Examining the Relationship Between Novel Data from Electronic Health Records (EHRs) and Traditional Public Health Surveillance Data for Influenza-like Illness among 12 U.S. Jurisdictions, 2016-2019. (March 2022)

• Record Type: Journal Article
• Authors: Palekar, R.; Schill, T.; Aldin, G.; Hoeksema, L.; Gildersleeve, R.; Guan, S.; Teixeira, C.; Kanter, A.; Luginbuhl, R.
• Is Part Of: International journal of infectious diseases. Volume 116(2022)Supplement
• Page Start: S101
• Publication Date: 2022-03
• Subjects: Communicable diseases -- Periodicals; Communicable Diseases -- Periodicals; Communicable diseases; Periodicals; Electronic journals; 616.9
• DOI: 10.1016/j.ijid.2021.12.238
• Languages: English
• ISSNs: 1201-9712

Abstract

Abstract : Purpose: Timely, geographically representative, syndromic surveillance is important for early threat detection. In the United States, the ambulatory-based influenza-like illness (ILI) surveillance system, ILINet, relies on data reported to public health authorities up to 10 days after clinical care is sought, making it a lagging indicator. Intelligent Medical Objects (IMO) has a search and select (SS) tool that is deployed within electronic health records (EHRs) to facilitate clinician searches using highly specific, clinically friendly terminology. Data from this SS tool are available in near real-time, throughout the United States, and at the ZIP code level. We examined the correlation between the novel SS data and the data from the U.S. Centers for Disease Control and Prevention's (CDC) ILINet, among 12 of the most populous U.S. jurisdictions. Methods & Materials: We mapped IMO influenza-related terms to SNOMED CT and International Classification of Diseases Tenth Revision (ICD-10) codes. We then queried the SS data for the proportion of influenza-related IMO terms, among all searches, during each epidemiologic week (EW) of the study period (2016-2019) by state. For ILINet data, we utilized the proportion of ILI visits, among all visits, each EW (2016-2019), by state, which are publicly available from CDC. We calculated Pearson's correlation coefficient between the two data streams overall, as well as, a rolling average over 52 weeks. Results: For the states of California, Georgia, Illinois, Massachusetts, Michigan, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Texas, and Virginia, which represent ∼53% of the U.S. population, the Pearson's correlation coefficient between ILI from ILINet and influenza SS, was >0. 83 for each state during the study period (0.90;0.96;0.86;0.90;0.87;0.92;0.84;0.93;0.94; 0.95;0.94, respectively). Among these 12 states, the correlation remained >0.70 for each EW during the study period. Conclusion: We found high correlations between ILINet data and SS influenza data across 12 U.S. jurisdictions. Given the timeliness and geographic representativeness of the SS data, the SS data might be able to provide additional information to support real-time public health decision-making, beyond what is available today. Future analyses should examine the ability to predict current and future ILINet data using SS data.