Virtual participation: Zoom details available here
Speaker: Dr Fiona Boland, Royal College of Surgeons in Ireland
Title: "Do psychometrics matter? Findings from antidepressant treatment trials"
Abstract: It has been contended that the sophisticated statistical techniques that are used to evaluate psychometric scales are vital to improving psychometric assessment. These techniques can often provide conflicting results and there is limited evidence that adopting these techniques actually makes important differences to ultimate outcomes. We therefore aimed to determine whether applying psychometric analyses to individual patient data would demonstrate important differences in depression trial treatment effects.
We conducted a secondary analysis of individual participant data from antidepressant treatment trials from Vivli.org that used the Montgomery-Asberg Depression Rating Scale (MADRS) or Hamilton Rating Scale for Depression-17 (HRSD-17). Data was analysed using confirmatory factor analysis, item response theory and network analysis, providing psychometrically-informed model scores to compare to original total scores in multilevel models. Differences in trial effect sizes was the outcome of interest.
There were mixed results for both MADRS and HRSD-17 and very little evidence that applying psychometric theory scores were better than simple abbreviated scale totals. Factor analysis increased effect sizes and may be the most effective method for identifying the items on which placebo and treatment group outcomes differ.
Biography: Dr Fiona Boland is a senior lecturer in Biostatistics and Research Methods based in the Data Science Centre in the Royal College of Surgeons in Ireland. Dr Boland specialises in statistical methodologies and their application to medical and health research. Dr Boland is particularly interested in randomised controlled trials, large cohort studies, clinical prediction rules and systematic reviews and meta-analysis. She is engaged in interdisciplinary research within a large network of affiliated research units. Areas of interest include medical care variation and quality of care, multimorbidity and comorbidity, substance use and addiction, depression and childhood obesity.