Systematic evaluation of patient-reported outcome (PRO) protocol content and reporting in cancer clinical trials: the EPiC study


Year:

Session type:

Theme:

Derek Kyte1,Ameeta Retzer1,Thomas Keeley2,Khaled Ahmed1,Jo Armes3,Julia M Brown4,Lynn Calman5,Chris Copland6,Fabio Efficace7,Anna Gavin8,Adam Glaser4,Diana M Greenfield9,Anne Lanceley10,Rachel M Taylor11,Galina Velikova4,Michael Brundage12,Rebecca Mercieca-Bebber13,Madeleine T King13,Melanie Calvert1
1University of Birmingham,2Parexel International,3King's College London,4University of Leeds,5University of Southampton,6NCRI Psychosocial Oncology and Survivorship CSG,7Italian Group for Adult Dematologic Diseases (GIMEMA),8Queen's University Belfast,9Sheffield Teaching Hospitals NHS Foundation Trust,10University College London,11University College Hospitals NHS Foundation Trust,12Queen's University Cancer Research Institute,13University of Sydney

Abstract

Background

Patient-reported outcomes (PROs), including quality of life (QoL), are commonly collected in cancer trials. However, emerging evidence suggests trial protocols often omit PRO-specific content, potentially impairing trial execution and reporting of PROs, limiting their potential to inform cancer care. This study aimed to determine if these early findings are generalisable to cancer trials on the National Institute for Health Research (NIHR) portfolio. 

Method

NIHR portfolio randomised controlled cancer trials (all cancer specialities/ages) collecting a primary/secondary PRO (completed between 2001-2014) were included. Two independent investigators retrieved the most up-to-date trial protocol (final ethically approved version) and arising publications for review, extracted demographic trial data, and reviewed: (i) the general quality/completeness of protocols/publications using the SPIRIT and CONSORT checklists respectively; and (ii) the PRO-specific aspects using a PRO protocol checklist and the CONSORT-PRO Extension.

Results

n=251 trials were included and n=106 protocols and n=157 publications were sourced. Preliminary data based on n=79 protocols matched to their corresponding publication(s) suggests general protocol quality/completeness was above average (adjusted mean SPIRIT score (max 50)=30.96 (64.65%)), but PRO-specific content was frequently omitted (adjusted mean PRO protocol checklist score (max 33)=9.96 (31.53)). A similar picture was seen across the included trial publications (adjusted mean CONSORT score (max 37)=23.43 (66.17%); adjusted mean CONSORT PRO score (max 14)=3.14 (23.92%)). Worryingly, n=17 (21.52%) trials failed to report PRO results in either a primary or secondary publication, meaning that 14,989 participants provided QoL data that was not made available for future patients.

Conclusion

The PRO-specific content of cancer trial protocols/publications is sub-optimal, despite general aspects being of good quality. Potential non-reporting and poor reporting of PRO cancer trial results devalues the contribution of large numbers of trial participants and, crucially, means that future patients may not have access to PRO data to inform important treatment decisions.