BMJ 1998;317:1171-1172 ( 31 October )

Editorials

Cluster randomised trials: time for improvement

The implications of adopting a cluster design are still largely being ignored 

Cluster randomised trials, where groups of patients rather than individuals are randomised, are increasingly being used in health services research. Randomisation by individual is inappropriate for evaluating some interventions, such as organisational changes, where it may not be feasible to randomise at the patient level. In such cases cluster randomisation at the level of the health professional or organisation is necessary. Such randomisation can also minimise the potential for contamination between treatments when trial patients are managed within the same setting.

The main consequence of adopting a cluster design is that the outcome for each patient can no longer be assumed to be independent of that for any other patient (which is the case in an individually randomised trial). Patients within any one cluster are more likely to have similar outcomes. For example, the management of patients within a single general practice is more likely to be consistent than management across several practices.

This lack of independence has implications for the design and analysis of these trials.¹ The statistical power of a cluster randomised trial is greatly reduced in comparison with a similar sized individually randomised trial. Therefore standard sample size estimates have to be inflated to take account of the cluster design.

The impact on sample size can be substantial and depends on the size of the clustering effect and the number of clusters available. The clustering effect would be high if, for example, management of patients within individual hospitals was very consistent but there was wide variation across hospitals. Consider a trial of an educational intervention to implement a clinical guideline. A patient randomised trial would require 194 patients to detect a change from 40% to 60% in the proportion of patients who are managed appropriately (with 80% power and 5% significance). However, this design would be inappropriate because of the potential for contamination. For a cluster randomised trial with a moderate clustering effect and 10 available patients per cluster the equivalent sample size adjusting for clustering is 38 clusters or 380 patientsthat is, almost double.¹

The analysis of cluster randomised trials must also take into account the clustered nature of the data. Standard statistical techniques are no longer appropriate, unless an aggregated analysis is performed at the level of the cluster,² as they require data to be independent. If the clustering effect is ignored P values will be artificially extreme, and confidence intervals will be over-narrow, increasing the chances of spuriously significant findings and misleading conclusions.

Although an aggregated analysis can be performed at the cluster level using standard statistical tests, this approach is statistically inefficient. Furthermore, it does not allow variation at the patient level to be exploredfor example, it cannot take account of patient characteristics such as disease severity. More advanced techniques have now been developed to analyse patient level data arising from a clustered design, which allow the hierarchical nature of the data to be modelled appropriately.³ They essentially allow variation to be modelled at each level of the datafor example, at both the practice and the patient level.

Despite the increased use of cluster randomised trials, the implications of adopting such a design continue to be largely ignored. For example, a review by Devine et al which examined studies of physicians' patient care practices observed that 70% of studies identified had not appropriately accounted for the clustered nature of their study data.⁴

Many trials do not take cluster randomisation into account when calculating the required sample size, resulting in studies which are underpowered. This may, in part, be explained by the lack of published information on the likely size of the clustering effect, known as the intracluster correlation coefficient.⁵ Reliable estimates of this coefficient are required to ensure robust sample size calculations, yet publication of intracluster correlation coefficients in trial reports is rare.

We need a standardised approach to the reporting of all aspects of cluster randomised trials, including intracluster correlation coefficients. The CONSORT statement for standards of reporting of patient randomised trials⁶ has improved the reporting of such trials, and a similar approach for the reporting of cluster randomised trials would be beneficial. A letter to the BMJ highlighted the need for such an approach,⁷ and a proposed amendment to the CONSORT statement is being formulated. We believe that this will aid the appraisal of published trial reports and the planning of future research. It deserves the full support of the medical research community.

Marion K Campbell, Senior statistician. 
Jeremy M Grimshaw, Programme director. 

