Assessing statistical analyses of clinical trials: primer in statistical methods for thoracic surgeons

Each year there are many randomized controlled trials published for thoracic surgeons. To apply the results of these articles to their patients, surgeons need to have an understanding of the criteria for assessing the validity of these articles. They need to be able to recognize both methodological and statistical flaws a study may have. This article presents a published "validity" tool that can be used for assessment, gives flowcharts to show which statistical tests can be used, and also presents the assessment findings of a review of 52 published randomized controlled trials. Key issues discussed include documentation of validity items in the methods section, overuse of P values, and correct statistical analyses for patients measured at various times. The article concludes with suggestions for future assessment and publication.