Format

Send to

Choose Destination
Contemp Clin Trials. 2015 May;42:41-50. doi: 10.1016/j.cct.2015.02.011. Epub 2015 Mar 9.

Sample size calculations for the design of cluster randomized trials: A summary of methodology.

Author information

1
National Heart Research Institute Singapore, National Heart Centre Singapore, 5 Hospital Drive, Singapore 169609; Center for Quantitative Medicine, Duke-NUS Graduate Medical School, 8 College Road, Singapore 169857. Electronic address: gao.fei@nhcs.com.sg.
2
Center for Quantitative Medicine, Duke-NUS Graduate Medical School, 8 College Road, Singapore 169857; Department of Epidemiology and Preventive Medicine, Monash University, Level 6, The Alfred Centre, 99 Commercial Road, Melbourne, Victoria 3004, Australia. Electronic address: arul_earnest@hotmail.com.
3
Health Services & Systems Research, Duke-NUS Graduate Medical School, 8 College Road\, 169857, Singapore; Department of Medicine (General Internal Medicine), Duke University Medical Center, 411 W. Chapel Hill Street, Suite 500, Durham, NC 27701, USA. Electronic address: david.matchar@duke-nus.edu.sg.
4
Medical Statistics Group, School of Health and Related Research, University of Sheffield, Regents Court, 30 Regent Street, Sheffield S1 4DA, UK. Electronic address: m.j.campbell@sheffield.ac.uk.
5
Medical Statistics Group, School of Health and Related Research, University of Sheffield, Regents Court, 30 Regent Street, Sheffield S1 4DA, UK; Department of Cancer Studies and Molecular Medicine, Clinical Sciences Building, University of Leicester, Leicester Royal Infirmary, Leicester LE2 7LX, UK. Electronic address: dm113@leicester.ac.uk.

Abstract

Cluster randomized trial designs are growing in popularity in, for example, cardiovascular medicine research and other clinical areas and parallel statistical developments concerned with the design and analysis of these trials have been stimulated. Nevertheless, reviews suggest that design issues associated with cluster randomized trials are often poorly appreciated and there remain inadequacies in, for example, describing how the trial size is determined and the associated results are presented. In this paper, our aim is to provide pragmatic guidance for researchers on the methods of calculating sample sizes. We focus attention on designs with the primary purpose of comparing two interventions with respect to continuous, binary, ordered categorical, incidence rate and time-to-event outcome variables. Issues of aggregate and non-aggregate cluster trials, adjustment for variation in cluster size and the effect size are detailed. The problem of establishing the anticipated magnitude of between- and within-cluster variation to enable planning values of the intra-cluster correlation coefficient and the coefficient of variation are also described. Illustrative examples of calculations of trial sizes for each endpoint type are included.

KEYWORDS:

Cluster randomized trial; Intra-class correlation; Sample size

PMID:
25766887
DOI:
10.1016/j.cct.2015.02.011
[Indexed for MEDLINE]
Free full text

Supplemental Content

Full text links

Icon for Elsevier Science Icon for White Rose Research Online
Loading ...
Support Center