OBJECTIVES

In real-world oncology research, time to treatment discontinuation (TTD), time to next treatment (TTNT) and clinician-reported progression-free survival (rwPFS) are commonly used to assess systemic therapy benefits. As these endpoints have distinct definitions, it is important to understand how different methods of measurement influence assessment of clinical outcomes. The aim of this study was to evaluate concordance of TTD, TTNT and rwPFS.

METHODS

Seven retrospective studies evaluated rwPFS with TTD and/or TTNT using Kaplan-Meier methods among patients with solid tumors who received treatment in the US Oncology Network (USON). For all studies, TTD was the duration between the start and stop dates of a single line of therapy (LOT); TTNT was the duration between the start of one LOT and the start of the next LOT; and rwPFS was the duration between the start of one LOT and clinician-reported date of progression or death. For these time-to-event endpoints, patients who did not experience the event were censored. Study data were captured from the electronic heath record system of the USON, including both structured data (which is programmatically extracted relatively quickly) and unstructured data (requiring manual chart review). Treatment start/stop dates and dates of death were sourced from structured and unstructured data, while dates of progression were only available in unstructured data.

RESULTS

Across all studies, median TTD durations were shorter than median rwPFS and TTNT durations. Results based on structured data only were similar to those yielded from those supplemented by unstructured for both the TTD and TTNT endpoints (within 95% confidence intervals). Median TTNT and rwPFS were most closely aligned, particularly when data were sourced from unstructured fields.

CONCLUSIONS

Based on these results, TTD and TTNT appear to be reliably captured with either structured or unstructured data. To estimate rwPFS without unstructured data, TTNT is a consideration.