Avoiding pitfalls in conducting hand surgery research – potential consequences of common errors in statistical analyses

Abstract

Although errors in statistical analyses can seriously distort scientific findings, they remain relatively common in hand surgery research. In this article we demonstrate potential consequences of four commonly made mistakes: unnecessary dichotomization, ignoring data clustering, applying prediction modelling methods for examining independent associations and violating the assumptions of logistic regression. We use real-life examples from our reviewing experience to demonstrate how these commonly made errors can influence results and conclusions, and therefore, impact clinical decision-making. Through these, we hope to further raise awareness of the importance of avoiding errors in statistical analyses, so we can strive towards providing reliable answers to research questions in hand surgery.