Abstract
Background: There is considerable interest in the concept of recovery from psychosis. Consumers describe recovery as a process as opposed to a clinical outcome. Measures of recovery have an important role in the development of recovery based mental health services. Objectives: To investigate the validity and reliability of the Stages of Recovery Instrument (STORI: Andresen et al, 2006). The STORI was developed in Australia as a measure of recovery from psychosis. The STORI was considered to be a promising measure because it is based upon a stage model of recovery developed from the perspectives of consumer accounts of the recovery process. Design: A questionnaire design was used to investigate the following aspects of validity: face validity and feasibility, concurrent validity, construct validity and test-retest reliability. Participants: Fifty people from the caseloads of three specialist mental health teams participated in the study. Twenty-two of these participated in the test-retest reliability stage of the research. Measures: The STORI yields scores for five stages of recovery. The highest stage score indicates the stage of recovery that the respondent is in. Participants also completed the STORI Feedback questionnaire (STORI-F) a seven item measure addressing feasibility, and the Recovery Assessment Scale (RAS: Giffort et al, 1995). The RAS is a 41 item measure that yields a single recovery score. Those who took part in the test-retest aspect of the study did so after completing a subtest of the Weschler Adult Intelligence scale as a distracter task for four minutes. Statistical Analysis: Descriptive statistics of responses to the STORI-F were used to investigate feasibility and face validity. Concurrent validity was investigated using Pearson correlations between the STORI stages and the RAS. Construct validity was investigated using Pearson correlations between stage scores and hierarchical cluster analysis. Test-retest reliability data was analysed for stage and cluster scores (using Pearson correlations) and stage and cluster allocations (using Spearman’s correlations). Results: Participants’ responses to the STORI-F were mainly positive. This was seen as evidence for face validity and feasibility of the STORI as a measure of recovery. There were significant positive correlations between the later stages of recovery (3, 4 and 5), and the RAS. There was a significant negative correlation between stage 1 and the RAS. This was seen as initial evidence of construct validity. Cluster analysis revealed that the STORI items formed three clusters rather than five. It was therefore concluded that the STORI does not have construct validity as a measure of a five stage model of recovery. Finally, there was initial evidence of test-retest reliability for stage and cluster scores. However, these analyses were limited by a small sample size. Conclusions: This study provided initial evidence for the validity of the STORI in the UK. Possible reasons for the failure to find five stage-based clusters were discussed. The results from this study were discussed in the wider context of attempts to measure recovery.