Informative censoring due to missing data in quality of life was inadequately assessed in most oncology randomized controlled trials

Abstract

Objective We aimed to systematically characterize reporting missing quality of life (QoL) data in oncology randomized controlled trials (RCTs) and to estimate prevalence of adequate reporting according to existing guidelines. Study Design and Setting This cross-sectional analysis includes all articles on anti-cancer drugs tested in RCTs in six high impact medical/oncology journals, published between January 2015 and May 2020, that reported QoL outcomes. From 1942 identified articles, 215 (11%) met inclusion criteria. Data abstracted included whether compliance for QoL assessment were reported, whether results from a missing data statistical analysis were reported, whether articles met current recommendations for reporting missing data in QoL assessments. Results The results from a missing data statistical analysis were available in 22 trials (10.2%). Overall, 16 trials (7.4%) met current recommendations for reporting missing data in QoL assessments. Articles specifically reporting on QoL or patient reported outcomes were more likely to meet recommendations than other reports (P < 0.0001). Conclusion This systematic cross-sectional study found that most oncology RCTs reporting on QoL do not report adequately on missing data in QoL, with only 7.4% of trials meeting current reporting guidelines. The possibility of informative censoring, therefore, cannot be assessed in most of trials.

Keywords: Imputation; Informative censoring; Missing data; Oncology; Quality of life.