Abstract
BACKGROUND: Statistical training across the continuum of medical education may not have advanced at the pace of statistical reporting in the medical literature, yet a comprehensive understanding of statistical concepts most commonly presented in current research is critical to the effective practice of Evidence Based Medicine. The objective of this content analysis was to describe statistical techniques used in a leading medical journal, JAMA, across a 20-year period, with a focus on implications for medical education. METHODS AND FINDINGS: Two issues of JAMA published each month in 1990, 2000, and 2010 were randomly selected; from these, 361 articles were reviewed. Primary focus, study design, and statistical components were abstracted and examined by year of publication. The number of published RCTs and cohort studies differed significantly across years of interest, with an increasing trend of publication. The most commonly reported statistics over the 20-year period of interest included measures of morbidity and mortality, descriptive statistics, and epidemiologic outcomes. However, between 1990 and 2010, there was an increase in reporting of more advanced methods, such as multivariable regression, multilevel modeling, survival analysis, and sensitivity analysis. While this study is limited by a focus on one specific journal, a strength is that the journal examined is widely read by a range of clinical specialties and is considered a leading journal in the medical field, setting standards for published research. CONCLUSIONS: The increases in frequency and complexity of statistical reporting in the literature over the past two decades may suggest that moving beyond basic statistical concepts to a more comprehensive understanding of statistical methods is an important component of clinicians' ability to effectively read and use the medical research. These findings provide information to consider as medical schools and graduate medical education training programs review and revise their statistical training components.