Factors influencing the statistical planning, design, conduct, analysis and reporting of trials in health care: A systematic review

Background: Trials in health care are prospective human research studies designed to test the effectiveness and safety of health care interventions, such as medications, surgeries, medical devices and other treatment or prevention interventions. Statistics is an important and powerful tool in trials. Inappropriately designed trials and/or inappropriate statistical analysis produce unreliable results and a lack of transparency when reported, with limited clinical use. Aim: This systematic literature review aimed to identify, describe and synthesise factors contributing to or influencing the statistical planning, design, conduct, analysis and reporting of trials. Methods: Information sources were retrieved from the following electronic citation databases: PubMed, Web of Science, PsycINFO, and CINAHL and the grey literature repository: OpenGrey. 90 articles and guidelines were included in this review. A narrative, thematic synthesis identified the key factors influencing the statistical planning, design, conduct, analysis and reporting of trials in health care. Findings and conclusion: We identified three analytical themes within which factors are grouped. These are: "what makes a statistician?", "the need for dynamic statistical involvement and collaboration throughout a trial - it's not just about the numbers", "and the "accountability of statisticians in ensuring the safety of trial participants and the integrity of trial data". While important insights emerged about the qualifications, training, roles, and responsibilities of statisticians and their collaboration with other team members in a trial, further empirical research is warranted to elicit the perceptions of trial team members at the centre of statistics in trials.