DATA QUALITY CONSIDERATIONS FOR DETERMINING THE IMPACT OF THE COVID-19 PANDEMIC ON OUTPATIENT VISITS AND PROCEDURES IN THE UNITED STATES

Objectives: To use real-world data to better understand the temporal impact of COVID-19 on frequency of outpatient visits in urology clinics and apply a data quality framework to identify and resolve issues that could influence the validity of analyses. Methods: We examined visit trends from 157 outpatient urologic practices and 3,165 providers included in the American Urological Association Quality (AQUA) Registry, a national Qualified Clinical Data Registry designed to measure, report and improve healthcare quality and patient outcomes. As of December 2020, 50 million patient visits from 7.3 million unique patients exist in the database. We used a data quality framework to assess issues related to data completeness and timeliness. Past data lag trends were assessed to determine a temporal cutoff point that minimized lag. To account for missing data, a comprehensive visit definition that went beyond E/M (Evaluation and Management) coding was applied and compared to results using the E/M-only definition. Results: Correlating with the onset of the COVID-19 pandemic in the US in mid-March 2020, outpatient visits declined for five weeks, resulting in a 47.7% drop from the February 2020 baseline, followed by an 83.1% recovery back to baseline by July 31. Based on timeliness scores, which consider the average time between visit date and the visit data processing date from the selected practices, a 90-day buffer between data cutoff and the end of the analysis’s observation period was necessary to minimize data lag. Total patient encounters increased by more than 20% when using a more comprehensive visit definition Conclusions: The COVID-19 pandemic contributed to a reduction in outpatient visits, followed by a recovery close to baseline. Our data quality framework identified and informed key study design decisions to account for inadequate data timeliness and completeness. These methods prevented an overestimation of the true decline in patient encounters and missed patient encounters.