Murtuza Bharmal1, Matthias Hunger2, Michael Schlichting3. 1. Global Evidence & Value Development, EMD Serono, Merck KGaA, Darmstadt, Germany. Electronic address: murtuza.bharmal@emdserono.com. 2. Commercialisation & Outcomes, ICON plc, Munich, Germany. 3. Global Biostatistics and Epidemiology, Merck Healthcare KGaA, Darmstadt, Germany.
Abstract
BACKGROUND: Currently, there are 4 EQ-5D-5L scoring algorithms available for the United Kingdom that can be used to derive a utility score. OBJECTIVES: To perform a psychometric validation of the EQ-5D utility score in patients with metastatic Merkel cell carcinoma (mMCC) and to compare 4 EQ-5D-5L value sets currently available for the United Kingdom. METHODS: Data collected in the phase 2 trial JAVELIN Merkel 200 were analyzed, in which 88 patients with mMCC were treated with avelumab. Utility scores were calculated using the EQ-5D-5L valuation set by Devlin et al and 3 further algorithms that are based on EQ-5D-3L valuations. Criterion validity was assessed by Pearson correlations between utility and other HRQoL scales. Responsiveness was assessed by correlating change in utility scores with change in HRQoL scales and percentage change in tumor size. Effect sizes for change from baseline in utility were calculated in subgroups of patients with improving and deteriorating health based on clinically relevant changes in various external anchors. RESULTS: Devlin utilities were up to 0.10 points higher than utilities calculated from EQ-5D-3L valuations. Large correlations (0.45-0.72; P < .01) between utility and the EuroQol Visual Analogue Scale and FACT-G total, physical, and functional scores were found for all 4 algorithms. Large correlations between these measures were also found for change from baseline scores. In patients with health improvements, effect sizes were larger for the 3 EQ-5D-3L-based algorithms, whereas the Devlin utilities had larger effect sizes for health deteriorations, probably because of different levels of baseline utility. CONCLUSIONS: The English National Institute for Health and Care Excellence currently recommends use of the 3L valuation sets. Our analyses show that all 4 scoring algorithms have good and similar psychometric properties in patients with mMCC.
BACKGROUND: Currently, there are 4 EQ-5D-5L scoring algorithms available for the United Kingdom that can be used to derive a utility score. OBJECTIVES: To perform a psychometric validation of the EQ-5D utility score in patients with metastatic Merkel cell carcinoma (mMCC) and to compare 4 EQ-5D-5L value sets currently available for the United Kingdom. METHODS: Data collected in the phase 2 trial JAVELIN Merkel 200 were analyzed, in which 88 patients with mMCC were treated with avelumab. Utility scores were calculated using the EQ-5D-5L valuation set by Devlin et al and 3 further algorithms that are based on EQ-5D-3L valuations. Criterion validity was assessed by Pearson correlations between utility and other HRQoL scales. Responsiveness was assessed by correlating change in utility scores with change in HRQoL scales and percentage change in tumor size. Effect sizes for change from baseline in utility were calculated in subgroups of patients with improving and deteriorating health based on clinically relevant changes in various external anchors. RESULTS: Devlin utilities were up to 0.10 points higher than utilities calculated from EQ-5D-3L valuations. Large correlations (0.45-0.72; P < .01) between utility and the EuroQol Visual Analogue Scale and FACT-G total, physical, and functional scores were found for all 4 algorithms. Large correlations between these measures were also found for change from baseline scores. In patients with health improvements, effect sizes were larger for the 3 EQ-5D-3L-based algorithms, whereas the Devlin utilities had larger effect sizes for health deteriorations, probably because of different levels of baseline utility. CONCLUSIONS: The English National Institute for Health and Care Excellence currently recommends use of the 3L valuation sets. Our analyses show that all 4 scoring algorithms have good and similar psychometric properties in patients with mMCC.