A secondary exploratory study of associations between patient- and clinician-reported clinical outcomes and fidelity for four evidence-based psychosis treatments

Abstract

Background: Implementation of evidence-based practices (EBPs), measured as fidelity to the EBP model, is generally expected to yield significant positive clinical outcomes. However, this association has only partially been established for EBPs used in psychosis treatment. From a cluster-randomized controlled trial (CRCT), we previously reported on the effects of implementation support for four EBPs for psychosis, using fidelity as the measure of implementation success. The current secondary, exploratory study used data from the non-blinded CRCT parent study to investigate the associations between patient- and clinician-reported outcomes and fidelity for these four EBPs. Methods: Clinical outcomes were measured in a cohort of 325 patients over three six-month periods. Primary outcomes were BASIS-24 (patient-reported) and HoNOS (clinician-reported). Secondary outcomes were selected subscales of these two measures. The EBPs were Physical Health Care, Antipsychotic Medication Management, Family Psychoeducation, and Illness Management and Recovery. In the CRCT, each of 39 clinical units across six health trusts selected two EBPs for implementation. Units were randomized to the intervention group (implementation support) for one EBP and the control group (written manual) for the other. Fidelity of the four EBPs was measured at baseline and every six months for 18 months. We analyzed the associations between outcomes and fidelity using linear mixed models. Results: BASIS-24 and HoNOS showed improvements for the total sample at 6 and 12 months, and two patient-reported subscales, Symptoms and Relationships, showed improvement at 6 months within two different EBP subsamples. However, no positive associations were found between secondary outcomes and EBP fidelity. Conclusions: Despite some improvements in primary and secondary outcomes over the first 6 to 12 months, we found no positive associations between outcomes and fidelity. Sample size, attrition, trial design, variance in variables, measurement properties, and low exposure, as well as interaction between such factors, might have contributed to our failure to find positive associations between outcomes and fidelity. Future studies of the association between outcomes and fidelity should involve large samples, use outcome and exposure measures closely related to the EBPs, and track cohorts from the beginning of treatment.