Data Sources for Clinical T1 Renal Masses and the Potential for Bias.

INTRODUCTION

The rising incidence of early-stage kidney cancer has driven comparative effectiveness research using cancer registry and administrative data. These sources may be biased for small renal masses (SRM), where histologic confirmation prior to treatment is not standard. To better understand these limitations, we compared characteristics of patient populations across three SRM data sources.

PATIENTS AND

METHODS

We identified patients diagnosed with clinical T1 renal masses from 2019 to 2020 at our institution. Data were obtained from the institutional cancer registry, a prospective clinical trial, and electronic health record (EHR) extraction. Demographic and clinical characteristics were compared across cohorts using chi-squared, Fisher's exact testing, and multivariable regression analysis. Radiologic reports were reviewed for terminology concordance with registry inclusion criteria.

RESULTS

Among 555 cases, 169 (30.5%) were in the clinical trial, 273 (49.2%) in the cancer registry only, and 113 (20.4%) from EHR extraction only. Active surveillance was more common in the EHR extraction cohort (85%) than in the registry (48%) or trial (33%) (P < .001). The registry and trial cohorts had higher Charlson Comorbidity Index scores (P < .001), and the clinical trial cohort included fewer Hispanic/Latino patients (P = .04) and non-English speakers (P < .01). Registry capture was limited by terminology in radiologic reports, with qualifying terms present in only 15% of EHR cases and 49% of delayed registry entries.

CONCLUSIONS

Cohort composition differed by data source, particularly for patients undergoing active surveillance. Clinicians must recognize potential bias when interpreting findings. Standardized radiologic reporting and revised registry criteria may ensure more complete data for early-stage kidney cancer.