The derivation of a preference-based measure for people with common mental health problems from the Clinical Outcomes in Routine Evaluation Outcome Measure (CORE-OM)

Background: Generic preference-based measures (PBMs), such as the EQ-5D and SF-6D, are widely used for the estimation of Quality Adjusted Life Years in cost-utility analyses of healthcare interventions. However, their relevance in some disease areas, including mental health, has been questioned. Objective of the thesis: To derive a PBM specific to mental health problems from an existing condition-specific measure (CSM) Methods: A systematic literature review was conducted to identify an appropriate CSM for the derivation of a health state classification. Derivation of the new measure was achieved using novel methodology developed for this purpose, due to the high correlation across the items of the original CSM. Selected health states were valued by members of the public. Regression analysis was employed to predict utility values for all states of the health state classification. Psychometric and qualitative assessments evaluated the performance of the new PBM compared with generic PBMs and the original CSM. Results: The Clinical Outcomes in Routine Evaluation – Outcome Measure (CORE-OM) was selected as the basis for the derivation of the new PBM. Application of novel methodology based primarily on Rasch analysis resulted in the development of CORE-6D, a health state classification that consists of a 5-item emotional component and a physical item. Rasch analysis was used to select plausible health states for valuation. A highly predictive regression model was used to attach utility values to all CORE-6D health states. The new PBM has shown promising results regarding its psychometric properties compared with generic PBMs and suffers from little loss of information relative to the original measure, CORE-OM. Further research needs to validate these findings. Conclusion: The CORE-6D preference-based index will enable cost-utility analysis of mental health interventions using existing and prospective CORE-OM datasets. The new methodology for deriving PBMs from existing instruments can be useful for the derivation of PBMs from other instruments with highly correlated dimensions