Researchers' views of risk of bias in cluster randomised trials: a qualitative interview study.

Abstract

Cluster randomised trials (CRTs) can be at risk of bias driven by differential identification and recruitment of participants across treatments, posing a threat to the validity of findings. We explored the awareness and importance, among CRT researchers, of the recommended bias mitigation measures. Qualitative interview study using semistructured interviews. Participants were researchers involved in conducting CRTs, including investigators, statisticians and trial coordinators. 24 participants, including statisticians (n=13, 54.2%), clinical investigators (n=9, 37.5%) and trial coordinators (n=2, 8.3%), were interviewed; with representation from the UK (n=10, 41.7%), Australia (n=5, 20.8%) and the USA (n=4, 16.7%). Participants exhibited differing levels of knowledge related to biases. Some participants demonstrated high levels of knowledge, but we also identified prevalent misconceptions, with some evidence of superficial knowledge. While some participants worked in collaborative teams, other teams' responsibilities were delineated, and this impacted on how knowledge of biases was shared and acted on. Logistical and practical issues could prevent known solutions to mitigate biases being implemented. Biases also manifested because of a perception from participant recruiters that the purpose of research is for participant benefit rather than producing generalisable knowledge; and a normalisation or expectation that CRTs produce a lower level of evidence. There is an urgent need to ensure that CRTs are free from risks of bias. Mitigation measures are either not known, not practical or unconsciously subverted. More education and collaborative working might help. Preventing subconscious bias during participant recruitment and dispelling the myth that CRTs produce lower levels of evidence would require a change in culture.