(178) EHR Continuity in a Multicenter EHR Data Repository: Potential Implications for Studies Evaluating COVID-19 Therapy

Publication Number: 1223

Background: Studies using electronic health records (EHR) typically rely on available records to establish patient history and baseline comorbidities. However, since EHR are limited to care within contributing healthcare systems, completeness of baseline medical history varies by whether patients receive out-of-system care.

Objectives: Assess differences in baseline history among patients hospitalized for COVID-19 using a large multicenter EHR database.

Methods: Using the National COVID Cohort Collaborative (N3C) EHR from 69 institutions, we identified adult patients (≥ 18 yrs) hospitalized for COVID-19 between March 1, 2020 and September 1, 2021. We defined the duration of each patient’s baseline history as the number of months between the earliest recorded visit in the N3C EHR and the first COVID-related admission. We calculated standardized mean differences (SMD) to compare demographic factors, medications, and comorbidities recorded on admission and during baseline between those with no prior visits, those with 1-23 months of history, and those with 24+ months of history.

Results: Among 43,462 patients who were hospitalized for COVID-19, 12,523 (28.8%) had no prior visit in the N3C EHR, 9,784 (22.5%) had 1-23 months of history, and 21,155 (48.7%) had 24+ months of history. Compared to patients with 24+ months of history, patients with no prior visits were younger (59, IQR: [45, 71] vs 65, IQR: [52, 77]), more likely to be male (61% vs 47%), less likely to be white (45% vs 59%), less likely to have recorded diagnoses for comorbidities such as hypertension (45% vs 74%), diabetes (33% vs 45%), and chronic obstructive pulmonary disease (7.4% vs 18%). The largest SMDs were for hypertension (0.50) and prescriptions (statins: 0.83; ACE inhibitors: 0.55; angiotensin II receptor blockers: 0.46). Patients with no prior visits were also twice as likely to be ventilated on admission (19% vs 8.8%) and 27% more likely to experience inpatient mortality (14% vs 11%).

Conclusions: Despite being younger with fewer recorded risk-factors for severe COVID-19, patients with no prior visits in the N3C EHR were more likely to be ventilated on admission and experience inpatient mortality. These patients may represent a heterogeneous mixture of individuals who were previously healthy (truly without comorbidities), patients who received routine care elsewhere (comorbidities present, but data are missing due to health system fragmentation), and patients with unmet medical need (comorbidities present but not documented in any health system due to poor access to healthcare). Researchers will need to be aware of the potential information bias in estimated treatment effects among hospitalized patients that may be present if differences in EHR continuity are not properly accounted for in analyses.