External validity is also an ethical consideration in cluster-randomised trials of policy changes

• Published in: BMJ quality & safety Vol. 28; no. 2; p. 167
• Main Authors: Bilimoria, Karl Y, Chung, Jeanette W, Hedges, Larry V
• Format: Journal Article
• Language: English
• Published: England BMJ Publishing Group LTD 01.02.2019
• Subjects: Clinical trials; Medical ethics; Mortality; Neurosurgery; Patient safety; Policy; Regulation; Research Design; Statistical analysis; Validity; United States > US; Nasca; Chicago Illinois; attitudes; accreditation; anaesthesia
• ISSN: 2044-5415; 2044-5423
• DOI: 10.1136/bmjqs-2018-008777

In the Statistical Analysis Plan (SAP), which was made available at www.nejm.org, we clearly and prospectively stated, ‘[t]his study is a trial-based evaluation of potential policy effects on patient safety and resident wellbeing… this study is intended to inform real-world policy decision-making with respect to resident duty hours regulation.’2 The SAP and online supplementary appendix (www.nejm.org) also provide all assumptions for our power calculations and cluster sizes, which were not large in the case of resident outcomes.3 As such, it was important that the trial closely resembles real-world conditions in which residency training occurs and duty hour policies are implemented. A shorter study would pose risks owing to non-standard, multiple policy shifts and would increase administrative/organisational burdens of staff workflow coordination. [...]a shorter period would threaten external validity, rendering evidence from the study potentially uninformative for policy decisions that could affect the lives of millions. [...]a shorter study would be of little social benefit while imposing unjustifiable risks and burden.