Ethics and Cluster trials Dr Sarah JL Edwards Senior Lecturer, Research Ethics and Governance

Cluster randomized trials are increasingly being used in health and health services research Illustrate the types of CRTs; Identify some ethical issues in review Refer to literature and guidance Ethics and study design: Cluster randomized trials in health research

Cluster randomized trials Complex studies Units of randomization are clusters or groups Study intervention may be delivered to: the cluster as a unit (cluster-cluster trial) health professionals (professional-cluster trial) individual cluster members (individual-cluster trial) Outcomes are measured on individuals. Ethics and study design: Cluster randomized trials in health research

COMMIT trial (cluster-cluster trial) Unit of randomisation: US and Canadian Communities Intervention: Multimodal community-level intervention to reduce cigarette consumption including media and billboard campaign and targeted messaging toward smokers from health professionals Data collection: Interviews with a random sample of smokers in each community, the amounts of tobacco purchased by people living in the intervention and control communities. Result: Intervention led to an improved quit rate for mild to moderate smokers, with no effect on the quit rate of heavy smokers. Ethics and study design: Cluster randomized trials in health research

Improving primary care prescribing (professional-cluster trial) Unit of randomisation: General practitioners in Ireland Intervention: Personalized summary by mail of prescribing practices of antiplatelet and lipid lowering drugs plus an educational visit versus personalized summary of prescribing practices alone Data collection: Data on prescribing practices from the national pharmacy insurance program database. Data on prescribing practices were aggregated by physician and contained no identifiable information Result: Both interventions led to similar improvements. Ethics and study design: Cluster randomized trials in health research

Cleansing of the umbilical stump in neonates (individual-cluster trial) Unit of randomisation: Geographical areas in Nepal Intervention: Community health workers provide cleaning the umbilical stump with chlorhexidine versus soap and water versus dry stump care Data collection: Incidence of infection through clinical examination during household visits (15,000 infants), neonatal mortality, questionnaires about the household and infant care Result: Chlorhexidine reduced infection of the umbilical stump by 75% and neonatal mortality by 24%. Ethics and study design: Cluster randomized trials in health research

Rationale important: -Test how generalisable previous results have been -Test cluster level intervention -Avoid contamination between trial arms e.g. educational or behavioural interventions -Capture ‘herd’ or group effects e.g. vaccines -Speed up community roll out of experimental medicine

Right to health and prior evidence of efficacy might render cluster trial unjustified Treatment Action Campaign and Others v. Minister of Health and Others Constitutional Court of South Africa; 2002

1. In whose interests? Choice and size of cluster Moral status of social groups is not well understood Determine group interests: average, aggregate? Individual interests: each and every best interests? Ethics and study design: Cluster randomized trials in health research

2. Who is research subject? CRTs are complex and have multiple levels: e.g., hospitals are randomized, health care workers receive experimental intervention, and patient outcomes are assessed Complicates the identification of research participants From whom is informed consent required, if anyone? Ethics and study design: Cluster randomized trials in health research

3. Who represents group? In a CRT, clusters may be randomized prior to the identification of individual cluster members Is consent to randomization required? If so from whom ought it be sought? May gatekeepers provide consent to randomization? Person politically represent group anyway? What are his/her duties? Ethics and study design: Cluster randomized trials in health research

4. Consent to receive intervention In a CRT, the study intervention may be directed at the individual or the cluster or both Cluster level interventions (e.g., public educational messages or fluoridation of water supply) may be difficult for individual cluster members to avoid In such cases, refusal of study participation may be meaningless. Ethics and study design: Cluster randomized trials in health research

5. Consent to submit data on outcomes In a CRT, the outcomes relate to individuals within clusters Individuals may not have consented to randomisation of group or to receive experimental intervention especially if in control group Risk of contamination may be rationale for using cluster design. Ethics and study design: Cluster randomized trials in health research

MRC 2001 Ethical and methodological issues following BMJ paper 1999 by Edwards et al.

Ethics and study design: Cluster randomized trials in health research

Acknowledgements HTA for funding in 1999 Canadian NIH, Ottowa Statement MRC PhD candidate, Sapfo Lignou, for work on ideas of community Wellcome Trust PhD candidate, Elizabeth Oduwo, right to health UCL BRC funded work on use of cluster designs during pandemic WHO conference 6 th June 2014