported were only 53% thus raising awareness and appropriate education to the investigators and providers are highly advised. It is critical to the control of LD that enhanced surveillance is maintained at a high level. Consequently, more consideration should be given for the more widespread use of Legionnaires confirming test when a patient presents with pneumonia.
Hospitals and other healthcare facilities often have large, complex water systems, making them potentially high-risk settings for transmission of legionellosis to vulnerable patients or residents. We recommend all healthcare facilities have a water management program to control Legionella.
Acknowledgement We thank the flollowing agencies for providing data for this study:
1. Texas Department of State Health Services.
2. Houston Health Department, Diivision of Disease Prevention and Control.
3. Houston Electronic Disease Surveillance System.
References 1. Centers for Disease Control and Prevention. (2018). Infection Control Assessment Tools. Retrieved October 5, 2018, from https://www.cdc.gov/legionella/
2. Texas Health and Human Services. (2018). Legionellosis. Retrieved October,5, 2018, from https://www.dshs.texas.gov/idcu/disease/legionnaires/
Brief bio for lead author/ presenter to be used by session moderators at the conference I am Razina Khayat, currently working as an Informatic, Biostatician at the City of Houston Health Department. I also worked as an Infectious Disease Epidemiologist at the same department. I'm interested in Emerging Infectious Disease Research and prevention.
Brief summary (100 words) of Presentation to be Used in Conference Program Legionellosis is a respiratory condition which is caused by the bacteria Legionella pneumophilia. Throughout the nation, there was an abrupt increase in incidence of legionnaires' disease since 2003. Data used in this study were extracted from Houston’s Electronic Disease Surveillance System (HEDSS) from January 1, 2014 to December 31, 2017. Among the confirmed cases, males, African American and adults over 65 years oldhad the highest proportion of cases. Large number of Legionnaire cases are still under reported. Thus, continued monitoring and surveillance activities are warranted for prevention and control of Legionnaire disease.


Surveillance for Lyme disease in Canada: 2009-2015

Jules Koffi, Salima Gasmi

Public Health Agency of Canada, Saint-Hyacinthe, Quebec, Canada


Objective This study aims to describe incidence over time, geographic and seasonal distribution, demographic and clinical characteristics of Lyme disease cases in Canada.
Introduction Lyme disease (LD), a multisystem infection that is manifested by progressive stages (1), is emerging in central and eastern provinces of Canada due to northward expansion of the geographic range of Ixodes scapularis, the main vector in these regions (2). In 2004, approximately 40 human cases of LD were reported in Canada. In 2009, LD disease became nationally notifiable, with provincial and territorial health departments reporting clinician-diagnosed cases to the Public Health Agency of Canada (PHAC). This study summarizes seven years (2009-2015) of national surveillance data for LD in Canada.
Methods National Lyme disease surveillance data is collected through two surveillance systems, the Canadian National Disease Surveillance System (CNDSS) and the Lyme disease enhanced surveillance system (LDES). The CNDSS collects only demographic data (age and sex), and information on episode date and case classification. The LDES system captures additional data, including: possible geographic location of infection (for both locally acquired and travel-related cases); clinical manifestations; and results of laboratory testing. Nine provinces out of ten participate to LDES that means they provide a part of or all the data elements of this surveillance system. The 2009 national Lyme disease case definition (3) that distinguishes confirmed and probable cases (Table 1) is used to classify and report cases diagnosed by clinicians.This study describes the incidence over time, seasonal and geographic distribution, demographic and clinical characteristics of reported LD cases. Logistic regression was used to explore variations among age groups, sex and year of reporting clinical manifestations to better understand potential demographic risk factors for the occurrence of LD. Different models were used with as outcomes absence or presence of: erythema migrans (early Lyme disease), neurologic and cardiac symptoms and multiple erythema, migrans (early disseminated Lyme disease); and arthritis (late disseminated Lyme disease). The most parsimonious multivariate models were sought by backward elimination of nonsignificant variables until all factors in the model were significant (P<0.05).
Results The number of reported LD cases increased more than six-fold, from 144 in 2009 to 917 in 2015, mainly due to an increase in infections acquired in Canada. For the provinces participating into the LDES system, the month of illness onset for Lyme disease cases acquired in Canada was available for 2010 cases. Most cases were reported during the summer months of June (20.7%), July (35.4%) and August (17.3%) (Figure 1). An increase in incidence of LD was observed in provinces from Manitoba eastwards (Figure 2). This is consistent with our knowledge of range expansion of the tick vectors in this region. In the western provinces the incidence has remained low and stable. All cases reported by Alberta, Saskatchewan and Newfoundland and Labrador were acquired outside of the province, either elsewhere in Canada or abroad. There was a bimodal distribution for LD by age with peaks at 5–9 and 45–74 years of age (Figure 3). The most common presenting symptoms were a single erythema migrans rash (74.2%) and arthritis (35.7%) (Figure 4). In the multivariate analysis for clinical manifestations, children aged 0–9 years had a greater number of cases reported as early LD (erythema migrans only) than patients aged 10–19 and 30–39 years (P<0.05). For early disseminated manifestations, young adults 20–29 years of age reported more neurologic manifestations, cardiac manifestations or multiple erythema migrans than the reference age group of 0–9 years (P<0.05). For late disseminated manifestations, children under 15 years of age were more frequently reported as having arthritis than other age groups.
Conclusions Lyme disease incidence continues to increase in Canada as does the geographic range of ticks that carry the LD bacteria. This increasing of LD incidence might also be due to changing in knowledge, attitudes, and practices of clinicians who diagnose the disease and or of the public health workers who collect and report the data. Ongoing surveillance, preventive strategies as well as early disease recognition and treatment will continue to minimize the impact of LD in Canada.
Acknowledgement The authors thank all the provincial and regional public health workers who collect and report data to the Public Health Agency of Canada.
References 1. Aguero-Rosenfeld ME, Wang G, Schwartz I, Wormser GP (2005) Diagnosis of Lyme borreliosis. Clin Microbiol Rev 18: 484–509.
2. Ogden NH, Koffi KJ, Pelcat Y, Lindsay LR. Environmental risk from Lyme disease in central and eastern Canada: a summary of recent surveillance information. Can Commun Dis Rep. 2014;40(5):74-82. http://www.phac-aspc.gc.ca/publicat/ccdr-rmtc/14vol40/dr-rm40-05/ assets/pdf/14vol40_05-eng.pdf.
3. Public Health Agency of Canada. Case definition for communicable diseases under National Surveillance. Ottawa: Public Health Agency of Canada; 2017. https://www.canada.ca/en/public-health/services/ reports-publications/Canada-communicable-disease-report-ccdr/ monthly-issue/2009-35/definitions-communicable-diseases-national-surveillance/lyme-disease.html [Accessed 2017 Aug 17].
Brief bio for lead author/ presenter to be used by session moderators at the conference Jules Koffi is an epidemiologist at the Centre for Food-borne, Environmental and Zoonotic Infectious Diseases of the Public Health
Agency of Canada. He is a family physician graduated from University of Abidjan Cocody ( Côte d'Ivoire), with a master in epidemiology at the University of Montreal. Dr Koffi is working at the Public Health Agency of Canada since 2009. He contributed to the design and the implementation of the Lyme disease Enhanced Surveillance system in Canada in partnership with the provincial public health organizations. Since March 2015 he is the lead of this surveillance system. He is working on operational activities of this surveillance system, analyzing the surveillance data and writing national annual report. Also, he conducts field tick surveillance to detect/identify emerging tick established populations and new Lyme risk areas in collaboration with provincial public health organizations and university researchers..
Brief summary (100 words) of Presentation to be Used in Conference Program This study summarized the first 7 years (2009-2015) of national surveillance for Lyme Disease (LD) in Canada. LD surveillance data for 2009-2015 were analyzed to describe the overtime incidence, geographic and seasonality distributions, demographic and clinical characteristics of . LD in Canada. The results of the study indicated that LD incidence is increasing in central and eastern Canada due to northwards expansion of the blackelleged tick (Ixodes scapularis); erythema migrans was the ,most common symptoms; childrens between 5–9 and adults between 45–74 years of age were the most affected sub-populations; and there were variations in the frequency of reported clinical manifestations.



Figure 1: Month of Lyme disease illness onset for locally-acquired infection: Canada, 2009-2015 (n=2,010)




Figure 2: Reported locations of Lyme disease acquisition, Canada, 2009–2015




Figure 3: Incidence of Lyme disease by age group and sex, Canada 2009-2015 (n=3,004)




Figure 4: Percentage of clinical manifestations for Lyme disease infections acquired in Canada, 2009-2015 (n=1,657)



Table 1: 2009 national Lyme disease case definition


Confirmed case Probable case
Clinical evidence of illness with
laboratory confirmation:
Clinical evidence of illness
without a history of residence in
or visit to an endemic area but
with laboratory evidence of
infection:
isolation of
Borrelia burgdorferi from an
appropriate clinical specimen
positive serologic test using
the two-tier ELISA and Western
Blot criteria
OR detection
of B. burgdorferi DNA by PCR
OR clinician-observed erythema
migrans without laboratory
evidence but with history
of residence in or visit to an
endemic area
OR clinical evidence of illness with
a history of residence in, or
visit to, an endemic area and
with laboratory evidence of
infection, i.e., positive serologic
test using the two-tier ELISA
and Western Blot criteria


Multimorbidity Network Surveillance: Chronic Disease Clusters and Social Disparities

Eun Kyong Shin1, Youngsang Kwon2, Arash Shaban-Nejad1

1Pediatrics, UTHSC, Memphis, Tennessee, United States, 2University of Memphis, Memphis, Tennessee, United States


Objective We study how multimorbidity prevalence is related to socio-economic conditions in Memphis, TN. In addition, we demonstrate that the accumulation of chronic conditions, which is measured by affinity in multimorbidity, is unevenly distributed throughout thecity. Our research shows that not only are socio-economic disadvantages linked to a higher prevalence in each major chronic condition, but also major chronic conditions are heavily clustered in socially disadvantaged neighborhoods.
Introduction Chronic diseases impose heavy burdens onhealth systems, economies, andsocieties (1). Half of all Americans live with at least one of the chronic conditions and more than 75% of health care cost is associated with people with chronic diseases (2). Multimorbidity, the coexistence of two ormore chronic conditions in an individual or a population, often require complex and ongoing care and a deep understanding of different risk factors, and their indicators.Multimorbidity has been increased over the past years and the trend is expected to continue across the U.S. Knowing how different chronic conditions are related to one another andwhat are the underlying socioeconomic factorsis crucial to design and implement effective health interventions. We introduce “multimorbidity network affinity”, which measures the degree of how multiple chronic conditions are clustered within a geographic unit. Accurate estimations of how chronic conditions are spatially clustered and linked to other sociomarkers(3) and socio-economic disadvantages facilitate designing effective interventions.
Methods Multiple datasets including major chronic condition data from the Center for Disease Control and Prevention (CDC) 500 cities, and socio-demographic data from the U.S. Census Bureau and the Environmental Systems Research Institute (ESRI) demographics data have been consistently integrated. Then, network analytics have been performed to examine the inter-relations among a selected number of major chronic conditions and their manifestations in Memphis. To checkwhether a distinctive geographic pattern in multimorbidity is present, we carried out a test using global Moran’s I and Getis-Ord Gi*statistics. If apattern is detected, we use robust regression to explore how affinity isassociatedwiththe socio-economic disadvantages of the area.
Results The network analysis confirms the existence of close relationships between various chronic conditions. Ourspatial analysisshowthat the geo-distinctive patterns of clustered comorbidities are associated with socio-economic deprivation. Statistical results suggest that neighborhoodswith higherrates of crime, poverty, and unemployment are associated with an increased likelihood of having dense clusters of chronic conditions.
Conclusions This study shows the importance of geospatialfactors in multimorbidity network surveillance. Moreover, it demonstrates how socio-economic disadvantages and multimorbidity network are connected. The healthdisadvantages are disproportionately accumulated in socially disadvantaged areas. Network analysis enables us to discover the links between commonly co-observed chronic diseases and explore the complexity of their interactions. This will improve the surveillance practice and facilitate timely response as well as public health planning and decision making.
Acknowledgement
References 1. Wu S-Y, Green A. The Growing Crisis of Chronic Disease in the United States. RAND Corporation. 2000.
2. Anderson G, Horvath J. The growing burden of chronic disease in America. Public health reports. 2004;119(3):263-70.
3. Shin EK, Mahajan RM, Akbilgic OA, Shaban-Nejad A. Sociomarkers and Biomarkers: Predictive Modeling in Identifying Pediatric Asthma Patients at Risk of Hospital Revisits. npj Digital Medicine (2018) 1:50; doi:10.1038/s41746-018-0056-y.
Brief bio for lead author/ presenter to be used by session moderators at the conference Dr. Eun Kyong Shin is currently a postdoctoral fellow at the University of Tennessee Health Science Center. By training, she's sociologist, holding PhD in Sociology from Columbia University and her work focuses on social determinants of health outcomes, using multiple analytical tools, such as stataitistics, network science, machine learning and GIS.
Brief summary (100 words) of Presentation to be Used in Conference Program The paper, Multimorbidity Network Surveillance: A Closer look at Chronic Disease Clusters and Social Disparities in Memphis, TN, shows that not only socio-economic disadvantages are significantly linked to higher prevalence in each major chronic condition, but also major chronic conditions are heavily clustered in socially disadvantaged neighborhoods.


Impact of a new diagnoses thesaurus on the French ED syndromic surveillance system

Cécile Forgeot1, Isabelle PONTAIS1, Emmanuel Dos Ramos4, Gilles Viudes3, Christophe Vincent-Cassy2, François Dubos3, Anne Fouillet1, Céline Caserio-Schönemann1

1DATA, Sante publique France, Saint-maurice, France, 2SFMU, French society for emergency healthcare, Paris, France, 3Fedoru, Emergency regional observatory Federation, Paris, France, 4ORUPACA, PACA Emergency Network Observatory, Hyères, France


Objective The study aims to evaluate the potential impact of the revision of the thesaurus used by ED physicians to code medical diagnoses, on the syndromic indicators used daily to achieve the detection objective of the French syndromic surveillance system.

