Reducing Interval Censoring in Time-to-event Analyses of Functional Recovery After Radical Prostatectomy Using Patient-estimated Moment of Recovery.

Abstract

Background and objective: Functional recovery after radical prostatectomy happens continually but is only measured at set times when questionnaires are given. This leads to what is known as interval censoring. In the urologic literature, patients are typically considered to have recovered function at the moment they complete a follow-up survey, although recovery may have occurred earlier, leading to systematic overestimation of recovery time. At our institution, patients completing function assessments are additionally asked to estimate when recovery occurred, and this estimate is incorporated into the analysis. We aimed to compare our approach with the standard method, which assigns recovery to the survey completion date.

Methods: We retrospectively identified 11 161 men who underwent radical prostatectomy between 2009 and 2025. Patients completed urinary and erectile function questionnaires at regular intervals after surgery; at the first response meeting recovery criteria (eg, pads no longer used), patients estimated the timing of recovery using predefined intervals. We used Kaplan-Meier analyses to estimate cumulative recovery over time, using both the survey completion date and the patient-reported estimate.

Key findings and limitations: We identified 7896 patients who recovered continence and 6521 who recovered erectile function. Patient-estimated recovery time was considerably earlier than recovery time based on the survey completion date. Median time to continence recovery was 5 mo using our approach versus 8 mo using the survey completion date approach, whereas median time to erectile function recovery was 18 mo versus 23 mo. At 6 mo, recovery rates were higher when based on patient-estimated dates compared with survey completion dates for both continence (57% vs 40%) and erectile function (35% vs 28%).

Conclusions and clinical implications: Functional questionnaires should include an additional item asking patients to report when recovery occurred. The reported time of recovery should be used in the analysis rather than the survey completion date.