Background rates of medical events of interest before and during the COVID-19 pandemic: a longitudinal cohort study using claims data.

IMPORTANCE

Background rates are critical for contextualising safety signals arising from COVID-19-related interventions in investigational or real-world settings.

OBJECTIVE

To estimate background rates of medical events of interest (MEI) for which COVID-19 infection and/or COVID-19 interventions may be risk factors in two US claims databases.

DESIGN, SETTING AND PARTICIPANTS: This retrospective cohort study spans the pre-COVID-19 (2018-2019) and COVID-19 (2020-2021) periods. We constructed three cohorts, in each of Inovalon/HealthVerity (Inovalon/HV) and Optum databases: a COVID-19-positive adult cohort (2020-2021), a paediatric cohort (2018-2021) and a high-risk cohort (2018-2021) comprising patients at increased risk for severe COVID-19. Participants were indexed on the day they first qualified to enter each cohort during the study period. Background rates of 17 MEI were estimated per 1000 person-years (PY) with 95% CIs.

MAIN OUTCOMES AND MEASURES

Annual incidence rates (IRs) of 17 MEI.

RESULTS

Overall, 758 414 (COVID-19-positive adults; 57.8% women), 12 513 664 (high-risk adults; 56.8% women) and 8 510 627 (paediatric patients; 49.1% women) patients were identified in the HV database. IRs of MEI varied substantially by year, data source, study cohort and duration of follow-up. The IRs of MEI were highest among COVID-19-positive adults and lowest among paediatric patients. For example, IR of myocarditis/pericarditis per 1000 PY was 3.0 (95% CI: 2.6 to 3.4) in the COVID-19-positive adult cohort vs 0.36 (95% CI: 0.34 to 0.37) among high-risk adults and 0.05 (95% CI: 0.05 to 0.06) among paediatric patients. In the COVID-19-positive adult cohort, we observed higher IRs during 90-day follow-up (eg, IR of acute myocardial infarction (AMI) 26.5 (95% CI: 25.3 to 27.7)) vs 365-day follow-up (eg, IR of AMI 20.0 (95% CI: 9.2 to 20.8)) and during 2020 compared with 2021. IRs were higher in the high-risk adult and paediatric populations during the pre-COVID-19 period than during the COVID-19 pandemic.

CONCLUSIONS

Substantial variability was observed in IRs of MEI by study cohort, year, data source and follow-up duration. When generating background rates for contextualising safety signals from COVID-19 interventions, careful consideration must be given to the indicated subpopulation of interest, COVID-19-related temporal variations and data sources.