Introduction As part of the French syndromic surveillance system SurSaUD®, the French Public Health Agency (Santé publique France) collects daily data from the emergency department (ED) network OSCOUR® [1]. The system aims to timely identify, follow and assess the health impact of unusual or seasonal events on emergency medical activity.
Individual ED data contain demographic (age, gender, residence zip code), administrative (dates of attendances and discharge, ED, etc.) and medical information (chief complaint, main and associated medical diagnoses, severity). Medical diagnoses are encoded using the ICD10 classification. Then syndromic groups are built based on these ICD10 codes for ensuring syndromic surveillance in routine.
Even if ICD10 is recommended on the national guidelines for coding ED attendances, this thesaurus offers a too large variety of codes. Particularly, it includes lots of diseases that may never be observed or confirmed in ED. This variety let selection of the appropriate codes difficult for physicians in a reactive use and could discourage them to code diagnoses.
In order to encourage appropriate and reactive coding practice, we decided in 2017 to produce a new diagnoses thesaurus with a limited list of ICD10 codes. Then a committee of medical and epidemiological experts was created by the Federation of regional emergency observatories (FedORU), to propose an operational thesaurus that includes relevant codes for both ED in a daily routine practice and syndromic surveillance.
Methods The committee has met 10 times since 2017. Since it would have been hard to work on the complete ICD10 list, the work was based on a more limited thesaurus already used by part of French ED. Only codes, which were pertinent regarding ED activity and interest for public health alert, have been considered. The main principles that have guided the selection were to 1) keep codes related to diagnoses that physicians are able to diagnose on a clinical basis or with rapid diagnostic tests, 2) remove diagnoses providing redundant information regarding other variables (such as circumstantial information) and 3) ensure that a substitution code was kept when a removed code was frequently used or was of interest for syndromic surveillance.
Among the 86 syndromic groups defined on the basis of a list of ICD10 codes selected in the complete thesaurus, 34 are daily analyzed by Santé publique France for outbreak detection and early assessment of public health events. Those 34 syndromic groups have been recalculated by considering the revised thesaurus on a three-year period (from 2015 to 2017) at national level.
In order to measure the potential impact of the revised thesaurus on the syndromic groups, we have considered three evaluation measures:
1. the proportion of ICD10 codes deleted (removal rate) from the initial definition of each syndromic group, due to the limitation of the thesaurus (calculated for the 86 syndromic groups);
2. the mean difference in the daily number of attendances between the initial and the new versions of each syndromic group (calculated for the 34 syndromic groups);
3. the linear correlation coefficient between the daily numbers of attendances of the initial and the new version of each syndromic group, in order to assess if the daily fluctuations of the new syndromic group are similar to those of the initial syndromic group (calculated for the 34 syndromic groups).
Results Among the 86 syndromic groups, 75 (85%) have been impacted by the revised thesaurus, which implied codes removal. Among those 75 syndromic groups, the number of ICD10 codes included in their definition has been reduced by 71% on average. This removal rate varied between 17% and 100%. Syndromic groups including initially more than 100 codes have been the most concerned by a limitation of the number of ICD10 codes.
Among the 34 syndromic groups daily analyzed for outbreak detection, 32 have been impacted by code removal with a mean removal rate of 68% (0%-97%). On average, 77% of daily attendances have been retained by the new version of syndromic groups, varying from 15% to 100%. Only 3 syndromic groups have kept less than 60% of attendances: Decrease of well-being (36%), Conjunctivitis (32%) and Hypothermia (15%).
On average, the correlation coefficient has been of 0.96, varying from 0.57 to 1. The lowest values have been observed for the same three syndromic groups listed above: Decrease of well-being (0.57), Conjunctivitis (0.91) and Hypothermia (0.59). 18 among the 34 syndromic groups had a correlation coefficient higher than 0.99.
Conclusions The study showed that most of the syndromic groups were impacted by the revised thesaurus, which resulted in a removal of about two thirds of the ICD10 codes usually considered in daily surveillance. However, more than three quarters of attendances were still retained in the new syndromic groups. This new thesaurus was conceived to rationalize the number of diagnoses codes but a substitution code was systematically proposed to replace removed codes.
Those results highlighted that a large number of codes included in the complete ICD10 thesaurus were rarely used and that the most frequent codes were kept in the revised thesaurus version.
However, this study showed that a few syndromic groups were strongly impacted by the revised thesaurus and can suffer of reduced performances to detect unusual variations. Based on those results, a second round of exploration of specific parts of the complete ICD10 thesaurus will be necessary to adapt either syndromic groups or the revised thesaurus.
Even if the number of attendances may be reduced due to the removal of ICD10 codes, temporal variations remain similar for the majority of syndromic groups. Syndromic surveillance system does not aim to provide exhaustive quantification of attendances for a pathology, but aims to be able to detect expected or unusual public health variations.
These evaluation results correspond to the worst-case scenario assuming that ED physicians will not modify their encoding habits by using the substitution codes but keep using their current thesaurus. However, we expect that this new and simplified version will facilitate diagnosis encoding task and lead toward a better diagnosis encoding rate. Once this new thesaurus will be widely used, we can expect a substantial improvement of the quality of ED medical data and then of syndromic surveillance results.
Finally, this study enhances the importance that both data providers and epidemiologists in charge of syndromic surveillance work closely, in order to improve system in shared objectives.
Acknowledgement
References [1] Fouillet A, Bousquet V, Pontais I, Gallay A, Caserio-Schönemann C. The French emergency department OSCOUR network: evaluation after a 10-year existence. Online J Public Health Inform. 2015; 7(1): e74.
Brief bio for lead author/ presenter to be used by session moderators at the conference After a master's degree, the author arrived in French National Public Health Agency in april 2011. She is actually in charge of the animation of the french ED network OSCOUR® and analysis of data collected by the French syndromic surveillance system SurSaUD®.
Brief summary (100 words) of Presentation to be Used in Conference Program In order to improve encoding practice of medical diagnosis in emergency departments (ED), a new thesaurus with a limited number of ICD10 codes has been produced by a committee of medical and epidemiological experts. This study showed that 68% of diagnosis codes from the initial thesaurus were removed in most of the daily analyzed syndromic groups, but three-quarters of attendances were still retained and temporal variations remained similar. A substitution code was proposed for each removed diagnoses in the new thesaurus. This new thesaurus will facilitate encoding practice and lead toward a better quality of ED medical.


Using Syndromic Surveillance and Climatic Data to Detect High Intensity HFMD Seasons

Arden Norfleet

Office of Epidemiology, Virginia Department of Health, Richmond, Virginia, United States


Objective To assess the relationship between seasonal increases in emergency department (ED) and urgent care center (UCC) visits for hand, foot, and mouth disease (HFMD) among children 0-4 years old and average dew point temperatures in Virginia. To determine if this relationship can be used to develop an early warning tool for high intensity seasons of HFMD, allowing for earlier targeted public health action and communication to the community and local childcare centers during these high intensity seasons.
Introduction Hand, foot, and mouth disease is a highly infectious disease common among early childhood populations caused by human enteroviruses (Enterovirus genus).1 The enteroviruses responsible for HFMD generally cause mild illness among children in the United States with symptoms of fever and rash/blisters, but have also been linked to small outbreaks of severe neurological disease such as meningitis, encephalitis, and acute flaccid myelitis.2

Enteroviruses circulate year-round but increase in the summer-fall months across much of the United States.3 The drivers of this seasonality are not fully understood, but research indicates climatic factors, rather than demographic ones, are most likely to drive the amplitude and timing of the seasonal peaks.3 A recent CDC study on nonpolio enteroviruses identified dew point temperature as a strong predictor of local enterovirus seasonality, explaining around 30% of the variation in intensity of transmission across the United States.3
Methods Syndromic surveillance data on ED and UCC visits among 0-4 year olds in Virginia were analyzed from January 1, 2012 to August 31, 2018. Visits for HFMD were identified using the following chief complaint and discharge diagnosis terms: hand, foot, and mouth; HFM; fever with rash, lesions, or blisters; ICD-10 code: B08.4; or SNOMED CT code: 266108008. Visits for HFMD among 0-4 year olds were aggregated by week and calculated as a proportion of all ED and UCC visits among this age group during the study period.

Hourly dew point readings from the Richmond International Airport from January 1, 2012 to August 31, 2018 were obtained from the National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center (NCDC). NOAA readings were averaged by week to establish a mean dew point for each week during the study period. Correlation analyses were performed on weekly dew point temperatures and weekly percent of HFMD visits. Weekly dew point averages were used to determine low-activity weeks at which to measure baseline percentages of HFMD visits. A low-activity week was defined as periods of two or more consecutive weeks in which each week had an average dew point temperature of less than 55.4 degrees Fahrenheit.3 To assess if HFMD seasons varied in intensity from year to year, a Kruskal-Wallis test was used to assess significant differences by year among the mean weekly percent of HFMD visits during high-activity weeks.

An early warning threshold for a high intensity season was developed by calculating the mean percent of HFMD visits during low-activity weeks for the previous three years and adding two standard deviations. Threshold rates were calculated for years 2015 through 2018 and compared to the percentage of 0-4 year old HFMD visits during high-activity weeks. The week where percent of HFMD visits crossed the early warning threshold in 2018 was assessed to determine when public health notifications could have been made to alert the community about a high intensity (above threshold) HFMD season if this early warning tool had been utilized.
Results Between January 1, 2012 and August 31, 2018, there were 27,181 visits for HFMD among children aged 0-4 years. Mean and median weekly percent of HFMD visits were 1.33% and 1.01% of total 0-4 year old visits, respectively, with a range from 0.18% to 5.32%. These visits were most prominent during the summer or fall each year, with annual peaks occurring between weeks 22-46.

Weekly percent of HFMD visits and average weekly dew point temperatures were significantly correlated (r=0.562, p<.0001). The mean weekly dew point temperature for high-activity weeks was 67.2 degrees Fahrenheit, with a range between 49.3 and 73.5 degrees. A Kruskal-Wallis test showed a significant difference in the mean weekly percent of visits by year for high-activity weeks (p<.0001).

Over the 4 years of data to which the threshold was applied, percent of HFMD visits crossed the threshold in 2016 and 2018, indicating both years experienced high intensity HFMD seasons (Fig. 1). Percent of HFMD visits never crossed the early warning threshold in 2015 nor 2017. In 2018, the threshold was met on Week 21 (week ending June 2, 2018) which was more than 3 weeks prior to when public health notifications were made using routine surveillance methods through ESSENCE.
Conclusions Visits for HFMD among the young childhood population (0-4 year olds) in Virginia exhibit annual summer-fall seasonality with significant differences between the percent of visits from year to year. Seasons exhibiting a significantly higher percent of HFMD visits during high-activity weeks warrant a greater level of public health communication and outreach to educate parents, physicians and childcare centers about the disease and prevention measures. It can be difficult to differentiate high intensity seasons from low intensity seasons in the early weeks of increasing disease activity. Traditional syndromic surveillance methods using ESSENCE identify significant increases in HFMD visits from the previous 90 days, but do not readily alert on differences in seasonality from year to year. These results support the use of dew point temperature data to develop an early warning tool for high intensity seasons of HFMD. This early warning tool will allow for more efficient use of resources and targeted outreach during years with particularly high HFMD activity within the young childhood population. This early warning tool will be implemented by the Virginia Department of Health in 2019 to evaluate its effectiveness at identifying high HFMD activity in real-time.
Acknowledgement I would like to thank Erin Austin, Jonathan Falk, and Tim Powell for their guidance and review.
References 1. Khetsuriani N, Lamonte-Fowlkes A, Oberst S, Pallansch MA. Enterovirus surveillance—United States, 1970–2005. MMWR Surveill Summ 2006;55(No. SS-8). https://www.ncbi.nlm.nih.gov/pubmed/16971890
2. Centers for Disease Control and Prevention (2018). Hand, Foot, and Mouth Disease (HFMD). Retrieved Sept 25, 2018, from https://www.cdc.gov/hand-foot-mouth/about/complications.html.
3. Pons-Salort M, Oberste MS, Pallansch MA, et al. The seasonality of nonpolio enteroviruses in the United States: patterns and drivers. Proc Natl Acad Sci U S A 2018;115:3078–83 http://doi.org/10.1073/pnas.1721159115
Brief bio for lead author/ presenter to be used by session moderators at the conference Arden Norfleet is the Enhanced Surveillance Epidemiologist for the Virginia Department of Health. In this role she analyzes syndromic surveillance data from emergency department and urgent care centers in Virginia using ESSENCE. She has been with the Virginia Department of Health for two and a half years. Arden earned her M.P.H. from Virginia Commonwealth University. She is originally from Mobile, Alabama.
Brief summary (100 words) of Presentation to be Used in Conference Program Visits for hand, foot, and mouth disease (HFMD) among 0-4 year olds in Virginia exhibit annual summer-fall seasonality with variations in intensity from year to year. Dew point temperature has recently been identified as a driver of local enterovirus seasonal patterns. The relationship between seasonal increases in emergency department and urgent care center visits for HFMD among children 0-4 years old and average dew point temperatures in Virginia is used to develop an early warning tool for high intensity seasons of HFMD. This early warning tool will allow for earlier public health communication and outreach during high intensity HFMD seasons.




Monitoring Heat-Related Illness through Syndromic Surveillance in Los Angeles County

Jimmy Duong, Michael Lim, Emily Kajita, Bessie Hwang

Public Health, Los Angeles County, Los Angeles, California, United States


Objective To analyze Los Angeles County’s (LAC) extreme heat season in 2018 and evaluate the Council of State and Territorial Epidemiologists’ (CSTE) syndrome query for heat-related-illness (HRI) in Los Angeles County (LAC)
Introduction LAC experienced several days of record-breaking temperatures during the summer of 2018. Downtown Los Angeles temperatures soared to 108°F in July with an average daily maximum of 92°F. Extreme heat events such as these can pose major risks to human health. Syndromic surveillance can be a useful tool in providing near real-time surveillance of HRI. In 2014, a working group was formed within the CSTE Climate Change Subcommittee to define and analyze HRI. The workgroup’s goal was to provide guidance to public health professionals in adapting and implementing an HRI syndrome surveillance query. The Acute Communicable Disease Control Program’s (ACDC) Syndromic Surveillance Unit utilized CSTE’s HRI query to provide surveillance during the extreme heat season in 2018 in LAC. Additional modifications to the CSTE query were evaluated for potential improvements towards characterizing HRI trends.
Methods From May 1 to September 30, 2018, Emergency Department (ED) data were queried for cases using the CSTEs definition for HRI. The queries consisted of key word searches within the chief complaint (CC) data field, and, if available, the diagnosis data fields. The query was derived from the CSTE HRI query published in 20161. In addition, ACDC explored the utility of expanding the CSTE syndrome definition to include additional chief complaints commonly associated with HRI such as dehydration and syncope. Both queries were applied on all participating syndromic EDs in LAC alongside daily high temperature data trends. Local temperature data for downtown Los Angeles weather station KCQT were taken from the Weather Underground website. Spearman correlation coefficients were calculated for each query during the heat season. Similarly, both queries were also applied during colder months from October 1, 2017 to April 30, 2018 for comparison. Lastly, results for dehydration and syncope were independently assessed apart from other HRI query terms during both heat seasons and colder months.
Results The CSTE HRI query and the query with the added terms yielded 1,258 and 63,332 ED visits, respectively, during the heat season. On July 6, the maximum daily temperature peaked at 108 °F; the HRI and the query with the added terms yielded 136 and 618 ED visits, respectively. The HRI query and the HRI query with the added terms had a correlation coefficient of 0.714 (p <0.0001) and 0.427 (p <0.0001), respectively. During colder months, the CSTE HRI query and the query with the added terms yielded 377 and 86,008, respectively, with correlation coefficients of 0.342 (p < 0.0001) and 0.133 (p < 0.052). The syncope-only query saw no variation in HRI classified encounters throughout the heat season (mean: 328; min: 228; max: 404) or colder months (mean: 328; min: 261; max: 404) with correlation coefficients of 0.238 (p = 0.003) and 0.155 (p = 0.024), respectively. Similarly, the dehydration-only query saw no variation in HRI classified encounters throughout the heat season (mean: 96; min: 58; max: 258) or colder months (mean: 94; min: 60; max: 160) with correlation coefficients of 0.596 (p < 0.0001) and -0.016 (p = 0.822).
Conclusions The CSTE HRI query proved to be a strong indicator for HRI, and the addition of terms associated with dehydration and syncope to the CSTE HRI query weakened the correlation with temperature. Compared to the original CSTE HRI query, the added terms yielded a 4934% increase in HRI classified encounters during the heat season; however, these were likely due to causes other than HRI -- adding the extra terms resulted in a weaker correlation with temperature. Additionally, the comparative analysis showed that, with the added terms, the volume of HRI encounters was larger during colder months than hotter months suggesting misclassification of non-HRI illnesses. Surveillance of HRI has proven to be difficult because many of the HRI symptoms are too commonly associated with non-HRI conditions which would explain the weaker correlations when adding additional chief complaints associated with HRI. In conclusion, the CSTE syndrome definition for HRI proved to be the most robust query for HRI during the heat season. Case counts of HRI are difficult due to symptom overlap with many other medical conditions. However, syndromic surveillance using the CSTE HRI query is useful for trend analysis in near real-time during heat events.
Acknowledgement
References 1. Council of State and Territorial Epidemiologists. Heat-Related Illness Syndrome Query: A Guidance Document for Implementing Heat-Related Illness Syndromic Surveillance in Public Health Practice. Version 1.0. 2016 Sep. 12 p.
Brief bio for lead author/ presenter to be used by session moderators at the conference Jimmy Duong (MPH, Loma Linda University School of Public Health 2016) is an epidemiology analyst for the Acute Communicable Disease Control program for the Los Angeles County Department of Public Health.
Brief summary (100 words) of Presentation to be Used in Conference Program Evaluating heat-related illness (HRI) using the Council of State and Territorial Epidemiologists’ (CSTE) syndrome definition to capture heat-related ED visits in Los Angeles County in 2018. Additional terms for HRI, such as dehydration and syncope, were evaluated for increased capture of HRI ED visits. While the additional terms did increase the volume of ED visits matching the key words, they also resulted in decreased specificity, spurious detections during colder months and a weaker correlation with temperature. In conclusion, the original CSTE syndrome definition for HRI proved be more robust and reliable for determining increased HRI trends due to hot temperatures.



