Meta-analysis provides evidence-based effect sizes for a cancer-specific quality-of-life questionnaire, the FACT-G

King, Madeleine T.; Stockler, Martin R.; Cella, David F.; Osoba, David; Eton, David T.; Thompson, Joanna; Eisenstein, Amy R.

doi:10.1016/j.jclinepi.2009.05.001

« Previous Next »Journal of Clinical Epidemiology
Volume 63, Issue 3 , Pages 270-281, March 2010

Meta-analysis provides evidence-based effect sizes for a cancer-specific quality-of-life questionnaire, the FACT-G

Madeleine T. King
- Quality of Life Office, Psycho-oncology Co-operative Research Group, School of Psychology, University of Sydney, Sydney, NSW 2006, Australia
- Centre for Health Economics Research and Evaluation, University of Technology, Sydney, NSW, Australia
- Corresponding author. Quality of Life Office, Psycho-oncology Co-operative Research Group, School of Psychology, University of Sydney, Brennan MacCallum Building (A18), Sydney, NSW 2006, Australia. Tel.: +61-2-9036-6114; fax: +61-2-9036-5292.
Martin R. Stockler
- Sydney Cancer Centre, Royal Prince Alfred & Concord Hospitals, University of Sydney, NSW, Australia
- NHMRC Clinical Trials Centre, University of Sydney, NSW, Australia
David F. Cella
- Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, USA
David Osoba
- QOL Consulting, Vancouver, BC, Canada
David T. Eton
- Mayo Clinic, Department of Health Sciences Research, Rochester, MN, USA
Joanna Thompson
- Centre for Health Economics Research and Evaluation, University of Technology, Sydney, NSW, Australia
Amy R. Eisenstein
- Center for Research on Health and Aging, University of Illinois, Chicago, IL, USA

Accepted 11 May 2009. published online 28 August 2009.

Abstract

Objective

To compare Cohen's guidelines for small (0.2), medium (0.5), and large (0.8) effect sizes with empirical estimates for a cancer-specific health-related quality-of-life questionnaire (HRQOL), the Functional Assessment of Cancer Therapy - General (FACT-G).

Methods

Seventy-one papers satisfied inclusion criteria for meta-analysis. Blinded to the HRQOL results, three “experts” (with expertise in interpreting the FACT-G questionnaire and managing cancer patients), predicted the relative magnitude of HRQOL mean differences. Size classes (small, medium, large) were defined in terms of relevance to clinical decision making. The experts worked independently and based their predictions on patient characteristics and clinical circumstances. Their judgments were linked with FACT-G results and inverse-variance–weighted mean effect sizes calculated for each size class.

Results

At least two experts were perfectly concordant and up to one was discordant by at most one size category for 833 of the mean differences; for these, weighted kappas were generally in the “substantial” range (0.60–0.79). Of these mean differences, 617 were cross-sectional; small, medium, and large mean effect sizes were physical well-being 0.42, 0.87, 1.6; functional well-being 0.37, 0.71, 1.6; emotional well-being 0.32, 0.40, no large differences; and social well-being 0.14, 0.23, no large differences. Two hundred and sixteen longitudinal mean differences yielded small and medium effect sizes: physical well-being 0.26, 0.34; functional well-being 0.14, 0.28; emotional well-being 0.27, 0.23; and social well-being 0.08, 0.01. There was virtually no evidence for large longitudinal effects.

Conclusion

These results provide specific, evidence-based alternatives to Cohen's generic guidelines, for use in sample-size calculations for the FACT-G and interpretation of the clinical significance of effects measured with FACT-G.

Keywords: Health-related quality of life, Meta-analysis, Interpretation guidelines, FACT-G, Cancer, Patient-reported outcomes, Minimum clinically important difference, MCID, Minimum important difference, MID