Invited Expert Opinion: Guidelines for improving the use and presentation of P values
P values were originally developed for hypothesis testing; however, the use of the P value in cardiovascular and thoracic research has expanded as it has become the most commonly reported summary measure of statistical results. A P value measures whether or not the actual results of a study or trial are consistent with what would be expected by chance or if the results more likely indicate a real difference between the treatment groups or surgical approaches being studied. A small P value is often regarded as evidence that the surgeon can rule out the chance explanation for the observed differences between the study groups.
Recently, there has been discussion in the literature regarding P values in medical research as it pertains to reporting standards.1,2 Although P values have value in providing a snapshot summary of the statistical evidence of group differences, treatment effects, or measures of association, there are times when P values have been misused or overused. The objective of this article is to provide a perspective regarding the use of P values in the Journal and to provide guidance in how to make the best use of P values particularly for common statistical study designs found in studies in the Journal. We provide a background regarding P values and guidance regarding the optimal use of P values for 3 specific common statistical study designs of manuscript submitted to the Journal.