Heat-related ED visits, defined by CSTE’s HRI syndrome definition, per day in Los Angeles County during the heat season from 5/1/2018 to 9/30/2018




Dehydration-related ED visits per day in Los Angeles County during the heat season from 5/1/2018 to 9/30/2018




Syncope-related ED visits per day in Los Angeles county during the heat season from 5/1/2018 to 9/30/2018



Evaluation of ESSENCE Syndromic Definitions for ED Visits Related to Falls in Icy Weather

Jessica Hensley1, 3, 4, Sandra Gonzalez1, 2, Derry Stover1, Thomas Safranek1, Ming Qu1

1Nebraska Department of Health and Human Services, Lincoln, Nebraska, United States, 2University of Nebraska-Lincoln, Lincoln, Nebraska, United States, 3USPHS, Omaha, Nebraska, United States, 4American Military University, Charlestown, West Virginia, United States


Objective This project evaluated and compared two ESSENCE syndromic surveillance definitions for emergency department (ED) visits related to injuries associated with falls in icy weather using 2016-2017 data from two hospitals in Douglas County, Nebraska. The project determined the validity of the syndromic surveillance definition as applied to chief complaint and triage notes and compared the chief complaint data alone to chief complaint plus triage notes definitions to find the most reliable definition for ED visits resulting from fall-related injuries.
Introduction Icy weather events increase the risk for injury from falls on untreated or inadequately treated surfaces. These events often result in ED visits, which represents a significant public health and economic impact1.

The goal of this project was to start the process toward an evaluation of the public health impact and the economic impact of falls associated to icy weather in Douglas County, NE for the ultimate purpose of designing and implementing injury prevention related public health protection measures. Additionally, the validated definition will be used by NE DHHS Occupational Health Surveillance Program to identify work related ice-related fall injuries that were covered by workers compensation. To achieve the goal, the first step was to identify a valid and reliable syndromic surveillance. Specifically, this project looked at the applicability of the ESSENCE syndromic surveillance definitions related to injuries associated with falls. Two syndromic surveillance definitions were compared, one that includes triage note and chief complaint search terms, and another that only includes chief complaint. The hypothesis was that the ESSENCE syndromic surveillance definition that includes triage note and chief complaint search terms, rather than the syndromic surveillance definition that only includes chief complaint, would be more effective at identifying ED visits resulting from fall-related injuries.
Methods This project included 751 EDs visits from two hospitals located in Douglas County Nebraska, during ice events on December 16-18, 2016, January 10-12, 2017, and January 15-18, 2017.

Two ESSENCE syndromic surveillance definitions, “Chief Complaint or Triage Note” and “Chief Complaint Only,” were used to identify fall-related ED visits from two participating EDs in Douglas County, NE. In the chief complaint and the triage note fields, the keywords selected were: fall, fell, or slip. In that the ESSENCE time series analysis indicated the increase in the number of falls were associated with ice events from baseline, an assumption was made that the increase was a result of the weather. Then, the Syndromic Surveillance Event Detection of Nebraska database was used to find the patient and visit identification numbers. These two identification numbers were used to identify the EHRs needed for a gold standard review. Chart data was used to evaluate the reliability and validity of the two syndromic surveillance definitions for the detection of falls on the study dates. This analysis was used to find the sensitivity, specificity and predictive value.
Results The sensitivity, specificity and positive predictive value for the “Chief Complaint Only” definition yielded 71.7%, 100%, and 100% respectively. The “Chief Complaint or Triage Note” definition results were 90.9%, 98.8%, and 95.5% for these analyses. Negative predictive value for both definitions was 97.5%.
Conclusions The sensitivity indicates both definitions are unlikely to give false positives, and the positive predictive value indicates both definitions successfully identify most of the true positives found in the visits. However, the “Chief Complaint Only” definition resulted in a minimally higher specificity and positive predictive value. Therefore, the results indicate that although both definitions have similar specificity and positive predictive value, the “Chief Complaint or Triage Note” definition is more likely than the “Chief Complaint Only” definition to correctly identify ED visits related to falls in icy weather.
Acknowledgement Brian Buss, DVM, MPH, DACVPM, CDC Career Epidemiology Field Officer
References 1. Beynon C, Wyke S, Jarman I, Robinson M, Mason J, Murphy K, Bellis MA, Perkins C. The cost of emergency hospital admissions for falls on snow and ice in England during winter 2009/10: a cross sectional analysis. Environmental Health 2011;10(60).
Brief bio for lead author/ presenter to be used by session moderators at the conference CDR Jessica Hensley is an Active Duty U.S. Public Health Service Commissioned Corps officer and the Compliance Branch Director for the FDA's Office of Human and Animal Food West - Division 3. CDR Hensley received her Bachelor's of Science in Environmental Health from Eastern Kentucky University and her Master's of Public Health from American Military University. CDR completed this work with the other authors at the Nebraska Department of Health and Human Services as her thesis project for her MPH.
Brief summary (100 words) of Presentation to be Used in Conference Program This project evaluated and compared two ESSENCE syndromic surveillance definitions for emergency department (ED) visits related to injuries associated with falls in icy weather using 2016-2017 data from two hospitals in Douglas County, Nebraska. The project determined the validity of the syndromic surveillance definition as applied to chief complaint and triage notes and compared the chief complaint data alone to chief complaint plus triage notes definitions to find the most reliable definition for fall-related injuries. Results of this project indicate that although both definitions have similar specificity and positive predictive value, the “Chief Complaint or Triage Note” definition is more likely than the “Chief Complaint Only” definition to correctly identify ED visits related to falls in icy weather.


Number of falls detected by the ESSENCE “Chief Complaint Only” Falls Definition and the Gold Standard Chart Review


Detected by ESSENCE Fall related visits Non-fall related visits Total
Yes 38 0 38
No 15 579 594
Total 53 579 632


Number of falls detected by the ESSENCE “Chief Complaint or Triage Note” Falls Definition and the Gold Standard Chart Review


Detected by ESSENCE Fall related visits Non-fall related visits Total
Yes 150 7 157
No 15 579 594
Total 165 586 751


Using syndromic surveillance to monitor response to cyanotoxin contamination event

Kelly E. Cogswell

Acute and Communicable Disease Prevention, Oregon Health Authority, Portland, Oregon, United States


Objective Examine healthcare seeking behavior in a population exposed to low levels of cyanotoxins in the public drinking water supply
and quantify how publicity of the event may have affected perceptions of risk in the affected population.
Introduction Cyanotoxins are unregulated, emerging contaminants that have been associated with adverse health effects, including gastroenteritis, when consumed at high levels1,2. In May and June of 2018 cyanotoxins were detected in the public drinking water system for Salem, OR at levels above Environmental Protection Agency (EPA) health advisory levels for sensitive groups3. Sensitive groups were defined as children under 6, elderly adults, pregnant women, nursing mothers, people with compromised immune systems, people receiving dialysis, people with pre-existing liver conditions, and pets. Several health advisories were issued, and there was substantial media coverage of the event. The Oregon Health Authority (OHA) organized an Incident Management Team (IMT), which coordinated activities with other state and local agencies. Oregon ESSENCE staff used syndromic surveillance to monitor the population for health effects and healthcare seeking behavior.
Methods Oregon ESSENCE staff developed syndromic surveillance queries to monitor visits made to local emergency departments (i.e., visits by hospital location), as well as visits made by residents of the affected area (i.e., visits by patient location). Specifically, Oregon ESSENCE staff monitored total visits, gastroenteritis syndrome, visits by age group, and mentions of the word ‘water’ daily during the relevant time period. OHA communications staff tracked media coverage of the event. After the event, Oregon ESSENCE staff reconciled syndromic surveillance visit data with water test data, health advisory status, and media coverage to characterize how messaging may have affected healthcare seeking behavior.
Results Cyanotoxins were detected at levels above EPA guidelines for sensitive groups on 9 days between May 23, 2018 and June 19, 2019. OHA identified 67 news articles related to the event published in May and 179 published in June. Additionally, there was an unquantified amount of activity on social media, and a mass text alert that was sent out by the Oregon Office of Emergency Management. Visits for gastroenteritis were highest on the days immediately following the issuance of the first drinking water advisory. The first drinking water advisory was issued three days after the first results that contained cyanotoxins at levels exceeding the EPA guidelines for sensitive groups were received. Visits where the word ‘water’ was mentioned were similarly elevated immediately after the first drinking water advisory was issued. However, visits for gastroenteritis were also above expected levels on one day that had a water sample above EPA guidelines for sensitive groups, but before the first drinking water advisory was issued.
Conclusions Because cyanotoxins are unregulated, limited federal guidance was available and it took several days for the Oregon Health Authority to develop state guidance and educational materials. This delay contributed to public confusion about the level of risk associated with drinking the water, as well as confusion about which groups of people should avoid drinking the water. Our data suggest that emergency department visit behavior was largely driven by publicity of the event. Visits to the emergency department for gastroenteritis and mentions of the word ‘water’ decreased as more public information and guidance became available. However, we cannot rule out a real health effect related to cyanotoxins in the drinking water for area residents. One lesson learned from this type of high profile event relates to tracking of media coverage; it is difficult to measure how many people media coverage actually reaches, and attempting to characterize media coverage becomes more difficult after the event.
Acknowledgement
References U.S. EPA (United States Environmental Protection Agency). 2015. Drinking Water Health Advisory for the Cyanobacterial Toxin Cylindrospermopsin. EPA 820R15101, Washington, DC; June, 2015. Available from: http://water.epa.gov/drink/standards/hascience.cfm

U.S. EPA (United States Environmental Protection Agency). 2015. Drinking Water Health Advisory for the Cyanobacterial Toxin Microcystin. EPA 820R15100, Washington, DC; June, 2015. Available from: http://water.epa.gov/drink/standards/hascience.cfm

U.S. EPA (United States Environmental Protection Agency). 2015. 2015 Drinking Water Health Advisories for Two Cyanobacterial Toxins. EPA 820F15003, Washington, DC; June, 2015. Available from: https://www.epa.gov/sites/production/files/2017-06/documents/cyanotoxins-fact_sheet-2015.pdf
Brief bio for lead author/ presenter to be used by session moderators at the conference Kelly Cogswell is an epidemiologist with the Acute and Communicable Disease Prevention section at the Oregon Health Authority. She has an MPH with a concentration in epidemiology. Her current role covers a variety of public health surveillance topics in communicable disease and preparedness. Previously, she worked as an epidemiologist in environmental health, worked for a private forensic epidemiology firm, and served as an HIV outreach volunteer in the Peace Corps.
Brief summary (100 words) of Presentation to be Used in Conference Program Cyanotoxins were detected in a major Oregon public drinking water system. Several health advisories were issued, and there was substantial media coverage of the event. Syndromic surveillance was used to monitor total emergency department visits, visits for gastroenteritis, visits by age group, and mentions of the word “water”. After the event, Oregon ESSENCE staff reconciled syndromic surveillance visit data with water test data, health advisory status, and media coverage to characterize how messaging may have affected healthcare seeking behavior.


Detection of a Salmonellosis Outbreak using Syndromic Surveillance in Georgia

Rene Borroto, Jessica Pavlick, Karl Soetebier, Bill Williamson, Patrick Pitcher, Cherie Drenzek

Georgia Department of Public Health, Atlanta, Georgia, United States


Objective Describe how the Georgia Department of Public Health (DPH) used data from its State Electronic Notifiable Disease Surveillance System (SendSS) Syndromic Surveillance (SS) module for early detection of an outbreak of salmonellosis in Camden County, Georgia.
Introduction Evidence about the value of syndromic surveillance data for outbreak detection is limited (1). In July 2018, a salmonellosis outbreak occurred following a family reunion of 300 persons held in Camden County, Georgia, where one meal was served on 7/27/2018 and on 7/28/2018.
Methods SendSS-SS and SAS were used for cluster detection of Emergency Department (ED) patients with similar Chief Complaint (CC), Triage Notes (TN), or Discharge Diagnoses (DDx) by facility, time of ED visit, and zip code / county of residence. A SAS-based free-text query related to food poisoning in the CC and DDx fields was also performed on a daily basis. County- and hospital-specific charting of the Diarrhea syndrome was also conducted in SendSS-SS, whereas county- and zip code-specific charting of the same syndrome were done in both SendSS-SS and SAS (2).
Results On Sunday July 29th, 2018, three children and three adults were seen within 18 hours at the ED of Hospital A in Camden County, Georgia. All patients complained of diarrhea, vomiting, and food poisoning, after a large family reunion that had been held the day before. This early cluster was detected by the SAS-based free-text query of ‘food poisoning’ and the SAS-based cluster detection tool for patients with Diarrhea syndrome. The District Epidemiologists (DE) in the Coastal Health District were notified on Monday, July 30th, 2018. One-year high daily spikes of the Diarrhea syndrome occurred from July 29th to July 31st, 2018 in a local hospital ED (Fig 1), Camden County, and zip code 31548. Two HIPAA-compliant line lists with a total of 27 patients seen at EDs were emailed to the DEs to support active case finding. No further spikes of the Diarrhea syndrome were detected in Camden County during the 2-week period after the 3-day spike.
Conclusions Syndromic surveillance was a useful surveillance tool for early detection of a salmonellosis outbreak, helping with the active search for outbreak cases, tracking the peak of the outbreak, and assuring that no further spikes were occurring.
Acknowledgement We are grateful to the Epidemiologists of the Coastal District, Robert Thornton, Meredith Avery, and Elizabeth Goff, for their hard work during this outbreak.
References 1.R Hopkins, C Tong, H Burkom, et al. A Practitioner-Driven Research Agenda for Syndromic Surveillance. Public Health Reports 2017; 132(Supplement1): 116S-126S.
2. G Zhang, A Llau, J Suarez, E O'Connell, E Rico, R Borroto, F Leguen. Using ESSENCE to Track a Gastrointestinal Outbreak in a Homeless Shelter in Miami-Dade County, 2008. Advances in Disease Surveillance. 2008; 5:139.
Brief bio for lead author/ presenter to be used by session moderators at the conference Rene Borroto is the Syndromic Surveillance Coordinator at the Georgia Department of Public Health.
Brief summary (100 words) of Presentation to be Used in Conference Program This presentation describes how syndromic surveillance was a useful tool for early detection of a salmonellosis outbreak in Camden County, Georgia, helping with the active search for outbreak cases, tracking the peak of the outbreak, and assuring that no further spikes were occurring.




