PSY192 - PATIENT HEALTH-RELATED QUALITY OF LIFE ASSOCIATED WITH REMISSION OF ATTP, A REGRESSION ANALYSIS USING NON-RANDOMISED OBSERVATIONAL DATA FROM THE OKLAHOMA TTP REGISTRY

Abstract

Acquired thrombotic thrombocytopenic purpura (aTTP) is a life-threatening, thrombotic microangiopathy characterized by thrombocytopenia, microangiopathic haemolytic anaemia, and organ ischemia. Currently, there is no published EQ-5D-3L-based health state utility value (HSUV) describing the health-related quality of life (HRQL) of patients with aTTP. To inform cost-effectiveness decisions for aTTP interventions, there is a need to use existing non-randomised observational data to investigate aTTP patient HRQL. Data from the subset of Oklahoma TTP registry patients with aTTP (ADAMTS13 activity < 10%) was mapped from SF-36 responses to EQ-5D-3L utility values. Mixed-effects (ME) regression correcting for patient heterogeneity was used to estimate an aTTP remission HSUV. The predictive power of covariates, including demographics, comorbidities, recovery time following aTTP events, and disease duration was evaluated, and the results used to inform the final model specification. Finally, confidence intervals for HSUVs were estimated using nonparametric bootstrapping. In the primary complete-case analysis (371/430 obs., 55/56 patients), the crude mean of the mapped patient level utility data was 0.736. A ME model including a random effect for patient had the best statistical fit, according to Akaike and Bayesian information criteria, compared to all ME models including demographic, recovery time or disease history covariates, estimating a remission HSUV of 0.707 [95% CI: 0.649–0.766]. A ME model including self-reported neurological problems and a patient random-effect had a better statistical fit than an equivalent ME model including no covariates yet included less observations (295/430). This estimated HSUVs with and without neurological problems of 0.682 [0.622–0.742] and 0.736 [0.674–0.798], respectively. Both models have very similar visual fit to the mapped EQ-5D-3L utility data. As it is unclear which model best describes the HSUV of aTTP patients in remission, both sets of values should be considered as scenarios in a cost-effectiveness analysis setting.