Evaluation of Electronic Ambulatory Care Data for Use in Influenza-Like Illness Surveillance Network (ILINet)

BACKGROUND: During summer 2012, Washington State Department of Health (WA DOH) surveyed active ILINet providers and found that half utilize electronic medical records (EMR) within their practice and 25% use these systems to identify visits for influenza-like illness (ILI).  Several different query methodologies were reported. WA DOH and Public Health – Seattle & King County sought to evaluate EMR from outpatient clinic networks to determine if a syndromic ILI definition previously validated for emergency department (ED) data accurately identified ILI visits in electronic ambulatory care data and to develop query guidance for sentinel providers.

METHODS: Two clinic networks in Washington State submit EMR to public health. Data include clinic name, visit date and time, patient age, sex, zip code, chief complaint, diagnoses, and a visit and patient key. In this ecological study, the following syndromic definition for ILI was applied to historical outpatient data: (1) ICD codes for influenza or mention of “flu” in chief complaint or diagnosis, or (2) a chief complaint or diagnosis of fever plus cough, or (3) a chief complaint or diagnosis of fever plus sore throat.  We assessed the correlation between the proportion of weekly visits for ILI and number of positive influenza tests from laboratories serving each clinic network.

RESULTS: Clinic Network A included seven family practice clinics totaling roughly 6,000 weekly visits. The proportion of weekly ILI visits strongly correlated with the number positive influenza tests during August 2007 – August 2012 (correlation coefficient (r) = 0.85). Clinic Network B included six urgent care clinics totaling roughly 3,000 visits per week. The proportion of weekly ILI visits strongly correlated with the number positive influenza tests during October 2010 – October 2012 (r = 0.89). Five Network A clinics and one Network B clinic were incorporated into ILINet.

CONCLUSIONS: Identified ILI visits strongly correlated with influenza activity in the community, highlighting electronic ambulatory data as a potential data source for ILINet. The data are robust and EMR query is less resource-intensive for providers than manual reporting. Results may not be generalized to all networks. WA DOH had an immediate need to develop EMR query guidance for sentinel providers and found the established syndromic definition a suitable first approach. It is important to assess correlations prior to use of data for ILI surveillance.