Precision public health through clinic-based syndromic surveillance in communities

Ta-Chien Chan, Yung-Chu Teng, Yen-Hua Chu, Tzu-Yu Lin

Academia Sinica, Taipei City, Taiwan


Objective Sentinel physician surveillance in the communities has played an important role in detecting early aberrations in epidemics. The traditional approach is to ask primary care physicians to actively report some diseases such as influenza-like illness (ILI), and hand, foot, and mouth disease (HFMD) to health authorities on a weekly basis. However, this is labor-intensive and time-consuming work. In this study, we try to set up an automatic sentinel surveillance system to detect 23 syndromic groups in the communites.
Introduction In December 2009, Taiwan’s CDC stopped its sentinel physician surveillance system. Currently, infectious disease surveillance systems in Taiwan rely on not only the national notifiable disease surveillance system but also real-time outbreak and disease surveillance (RODS) from emergency rooms, and the outpatient and hospitalization surveillance system from National Health Insurance data. However, the timeliness of data exchange and the number of monitored syndromic groups are limited. The spatial resolution of monitoring units is also too coarse, at the city level. Those systems can capture the epidemic situation at the nationwide level, but have difficulty reflecting the real epidemic situation in communities in a timely manner. Based on past epidemic experience, daily and small area surveillance can detect early aberrations. In addition, emerging infectious diseases do not have typical symptoms at the early stage of an epidemic. Traditional disease-based reporting systems cannot capture this kind of signal. Therefore, we have set up a clinic-based surveillance system to monitor 23 kinds of syndromic groups. Through longitudinal surveillance and sensitive statistical models, the system can automatically remind medical practitioners of the epidemic situation of different syndromic groups, and will help them remain vigilant to susceptible patients. Local health departments can take action based on aberrations to prevent an epidemic from getting worse and to reduce the severity of the infected cases.
Methods We collected data on 23 syndromic groups from participating clinics in Taipei City (in northern Taiwan) and Kaohsiung City (in southern Taiwan). The definitions of 21 of those syndromic groups with ICD-10 diagnoses were adopted from the International Society for Disease Surveillance (https://www.surveillancerepository.org/icd-10-cm-master-mapping-reference-table). The definitions of the other two syndromic groups, including dengue-like illness and enterovirus-like illness, were suggested by infectious disease and emergency medicine specialists.

An enhanced sentinel surveillance system named “Sentinel plus” was designed for sentinel clinics and community hospitals. The system was designed with an interactive interface and statistical models for aberration detection. The data will be computed for different combinations of syndromic groups, age groups and gender groups. Every day, each participating clinic will automatically upload the data to the provider of the health information system (HIS) and then the data will be transferred to the research team.

This study was approved by the committee of the Institutional Review Board (IRB) at Academia Sinica (AS-IRB02-106262, and AS-IRB02-107139). The databases we used were all stripped of identifying information and thus informed consent of participants was not required.
Results This system started to recruit the clinics in May 2018. As of August 2018, there are 89 clinics in Kaohsiung City and 33 clinics and seven community hospitals in Taipei City participating in Sentinel plus. The recruiting process is still ongoing. On average, the monitored volumes of outpatient visits in Kaohsiung City and Taipei City are 5,000 and 14,000 per day.

Each clinic is provided one list informing them of the relative importance of syndromic groups, the age distribution of each syndromic group and a time-series chart of outpatient rates at their own clinic. In addition, they can also view the village-level risk map, with different alert colors. In this way, medical practitioners can know what’s going on, not only in their own clinics and communities but also in the surrounding communities.

The Department of Health (Figure 1) can know the current increasing and decreasing trends of 23 syndromic groups by red and blue color, respectively. The spatial resolution has four levels including city, township, village and clinic. The map and bar chart represent the difference in outpatient rate between yesterday and the average for the past week. The line chart represents the daily outpatient rates for one selected syndromic group in the past seven days. The age distribution of each syndromic group and age-specific outpatient rates in different syndromic groups can be examined.
Conclusions Sentinel plus is still at the early stage of development. The timeliness and the accuracy of the system will be evaluated by comparing with some syndromic groups in emergency rooms and the national notifiable disease surveillance system. The system is designed to assist with surveillance of not only infectious diseases but also some chronic diseases such as asthma. Integrating with external environmental data, Sentinel plus can alert public health workers to implement better intervention for the right population.
Acknowledgement This research was supported by a grant titled “Multidisciplinary Health Cloud Research Program: Technology Development and Application of Big Health Data” from Academia Sinica and a grant titled “Implementing Integrated Surveillance Network of Dengue Fever” from the National Health Research Institute, Taiwan and a grant from the Department of Health, Taipei City government. We would like to thank the Departments of Health in Kaohsiung and Taipei, as well as Taipei Medical Association for help in recruiting the participating clinics and hospitals.
References 1. James W. Buehler AS, Marc Paladini, Paula Soper, Farzad Mostashari: Syndromic Surveillance Practice in the United States: Findings from a Survey of State, Territorial, and Selected Local Health Departments. Advances in Disease Surveillance 2008, 6(3).
2. Ding Y, Fei Y, Xu B, Yang J, Yan W, Diwan VK, Sauerborn R, Dong H: Measuring costs of data collection at village clinics by village doctors for a syndromic surveillance system — a cross sectional survey from China. BMC Health Services Research 2015, 15:287.
3. Kao JH, Chen CD, Tiger Li ZR, Chan TC, Tung TH, Chu YH, Cheng HY, Liu JW, Shih FY, Shu PY et al.: The Critical Role of Early Dengue Surveillance and Limitations of Clinical Reporting -- Implications for Non-Endemic Countries. PloS one 2016, 11(8):e0160230.
4. Chan TC, Hu TH, Hwang JS: Daily forecast of dengue fever incidents for urban villages in a city. International Journal of Health Geographics 2015, 14:9.
5. Chan TC, Teng YC, Hwang JS: Detection of influenza-like illness aberrations by directly monitoring Pearson residuals of fitted negative binomial regression models. BMC Public Health 2015, 15:168.
6. Ma HT: Syndromic surveillance system for detecting enterovirus outbreaks evaluation and applications in public health. Taipei, Taiwan: National Taiwan University; 2007.
Brief bio for lead author/ presenter to be used by session moderators at the conference Dr. Ta-Chien Chan is an interdisciplinary scholar in both health and spatial science. He graduated with a doctoral degree (Ph.D) from the Institute of Epidemiology, National Taiwan University in 2010. After that, he passed the national civil service exam and worked in the Department of Health during 2010-2012. In August 2012, he joined Academia Sinica as an assistant research fellow. In May 2016, he was promoted to associate research fellow. The main focuses of his recent research work have been on understanding the effects of the environment on infectious diseases in East Asia, applying health informatics to improve disease surveillance systems, and visualization of health data. He has devoted his career not only to research but also to improving first-line public health systems.
Brief summary (100 words) of Presentation to be Used in Conference Program The paper set up an automated syndromic surveillance system to collect data from clinics and community hospitals. The system overcomes the limitations in the traditional sentinel surveillance system. Primary care physicians can use this information to aid their clinical diagnosis and know the situations in their communities. Public health workers can view different spatial resolutions of signals and do intervention earlier. Communicable and non-communicable diseases can be simultaneously monitored, giving the whole picture of disease surveillance in communities.



Figure 1. The Department of Health dashboard in Sentinel plus



Customizing ESSENCE Queries for Select Mental Health Sub-indicators

Achintya N. Dey, Michael Coletta, Hong Zhou, Nelson Adekoya, Deborah Gould

CCSELS/DHIS, CDC, Atlanta, Georgia, United States


Objective Emergency department (ED) visits related to mental health (MH) disorders have increased since 2006 (1), indicating a potential burden on the healthcare delivery system. Surveillance systems has been developed to identify and understand these changing trends in how EDs are used and to characterize populations seeking care. Many state and local health departments are using syndromic surveillance to monitor MH-related ED visits in near real-time. This presentation describes how queries can be created and customized to identify select MH sub-indicators (for adults) by using chief complaint text terms and diagnoses codes. The MH sub-indicators examined are mood and depressive disorders, schizophrenic disorders, and anxiety disorders. Wider adoption of syndromic surveillance for characterizing MH disorders can support long-term planning for healthcare resources and service delivery.
Introduction Syndromic surveillance systems, although initially developed in response to bioterrorist threats, are increasingly being used at the local, state, and national level to support early identification of infectious disease and other emerging threats to public health. To facilitate detection, one of the goals of CDC’s National Syndromic Surveillance Program (NSSP) is to develop and share new sets of syndrome codes with the syndromic surveillance Community of Practice. Before analysts, epidemiologists, and other practitioners begin customizing queries to meet local needs, especially monitoring ED visits in near-real time during public health emergencies, they need to understand how syndromes are developed.
More than 4,000 hospital routinely send data to NSSP’s BioSense Platform, representing about 55 percent of ED visits in the United States (2). The platform’s surveillance component, ESSENCE,* is a web-based application for analyzing and visualizing prediagnostic hospital ED data. ESSENCE’s Chief Complaint Query Validation (CCQV) data source, which is a national-level data source with access to chief complaint (CC) and discharge diagnoses (DD) from reporting sites, was designed for testing new queries.
Methods We used ESSENCE CCQV to query weekly data for the nine week period from the first quarter of 2018 and looked at three common MH sub-indicators: mood and depressive disorders, schizophrenic disorders, and anxiety disorders. We developed four query types for each MH sub-indicator. Query-1 focused on DD codes; query-2 focused on CC text terms; query-3 focused on a combination of CC, DD, and no exclusion for mental health co-morbidity; and query-4 focused on a combination of CC and DD and excluded mental health co-morbidity. We also examined the summary distribution of CC texts to identify keywords related to MH sub-indicators.
For mood and depressive disorders, we queried ICD-9 codes 296, 311; ICD-10 codes F30–F39; CC text terms for words “depressive disorder,” bipolar disorder,” “mood disorder,” “depression,” “manic episodes,” and “psychotic.” For schizophrenic disorders, we queried ICD-9 codes 295; ICD-10 codes F20–F29; CC text terms for words “psychosis,” “psychotic,” “schizo,” “delusional,” “paranoid,” “auditory,” “hallucinations,” and “hearing voices.” For anxiety disorders, we queried ICD-9 codes 300, 306, 307, 308, 309; ICD-10 codes F40–F48; CC text terms for words “anxiety,” “anexiy,” “aniety,” “aniexty,” “ansiety,” “anxety,” “anxity,” “anxiety,” “phobia,” and “panic attack.”
Results We identified 2.3 million average weekly ED visits for the 9-week period queried. Table 1 shows average weekly ED visits of select MH sub-indicators from the four query types. Because query 4 focused on specific MH outcomes and excluded MH co-morbidities, the average weekly ED visit for all three sub-indicators was almost half that of query 3, which focused on broader concepts by including MH co-morbidities. Among mood and depressive disorders, query 4 identified on average 23,352 ED visits per week versus 45,504 visits per week for query 3. Similarly, for schizophrenic disorders and anxiety disorders, query 4 identified on average 4,988 and 32,790 visits per week compared with 9,816 and 53,868 visits, respectively, for query 3. Further, more MH-related visits were identified using the DD-coded query (query 1) than CC-based text terms (query 2).
Conclusions
Analysts can benefit from having queries on select sub-indicators readily available and can use these to facilitate routine MH-related monitoring of ED visits, or customize the queries by including local text terms. Consistent with our previous work (3), this analysis demonstrated that MH-related ED visits are more likely to be found in DD codes than in CC alone.
* Electronic Surveillance for the Early Notification of Community-based Epidemics
Acknowledgement
References [1] Weiss AJ, Barrett ML, Heslin KC , Stocks C. Trends in Emergency Department Visits Involving Mental and Substance Use Disorders, 2006–2013. HCUP Statistical Brief #216 [Internet]. Rockville (MD): Agency for Healthcare Research and Quality; 2016 Dec [cited 2018 Aug 14]. Available from: http://www.hcup-us.ahrq.gov/reports/statbriefs/sb216-Mental-Substance-Use-Disorder-ED-Visit-Trends.pdf.
[2] Gould DW, Walker D, Yoon PW. The Evolution of BioSense: Lessons Learned and Future Directions. Public Health Reports. 2017 Jul/Aug;132(Suppl 1):S7–S11.
[3] Dey AN, Gould D, Adekoya N, Hicks P, Ejigu GS, English R, Couse J, Zhou H. Use of Diagnosis Code in Mental Health Syndrome Definition. Online Journal of Public Health Informatics [Internet]. 2018 [cited 2018 Aug 14];10(1). Available from: https://doi.org/10.5210/ojphi.v10i1.8983
Brief bio for lead author/ presenter to be used by session moderators at the conference Achintya N. Dey is an epidemiologist in the Division of Health Informatics and Surveillance at the Center for Surveillance, Epidemiology and Laboratory Services, Centers for Disease Control and Prevention. For the last twenty six years, he has been working in the field of health care research. His current research focused on creating and evaluating syndromes for syndromic surveillance.
Brief summary (100 words) of Presentation to be Used in Conference Program This presentation describes how queries can be created and customized to identify select MH sub-indicators (for adults) by using chief complaint text terms and diagnoses codes. The MH sub-indicators examined are mood and depressive disorders, schizophrenic disorders, and anxiety disorders. Wider adoption of syndromic surveillance for characterizing MH disorders can support long-term planning for healthcare resources and service delivery.


