COMPARISON OF EQ-5D AND SF-6D BASED COST-EFFECTIVENESS ANALYSIS OF HAEMODIALYSIS AND PERITONEAL DIALYSIS TREATMENTS
This study aimed to investigate whether the multi-attribute EuroQol-5D (EQ-5D) and Short Form-6D (SF-6D) utility instruments would lead to consistent cost-effectiveness outcomes when they are used to evaluate dialysis treatments for patients with end-stage renal disease (ESRD).
A Markov model was constructed to compare haemodialysis (HD) and peritoneal dialysis (PD) for Singaporean ESRD patients with and without type 2 diabetes. Probabilistic sensitivity analysis (PSA) was used to determine cost-effectiveness by comparing the incremental cost-effectiveness ratio (ICER) based on costs incurred (in 2015 Singaporean dollars [S$]) and the quality-adjusted life years (QALYs) gained over a 10-year time horizon with a pre-determined maximum willingness-to-pay of S$60,000 per QALY. Cost and clinical inputs were estimated using local data and utility inputs were from Singaporean dialysis patients (HD: n=75; PD: n=75) interviewed using the 5-level EQ-5D (EQ-5D-5L) and Short Form-12 (SF-12) questionnaires.
EQ-5D scores were calculated using available EQ-5D-5L value sets (Canada, UK, and Japan), selected 3-level EQ-5D (EQ-5D-3L) value sets via crosswalk (Singapore, UK, Japan, and Thailand), and a crosswalk algorithm from SF-12 responses to UK EQ-5D-3L values. SF-6D scores were derived from SF-12 using a recommended algorithm.
For non-diabetic ESRD patients, the ICER of HD compared to PD was S$83,602 using SF-6D and ranged from 58,158 to 91,478 using EQ-5D scores. PSA showed that PD was more likely to be cost-effective using SF-6D and EQ-5D scores generated from SF-12, while HD was more likely to be preferable using EQ-5D scores generated from EQ-5D-5L.
For diabetic ESRD patients, the ICER was 82,365 using SF-6D and ranged from 67,984 to 96,110 using EQ-5D scores. PSA showed that two options were equally cost-effective using the Canada and UK EQ-5D-5L scores, while PD was more likely to be optimal using other scores.
We demonstrated that the choice of the EQ-5D and SF-6D instruments may affect the outcome of cost-effectiveness analysis, and so may be the case with the use of EQ-5D values from different sources or generated using different methods.