Interpreting small treatment differences from quality of life data in cancer trials: an alternative measure of treatment benefit and effect size for the EORTC-QLQ-C30

Abstract

Background: The EORTC-QLQ-C30 is a widely used health related quality of life (HRQoL) questionnaire in lung cancer patients. Small HRQoL treatment effects are often reported as mean differences (MDs) between treatments, which are rarely justified or understood by patients and clinicians. An alternative approach using odds ratios (OR) for reporting effects is proposed. This may offer advantages including facilitating alignment between patient and clinician understanding of HRQoL effects.

Methods: Data from six CRUK sponsored randomized controlled lung cancer trials (2 small cell and 4 in non-small cell, in 2909 patients) were used to HRQoL effects. Results from Beta-Binomial (BB) standard mixed effects were compared. Preferences for ORs vs MDs were determined and Time to Deterioration (TD) was also compared.

Results: HRQoL effects using ORs offered coherent interpretations: MDs >0 resulted in ORs >1 and vice versa; effect sizes were classified as 'Trivial' if the OR was between 1 ± 0.05 (i.e. 0.95 to 1.05); 'Small': for 1 ± 0.1; 'Medium': 1 ± 0.2 and 'Large': OR <0.8 or >1.20. Small HRQoL effects on the MD scale may translate to important treatment differences on the OR scale: for example, a worsening in symptoms (MD) by 2.6 points (p = 0.1314) would be a 17 % deterioration (p < 0.0001) with an OR. Hence important differences may be missed with MD; conversely, small ORs are unlikely to yield large MDs because methods based on OR model skewed data well. Initial evidence also suggests oncologists prefer ORs over MDs since interpretation is similar to hazard ratios.

Conclusion: Reporting HRQoL benefits as MDs can be misleading. Estimates of HRQoL treatment effects in terms of ORs are preferred over MDs. Future analysis of QLQ-C30 and other HRQoL measures should consider reporting HRQoL treatment effects as ORs.

Fig. 1

Distribution of QLQ-C30 responses: TOPICAL (x-axis is QLQ-C30 score on a scale of 0 to 1 and - y axis is relative frequency)

Fig. 2

Cumulative Frequency Plot of Effect Sizes for MDs and ORs. Horizontal reference lines are MDs effect sizes of 3, 10 and 15 points; circles refer to ORs and squares refer to MDs