Table 1. Average weekly emergency department visits for 9-week period during first quarter of 2018 for select mental health sub-indicators


Mental Health Sub-indicators Emergency Department Visits Mean (SD)
  Query 1 Query 2 Query 3 Query 4
Mood and depressive disorders 42,533 (1,982) 6,867 (387) 45,504 (2,098) 23,762 (521)
Schizophrenic disorders 7,436 (305) 2,303 (88) 9,816 (353) 4,988 (164)
Anxiety disorders 48,050 (2,445) 11,710 (650) 53,868 (2,745) 32,790 (1,925)


Enhancing Syndromic Surveillance with Procedure Data: A 2017-8 Influenza Case Study

Andrew Walsh

Health Monitoring, Pittsburgh, Pennsylvania, United States


Objective To identify additional data elements in existing syndromic surveillance message feeds that can provide additional insight into public health concerns such as the influenza season.
Introduction Syndromic surveillance achieves timeliness by collecting prediagnostic data, such as emergency department chief complaints, from the start of healthcare interactions. The tradeoff is less precision than from diagnosis data, which takes longer to generate. As the use and sophistication of electronic health information systems increases, additional data that provide an intermediate balance of timeliness and precision are becoming available.
Information about the procedures and treatments ordered for a patient can indicate what diagnoses are being considered. Procedure records can also be used to track the use of preventive measures such as vaccines that are also relevant to public health surveillance but not readily captured by typical syndromic data elements. Some procedures such as laboratory tests also provide results which can provide additional specificity about which diagnoses will be considered. If procedure and treatment orders and test results are included in existing syndromic surveillance feeds, additional specificity can be achieved with timeliness comparable to prediagnostic assessments.
Methods HL7 messages were collected for syndromic surveillance using EpiCenter software. They were retroactively scanned for PR1 procedure segments; procedure codes and descriptions were extracted when available. Influenza-related procedures were identified and classified as either a test for the virus or an administration of a vaccine. Classification was based on the procedure code when a standard code set was used and could be identified, otherwise it was based on the text description of the procedure.
Messages were also scanned for the presence of ‘influenza’ in text fields. Influenza test results were identified first by selecting messages with ‘influenza’ in an OBX segment and then further refining based on the test code and description.
Results A total of 443,074,748 messages from 2,577 healthcare facilities received between July 1, 2017 and August 31, 2018 were scanned for procedure information. Procedure codes were present in 39,142,670 messages from 287 facilities. The most common procedures included blood glucose measurements and other diabetes maintenance activities, incentive spirometry, blood count and metabolic panels, safety observation, and vital signs.
Of those, 995,754 messages from 142 facilities contained influenza-related procedure codes for 106,610 visits. 14,672 visits from 62 facilities had one of 48 vaccine procedure codes, and 91, 948 visits from 127 facilities had one of 66 test codes. Time series of both types of procedures showed a seasonal trend consistent with the influenza season. Figure 1 shows the daily counts of influenza test orders and vaccine administrations. Figure 2 breaks out the test orders by test type (antibody assay, antigen assay, PCR, or unspecified).
Seven facilities sent a total of 58,182 messages containing influenza test results. These included both positive and negative results. These results distinguished between influenza A and influenza B. Figure 3 shows the daily counts of both positive and negative results by virus type; this also follows the expected seasonal pattern.
Conclusions Since procedure information was not specifically requested from healthcare facilities, the overall representation of procedure data elements was low. These initial results indicate that such data would be useful both as a supplement to syndromic surveillance activities and as a new data source for other surveillance activities such as vaccine uptake tracking. Given the frequency of procedures and treatments for chronic diseases such as diabetes and heart disease, these data may be relevant for understanding the prevalence of those conditions as well. Tests and treatments relevant to other public health concerns like opioid use disorder were also present, suggesting a wide range of potential applications.
It is also possible to obtain and extract influenza test results from these syndromic surveillance messages. Both positive and negative results were present, providing information not just on the number of positive cases but also the rate of testing and rate of positive results. The pattern of testing and results also indicates that at least some facilities test for influenza throughout the season, contrary to some conventional wisdom about testing patterns.
Acknowledgement Health Monitoring would like to thank our customers for financial support of this work.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Andy Walsh is the Chief Science Officer at Health Monitoring where he oversees research for EpiCenter, a syndromic surveillance system for a large portion of the United States. Previously, he developed software for visualizing and analyzing viral genome data, as part of a research program for HIV and influenza vaccines. He has a PhD in molecular microbiology from the Bloomberg School of Public Health.
Brief summary (100 words) of Presentation to be Used in Conference Program Information about procedures ordered for a patient and their results can indicate what diagnoses are being considered. Including these data elements in existing syndromic surveillance feeds could provide additional specificity with timeliness comparable to prediagnostic assessments.
Scanning 443,074,748 messages from 2,577 healthcare facilities revealed procedure orders and test results were already present in many messages. Data from 142 facilities contained influenza-related procedure codes including test orders and vaccine administration. Messages from seven facilities included positive and negative influenza test results. The pattern of testing and results follow a typical seasonal pattern and indicate facilities test for influenza throughout the season.



Figure 1: Time Series of Influenza-Related Procedure Orders by Procedure Type




Figure 2: Time Series of Influenza-Related Test Orders by Test Type




Figure 3: Time Series of Influenza Test Results by Result and Influenza Type



Using Syndromic Surveillance to Quantify ED Visits for Coagulopathy Cases

Kelly Walblay, Megan Patel, Stacey Hoferka

Communicable Diseases, Illinois Department of Public Health, Chicago, Illinois, United States


Objective To determine whether emergency department (ED) visits were captured in syndromic surveillance for coagulopathy cases associated with an outbreak linked to synthetic cannabinoid (SC) use and to quantify the number of ED visits and reasons for repeat visits.
Introduction In March 2018, the Illinois Department of Public Health (IDPH) was informed of a cluster of coagulopathy cases linked to SC use. By June 30, 2018, 172 cases were reported, including five deaths, where 74% were male and the mean age was 35.3 years (range: 18–65 years). All cases presented to an emergency department (ED) at least once for this illness. Ninety-four cases provided clinical specimens and all tested positive for brodifacoum, a long-acting anticoagulant used in rodenticide. Cases were reported to the health department by the Illinois Poison Control Center and direct reporting from hospitals. REDCap was the primary database for tracking cases and collecting demographic information, risk factor data and healthcare facility utilization, including number of ED visits. Syndromic surveillance was utilized to monitor ED visits related to the cluster, assist with case finding and provide situational awareness of the burden on the EDs and geographic spread. In this study, we retrospectively used syndromic surveillance along with the data in REDCap to quantify the number of ED visits per coagulopathy case, understand the reasons for repeat visits, and determine whether visits were captured in syndromic surveillance.
Methods Illinois hospital ED data submitted to the National Syndromic Surveillance Platform instance of ESSENCE (ESSENCE), was compared to data present in our primary REDCap database. A subset of the cases, males 18-44 years of age (n=105; 61% of cases), were included in this analysis. Illinois ESSENCE data in males aged 18-44 years from March 10, 2018–June 30, 2018 were matched to cases in the REDCap database by age, zip code, initial visit date, facility, and reason for visit including: chief complaint, discharge diagnosis, and triage note. If the initial visit was found, the matching criteria and medical record number were used to search for additional related visits. The number of visits in ESSENCE and reasons for visits were totaled for each patient. Reasons for repeat visits were categorized into four categories: continued gross bleeding or symptoms associated with coagulopathy, medical evaluation or follow-up, laboratory work and prescription refill. Repeat visits may fall into more than one category. The number and dates of ED visits captured in ESSENCE per case were compared to that reported in REDCap. An epidemic curve was constructed to display the number of ED visits and type (i.e. primary visit or repeat visit) captured by REDCap only, ESSENCE only or both by visit date.
Results Of the 105 cases in REDCap, 89 (85%) were matched to at least one ED visit in ESSENCE from March 10, 2018–June 30, 2018. The mean number of ESSENCE ED visits per case was 1.9 visits and the median was one visit (range: 1–11 visits). The main chief complaints for the primary visit included hematuria (n=31), abdominal pain (n=20), back pain/flank pain (n=13), K2 (n=11), bleeding from multiple sites (n=8), vomiting blood (n=7), and urinary tract infection or kidney stones (n=7). Of the 89 cases matched to a visit in ESSENCE, 84 (94%) cases, representing 142 (79%) of ED visits, were captured by syndrome definitions that were being utilized to monitor the cluster. Forty-three cases (48%) had at least two visits in ESSENCE. The reasons for return visits captured in ESSENCE (n=84) were continued gross bleeding or symptoms associated with coagulopathy (n=53), medical evaluation or follow-up (n=14), laboratory work (n=13), prescription refill (n=7) or unknown (n=2). Of the 105 cases, the number of ED visits reported in REDCap matched the number of visits found in ESSENCE for 49 cases (47%). For 24 cases (23%), ESSENCE identified more visits than REDCap and for 16 cases (15%), REDCap had more ED visits reported than captured in ESSENCE. Sixteen cases (15%) in REDCap were not found in ESSENCE. All of the unmatched visits were due to ESSENCE data quality, including a lack of reporting hospital admissions, lack of submitting data to ESSENCE, and missing data including: date of birth, medical record number, and triage notes.
Conclusions Syndromic surveillance was a useful tool in describing the burden of ED visits for patients in the Illinois coagulopathy outbreak linked to SC use. ESSENCE data helped to quantify the number of ED visits per patient and identify patients that re-presented for the same illness. The most common reason for repeat ED visits was continued symptoms, which may be attributed to misdiagnosis at the initial healthcare visit. ED visits that were not picked up by ESSENCE were a result of data quality issues from select facilities that were not reporting hospitalizations or key information such as date of birth, medical record number or triage notes. Engagement with healthcare facilities to provide this information will improve the data quality of syndromic surveillance.
Acknowledgement This study was supported in part by an appointment to the Applied Epidemiology Fellowship Program administered by the Council of State and Territorial Epidemiologists (CSTE) and funded by the Centers for Disease Control and Prevention (CDC) Cooperative Agreement Number 1U38OT000143-05.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Kelly Walblay is the current CDC/CSTE Applied Epidemiology Fellow at the Illinois Department of Public Health in Chicago, Illinois in the Division of Communicable Diseases. Kelly holds a Master’s degree in Epidemiology from the University of Michigan and a Bachelor’s degree in Biology and Spanish from Hope College. Her responsibilities include conducting surveillance of Shiga toxin-producing E. coli, Typhoid Fever and Vibriosis cases in Illinois and she was actively involved in the response to the coagulopathy outbreak linked to synthetic cannabinoid use in the spring of 2018.
Brief summary (100 words) of Presentation to be Used in Conference Program In March 2018, the Illinois Department of Public Health was informed of a cluster of cases with severe bleeding linked to synthetic cannabinoid use. By the end of June, over 170 cases were identified in this outbreak. Cases were tracked in REDCap and syndromic surveillance was utilized to monitor emergency department (ED) visits and provide situational awareness. This presentation will provide context on this unique outbreak and discuss how syndromic surveillance was used to quantify the number of ED visits for cases, understand the reasons for repeat visits and determine whether ED visits were captured in syndromic surveillance.




Monitoring suicide-related events using National Syndromic Surveillance Program data

Marissa L. Zwald, Kristin M. Holland, Francis Annor, Aaron Kite-Powell, Steven A. Sumner, Daniel Bowen, ALANA M. VIVOLO-KANTOR, Deborah M. Stone, Alex E. Crosby

CDC, Atlanta, Georgia, United States


Objective To describe epidemiological characteristics of emergency department (ED) visits related to suicidal ideation (SI) or suicidal attempt (SA) using syndromic surveillance data.
Introduction Suicide is a growing public health problem in the United States.1 From 2001 to 2016, ED visit rates for nonfatal self-harm, a common risk factor for suicide, increased 42%.2–4 To improve public health surveillance of suicide-related problems, including SI and SA, the Data and Surveillance Task Force within the National Action Alliance for Suicide Prevention recommended the use of real-time data from hospital ED visits.5 The collection and use of real-time ED visit data on SI and SA could support a more targeted and timely public health response to prevent suicide.5 Therefore, this investigation aimed to monitor ED visits for SI or SA and to identify temporal, demographic, and geographic patterns using data from CDC’s National Syndromic Surveillance Program (NSSP).
Methods CDC’s NSSP data were used to monitor ED visits related to SI or SA among individuals aged 10 years and older from January 1, 2016 through July 31, 2018. A syndrome definition for SI or SA, developed by the International Society for Disease Surveillance’s syndrome definition committee in collaboration with CDC, was used to assess SI or SA-related ED visits. The syndrome definition was based on querying the chief complaint history, discharge diagnosis, and admission reason code and description fields for a combination of symptoms and Boolean operators (for example, hang, laceration, or overdose), as well as ICD-9-CM, ICD-10-CM, and SNOMED diagnostic codes associated with SI or SA. The definition was also developed to include common misspellings of self-harm-related terms and to exclude ED visits in which a patient “denied SI or SA.”

The percentage of ED visits involving SI or SA were analyzed by month and stratified by sex, age group, and U.S. region. This was calculated by dividing the number of SI or SA-related ED visits by the total number of ED visits in each month. The average monthly percentage change of SI or SA overall and for each U.S. region was also calculated using the Joinpoint regression software (Surveillance Research Program, National Cancer Institute).6
Results Among approximately 259 million ED visits assessed in NSSP from January 2016 to July 2018, a total of 2,301,215 SI or SA-related visits were identified. Over this period, males accounted for 51.2% of ED visits related to SI or SA, and approximately 42.1% of SI or SA-related visits were comprised of patients who were 20-39 years, followed by 40-59 years (29.7%), 10-19 years (20.5%), and ≥60 years (7.7%).

During this period, the average monthly percentage of ED visits involving SI or SA significantly increased 1.1%. As shown in Figure 1, all U.S. regions, except for the Southwest region, experienced significant increases in SI or SA ED visits from January 2016 to July 2018. The average monthly increase of SI or SA-related ED visits was 1.9% for the Midwest, 1.5% for the West (1.5%), 1.1% for the Northeast, 0.9% for the Southeast, and 0.5% for the Southwest.
Conclusions ED visits for SI or SA increased from January 2016 to June 2018 and varied by U.S. region. In contrast to previous findings reporting data from the National Electronic Injury Surveillance Program – All-Injury Program, we observed different trends in SI or SA by sex, where more ED visits were comprised of patients who were male in our investigation.2 Syndromic surveillance data can fill an existing gap in the national surveillance of suicide-related problems by providing close to real-time information on SI or SA-related ED visits.5 However, our investigation is subject to some limitations. NSSP data is not nationally representative and therefore, these findings are not generalizable to areas not participating in NSSP. The syndrome definition may under-or over-estimate SI or SA based on coding differences and differences in chief complaint or discharge diagnosis data between jurisdictions. Finally, hospital participation in NSSP can vary across months, which could potentially contribute to trends observed in NSSP data. Despite these limitations, states and communities could use this type of surveillance data to detect abnormal patterns at more detailed geographic levels and facilitate rapid response efforts. States and communities can also use resources such as CDC’s Preventing Suicide: A Technical Package of Policy, Programs, and Practices to guide prevention decision-making and implement comprehensive suicide prevention approaches based on the best available evidence.7
Acknowledgement Rasneet Kumar, Zachary Stein, and members of the ISDS Syndrome Definition Committee
References 1. Stone DM, Simon TR, Fowler KA, et al. Vital Signs: Trends in State Suicide Rates — United States, 1999–2016 and Circumstances Contributing to Suicide — 27 States, 2015. Morb Mortal Wkly Rep. 2018;67(22):617-624.
2. CDCs National Center for Injury Prevention and Control. Web-based Injury Statistics Query and Reporting System (WISQARS). https://www.cdc.gov/injury/wisqars/index.html. Published 2018. Accessed September 1, 2018.
3. Mercado M, Holland K, Leemis R, Stone D, Wang J. Trends in emergency department visits for nonfatal self-inflicted injuries among youth aged 10 to 24 years in the United States, 2005-2015. J Am Med Assoc. 2017;318(19):1931-1933. doi:10.1001/jama.2017.13317
4. Olfson M, Blanco C, Wall M, et al. National Trends in Suicide Attempts Among Adults in the United States. JAMA Psychiatry. 2017;10032(11):1095-1103. doi:10.1001/jamapsychiatry.2017.2582
5. Ikeda R, Hedegaard H, Bossarte R, et al. Improving national data systems for surveillance of suicide-related events. Am J Prev Med. 2014;47(3 SUPPL. 2):S122-S129. doi:10.1016/j.amepre.2014.05.026
6. National Cancer Institute. Joinpoint Regression Software. https://surveillance.cancer.gov/joinpoint/. Published 2018. Accessed September 1, 2018.
7. Centers for Disease Control and Prevention. Preventing Suicide: A Technical Package of Policy, Programs, and Practices.
Brief bio for lead author/ presenter to be used by session moderators at the conference Dr. Marissa Zwald is an epidemiologist with CDC's Division of Violence Prevention.
Brief summary (100 words) of Presentation to be Used in Conference Program Suicide is a growing public health problem in the United States. Using data from CDC’s National Syndromic Surveillance Program (NSSP), ED visits related to suicidal ideation (SI) and suicidal attempt (SA) among individuals aged 10 years and older from January 1, 2016 through July 31, 2018 were examined by sex, age group, and U.S. region. From January 2016 through July 2018, ED visits for SI/SA assessed in NSSP increased on average 1.1% per month. Monitoring SI/SA using ED syndromic data might assist states and communities in the development of timely interventions to address suicide-related problems.


