Utility of Outpatient Syndromic Data for Monitoring Influenza-like Illness

How to Cite

Baber, J., & Feist, M. (2017). Utility of Outpatient Syndromic Data for Monitoring Influenza-like Illness. Online Journal of Public Health Informatics, 9(1). https://doi.org/10.5210/ojphi.v9i1.7629


ObjectiveTo explore how outpatient and urgent care syndromic surveillancefor influenza-like illness (ILI) compare with emergency departmentsyndromic ILI and other seasonal ILI surveillance indicators.IntroductionThe North Dakota Department of Health (NDDoH) collectsoutpatient ILI data through North Dakota Influenza-like IllnessNetwork (ND ILINet), providing situational awareness regardingthe percent of visits for ILI at sentinel sites across the state. Becauseof increased clinic staff time devoted to electronic health initiativesand an expanding population, we have found sentinel sites have beenharder to maintain in recent years, and the number of participatingsentinel sites has decreased. Outpatient sentinel surveillance forinfluenza is an important component of influenza surveillance becausehospital and death surveillance does not capture the full spectrum ofinfluenza illness.Syndromic surveillance (SyS) is another possible source ofinformation for outpatient ILI that can be used for situationalawareness during the influenza season; one benefit of SyS is that itcan provide more timely information than traditional outpatient ILIsurveillance [1,2]. The NDDoH collects SyS data from hospitals(emergency department and inpatient visits) and outpatient clinics,including urgent and primary care locations. Visits include chiefcomplaint and/or diagnosis code data. This data is sent to theBioSense 2.0 SyS platform. We compared our outpatient SyS ILI withour ND ILINet and reported influenza cases, and included hospitaland combined SyS ILI for comparison.MethodsWeekly rates from ND ILINet, SyS ILI, and counts of reportedcases from the influenza season (annual weeks 40 through 20) forthe 2014-2015 and 2015-2016 seasons were compiled. Syndromiccategories for outpatient, hospital (emergency department andinpatient), and combined hospital and outpatient data were created,and the BioSense 2.0 definition for ILI was used. These includeddata from 127,050 outpatient and 323,318 hospital visits for2014-15, and 124,597 outpatient and 424,097 hospital visits for2015-16. Because influenza is a reportable condition in North Dakota,case data is routinely used to represent the seasonal influenza trend,and is useful when other respiratory viruses are circulating. A PearsonCorrelation Coefficient was calculated on all variables using SAS 9.4.Alpha was set to 0.05. There was no overlap between the outpatientclinics providing syndromic surveillance data and clinics participatingin ND ILINet.ResultsAll outpatient, hospital, and combined outpatient and hospitalILI rates from SyS data were positively and significantly correlatedwith both ND ILINet rates and influenza case counts (Table 1). Thecorrelation between outpatient SyS ILI rates and traditional influenzaindicators was lower than for hospital SyS ILI rates for both years,with correlation coefficients ranging from 0.38-0.48 and 0.56-0.92,respectively. Generally SyS data was more highly correlated withcase counts than ND ILINET rates. For the 2014-15 season, hospitalSyS data was the most strongly correlated with traditional influenzaindicators. For 2015-16, combined SyS data was the most stronglycorrelated. Visual inspection of the chief complaint data for ILI visitsfound a significant number of gastrointestinal visits that included thephrase “flu-like illness” in both outpatient and hospital SyS data.ConclusionsAlthough correlation coefficients were lower for outpatient SySILI rates, they are significant enough to be included in our ongoinginfluenza surveillance. One possible confounding factor for therelationship between ED surveillance and reported cases is that peoplewith more severe illness may be more likely to be tested for influenza,and may be more likely to seek medical attention at a hospital setting.This may explain why hospital SyS data provided the strongestcorrelation during the 2014-15 season, a season with higher rates ofmore severe illness than 2015-16. The combination of outpatient dataand hospital data provided the strongest correlation for the 2015-16influenza season, indicating the addition of outpatient data, which mayincrease representativeness of ILI data, may be beneficial to SyS ILIsurveillance. We used an existing ILI syndrome from the BioSense2.0 tool, and revising this syndrome may improve correlationsbetween SyS ILI and ND ILINet and case count data. Negation termsto remove visits for GI illness incorrectly referred to as “flu-like”would be one useful change. The nature of visits for influenza atoutpatient clinic versus hospitals is different, and it is possible thismay account for the difference in the strength of correlations betweenthe two data sources. Use of a different ILI syndrome definition foroutpatient SyS data should be investigated.Table 1. Pearson correlation coefficient values for influenza-like illness in threesyndromic surveillance categories compared with ND ILINET and influenzacase counts.
Authors own copyright of their articles appearing in the Online Journal of Public Health Informatics. Readers may copy articles without permission of the copyright owner(s), as long as the author and OJPHI are acknowledged in the copy and the copy is used for educational, not-for-profit purposes. Share-alike: when posting copies or adaptations of the work, release the work under the same license as the original. For any other use of articles, please contact the copyright owner. The journal/publisher is not responsible for subsequent uses of the work, including uses infringing the above license. It is the author's responsibility to bring an infringement action if so desired by the author.