Health Services Research Unit, University of Aberdeen, Aberdeen AB25 2ZD (m.k.campbell{at}abdn.ac.uk)

---

1. Donner A. Some aspects of the design and analysis of cluster randomization trials. Appl Statist 1998; 47: 95-113.
2. Kerry SM, Bland JM. Analysis of trials randomised in clusters. BMJ 1998; 316: 54[Free Full Text].
3. Rice N, Leyland A. Multilevels models: applications to health data. JHSRP 1996; 1: 154-164.
4. Divine GW, Brown JT, Frazier LM. The unit of analysis error in studies about physicians' patient care behavior. J Gen Intern Med 1992; 7: 623-629[Medline].
5. Kerry SM, Bland JM. The intracluster correlation coefficient in cluster randomisation. BMJ 1998; 316: 1455[Free Full Text].
6. Begg C, Cho M, Eastwood S, Horton R, Moher D, Olkin I, et al. Improving the quality of reporting of randomized controlled trials: the CONSORT statement. JAMA 1996; 276: 637-639[Medline].
7. Elbourne D. Guidelines are needed for evaluations that use cluster approach. BMJ 1997; 315: 1620-1621[Free Full Text].

---

© BMJ 1998
CiteULike    Complore    Connotea    Del.icio.us    Digg    Reddit    Technorati    What's this?

Related Articles

Internal and external validity of cluster randomised trials: systematic review of recent trials

Sandra Eldridge, Deborah Ashby, Catherine Bennett, Melanie Wakelin, and Gene Feder
BMJ 2008 336: 876-880. [Abstract] [Full Text] [PDF]

Cluster randomised trials

Nick Freemantle, John Wood, Marion K Campbell, and Jeremy M Grimshaw
BMJ 1999 318: 1286. [Extract] [Full Text]

This article has been cited by other articles:

