ISDS Annual Conference Proceedings 2017. This is an Open Access article distributed under the terms of the Creative Commons Attribution-
Noncommercial 3.0 Unported License (http://creativecommons.org/licenses/by-nc/3.0/), permitting all non-commercial use, distribution,  
and reproduction in any medium, provided the original work is properly cited.

ISDS 2016 Conference Abstracts

“That was then, this is now” improving public health 
syndromic surveillance baselines
Roger Morbey*, Alex J. Elliot, Paul Loveridge, Helen Hughes, Sally Harcourt and 
Sue Smith
Public Health England, Birmingham, United Kingdom

Objective
To improve the ability of syndromic surveillance systems to detect 

unusual events.

Introduction
Syndromic surveillance systems are used by Public Health England 

(PHE) to detect changes in health care activity that are indicative of 
potential threats to public health. By providing early warning and 
situational awareness, these systems play a key role in supporting 
infectious disease surveillance programmes, decision making and 
supporting public health interventions.

In order to improve the identification of unusual activity, we 
created new baselines to model seasonally expected activity in 
the absence of outbreaks or other incidents. Although historical 
data could be used to model seasonality, changes due to public 
health interventions or working practices affected comparability. 
Specific examples of these changes included a major change in the 
way telehealth services were provided in England and the rotavirus 
vaccination programme introduced in July 2013 that changed the 
seasonality of gastrointestinal consultations. Therefore, we needed to 
incorporate these temporal changes in our baselines.

Methods
We used negative binominal regression to model daily syndromic 

surveillance, allowing for day of week and public holiday effects. 
To account for step changes in data caused by changes in healthcare 
system working practices or public health interventions we introduced 
specific independent variables into the models. Finally, we smoothed 
the regression models to provide short term forecasts of expected 
trends.

The new baselines were applied to PHE’s four syndromic 
surveillance systems for daily surveillance and public-facing weekly 
bulletins.

Results
We replaced traditional surveillance baselines (based on simple 

averages of historical data) with the regression models for daily 
surveillance of 53 syndromes across four syndromic surveillance 
systems. The improved models captured current seasonal trends and 
more closely reflected actual data outside of outbreaks.

Conclusions
Syndromic surveillance baselines provide context for 

epidemiologists to make decisions about seasonal disease activity and 
emerging public health threats. The improved baselines developed 
here showed whether current activity was consistent with expected 
activity, given all available information, and improved interpretation 
when trends diverged from expectations.

Keywords
syndromic; surveillance; Public Health

Acknowledgments
RAM, AJE & GES are partly funded by the National Institute for 
Health Research (NIHR) Health Protection Research Unit in Emergency 
Preparedness and Response. The views expressed are those of the authors 
and not necessarily those of the National Health Service, the NIHR, the 
Department of Health or Public Health England.

*Roger Morbey
E-mail: roger.morbey@phe.gov.uk

Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * 9(1):e22, 2017