OBJECTIVES:

The validity of EQ-5D-3L has been demonstrated in many skin conditions, notably psoriasis and psoriatic arthritis. We investigate the construct validity of EQ-5D-3L in NMSC.

METHODS:

Patients (n=92) referred to the Dermatology Department of a tertiary care centre in Ireland for the removal of NMSC were eligible for inclusion. Consenting patients completed two condition-specific non-preference-based measures (CSMs) previously validated in NMSC, the Skindex-16 and Skin Cancer Index (SCI), and the EQ-5D-3L prior to the removal procedure. Data analysis is conducted in ‘R’, using published scoring algorithms; the UK tariff is used for EQ-5D-3L values. Concurrent validity is evaluated using Spearman correlations. Known group analysis compares effect sizes (ES) and uses Wilcoxon rank-sum tests to assess differences within groups. Pre-specified groups were based on gender, age (under or over 50 years) and history of NSMC.

RESULTS:

There was poor correlation between the EQ-5D-3L utility score and the composite scores of Skindex-16 (ρ=-0.36) and SCI (ρ=0.29). By contrast correlations were strong between SCI and Skindex-16 composite scores (ρ=-0.65) and across most domains. Known group analysis failed to show statistically significant differences according to gender or age using any of the measures. ES were markedly different according to age (EQ-5D-3L ES = 0.4, Skindex-16 ES= 0.28 and SCI ES = 0.21). SCI detected a statistically significant higher quality of life (QOL) (p=0.01) in those without history of NMSC with a moderate ES (0.49) which neither Skindex-16 (ES=0.34, p=0.099) or EQ-5D-3L (ES=0.22, p=0.1) detected.

CONCLUSIONS:

In this cohort, correlation between EQ-5D-3L and CSMs is weak. ES between the known groups was mostly small by all measures, but EQ-5D failed to detect clinically meaningful and statistically significant differences in QOL in those with and without a previous history of NMSC. We conclude there is no evidence to support construct validity of EQ-5D-3L in NMSC.