Statistically significant papers in psychiatry were cited more often than others

Abstract

Objective: Citations by other researchers are important in the dissemination of research findings. We aimed to investigate whether preferential citation of statistically significant articles exists in the psychiatric literature. Study Design and Settings: We analyzed all original research papers published in 1996 in four psychiatric journals. Using a standardized questionnaire, from each paper, we extracted the primary outcome and its statistical significance. The number of citations, excluding authors' "self-citations," received by April 2005 was obtained. Regression analysis was used to relate citation frequency to statistical significance, adjusting for confounders. Results: Of 448 extracted papers, 368 used statistical significance testing and 287 (77.8%) reported P < 0.05. The median number of citations for papers reporting "significant" and "nonsignificant" results was 33 vs. 16. After adjustment for journal, study design, reporting quality, whether outcome confirmed previous findings and study size, the ratio of the number of citations per article for articles reporting "P < 0.05" on the primary outcome to those reporting "P > 0.05" was 1.63 (95% confidence interval [CI]: 1.32, 2.02, P < 0.001). Conclusion: Authors cite studies based on their P-value rather than intrinsic scientific merit. This practice skews the research evidence. Systematic study registration and inclusion in meta-analysis should be encouraged. © 2007 Elsevier Inc. All rights reserved.