Meta-analysis of preference-based quality of life values in heart failure

Objectives: A large number of studies measuring quality of life (QoL) in heart failure (HF) patients is available in the literature. However, only the values measured with the preference-based instruments (e.g. EQ-5D UK and US “tariff” version, time trade-off (TTO), standard gamble (SG)) can be directly applied in a cost-utility analysis (CUA). Our aim was to summarize instrument-specific values in HF while accounting for study-level covariates and for the correlation between multiple values both within and between studies. Methods: The instrument-specific values in HF were identified through a systematic search. Study-level covariates recorded from the eligible studies were: disease-severity level, publication year and mean population age. Finally, a multivariate meta-regression model was used to estimate pooled, instrument-specific values in HF. Results: Our systematic search identified 22 studies reporting 28 values in HF. Instrument-specific estimates pooled on the base-case dataset were: 0.6385 (EQ-5D UK “tariff”), 0.7593 (EQ-5D US “tariff”), 0.6382 (SF-6D), 0.6534 (SG) and 0.7287 (TTO). Given that the EQ-5D (UK “tariff”) was the most commonly applied instrument for measuring preference-based values among the included studies, the impact of study-level covariates was examined on those instrument-specific values. Age and diseaseseverity level were the study-level covariates that showed a significant impact on pooling QoL estimates. Yet, meta-regression modelling also identified a large amount of unexplained heterogeneity remaining. Conclusions: This is the first meta-analysis of QoL values in HF. Given the abundance of QoL measurements in HF and the requirement for applying single, accurate QoL values in a CUA, pooled estimates could be highly applicable in a CUA