PhD Candidate Columbia University NEW YORK, New York, United States
Background: Surveillance, a cornerstone of disease control and prevention, may be strengthened by leveraging data-rich sources such as health information exchanges (HIEs), platforms that share electronic health record (EHR) data across multiple health systems. HIEs offer robust, near real-time examination of clinical information, including testing patterns not typically available in public health case reporting. We sought to establish an HIE-based data pipeline for regional surveillance of sexually transmitted infections (STIs).
Methods: Using de-identified data from Healthix, we conducted a retrospective observational study of Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (GC) testing and incident case diagnoses in New York City (NYC) between 01/01/2018 and 12/31/2021. Healthix is the largest public HIE in the United States, serving more than 24 million patients and 8,000+ healthcare facilities. Data from different health systems and EHR platforms were standardized using the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). We used generalized estimating equations with a logit link function to estimate the likelihood of a patient receiving a test for an STI and for receiving a positive test result, adjusting for patient sex, race, ethnicity, and age.
Results: Data conversion using the OMOP CDM supported a robust, scalable data pipeline. The HIE dataset comprised 4.7 million patients and contained 2.4 million CT tests and 2.5 million GC tests; the proportion of positive results were 2.9% (69,240 cases) and 1.1% (28,581 cases), respectively. We identified population-level trends concerning testing and positive cases. For example, men were half as likely to test for CT or GC compared to women, but two to six times more likely to test positive (Table 1). In contrast, patients who were Hispanic/Latinx were more likely to test for GC compared to non-Hispanic white patients (adjusted odds ratio [aOR]: 2.48, 95%CI: 2.47-2.50), but were less likely to test positive for GC (aOR: 0.80, 95%CI: 0.71-0.90).
Conclusions: HIEs show promise for STI surveillance when data is standardized across EHRs. HIE-based surveillance suggests potential incongruence between testing patterns and STI risk in certain populations. Future work incorporating patient histories and other HIE data may offer further insight.