• Eldridge, S., Ashby, D., Bennett, C., Wakelin, M., Feder, G. (2008). Internal and external validity of cluster randomised trials: systematic review of recent trials. BMJ 336: 876-880 [Abstract] [Full text]  
• Zimmermann, C., Riechelmann, R., Krzyzanowska, M., Rodin, G., Tannock, I. (2008). Effectiveness of Specialized Palliative Care: A Systematic Review. JAMA 299: 1698-1709 [Abstract] [Full text]  
• Peri, K., Kerse, N., Robinson, E., Parsons, M., Parsons, J., Latham, N. (2008). Does functionally based activity make a difference to health status and mobility? A randomised controlled trial in residential care facilities (The Promoting Independent Living Study; PILS). Age Ageing 37: 57-63 [Abstract] [Full text]  
• Ross, L., Simkhada, P., Smith, W. C. S. (2005). Evaluating effectiveness of complex interventions aimed at reducing maternal mortality in developing countries. J Public Health (Oxf) 27: 331-337 [Abstract] [Full text]  
• Turner, R. M, Thompson, S. G, Spiegelhalter, D. J (2005). Prior distributions for the intracluster correlation coefficient, based on multiple previous estimates, and their application in cluster randomized trials. Clin Trials 2: 108-118 [Abstract]  
• Farrin, A., Russell, I., Torgerson, D., Underwood, M. (2005). Differential recruitment in a cluster randomized trial in primary care: the experience of the UK Back pain, Exercise, Active management and Manipulation (UK BEAM) feasibility study. Clin Trials 2: 119-124 [Abstract]  
• White, I. R, Thomas, J. (2005). Standardized mean differences in individually-randomized and cluster-randomized trials, with applications to meta-analysis. Clin Trials 2: 141-151 [Abstract]  
• Goodacre, S, McQuillan, P, Higgins, B (2005). Appropriate analysis and reporting of cluster randomised trials * Authors' response. Emerg. Med. J. 22: 156-156 [Full text]  
• Zyzanski, S. J., Flocke, S. A., Dickinson, L. M. (2004). On the Nature and Analysis of Clustered Data. Ann Fam Med 2: 199-200 [Full text]  
• Murray, D. M., Varnell, S. P., Blitstein, J. L. (2004). Design and Analysis of Group-Randomized Trials: A Review of Recent Methodological Developments. Am. J. Public Health 94: 423-432 [Abstract] [Full text]  
• Eldridge, S. M, Ashby, D., Feder, G. S, Rudnicka, A. R, Ukoumunne, O. C (2004). Lessons for cluster randomized trials in the twenty-first century: a systematic review of trials in primary care. Clin Trials 1: 80-90 [Abstract]  
• Martin, C. M., Doig, G. S., Heyland, D. K., Morrison, T., Sibbald, W. J. (2004). Multicentre, cluster-randomized clinical trial of algorithms for critical-care enteral and parenteral therapy (ACCEPT). CMAJ 170: 197-204 [Abstract] [Full text]  
• Laopaiboon, M (2003). Meta-analyses involving cluster randomization trials: a review of published literature in health care. Stat Methods Med Res 12: 515-530 [Abstract]  
• Isaakidis, P., Ioannidis, J. P. A. (2003). Evaluation of Cluster Randomized Controlled Trials in Sub-Saharan Africa. Am J Epidemiol 158: 921-926 [Abstract] [Full text]  
• Treweek, S. P., Glenton, C., Oxman, A. D., Penrose, A. (2002). Computer-generated Patient Education Materials: Do They Affect Professional Practice? : A Systematic Review. J. Am. Med. Inform. Assoc. 9: 346-358 [Abstract] [Full text]  
• GILBODY, S., WHITTY, P. (2002). Improving the delivery and organisation of mental health services: beyond the conventional randomised controlled trial. Br. J. Psychiatry 180: 13-18 [Abstract] [Full text]  
• Foy, R., Eccles, M., Grimshaw, J. (2001). Why does primary care need more implementation research?. Fam Pract 18: 353-355 [Full text]  
• Bower, P., Garralda, E., Kramer, T., Harrington, R., Sibbald, B. (2001). The treatment of child and adolescent mental health problems in primary care: a systematic review. Fam Pract 18: 373-382 [Abstract] [Full text]  
• Localio, A. R., Berlin, J. A., Ten Have, T. R., Kimmel, S. E. (2001). Adjustments for Center in Multicenter Studies: An Overview. ANN INTERN MED 135: 112-123 [Abstract] [Full text]  
• Gilbody, S. M, House, A. O, Sheldon, T. A (2001). Routinely administered questionnaires for depression and anxiety: systematic review. BMJ 322: 406-409 [Abstract] [Full text]  
• Torgerson, D. J (2001). Contamination in trials: is cluster randomisation the answer?. BMJ 322: 355-357 [Full text]  
• Plant, H., Bredin, M., Krishnasamy, M., Corner, J. (2000). Working with resistance, tension and objectivity: Conducting a randomised controlled trial of a nursing intervention for breathlessness. Journal of Research in Nursing 5: 426-434 [Abstract]  
• Eekhof, J., De Bock, G., Schaapveld, K, Springer, M. (2000). Effects of screening for disorders among the elderly: an intervention study in general practice. Fam Pract 17: 329-333 [Abstract] [Full text]  
• Campbell, M. K, Mollison, J., Steen, N., Grimshaw, J. M, Eccles, M. (2000). Analysis of cluster randomized trials in primary care: a practical approach. Fam Pract 17: 192-196 [Abstract] [Full text]  
• Grande, G E, Todd, C J (2000). Why are trials in palliative care so difficult?. Palliat Med 14: 69-74  
• Reading, R., Harvey, I., Mclean, M. (2000). Cluster randomised trials in maternal and child health: implications for power and sample size. Arch. Dis. Child. 82: 79-83 [Abstract] [Full text]  
• Grande, G. E, Todd, C. J, Barclay, S. I G, Farquhar, M. C (1999). Does hospital at home for palliative care facilitate death at home? Randomised controlled trial. BMJ 319: 1472-1475 [Abstract] [Full text]  
• Freemantle, N., Wood, J., Campbell, M. K, Grimshaw, J. M (1999). Cluster randomised trials. BMJ 318: 1286b-1286 [Full text]  
• Glazener, C. M A, Ramsay, C. R, Campbell, M. K, Booth, P., Duffty, P., Lloyd, D. J, McDonald, A., Reid, J A., Deeks, J. J (1999). Neonatal examination and screening trial (NEST): a randomised, controlled, switchback trial of alternative policies for low risk infants • Commentary: "Switchback" allocation---dangerous bends ahead!. BMJ 318: 627-632 [Abstract] [Full text]  

Rapid Responses:

Read all Rapid Responses

Cluster randomised trials: can empirical estimates be used for sample size calculations

Petra Denig
bmj.com, 23 Jul 1999 [Full text]