Using a statewide health information exchange to enhance COVID research and surveillance: The impact of information fragmentation.

Objective: Use of electronic health data for public health surveillance and research has become almost ubiquitous. However, since many patients routinely receive care from multiple healthcare systems, not all information for a patient may be available in a single system. Thus, using data from only a single healthcare system may unintentionally result in inaccuracies. This study compared the variation in the number of inpatient and emergency department encounters, diagnoses, and laboratory results, as well as COVID prevalence, when data about patients was (1)drawn from a single, major health system; (2)augmented with health information exchange (HIE) data; or (3)retrieved entirely from the HIE.

Materials And Methods: This study leveraged a registry of 1,343,703 COVID-positive patients covering the state of Indiana containing data from March 6, 2020 - February 14, 2022. The registry included data from the state-wide health information exchange as well as granular data from an individual health system. Data elements included encounters, diagnoses, laboratory testing, and death records.

Results: Enhancing the cohort with data from the HIE increased the average number of encounters per person from 20.25 to 37.05, unique diagnoses from 16.06 to 30.83, and laboratory results from 0.32 to 7.06. The absolute difference between the single-system and HIE cohorts was 22.5% for the estimated prevalence of COVID, 4% for the number of all-cause hospitalizations and -9.4% for emergency department visits.

Conclusion: Information fragmentation can lead to an incomplete picture, or even errors, in research or clinical settings. This study demonstrates that combining clinical data sources creates variation in results. Understanding the issues presented by information fragmentation, such as bias it may create, is critical to enabling accurate disease surveillance and crisis response planning.