218-POS

Abstract

Objectives Small clinical studies, animal studies and laboratory experiments are often used to investigate the pathophysiology of preeclampsia. The NIH recently highlighted the critical problem of irreproducibility of published studies, which may be partially due to poor data presentation. Bar graphs are intended for categorical data, but are frequently used to present continuous data in small sample size studies. Bar graphs provide little information about the distribution of continuous variables and often distort data in small studies. Methods We systematically reviewed all full-length original research articles (n = 698) published in the top 20 physiology journals between January 1 and March 31, 2014. We examined the types of figures used to present continuous outcomes, sample size, and statistical methods. Results 626 articles included figures depicting continuous outcomes. 85% of these articles included a bar graph for a continuous outcome, with 77% showing mean ± standard error. Line graphs or graphs showing means and error bars were common (61%). Scatterplots (13%), boxplots (5%) and histograms (8%) were used infrequently. The minimum and maximum sample sizes (median, interquartile range) for any group presented in a figure were 4 (3–6) and 10 (6–15). Despite these small sample sizes, most articles used only parametric statistics (79%). Among the 17% of articles that performed any non-parametric analyses, 58% showed data that were analysed non-parametrically as mean ± standard error or standard deviation. Conclusions Journal policy changes are urgently needed to encourage investigators to use appropriate statistical methods, and to select figures that show the distribution of small sample size data to ensure the validity of conclusions. Studies of preeclampsia are often performed on a limited number of animals or biological samples from preeclamptic women (i.e. placenta or renal tissue). Appropriate presentation of results is essential to advance our understanding of preeclampsia. Disclosures T.L. Weissgerber: None. N. Milic: None. S.J. Winham: None. V.D. Garovic: None.