295: PUBLICATION BIAS: A PILOT STUDY ON THE LAST DECADE OF CRITICAL CARE MEDICINE LITERATURE

Abstract

Introduction: Publication bias exists when the results of a study, based on the strength of the study findings influence the decision to publish it. This means that studies that have statistically significant or positive results get published more when compared to studies with statistically insignificant or negative results. Although publication bias has been reported in the literature for decades, there is evidence suggesting that this bias may be on the rise. Our study aims to uncover the inherent bias that exists in critical care medicine within the last 10 years. Methods: This is an observational study. Articles from renowned Critical Care Journals with high impact factors were randomly selected. These articles proposed a new intervention, care modality, or biomarker in critical care medicine. We analyzed a total of 500 studies within the last 10 years from the New England Journal of Medicine (NEJM), American Journal of Respiratory and Critical Care Medicine (AJRCCM), and Critical Care Forum (ccforum), Chest journal, and Critical Care Medicine (CCM). The articles were categorized into positive and negative results depending on the results. Case reports, Case series, Systemic reviews, and Meta-analyses were excluded. Results: We evaluated total of 500 articles (NEJM = 141, AJRCCM = 87, ccforum = 100, Chest Journal = 25, CCM = 147). The number of studies published with positive results were 85, 56, 67, 18, and 114 respectively with a total of (n = 340/500, 68 %). The studies published with negative results were 56, 31, 33, 7, and 33 with a total of (n = 160/500, 32%). We confirmed our result with a 5x2 chi contingency table. Our chi-square was 10.8 with a two-tailed p-value of .03 (significant at p < .05). We re-confirmed our result with a sign test, as repeated measures with paired values (positive and negative) of five journals. The z-value was 2.2 and the two-tailed p-value was .025 (significant at p < 0.05). Our results were limited by the small sample size, one specialty, and an uneven number of articles from each journal. Conclusions: Studies with positive results have higher rates of publication when compared to studies with negative results. Publication bias was found in renowned journals of critical care medicine. There is a need to eliminate this to minimize the impact of grey literature.