Article type
Abstract
Background: Numerous systematic reviews (SRs) have analyzed the efficacy of talking therapies and psychiatric drugs for the treatment of depression. Evidence suggests that trials of pharmacotherapy and psychotherapy differ in terms of conduct, participants’ characteristics, and outcomes measurement, among others. This results in considerable clinical and methodological heterogeneity. Statistics describing heterogeneity are ubiquitous in SRs, but their interpretation can be challenging and is sometimes technically incorrect. Prediction intervals (PIs) estimate the likely range where the effects of treatments in future studies could fall, helping to understand the variability in treatment outcomes and enabling a more personalized perspective on clinical decision-making. Despite providing information relevant to patient management and future research, PIs are rarely reported in meta-analyses.
Objectives: To determine whether the use of PIs in meta-analyses on treatments for depression would change or add to their overall conclusion on the clinical use of interventions or recommendations for future research.
Methods: We included meta-analyses of trials comparing pharmacotherapy or psychotherapy to inactive control conditions. Wherever appropriate, the PIs will be calculated for the main depression outcome using CMA Prediction Intervals software by Biostat and the efficacy data will be reinterpreted.
Results: Preliminary data suggest that including PIs in meta-analyses can substantially change the interpretation of the treatments’ effect. This may be due to high clinical and methodological heterogeneity in clinical trials of depression.
Conclusions: These results are in line with similar analyses in other fields of medicine, highlighting the necessity for measuring and exploring heterogeneity in meta-analyses using clinically useful statistics like PIs to allow patients and clinicians for better-informed decision-making and researchers to design more informative trials.
Relevance to patients: This analysis will improve the clinical significance of future evidence on the treatments for depression and facilitate its interpretation.
Public and/or healthcare consumer involvement: Literature on patient-identified priorities in depression research and public mental health literacy was reviewed and informed this analysis.
This research was funded by the National Science Centre, Poland (grant number 2021/41/B/HS6/02844). For the purpose of Open Access, the author has applied a CC-BY public copyright license to any Author Accepted Manuscript (AAM) version arising from this submission
Objectives: To determine whether the use of PIs in meta-analyses on treatments for depression would change or add to their overall conclusion on the clinical use of interventions or recommendations for future research.
Methods: We included meta-analyses of trials comparing pharmacotherapy or psychotherapy to inactive control conditions. Wherever appropriate, the PIs will be calculated for the main depression outcome using CMA Prediction Intervals software by Biostat and the efficacy data will be reinterpreted.
Results: Preliminary data suggest that including PIs in meta-analyses can substantially change the interpretation of the treatments’ effect. This may be due to high clinical and methodological heterogeneity in clinical trials of depression.
Conclusions: These results are in line with similar analyses in other fields of medicine, highlighting the necessity for measuring and exploring heterogeneity in meta-analyses using clinically useful statistics like PIs to allow patients and clinicians for better-informed decision-making and researchers to design more informative trials.
Relevance to patients: This analysis will improve the clinical significance of future evidence on the treatments for depression and facilitate its interpretation.
Public and/or healthcare consumer involvement: Literature on patient-identified priorities in depression research and public mental health literacy was reviewed and informed this analysis.
This research was funded by the National Science Centre, Poland (grant number 2021/41/B/HS6/02844). For the purpose of Open Access, the author has applied a CC-BY public copyright license to any Author Accepted Manuscript (AAM) version arising from this submission