Using Evaluation to Inform the BioSense Platform: Results from a 2018 Survey

Cassandra N. Davis

Division of Health Informatics and Surveillance, Centers for Control and Disease Prevention, Atlanta, Georgia, United States


Objective To assess the present status of utility, functionality, usability and user satisfaction of the BioSense Platform.
Introduction Since 2015, CDC’s Division of Health Informatics and Surveillance staff have conducted evaluations to provide information on the utility, functionality, usability and user satisfaction associated with the National Syndromic Surveillance Program’s BioSense Platform tools. The BioSense Platform tools include: 1) Access and Management Center (AMC), a tool that enables site administrators to manage users and data permissions; 2) Electronic Surveillance System for Early Notification of Community-based Epidemics (ESSENCE), a software application that enables syndromic surveillance related data visualization and analysis; 3) Adminer, a tool that allows users to access site data on the datamart; and 4) Rstudio, an application that can be used for data analysis and visualization. The evaluation findings have informed activities that led to improvements in functionality, development or procurement of platform associated tools, and development of resource materials. In May 2018, NSSP conducted an evaluation with eight jurisdictions that participated in the first user acceptance testing (UAT) evaluations in 2015. The purpose of the evaluation was to assess the present status of utility, functionality, usability and user satisfaction of the tools on the BioSense Platform, and delineate progress since 2015.
Methods CDC’s evaluation framework and utilization-focused evaluation were used to inform and engage stakeholders, develop the evaluation questions, metrics, and methodology. Eight selected jurisdictions participated in an online, Epi-Info survey that captured quantitative and qualitative information. Prior to the survey, participants received a presentation about the evolution of the BioSense Platform since 2015, and were provided an overview of components to evaluate. The participants were asked to assess the following key areas based on use of the BioSense Platform within the past 30 days: 1) the utility, functionality and usability of the AMC, ESSENCE, Adminer and Rstudio; 2) how well the enhanced data flow has enabled them to conduct syndromic surveillance activities; 3) usefulness of the quick start guides. Additionally, participants were asked to provide suggestions for other improvements to the BioSense Platform, and to indicate their overall satisfaction. Descriptive statistics were generated and thematic analysis was conducted to identify themes from qualitative responses.
Results Overall, participant’s responses remained positive about the utility, functionality, usability and overall satisfaction of the BioSense Platform. Participants indicated using the BioSense Platform regularly (e.g. daily, weekly and/or monthly) within those 30 days. Certain functions have been used more than others across the various tools to conduct syndromic surveillance, with at least 50% of participants reporting use. These included creating data access rules, viewing and verifying raw and processed data, running time series, conducting free-text queries, and assessing data details and total ER visit counts by hospital, county/region, or state. The challenges ranged from tool performance to user interpretation of the function. Participants reported that the enhanced data flow improved their data quality and helped identify issues. Although participants scored ESSENCE to have average usability per the system usability scale (SUS score=63.5 in 2018), the BioSense Platform and its tools were reported as useful by 88% of participants. Further, participants continue to be comfortable using the AMC, however creating data access rules that are outside of simple use cases continue to be a challenge. Participants comfort level with Adminer improved from 2016 to 2018 with all participants reporting comfortable in using the tool. The use of each tool’s quick start guides varied. Of those who used the guides, all of the participants agreed that the Adminer and Data Dictionary guides were useful. There was a smaller number of participants agreeing that the other guides were useful. Lastly, participants provided recommendations to improving the BioSense Platform. The most frequent recommendations were improving the data access control architecture, and sharing aggregate data with hospitals in their state.
Conclusions The development and operationalization of the BioSense Platform and associated tools has been in an environment of continuing advancements in technology and changing public health needs and priorities. Up-to-date evaluation activities have helped to ensure that BioSense is best suited to address these challenges and meet the syndromic surveillance needs of users. Overall, the findings outlined above indicate that the functionality and utility of BioSense are well suited to meet user needs.
Acknowledgement Harold Gil, Martha Sanchez, Michele Vickers, Caleb Wiedeman, Erin Austin, Yushuian Chen, Natasha Close, Katie Arends, Jennifer Broad, Amanda Thomas, Michael Coletta, Alan Davis, Robert Brown, David Walker, Hussain Yusuf, Violanda Grigorescu, Roseanne English
References Bangor A, Kortm P, Miller J. Determining What Individual SUS Scores Mean: Adding an Adjective Rating Scale. Journal of Usability Studies. 2009; 4:114-123.
Brief bio for lead author/ presenter to be used by session moderators at the conference Cassandra Davis is a health scientist and a program evaluator in the Division of Health Informatics and Surveillance at CDC. Ms. Davis’ work focuses on evaluation activities of the National Syndromic Surveillance Program cooperative agreement and the BioSense Platform. Ms. Davis has over nine years of public health experience focused on planning, implementing and evaluating surveillance programs and other large initiatives at the local, state and federal level. Ms. Davis graduated from Tulane University School of Public Health and Tropical Medicine with a MPH in Epidemiology and Maternal and Child Health in 2009 and is an alumnus of the CDC’s Public Health Prevention Service.
Brief summary (100 words) of Presentation to be Used in Conference Program An evaluation of the present status of the utility, functionality, usability and customer satisfaction of the BioSense Platform was conducted in 2018 among a small group of BioSense Platform users. The findings from the evaluation provide information about the user experience of the BioSense Platform and recommendations to the tools and associated guidance documents.


Syndrome Development to Assess IDU, HIV, and Homelessness in MA Emergency Departments

Stefanie P. Albert, Rosa Ergas, Sita Smith, Gillian Haney, Monina Klevens

Bureau of Infectious Disease and Laboratory Sciences, Massachusetts Department of Public Health, Boston, Massachusetts, United States


Objective We sought to measure the burden of emergency department (ED) visits associated with injection drug use (IDU), HIV infection, and homelessness; and the intersection of homelessness with IDU and HIV infection in Massachusetts via syndromic surveillance data.
Introduction In Massachusetts, syndromic surveillance (SyS) data have been used to monitor injection drug use and acute opioid overdoses within EDs. Currently, Massachusetts Department of Public Health (MDPH) SyS captures over 90% of ED visits statewide. These real-time data contain rich free-text and coded clinical and demographic information used to categorize visits for population level public health surveillance.

Other surveillance data have shown elevated rates of opioid overdose related ED visits, Emergency Medical Service incidents, and fatalities in Massachusetts from 2014-20171,2,3. Injection of illicitly consumed opioids is associated with an increased risk of infectious diseases, including HIV infection. An investigation of an HIV outbreak among persons reporting IDU identified homelessness as a social determinant for increased risk for HIV infection.
Methods To accomplish our objectives staff used an existing MDPH SyS IDU syndrome definition4, developed a novel syndrome definition for HIV-related visits, and adapted Maricopa County's homelessness syndrome definition. Syndromes were applied to Massachusetts ED data through the CDC’s BioSense Platform. Visits meeting the HIV and homelessness syndromes were randomly selected and reviewed to assess accuracy; inclusion and exclusion criteria were then revised to increase specificity. The final versions of all three syndrome definitions incorporate free-text elements from the chief complaint and triage notes, as well as International Statistical Classification of Diseases and Related Health Problems, 9th (ICD-9) and 10th Revision (ICD-10) diagnostic codes. Syndrome categories were not mutually exclusive, and all reported visits occurring at Massachusetts EDs were included in the analysis.

Syndromes Created

For the HIV infection syndrome definition, we incorporated the free-text term “HIV” in both the chief complaint and triage notes. Visit level review demonstrated that the following exclusions were needed to reduce misspellings, inclusion of partial words, and documentation of HIV testing results: “negative for HIV”, “HIV neg”, “negative test for HIV”, “hive”, “hivies”, and “vehivcle”. Additionally, the following diagnostic codes were incorporated: V65.44 (Human immunodeficiency virus [HIV] counseling), V08 (asymptomatic HIV infection status), V01.79 (contact with or exposure to other viral diseases), 795.71 (nonspecific serologic evidence of HIV), V73.89 (special screening examination for other specified viral diseases), 079.53 (HIV, type 2 [HIV-2]), Z20.6 (contact with and (suspected) exposure to HIV), Z71.7 (HIV counseling), B20 (HIV disease), Z21 (asymptomatic HIV infection status), R75 (inconclusive laboratory evidence of HIV), Z11.4 (encounter for screening for HIV), and B97.35 (HIV-2 as the cause of diseases classified elsewhere).

Building on the Maricopa County homeless syndrome definition, we incorporated a variety of free-text inclusion and exclusion terms. To meet this definition visits had to mention: “homeless”, or “no housing”, or, “lack of housing”, or “without housing”, or “shelter” but not animal and domestic violence shelters. We also selected the following ICD-10 codes for homelessness and inadequate housing respectively, Z59.0 and Z59.1.

We analyzed MDPH SyS data for visits occurring from January 1, 2016 through June 30, 2018. Rates per 10,000 ED visits categorized as IDU, HIV, or homeless were calculated. Subsequently, visits categorized as IDU, HIV, and meeting both IDU and HIV syndrome definitions (IDU+HIV) were stratified by homelessness.
Results Syndrome Burden on ED

The MDPH SyS dataset contains 6,767,137 ED visits occurring during the study period. Of these, 82,819 (1.2%) were IDU-related, 13,017 (0.2%) were HIV-related, 580 (<0.01%) were related to IDU + HIV, and 42,255 visits (0.6%) were associated with homelessness.

The annual rate of IDU-related visits increased 15% from 2016 through June of 2018 (from 113.63 to 130.57 per 10,000 visits); while rates of HIV-related and IDU + HIV-related visits remained relatively stable. The overall rate of visits associated with homelessness increased 47% (from 49.99 to 73.26 per 10,000 visits).

Rates of IDU, HIV, and IDU + HIV were significantly higher among visits associated with homelessness. Among visits that met the homeless syndrome definition compared to those that did not: the rate of IDU-related visits was 816.0 versus 118.03 per 10,000 ED visits (X2= 547.12, p<0. 0001); the rate of visits matching the HIV syndrome definition was 145.54 versus 18.44 per 10,000 ED visits (X2= 99.33, p<0.0001); and the rate of visits meeting the IDU+HIV syndrome definition was 15.86 versus 0.76 per 10,000 visits (X2= 13.72, p= 0.0002).
Conclusions Massachusetts is experiencing an increasing burden of ED visits associated with both IDU and homelessness that parallels increases in opioid overdoses. Higher rates of both IDU and HIV-related visits were associated with homelessness. An understanding of the intersection between opioid overdoses, IDU, HIV, and homelessness can inform expanded prevention efforts, introduction of alternatives to ED care, and increase consideration of housing status during ED care.

Continued surveillance for these syndromes, including collection and analysis of demographic and clinical characteristics, and geographic variations, is warranted. These data can be useful to providers and public health authorities for planning healthcare services.
Acknowledgement Special thanks to the Maricopa County Department of Public Health, and the following Massachusetts Department of Public Health Bureau of Infectious Disease and Laboratory Sciences staff: Mark Bova, Katherine Brown, Dan Church, Kevin Cranston, Alfred DeMaria, and Shauna Onofrey.
References 1. Vivolo-Kantor AM, Seth P, Gladden RM, et al. Vital Signs: Trends in Emergency Department Visits for Suspected Opioid Overdoses — United States, July 2016–September 2017. MMWR Morbidity and Mortality Weekly Report 2018; 67(9);279–285 DOI: http://dx.doi.org/10.15585/mmwr.mm6709e1
2. Massachusetts Department of Public Health. Chapter 55 Data Brief: An assessment of opioid-related deaths in Massachusetts, 2011-15. 2017 August. Available from: https://www.mass.gov/files/documents/2017/08/31/data-brief-chapter-55-aug-2017.pdf
3. Massachusetts Department of Public Health. MA Opioid-Related EMS Incidents 2013-September 2017. 2018 Feb. Available from: https://www.mass.gov/files/documents/2018/02/14/emergency-medical-services-data-february-2018.pdf
4. Bova, M. Using emergency department (ED) syndromic surveillance to measure injection-drug use as an indicator for hepatitis C risk. Powerpoint presented at: 2017 Northeast Epidemiology Conference. 2017 Oct 18 – 20; Northampton, Massachusetts, USA.
Brief bio for lead author/ presenter to be used by session moderators at the conference Stefanie Albert is a Senior Research Analyst at the Massachusetts Department of Public Health. She is relatively new to the syndromic surveillance world. She has a background working with public health agencies in the non-profit and government sectors focusing on program evaluation, data analysis, and data dissemination. Her interests include substance use, communicable diseases, and understanding health disparities. Stefanie received her MPH from the University of Virginia and her BS in Microbiology from the University of South Florida.
Brief summary (100 words) of Presentation to be Used in Conference Program There was an increase in rates of opioid overdoses in Massachusetts from 2014-2017. Injection of illicitly consumed opioids is associated with an increased risk of infectious diseases including HIV infection. A highly visible investigation of HIV infection among persons reporting injection drug use (IDU) identified homelessness as a potential social determinant for increased risk of HIV infection. We developed and adapted syndrome definitions prior to analyzing syndromic surveillance data to measure the burden of Massachusetts emergency department (ED) visits associated with IDU, HIV, and homelessness and evaluate the relationship of homelessness with IDU and HIV infection.


