Health-related quality of life in randomized controlled trials: a systematic review of prognostic significance

Date 22 October 2018
Event ESMO 2018 Congress
Session Poster Discussion session - Supportive and palliative care 2
Topics Psychosocial Aspects of Cancer
Presenter Justyna Mierzynska
Citation Annals of Oncology (2018) 29 (suppl_8): viii557-viii561. 10.1093/annonc/mdy296
Authors J. Mierzynska1, M. Pe1, C. Piccinin1, F. Martinelli1, C. Coens1, C. Gotay2, A.M.M. Eggermont3, M. Groenvold4, J. Reijneveld5, L. Dirven6, A. Bottomley1
  • 1Qol, EORTC - European Organisation for Research and Treatment of Cancer, 1200 - Brussels/BE
  • 2School Of Population And Public Health, University of British Columbia, Vancouver/CA
  • 3Oncology, Institut Gustave Roussy, 94800 - Villejuif/FR
  • 4Palliative Medicine, Bispebjerg Hospital, 2400 NV - Copenhagen/DK
  • 5Department Of Neurology, VU University Medical Center & Brain Tumor Center, Amsterdam/NL
  • 6Department Of Neurology, Leiden University Medical Center & Haaglanden Medical Center, Leiden/NL



A landmark systematic review (Gotay et al., 2008) found that baseline patient-reported physical functioning (PF) and global quality of life (QL) independently predicted overall survival (OS) in cancer patients. Moreover, the prognostic significance of PF was supported by a meta-analysis of 10,108 patients (Quinten et al., 2009). Despite these results, health-related quality of life (HRQOL) is rarely used as predictive or stratification factor in randomized clinical trials (RCTs) and practice. The prior review was updated by examining the extent to which previously reported and possibly other HRQOL domains show prognostic value across different cancer types. A methodological evaluation assessing the implementation of analysis methods was also undertaken.


A systematic review was conducted following the Cochrane methodology and including RCTs from 2006-present. Inclusion criteria were phase II, III or IV RCTs including at least one multivariate analysis examining the relationship between baseline HRQOL and OS while controlling for other clinical factors. Studies were reviewed and assessed by two independent raters using predefined criteria.


Forty-eight RCTs (n = 24,777) were included. These were mostly phase III studies (77%) across 13 cancer types. Studies of lung (21%) and head and neck (12%) cancers were the most prevalent. Cox proportional hazards models were most frequently used to assess the prognostic value of HRQOL (96%). In the majority of the RCTs (94%), at least one HRQOL domain was significantly associated with OS (p < .05) even after controlling for other clinical variables. PF (39%) and QL (35%) were commonly reported as independent prognostic factors in nine and eight types respectively. Methodological evaluation found that few studies followed rigorous methods.


Our results build upon previous findings confirming the prognostic significance of PF and QL and further highlight its prognostic value across the majority of disease types. Given this, we recommend that these scales be used to supplement available clinical staging and for stratification. A more consistent use of methods would allow for stronger conclusion regarding predictive accuracy.

Clinical trial identification

Legal entity responsible for the study



Fonds Cancer (FOCA).

Editorial Acknowledgement


A. Bottomley: Unrestricted education grants from Merck and Boehringer-Ingelheim. All other authors have declared no conflicts of interest.