Health-related quality-of-life (HRQoL) reporting by randomized cancer trials (RCT): Pitfalls for interpretation.

Abstract

6102 Background: Overall survival and HRQoL are well-recognized direct measures of patient benefit. While survival is an easily defined endpoint, the definition and measure of quality of life is less clear. Hence, the reporting of HRQoL may impact on result interpretation. Here, we evaluated the quality of reporting of HRQoL by RCT of anticancer therapy. Methods: We performed a systematic review of phase III RCT of cancer-directed therapy published in PubMed in the last 10 years, whose primary or secondary endpoint was HRQoL. Data on study characteristics were extracted and authors’ conclusions were independently analyzed and classified (positive, negative, neutral, not reported) by two blinded investigators. Articles were analyzed for HRQoL methodological quality, using a checklist previously published (Tannock et al. Ann Oncol 2004). Summary statistics were used to describe the results. Results: 347 of 922 trials were eligible. 136 (39%) cited HRQoL as a study endpoint. In such studies, the most common tumor site was Lung, 84% were superiority and 16% were non-inferiority RCT, and 52 (38%) trials were positive with respect to primary endpoint. 47 (34%) presented HRQoL results, 29 (21%) reported positive results, 30 (22%) clearly defined HRQoL as a study endpoint, 36 (26%) had a-priori hypothesis testing, and only 9 (7%) computed a sample size based on HRQoL hypothesis. Most studies reported negative conclusions on HRQoL (73; 53%), while 30 (22%) were considered positive, 17 (12%) neutral and in 66 (48%), no results were presented. There was a positive association between positive conclusions and authors’ declared financial conflict of interest with industry (p< .05). Despite the fact that 95 (70%) presented negative results on HRQoL, only 75 (55%) actually reported negative conclusions. Conclusions: The quality of reporting HRQoL data by RCT in oncology is poor, with the majority not presenting sufficient information for results interpretation. Also, authors' conclusions not always reflect HRQoL results. Improvement in conducting and reporting patient-reported outcomes are urgently needed to help clinicians to put such information in context and decide what is best for their patients.