Rate of IDU, HIV, and IDU+HIV Related Visits Captured by MDPH SyS per 10,000 ED Visits by Year (2016 through 2018*)


Year IDU-Related Visits HIV-Related Visits Meets IDU and HIV Definitions
(IDU+HIV)
Homeless Visits All MDPH SyS ED Visits (N)
2016 113.63 16.89 0.59 49.99 2,367,004
2017 125.19 20.56 0.95 66.88 2,842,048
2018
Jan-June*
130.57 20.39 1.08 73.26 1,558,085
TOTAL 122.38 19.24 0.86 62.44 6,767,137



Optimization of Linkage between North Carolina EMS and ED Data: EMS Naloxone Cases

Jonathan Fix1, Dennis Falls2, Scott Proescholdbell3, Amy Ising2, Tony Fernandez4, 2, Anna E. Waller2

1Epidemiology, UNC - Chapel Hill, Carrboro, North Carolina, United States, 2Emergency Medicine, UNC-CH, Chapel Hill, North Carolina, United States, 3North Carolina Division of Public Health, Raleigh, North Carolina, United States, 4EMS Performance Improvement Center, Chapel Hill, North Carolina, United States


Objective To improve linkage between North Carolina’s Emergency Medical Services (EMS) and Emergency Department (ED) data using an iterative, deterministic approach.
Introduction The opioid overdose crisis has rapidly expanded in North Carolina (NC), paralleling the epidemic across the United States. The number of opioid overdose deaths in NC has increased by nearly 40% each year since 2015.1 Critical to preventing overdose deaths is increasing access to the life-saving drug naloxone, which can reverse overdose symptoms and progression. Over 700 EMS agencies across NC respond to over 1,000,000 calls each year; naloxone administration was documented in over 15,000 calls in 2017.2

Linking EMS encounters with naloxone administration to the corresponding ED visit assists in understanding the health outcomes of these patients. However, less than 66% of NC EMS records with naloxone administration in 2017 were successfully linked to an ED visit record. This study explored methods to improve EMS and ED data linkage, using a multistage process to maximize the number of correctly linked records while avoiding false linkages.
Methods EMS data were provided by the EMS Performance Improvement Center2 (EMSPIC); ED data were provided by NC DETECT.3 Optimization of current EMS/ED linkage methods began by extracting a non-random subset of EMS encounters with naloxone administration between January 1, 2017 and November 30, 2017 from 12 NC counties, representing eastern, central and western regions and the overall linkage performance of the larger dataset. Records were eligible for linkage if EMS recorded that the patient was “treated and transported” to the ED. All records in the subset were manually reviewed in NC DETECT to identify corresponding ED visit records. This produced a “gold standard” dataset of linked EMS/ED records.

To evaluate linkage performance, we first identified all records eligible for linkage. Any EMS transport to either a hospital outside of NC or an NC ED not included in NC DETECT (e.g., military, VA and tribal hospitals) was excluded. Since existing linkage is performed daily and both EMS and ED records are updated over time to correct errors and missing data, existing linkage methods were re-run on updated data to evaluate the improvement provided solely by linking the most up-to-date data. Unlinked EMS records for which the encounter was an inter-facility transfer, transfer to helicopter transport, or the patient died during transfer were deemed ineligible for linkage, as these patients likely either bypassed or never made it to the ED.

To initially improve linkage quality, we updated the mapping file of EMS/ED destinations. An exact destination match was required for linkage and the EMS destination variable is recorded as free-text; thus, all variations of a destination name and spelling were identified and mapped to a standardized name. The maximum time difference between EMS drop-off and ED intake was then allowed to exceed 60 minutes, in iterations of 90, 120, 240, and 360 minutes. With each iteration, we compared the linked IDs with the gold standard dataset to identify false links.

Finally, a multistage linkage process was applied. First, deterministic linkage was run requiring exact matches for date of birth (DOB), sex/gender, and destination, and up to 360-minute difference between EMS/ED times. The unlinked records were then processed a second time, requiring exact matches for sex/gender and destination, DOB to be within +/- 10 days or +/- 1 year, and up to 60-minute difference between EMS/ED times.

This multistage process was then run for all 2017 EMS encounters with naloxone administration to ensure that the new method was not over fit to the data subset. Potential bias in the linkage was assessed by comparing the distributions of age (mean and median) and gender (% male) among the linked and unlinked records in each dataset.

Statistical analyses were completed using SAS 9.4 (Cary, NC). Linkage was executed using SQL Server.
Results Between 1/1/2017 and 11/30/2017, there were 14,793 EMS encounters with documented naloxone administration. Of these, 12,089 (81.7%) were recorded as “treated and transported”; 1,906 EMS encounters were included in the 12-county subset. The average age of patients was 45.1 years among all naloxone encounters and 45.2 years in the subset. 57.5% of all encounters were male; 58.1% were male in the subset.

After removing EMS transports to non-NC or non-NC DETECT hospitals, the existing subset linkage was 61.8% (1,154/1,866). This included 38 (2.0%) false positives, apparently caused by ED records purged since this linkage was conducted. When the existing methods were run against the most current data, linkage improved to 72.2% (1,389/1,866), reflecting an absolute improvement of 10.4% by simply using updated data. Only 1 (0.05%) false positive was identified in this process.

Following removal of unlinked inter-facility transfers, deaths during EMS transport, and transfers to helicopters, the records eligible for linkage dropped to 1,781. Linkage improved to 79.5% (1,417/1,781) when hospital names were standardized. Linkage using standardized hospital names and relaxing the EMS/ED time difference performed at the following levels: 82.3% at 90 minutes, 83.3% at 120 minutes, 87.9% at 240 minutes, and 89.4% at 360 minutes. Even when using the most relaxed time difference (+/- 360 minutes), only one false positive was identified, the same produced during initial linkage at +/- 60 minutes. The final multistage method produced linkage of 91.0% (1,620/1,781), with no additional false positives.

Applying the initial methods to the statewide EMS dataset produced linkage of 64.8%. The multistage linkage process performed nearly identically on statewide data as observed for the subset, at 91.1%. For statewide data, the age of linked patients was younger (mean = 44.7 years [SD = 18.4], median = 41.0 years) than that of unlinked patients (mean = 48.0 years [SD = 19.3], median = 47.0 years). Additionally, linked patients were more likely to be male (58.1%) when compared to unlinked patients (54.2%).
Conclusions High quality linkage between EMS and ED records is essential for research and public health surveillance examining health outcomes. Using a multistage process, we improved the linkage of EMS encounters with documented naloxone administration to ED visits in North Carolina in 2017 from 64.8% to 91.1%, with less than 0.05% false positive rate. This improved linkage will facilitate future analyses of relationships between exposures during EMS encounters and outcomes experienced in hospitals. Future research should evaluate the generalizability of this linkage methodology to all EMS records, not just those with naloxone administration, as well as to pre-2017 data. Implementation of probabilistic linkage or machine learning as a final stage in a multistage process may further improve linkage outcomes, overcoming missing data or unpredictable errors in the data.
Acknowledgement Funding provided by the CDC National Center for Injury Prevention and Control Enhanced State Opioid Overdose Surveillance (ESOOS) grant to the NC Division of Public Health (grant 5NU17CE924902).
References 1. Kansagra SM, Cohen MK. The Opioid Epidemic in NC: Progress, Challenges, and Opportunities. N C Med J 2018; 79(3): 157-62.
2. EMS Performance Improvement Center. About EMSPIC. https://www.emspic.org/about.
3. NC DETECT. Background. http://ncdetect.org/background/
Brief bio for lead author/ presenter to be used by session moderators at the conference Jonathan Fix is an MSPH/PhD student at the University of North Carolina at Chapel Hill, with a focus in Infectious Disease Epidemiology. His prior research has focused on vaccine efficacy, effectiveness, and policy, both in the U.S. and around the world. He works part-time at the Carolina Center for Health Informatics to improve the linkage between North Carolina EMS and ED data, and to explore associations between exposures and health outcomes among the populations using these services. Upon completion of his doctoral program, Jonathan hopes to contribute to the design and execution of clinical trials in international settings.
Brief summary (100 words) of Presentation to be Used in Conference Program Among EMS encounters with documented naloxone administration in North Carolina 2017, the current methods for linking EMS and ED records perform at only 64.8%. We investigated how use of a multi-staged, deterministic linkage approach could improve linkage quality while limiting the inclusion of false links. Following creation of a subset of all NC EMS encounters with naloxone administrations and manual record review to create a “gold standard dataset” we achieved linkage quality of 91.1%, with only 0.05% false-linkages. Application of this linkage method to all state-wide EMS encounters documented with naloxone administration in 2017 similarly performed at 91.0%.


Enhanced Surveillance of Nonfatal Emergency Department Opioid Overdoses in California

Natalie Demeter, Jaynia A. Anderson, Mar-y-sol Pasquires, Stephen Wirtz

California Department of Public Health, Sacramento, California, United States


Objective To track and monitor nonfatal emergency department opioid overdoses in California for use in the statewide response in the opioid epidemic.
Introduction The opioid epidemic is a multifaceted public health issue that requires a coordinated and dynamic response to address the ongoing changes in the trends of opioid overdoses. Access to timely and accurate data allows more targeted and effective programs and policies to prevent and reduce fatal and nonfatal drug overdoses in California. As a part of a Centers for Disease Control and Prevention Enhanced State Opioid Overdose Surveillance grant, the goals of this surveillance are to more rapidly identify changes in trends of nonfatal drug overdose, opioid overdose, and heroin overdose emergency department visits; identify demographic groups or areas within California that are experiencing these changes; and to provide these data and trends to state and local partners addressing the opioid crisis throughout California. Emergency department (ED) visit data are analyzed on an ongoing quarterly basis to monitor the proportion of all ED visits that are attributed to nonfatal drug, opioid, and heroin overdoses as a portion of the statewide opioid overdose surveillance.
Methods California emergency department data were obtained from the California Office of Statewide Health Planning and Development. Data were (and continue to be) analyzed by quarter as the data become available, starting in quarter 1 (Q1) 2016 through Q1 2018. Quarters were defined as standard calendar quarters; January-March (Q1), April-June (Q2), July-September (Q3), and October-December (Q4). Counts of nonfatal ED visits for all drug overdoses, all opioid overdoses, and heroin overdoses were defined by the following ICD-10 codes in the principle diagnosis or external cause of injury fields respectively; T36X-T50X (all drug), T40.0X-T40.4X T40.6 and T40.69 (all opioid), and T40.1X (heroin). Eligible ED visits were limited to CA residents, patients greater than 10 years of age, initial encounters, and were classified as unintentional overdoses or overdoses of undetermined intent. Overdose ED visits are described by quarter, drug, sex, and age for Q1 2016 – Q1 2018.
Results On average, 6,450 emergency department visits in California are attributed to drug overdose every quarter. Between Q1 2016 and Q1 2018, on average 1,785 (range: 1,559-2,011 ED visits) of those visits were due to opioid overdoses and a further 924 (52%) of those ED visits were due to heroin overdoses. About 26-30% of all drug overdose ED visits were for opioid overdoses in California during Q1 2016 – Q1 2018. Quarterly, that is around 6.00-7.64 opioid overdose ED visits for every 10,000 ED visits (Table 1), with about half those (3.09-4.30 ED visits) being heroin overdose ED visits. Males accounted for approximately 52% of all drug overdose ED visits, 65% of all opioid overdose ED visits, and 76% of all heroin overdose ED visits per quarter. Across all quarters, 25-34 year olds had the highest proportion of emergency department visits attributed to opioid and heroin overdose compared to all other age groups. However, 11-24 year olds had the highest proportion of emergency department visits attributed to all drug overdoses compared to all other age groups for all quarters except one. Between Q1 2016 and Q1 2018, the proportion of emergency department visits attributed to all drug overdoses increased by 1.8%, all opioid overdoses increased 3.1%, and heroin overdoses increased by 13.5%.
Conclusions Overall trends for the proportion of all emergency department visits for all drug overdoses and all opioid overdoses are relatively stable over this time period, however the proportion of heroin overdose ED visits shows a more substantial increase between Q1 2016 and Q1 2018. In addition, heroin overdose ED visits account for over half of all opioid overdose ED visits during this time in California. Ongoing surveillance of drug, opioid, and heroin overdose ED visits is a crucial component of assessing and responding to the opioid overdose crisis in California and helps to better understand the demographics of those who could be at risk of a future fatal opioid overdose. Timely data such as these (in addition to prescribing, hospitalization, and death data) can inform local and statewide efforts to reduce opioid overdoses and deaths.
Acknowledgement This project was supported by the U.S. Centers for Disease Control and Prevention, NU17CE924903-02-00.
References
Brief bio for lead author/ presenter to be used by session moderators at the conference Natalie Demeter is an epidemiologist working in the Safe and Active Communities Branch at the California Department of Public Health. She is the epidemiologist on the Enhanced State Opioid Overdose Surveillance project.
Brief summary (100 words) of Presentation to be Used in Conference Program The goals of this surveillance are to more rapidly identify changes in trends of nonfatal drug overdose, opioid overdose, and heroin overdose emergency department visits; identify demographic groups or areas within California that are experiencing these changes; and to provide these data and trends to partners addressing the opioid crisis throughout California. Emergency department visit data are analyzed on an ongoing quarterly basis to monitor the proportion of all ED visits that are attributed to nonfatal drug, opioid, and heroin overdoses as a portion of the statewide opioid overdose surveillance.


Table 1. Proportion of all CA Emergency Department Visits attributed to Drug Overdose (per 10,000 visits)


Quarter and Year All Drug All Opioid Heroin
Q1 2016 24.05 6.57 3.09
Q2 2016 26.19 7.05 3.34
Q3 2016 24.69 6.77 3.49
Q4 2016 24.55 6.82 3.56
Q1 2017 23.45 6.00 3.19
Q2 2017 25.73 7.44 4.11
Q3 2017 25.81 7.64 4.30
Q4 2017 24.38 6.73 3.39
Q1 2018 24.49 6.77 3.51


Opioid Seizures by Law Enforcement in Relation to Emergency Room Visits

Tanner Turley, Evan Mobley, Andrew Hunter

Health & Senior Services, State of Missouri, Jefferson City, Missouri, United States


