Psychometric Assessment of the EQ-5D-3L And ReQoL Measures in Secure Care Patients with an Intellectual Disability: Reliability, Agreement, Construct Validity, and Responsiveness
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
PURPOSE. People with intellectual disabilities (PID) are a vulnerable, marginalised group for whom mental ill-health is common. PID who commit criminal offences may require secure inpatient care, within which aggression and violence cause major problems stemming in-part from patients’ psychological distress. To conduct economic evaluations of interventions to aid secure care PID, utility-based measures (e.g., EQ-5D, ReQoL-UI) must be sensitive and responsive to changes in aggression and psychological distress. Within secure care PID, we assess the psychometric properties of the self-reported and proxy-reported EQ-5D-3L (generic health) and ReQoL-UI (recovery-focussed quality-of-life) alongside the clinician-reported MOAS (aggression) and self-reported BSI (psychological distress). We use the self-reported EQ-5D-3L adapted for PID; the ReQoL-10 was also assessed. METHODS. The SCHEMA trial collected measures at baseline, 19- and 38-weeks post-baseline. EQ-5D-3L and ReQoL measures’ inter-rater reliability and agreement between self- and proxy-responses was assessed, with construct validity and responsiveness judged against MOAS and BSI scores and cut-offs, e.g., with/out aggression or psychological distress. RESULTS. Forty-seven adults were randomised (intervention, 24; control, 23). The psychometric results were notably influenced by using self- or proxy-responses. The EQ-5D-3L suggested little heterogeneity among responders. Inter-rater reliability/agreement was: EQ-5D-3L, poor/none; ReQoL-UI, poor/minimal; ReQoL-10, moderate/weak-moderate. The ReQoL-UI/-10 had better construct validity with the MOAS and BSI than the EQ-5D-3L. Responsiveness tended to be larger for the proxy-reported ReQoL-10, albeit trivial to small across all measures. CONCLUSION. Despite the small sample size common with PID research, this study suggests the ReQoL measures might be preferred over the EQ-5D-3L in secure care PID.