Background Rheumatoid arthritis (RA) deeply affects the quality of life (QoL) of patients. The preferred approach to evaluate treatment efficiency is to value health as patient preferences known as utilities, and subsequently, calculate Quality-Adjusted Life Years gained. A new 5-level of severity EQ5D has recently released and a new tariff proposed for Spain (Ramos-Goñi,2016). Although QoL questionnaires are not of routine use in clinical practice, it is possible to estimate it using the Health Assessment Questionnaire Disability Index (HAQ-DI)

Objectives To develop a function that allows the estimation of EQ5D-5L utility values from HAQ-DI updated to the newest proposed tariff for Spain

Methods Patients with RA from two teaching hospitals, participating in a prospective observational study completed the HAQ-DI and EQ5D-5L at 0–6-12 month follow-up visits. Inclusion criteria: ACR RA diagnosed patients, on biologic treatment and whose disease activity remained stable at least for 3 months EQ5D-5L is a standardized, generic instrument for describing and valuing health and QoL, consisting in a five-dimensional descriptive system (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) and a visual analogue scale. A country-specific tariff converts patient's answer to a 0–1 (full health) utility index. HAQ-DI is a self-completed questionnaire used to assess the functional ability using 20 items, distributed across 8 dimensions and resulting in a four-level disability scale (0–3). In addition, socio-demographic and clinical data where recorded.To estimate the EQ5D-5L utility index OLS models were built. As this index is bounded to the [-0.416, 1] interval, Tobit models were also considered. Hereafter, the index was transformed to the open interval (0,1) and estimated through beta regression with a logit link. To determine the relationship grade between the index and the HAQ-DI scale and obtain residuals without trend, GAM models were used. Best fitting models were determined by AIC, MAE and RMSE. All analyses were performed using R software

Results 217 questionnaires fulfilled by 77 patients. Mean (SD) age was 57.0 (12.9), 87% women, AR duration 13.7 (7.1), mean DAS28 2.72 (1.00) and HAQ-DI 0.77 (0.60). Baseline EQ5D index: 0.768 (0.182). All the OLS estimation models resulted in the interval limits defined by the index, so Tobit models were not considered. When considering the linear model we obtained the best results with the HAQ-DI term and its third power: EQ5D5L = 0.9232 − 0.1760×HAQ − 0.0172×HAQ³ (AIC=−221.62; MAE=0.0974; RMSE=0.1363); for beta regression, we obtained the best model with the HAQ-DI to the first power alone: logit (EQ_01) = 2.5821 − 1.1165×HAQ (AIC=−444.4; MAE=0.0691; RMSE=0.0958). Considering the AIC and the residuals together, we obtained the best fitting model with the beta regression approach, with neither age nor sex

Conclusions So far, only a utility function using HAQ-DI and an older EQ5D-3L version was available for Spain (Carreño,2011). This updated utility function can be used as a practical approach to predict RA patients' QoL and EQ5D utility score for Spain when clinicians/researchers need them for clinical practice or cost-effectiveness analyses and generic QoL measurements are not available

Disclosure of Interest None declared