Mapping From the Health Assessment Questionnaire (HAQ) to the EQ-5D-5L Questionnaire in Patients With Rheumatoid Arthritis

Speaker(s)

Bilbao A1, Garmendia L2, Ramallo Y3, Gorostiza I4
1Osakidetza Basque Health Service - Basurto University Hospital, Research and Innovation Unit; Network for Research on Chronicity, Primary Care, and Health Promotion (RICAPPS); Biosistemak Institute for Health System Research, Barakaldo, BI, Spain, 2Basque Center for Applied Mathematics, BCAM, Bilbao, Bizkaia, Spain, 3Canary Islands Health Research Institute Foundation; Network for Research on Chronicity, Primary Care, and Health Promotion (RICAPPS), El Rosario, Tenerife, Spain, 4Osakidetza Basque Health Service - Basurto University Hospital, Research and Innovation Unit; Network for Research on Chronicity, Primary Care, and Health Promotion (RICAPPS); Biosistemak Institute for Health System Research, Bilbao, Spain, Bilbao, Spain

Presentation Documents

OBJECTIVES: Rheumatoid arthritis (RA) is a chronic and systematic inflammatory disease associated with progressive joint destruction, which may lead to significant disabilities causing a significant decrease in health-related quality of life (HRQoL). Therefore, studies of treatment efficiency, usually based on health utilities are of great interest. The EQ-5D-5L is one of the most widely used generic questionnaire to derive these utilities. However, in clinical practice, the use of specifics questionnaires is more frequent. Our objective was to develop mapping functions to estimate the EQ-5D-5L utility index from the specific Health Assessment Questionnaire (HAQ).

METHODS: A prospective observational study, including 103 patients from Spain with RA who completed the EQ-5D-5L and HAQ questionnaires. Of these, 95 responded to the 6-months follow-up. The baseline data was used to derive the mapping functions from the HAQ disability index. Three strategies were used for the modelling: linear, tobit and beta regression. To select the best model per approach the AIC, BIC and adjusted R-square were used. The predictive performance of the models were compared by MAE and RMSE. These functions were validated in the follow-up data using MAE and RMSE.

RESULTS: The mean EQ-5D-5L index was 0.715 (SD=0.232, range=-0.297 to 1). The three derived models obtained similar predictive accuracy, although the linear and beta models showed slightly lower RMSE. However, the validation of these functions in the follow-up sample showed slightly lower MAE and RMSE in the linear model. Based on linear model the function was: EQ-5D-5L=0.9067–0.1345·HAQ–0.1652·HAQ2 (R2=0.567, AIC=-89.84, BIC=-79.30, MAE=0.107, RMSE=0.151).

CONCLUSIONS: As far as we know, this is the first mapping function from the HAQ to the Spanish EQ-5D-5L in patients with RA. It could be very useful for clinicians and researchers when cost-effectiveness studies are needed, and generic HRQoL instruments to derive utility indexes are not available.

Code

MSR5

Topic

Methodological & Statistical Research, Patient-Centered Research

Topic Subcategory

Artificial Intelligence, Machine Learning, Predictive Analytics, Health State Utilities, PRO & Related Methods

Disease

Musculoskeletal Disorders (Arthritis, Bone Disorders, Osteoporosis, Other Musculoskeletal)