Effect of statistical review on manuscript quality in Medicina Clínica (Barcelona): a randomized study

The statistical review of biomedical articles should result in an improved quality. The objective of this study was to compare the effects of clinical review and joint clinical and statistical review on manuscript quality, in articles submitted to Medicina Clínica (Barcelona), a Spanish weekly journal of internal medicine. Original papers arriving between May 2000 and February 2001 were randomized either to a clinical review group or a clinical and statistical review group. Two evaluators, blinded to the paper's group, assessed the quality improvement in both groups, from submission to publication using a modified version ot the Goodman et al. scale. The protocol required that final versions arrived before the end of May 2001. Final sample size was 43 manuscripts, evaluated before and after peer review. On the intention to treat analysis, the estimated effect of statistical review was 1.35 (95% CI: -0.45 to 3.16) positive, but not statistically significant. The analysis of the reviewers' comments revealed some protocol deviations. Taking into account the spontaneous inclusion of statistical experts in the clinical group, the estimated effect was statistically significant, with a confidence interval of 0.3 to 3.7. The inclusion of a statistical expert in the peer review process improves manuscript quality, although in the intention to treat analysis the improvement was not statistically significant.