Related Experiment Videos
Common statistical errors in the design and analysis of subfertility trials
1Biostatistics Group, University of Manchester and Salford Royal Hospitals NHS Trust, R&D Support Unit, Hope Hospital, Stott Lane, Salford M6 8HD, UK. andy.vail@man.ac.uk
Human Reproduction (Oxford, England)
|May 2, 2003
Summary
Many subfertility trials contain critical design and analysis errors, failing peer review standards. These flaws hinder reliable interpretation and inclusion in systematic reviews and meta-analyses.
Area of Science:
- Reproductive Medicine
- Clinical Trial Methodology
- Evidence-Based Medicine
Background:
- Growing emphasis on evidence-based medicine and systematic reviews highlights the need for high-quality clinical trials.
- Concerns exist regarding persistent errors and omissions in published clinical trials, even after peer review.
Purpose of the Study:
- To assess if common errors and omissions in subfertility trials are still passing peer review.
- To evaluate the methodological quality of recently published subfertility trials.
Main Methods:
- A review of clinical trials published in 2001 in two major medical journals was conducted.
- Trials comparing pregnancy rates between two interventions using group allocation were selected from Medline.
Main Results:
- Of 39 eligible trials, six had fatal design flaws (e.g., inappropriate crossover design, flawed randomization).
- Only six trials claimed adherence to the intention-to-treat principle, with four misunderstanding it.
- Five trials reported sufficient live birth rates for meta-analysis; 82% had 'unit of analysis' errors.
Conclusions:
- Despite selection from respected journals, many subfertility trials exhibit insufficient methodological quality.
- Inadequate design, analysis, and reporting prevent reliable interpretation and meta-analysis of numerous subfertility trials.