Objective To evaluate the relationship between heroin and non-heroin opioid seizures reported by law enforcement and the number of ER visits due to heroin and non-heroin opioid poisoning in selected counties in Missouri.
Introduction In 2016, there were approximately 63,000 deaths nationally due to drug overdose. This trend continues to increase with the provisional number of US deaths for 2017 being approximately 72,000 (1). This increase in overdose deaths is fueled largely by the opioid class of drugs. The opioid epidemic began in the 1990s with a steady rise in prescription opioid overdoses. However, after 2010 a rise in heroin overdose deaths also began to occur. In addition to the heroin deaths increasing, there was a sharp rise in overdose deaths due to synthetic opioids including illicitly manufactured fentanyl beginning in 2013 (2). In Missouri, ER visits follow similar trends with heroin overdose visits greatly increasing after 2011. While PDMPs help function as data sources that provide information on the licit drug supply, they cannot give much knowledge on the illicit supply. Because of this, drug seizure data from law enforcement can provide a much-needed tool in understanding the supply of illicit substances and their impact on a county’s morbidity.
Methods Data sources used in this analysis include the El Paso Intelligence Center (EPIC) drug seizure database thanks to cooperation by the Midwest HIDTA (High Intensity Drug Trafficking Area) office and Missouri Highway Patrol. ER Visit Data was retrieved from the Missouri Patient Abstract System, which includes ER visits for non-federal hospitals. Data was aggregated on a quarterly basis from 2014-2016 resulting in 12 observations (n) for every county observed.
A subset of counties were selected and reviewed based on both high counts and high rates of ED visits for opioid overdoses (3). The counties reviewed were Franklin, Greene, Jefferson, St. Francois, St. Louis City and St. Louis County. The majority of these counties were located in the greater St. Louis Are with Greene and St. Francois counties being notable exceptions. Greene County contains the city of Springfield and is located in southwest Missouri. St. Francois is the most rural county in our subset and is located south of the St. Louis area. For each county, the number of ER Visits were compared to the number of drug seizures reported by law enforcement facilities in EPIC. Numbers were compared for both heroin and non-heroin opioids. Records were identified as a heroin overdose or non-heroin opioid overdose based on CDC drug poisoning guidance (4). If an ER discharge record contained codes for both heroin and a non-heroin opioid, the record was counted in the heroin column only. This method avoided counting records twice.
The Spearman correlation coefficient was calculated in SAS to determine if there was a possible relationship between seizures and ED visits at the county level due to the relatively few data points, the presence of outlier observations in the seizure numbers, as well as violations of statistical normality among the county seizure data. The Spearman Correlation Coefficient is a better alternative in this case to the commonly used Pearson Correlation Coefficient due to its ability to handle skewed data and outliers (5). As with the Pearson Correlation Coefficient, a score of 0 is read as the variables have no discernable relationships, and scores of 1 or -1 denote a perfect linear relationship between the observed variables (positive and negative respectively).
Results Initial results showed correlational effects between ED visits and seizures to be generally moderate or weak on the county level. The strongest relationships observed were found in St. Louis City for both heroin (R=-0.455) and non-heroin opioids (R=-0.51) as well as Jefferson County for both heroin (R=0.536) and non-heroin (R=0.50). St. Louis County also had a notable relationship for heroin seizures and heroin ED visits with R=-0.55. P values were also calculated to test if correlation values differed significantly from the null hypothesis of R=0 (i.e. no correlation). In all examined cases, there was no p value that was less than the standard cutoff of 0.05 which indicates none of the results are markedly different given the null hypothesis of R=0 is true (6).
Of particular interest is the contrast in results between St. Louis City and Jefferson County. St. Louis City had a moderate negative relationship with seizures and ED visits with ED Visits tending to decrease as drug seizures increased. Whereas, Jefferson County had a moderate positive relationship with ED Visits increasing alongside drug seizures. Due to their close geographic proximity, it is likely that both counties influence one another. Further evaluation is required to gauge regional effects.
Conclusions Due to the complexity of the opioid epidemic, the value of having varied data sources cannot be understated. While the correlational effects observed here are not indicative of a strong relationship between ED visits and drug seizures, further evaluation and research of both data sources is highly recommended. As additional data is gathered in the future, stronger analyses than the Spearman Correlation Coefficient may be used to further explore the relationship between drug overdose morbidity and law enforcement seizure data. Other relationships may also be explored such as drug seizures in relation to drug overdose mortality.
Acknowledgement
References 1. National Center for Health Statistics. (2018, September 12). Retrieved September 19, 2018, from https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm
2. Opioid Overdose. (2017, August 30). Retrieved September 19, 2018, from https://www.cdc.gov/drugoverdose/epidemic/index.html
3. Bureau of Health Care Analysis and Data Dissemination, Missouri Department of Health and Senior Services. (2018, June 27). ER Visits Due to Opioid Misuse. Retrieved September 19, 2018 from https://health.mo.gov/data/opioids/pdf/opioid-dashboard-slide-16.pdf
4. CDC Prescription Drug Overdose Team. (2013, August 12). GUIDE TO ICD-9-CM AND ICD-10 CODES RELATED TO POISONING AND PAIN. Retrieved September 2018 from https://www.cdc.gov/drugoverdose/pdf/pdo_guide_to_icd-9-cm_and_icd-10_codes-a.pdf
5.
Mukaka, M. (2012, September). A guide to appropriate use of Correlation coefficient in medical research. Retrieved September 20, 2018, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3576830/
6. Ronald L. Wasserstein & Nicole A. Lazar (2016) The ASA's Statement on p-Values: Context, Process, and Purpose, The American Statistician, 70:2, 129-133, DOI:
10.1080/00031305.2016.1154108 from https://amstat.tandfonline.com/doi/pdf/10.1080/00031305.2016.1154108
Brief bio for lead author/ presenter to be used by session moderators at the conference Tanner Turley is a Research Analyst III with the Bureau of Health Care Analysis and Data Dissemination within the Department of Health and Senior Services for the State of Missouri. In his current role, Turley reports non-fatal opioid overdose data to the Center to Disease Control on a quarterly basis under the Enhanced State Opioid Overdose Surveillance program. He also share somes responsibility in disseminating opioid overdose data to various stakeholder, and has previously worked with fatal opioid overdose data under the same program.
Brief summary (100 words) of Presentation to be Used in Conference Program This presentation will go over the initial analysis results of the relationship between the number of opioid seizures reported by law enforcement and the number of emergency room visits for opioid overdoses in a subset of counties in Missouri. Analysis methods used, as well as possibilities for the data moving forward, will also be covered.




Improving risk factor identification for opioid overdose deaths in Tennessee

Sarah J. Nechuta, Jenna Moses, Molly Golladay, Adele Lewis, Julia Goodin, Melissa McPheeters

Tennessee Department of Health, Nashville, Tennessee, United States


Objective To examine specific drugs present based on postmortem toxicology for prescription opioid, heroin, and fentanyl overdoses classified based on ICD-10 coding. To compare drugs identified from postmortem toxicology with those listed on the death certificate for opioid overdoses.
Introduction Using death certificates alone to identify contributing substances in drug overdose deaths may result in misclassification and underestimation of the burden of illicit and prescription opioids and other drugs in drug-related deaths. To enable timely and targeted prevention in Tennessee (TN), the identification and monitoring of new drugs and trends in use should utilize toxicology and medicolegal death investigation data directly, as recommended by others 1-3. These data can inform mortality outcome definitions for improved surveillance and risk factor identification 4-7. To our knowledge, this is the first analysis to use statewide linked toxicology and death certificate data in TN.
Methods We identified 615 opioid involved overdose deaths in TN of unintentional (underlying ICD-10 codes: X40-X44) or undetermined (underlying ICD-10 codes: Y10-Y14) intent during June 1st to December 31st 2017. Utilizing the Interim Medical Examiner Database (I-MED), we identified postmortem toxicology reports for 454 cases, which were from one of three national laboratories used by a state Regional Forensic Center. Toxicology data were abstracted and independently verified by two co-authors and linked to the TN death statistical file that included cause of death information (literal text and ICD-10 codes) and demographics. The analysis focuses on cases with an available toxicology report.
Results We identified 171 prescription opioid overdoses, 221 fentanyl overdoses, and 113 heroin overdoses. Table 1 displays postmortem toxicology profiles for major drugs/classes. For prescription opioid deaths (excluding fentanyl and heroin), positive toxicology results for prescription opioids were as follows: methadone (11%), buprenorphine (14%), hydrocodone (14%), oxycodone (36%) and oxymorphone (also a metabolite, 47%). Benzodiazepines were present in close to 58% of prescription opioid overdoses; stimulants (cocaine, amphetamines, methamphetamines) in about 25%. For fentanyl and heroin deaths, prescription opioids were detected in about 26% and 34%, respectively; stimulants in about 57.9% and 52.2%, respectively, and benzodiazepines 36-37%. Fentanyl was present on toxicology in about half of heroin overdoses, and 6–monoacetylmorphine in 72.6%.

Table 2 displays a comparison between death certificate (DC) listed drugs and drugs identified via toxicology. Close to all fentanyl deaths identified from the DC were identified via toxicology (98.7%). Benzodiazepines were involved in 34% of deaths based on DC, and 46% based on toxicology. Stimulants were involved in about in 39% of deaths based on DC, and 45% based on toxicology. Based on toxicology, about 20% of decedents were using antihistamines at overdose and 10% were using antidepressants.
Conclusions Using medical examiners’ data, including toxicology data, improves estimation of contributing drugs involved in opioid deaths. This analysis provides jurisdiction-specific data on drugs that can help with monitoring trends and informs risk factor identification. Future work includes adding information on prescribed opioid and benzodiazepines using TN’s Prescription Drug Monitoring Database and evaluating demographic variation in contributing drugs between toxicology and DC data to identify susceptible populations.
Acknowledgement This work was supported by funding from the Centers for Disease Control & Prevention (NU17CE924899-02-00) and (5 NU17CE002731-02-00). The funder had no role in study design, data collection and analysis, or decision to publish.
References 1. Slavova S, O'Brien DB, Creppage K, Dao D, Fondario A, Haile E, Hume B, Largo TW, Nguyen C, Sabel JC, Wright D, Council of S, Territorial Epidemiologists Overdose S. Drug Overdose Deaths: Let's Get Specific. Public Health Rep.
2. Horon IL, Singal P, Fowler DR, Sharfstein JM. Standard Death Certificates Versus Enhanced Surveillance to Identify Heroin Overdose-Related Deaths. Am J Public Health. 2018;108(6):777-81.
3. Mertz KJ, Janssen JK, Williams KE. Underrepresentation of heroin involvement in unintentional drug overdose deaths in Allegheny County, PA. J Forensic Sci. 2014;59(6):1583-5.
4. Landen MG, Castle S, Nolte KB, Gonzales M, Escobedo LG, Chatterjee BF, Johnson K, Sewell CM. Methodological issues in the surveillance of poisoning, illicit drug overdose, and heroin overdose deaths in new Mexico. Am J Epidemiol. 2003;157(3):273-8.
5. Davis GG, National Association of Medical E, American College of Medical Toxicology Expert Panel on E, Reporting Opioid D. Complete republication: National Association of Medical Examiners position paper: Recommendations for the investigation, diagnosis, and certification of deaths related to opioid drugs. J Med Toxicol. 2014;10(1):100-6.
6. Slavova S, Bunn TL, Hargrove SL, Corey T. Linking Death Certificates, Postmortem Toxicology, and Prescription History Data for Better Identification of Populations at Increased Risk for Drug Intoxication Deaths. Pharmaceutical Medicine. 2017;31(3):155-65.
7. Hurstak E, Rowe C, Turner C, Behar E, Cabugao R, Lemos NP, Burke C, Coffin P. Using medical examiner case narratives to improve opioid overdose surveillance. Int J Drug Policy. 2018;54:35-42.
Brief bio for lead author/ presenter to be used by session moderators at the conference Dr. Nechuta is Chief Scientist at the Tennessee Department of Health in the Office of Informatics and Analytics. She oversees analytics using public health data for a variety of projects, including investigations of risk factors for all drug and opioid overdose deaths, understanding opioid and benzodiazepine prescribing patterns during pregnancy in association with infant birth outcomes, and improving measures of opioid and drug-related health outcomes for mortality and morbidity statistics for public health surveillance. She has a strong interest in methods to improve entity management methodology for big data, case definitions for health outcomes, and identification of susceptible populations for improved population health monitoring and targeted prevention.
Brief summary (100 words) of Presentation to be Used in Conference Program The objectives of this study were: 1) to examine specific drugs present based on postmortem toxicology for prescription opioid (n=171), heroin (n=113), and fentanyl (n=221) overdoses classified based on ICD-10 coding and 2) to compare drugs identified from postmortem toxicology with the those listed on death certificate. To our knowledge, this is the first analysis to use statewide linked toxicology and death certificate data in Tennessee. Results demonstrate that utilization of toxicology data from Tennessee’s Interim Medical Examiners Database improves estimation of contributing drugs and provides key data to improve monitoring of trends and risk factor identification.


Table 1. Postmortem toxicology results among prescription opioid, fentanyl, and heroin overdose deaths in Tennessee, n (%)


  Prescription Opioid*
(n = 171)
Fentanyl*
(n = 221)
Heroin*
(n = 113)
Positive Toxicology Positive Toxicology Positive Toxicology
Yes No Yes No Yes No
Fentanyl 3 (1.8) 168 (98.2) 217 (98.2) 4 (1.8) 58 (51.3) 55 (48.7)
6–monoacetylmorphine 1 (0.6) 170 (99.4) 37 (16.7) 184 (83.3) 82 (72.6) 31 (27.4)
Morphine alone 32 (18.7) 139 (81.3) 89 (40.3) 132 (59.7) 112 (99.1) 1 (0.9)
Morphine and codeine 3 (1.8) 168 (98.2) 9 (4.1) 212 (95.9) 51 (45.1) 62 (54.9)
Codeine 5 (2.9) 166 (97.1) 9 (4.1) 212 (95.9) 52 (46.0) 61 (54.0)
Oxycodone 62 (36.3) 109 (63.7) 20 (9.1) 201 (90.9) 15 (13.3) 98 (86.7)
Hydrocodone 24 (14.0) 147 (86.0) 9 (4.1) 212 (95.9) 11 (9.7) 102 (90.3)
Oxymorphone** 81 (47.4) 90 (52.6) 17 (7.7) 204 (92.3) 5 (4.4) 108 (95.6)
Methadone 19 (11.1) 152 (88.9) 6 (2.7) 215 (97.3) 3 (2.6) 110 (97.4)
Buprenorphine 24 (14.0) 147 (86.0) 6 (2.7) 215 (97.3) 4 (3.5) 109 (96.5)
Benzodiazepines 99 (57.9) 72 (42.1) 80 (36.2) 141 (63.8) 42 (37.2) 71 (62.8)
Cocaine 18 (10.5) 153 (89.5) 74 (33.5) 147 (66.5) 34 (30.1) 79 (69.9)
Other Stimulants 25 (14.6) 146 (85.4) 54 (24.4) 167 (75.6) 25 (22.1) 88 (77.9)
*Defined using death certificate data.
**Also a pharmacologically active metabolite of oxycodone.


Table 2. Comparing postmortem toxicology results with death certificate listed drugs for opioid-involved overdose deaths in Tennessee, n


  Death
Certificate
Positive
Toxicology
Fentanyl 220 223
Heroin* 114 87
Morphine alone 56 178
Morphine and codeine 1 8
Codeine 6 59
Oxycodone 83 93
Hydrocodone 41 42
Methadone 23 26
Buprenorphine 28 32
Oxymorphone** 70 101
Tramadol 10 10
Benzodiazepines 155 201
Cocaine 91 108
Amphetamines/Methamphetamine 87 93
Antihistamines 14 90
Antidepressants 27 45
*Heroin or 6-monoacetylmorphine.
**Also a pharmacologically active metabolite of oxycodone.


Identification and Assessment of Repeat Drug Overdose Visits at EDs in Virginia

Inderbir Sohi, Erin E. Austin, Jonathan Falk

Virginia Department of Health, Richmond, Virginia, United States


Objective To identify and assess the characteristics of individuals with repeated emergency department (ED) visits for unintentional opioid overdose, including heroin, and how they differ from individuals with a single overdose ED visit.
Introduction The Virginia Department of Health (VDH) utilizes syndromic surveillance ED data to measure morbidity associated with opioid and heroin overdoses among Virginia residents. Understanding which individuals within a population use ED services for repeated drug overdose events may help guide the use of limited resources towards the most effective treatment and prevention efforts.
Methods VDH classified syndromic surveillance visits received from 98 EDs (82 hospitals and 16 emergency care centers) between January 2015 and July 2018. An unintentional opioid overdose, which included heroin, was classified based on the chief complaint and/or discharge diagnosis (ICD-9 and ICD-10) using Microsoft SQL Server Management Studio. ED visits were categorized as either a single or a repeat visit, where a repeat visit was defined as two or more separate ED visit records from the same individual. ED visit records were matched to individuals using medical record number. Each match represented a repeat visit for one person. RStudio was used to conduct Pearson’s chi-square tests for sex, race, and 10-year age groups among both visit groups and to

Contact Us

INTERNATIONAL SOCIETY FOR
DISEASE SURVEILLANCE

288 Grove St, Box 203
Braintree, MA 02184
(617) 779 - 0880
Email: syndromic@syndromic.org

 

Subscribe to ISDS News