Evaluation of COVID-19 Diagnosis Codes for Identification of SARS-CoV-2 Infections in a Nursing Home Cohort, 2022-2023.

OBJECTIVES

This study aimed to evaluate the utility of electronic health record (EHR) diagnosis codes for monitoring SARS-CoV-2 infections among nursing home residents.

DESIGN

A retrospective cohort study design was used to analyze data collected from nursing homes operating under the tradename Signature Healthcare between January 2022 and June 2023.

SETTING AND PARTICIPANTS

Data from 31,136 nursing home residents across 76 facilities in Kentucky, Tennessee, Indiana, Ohio, North Carolina, Georgia, Alabama, and Virginia were included.

METHODS

Resident demographics, diagnosis codes associated with clinical diagnoses (including COVID-19), and SARS-CoV-2 testing information were collected from the EHR and supplemental testing data sources. We described the rates of infection and the clinical characteristics of residents with incident-positive SARS-CoV-2 tests and new-onset COVID-19 diagnoses. Positive predictive values (PPVs) of COVID-19 diagnosis codes were calculated for residents stratified by whether a resident was continuously present in a facility for ±3 days from the diagnosis onset date listed in EHRs, using positive SARS-CoV-2 tests to confirm infection.

RESULTS

A total of 4876 incident-positive SARS-CoV-2 tests and 6346 new-onset COVID-19 diagnoses were recorded during the study period. Weekly rates of new-onset diagnoses were significantly higher than positive test rates, although trends followed similar trajectories. Among residents continuously present in the nursing home ±3 days from the diagnosis onset date, the PPV of COVID-19 diagnosis codes was high (3395 of 3685 = 92%; 95% CI, 91%-93%). The PPV among this group significantly varied by study quarter (P < .001). The PPV was substantially lower for 2661 diagnoses among residents not continuously present in the nursing home (24%; 95% CI, 22%-26%).

CONCLUSIONS AND IMPLICATIONS

This study demonstrates the utility of diagnosis codes for assessment of COVID-19 epidemiology and trends when testing data are unavailable for residents during their stay in a nursing home. Future research should explore strategies to evaluate the utility of diagnosis codes at admission and discharge to nursing homes to enhance surveillance efforts.