Assessing quality of life in palliative care settings: head-to-head comparison of four patient-reported outcome measures (EORTC QLQ-C15-PAL, FACT-Pal, FACT-Pal-14, FACT-G7).

PURPOSE

Head-to-head comparison of reliability, validity and responsiveness of four patient-reported outcome measures (PROMS) suitable for assessing health-related quality of life (HRQOL) in palliative care settings: EORTC QLQ-C15-PAL, FACT-G7, FACIT-Pal and short-form FACIT-Pal-14.

METHODS

Secondary analysis of two phase III randomised trials: ketamine for chronic cancer pain, octreotide for vomiting in inoperable malignant bowel obstruction. Sub-groups were defined by Australia-modified Karnofsky performance status (AKPS) and participants' global impression of change (GIC). Two aspects of reliability were assessed: internal consistency (Cronbach alpha, α); test-retest reliability (intra-class correlation coefficient (ICC)) of patients with stable AKPS and those who self-reported 'no change' on GIC. Construct validity was assessed via pre-determined hypotheses about sensitivity of PROM scores to AKPS groups and responsiveness of PROM change scores to GIC groups using analysis of variance.

RESULTS

FACIT-Pal had better internal consistency (α ranged 0.59-0.80, 15/18 ≥ 0.70) than QLQ-C15-PAL (0.51-0.85, 4/8 ≥ 0.70) and FACT-G7 (0.54-0.64, 0/2 ≥ 0.70). FACIT scales had better test-retest reliability (FACIT-Pal 11/27 ICCs ≥ 0.70, FACT-G7 2/3 ICCs ≥ 0.70) than QLQ-C15-PAL (2/30 ICCs ≥ 0.70, 18/30 ≤ 0.5). Four scales demonstrated sensitivity to AKPS: QLQ-PAL-15 Physical Functioning and Global QOL, FACT-G Functional Wellbeing and FACIT-Pal Trial Outcome Index (TOI). Nine scales demonstrated responsiveness: three in the ketamine trial population (QLQ-C15-PAL Pain, FACIT-Pal-14, FACT-G7), six in the octreotide trial population (QLQ-C15-PAL Fatigue; FACIT-Pal PalCare, TOI, Total; FACT-G Physical Wellbeing and Total).

CONCLUSIONS

No PROM was clearly superior, confirming that choosing the best PROM requires careful consideration of the research goals, patient population and the domains of HRQOL targeted by the intervention being investigated.