Power and sample size requirements for GEE analyses of cluster randomized crossover trials

Fan Li, Andrew B. Forbes, Elizabeth L. Turner, John S. Preisser

Research output: Contribution to journalArticleResearchpeer-review

Abstract

The cluster randomized crossover design has been proposed to improve efficiency over the traditional parallel cluster randomized design, which often involves a limited number of clusters. In recent years, the cluster randomized crossover design has been increasingly used to evaluate the effectiveness of health care policy or programs, and the interest often lies in quantifying the population-averaged intervention effect. In this paper, we consider the two-treatment two-period crossover design, and develop sample size procedures for continuous and binary outcomes corresponding to a population-averaged model estimated by generalized estimating equations, accounting for both within-period and interperiod correlations. In particular, we show that the required sample size depends on the correlation parameters through an eigenvalue of the within-cluster correlation matrix for continuous outcomes and through two distinct eigenvalues of the correlation matrix for binary outcomes. We demonstrate that the empirical power corresponds well with the predicted power by the proposed formulae for as few as eight clusters, when outcomes are analyzed using the matrix-adjusted estimating equations for the correlation parameters concurrently with a suitable bias-corrected sandwich variance estimator.

Original languageEnglish
Pages (from-to)636-649
Number of pages14
JournalStatistics in Medicine
Volume38
Issue number4
DOIs
Publication statusPublished - 20 Feb 2019

Keywords

  • cluster randomized trials
  • crossover
  • finite-sample correction
  • generalized estimating equations (GEE)
  • matrix-adjusted estimating equations (MAEEs)
  • sandwich variance estimator

Cite this

Li, Fan ; Forbes, Andrew B. ; Turner, Elizabeth L. ; Preisser, John S. / Power and sample size requirements for GEE analyses of cluster randomized crossover trials. In: Statistics in Medicine. 2019 ; Vol. 38, No. 4. pp. 636-649.
@article{c701d588785640a09a8ec54cfcf5cd18,
title = "Power and sample size requirements for GEE analyses of cluster randomized crossover trials",
abstract = "The cluster randomized crossover design has been proposed to improve efficiency over the traditional parallel cluster randomized design, which often involves a limited number of clusters. In recent years, the cluster randomized crossover design has been increasingly used to evaluate the effectiveness of health care policy or programs, and the interest often lies in quantifying the population-averaged intervention effect. In this paper, we consider the two-treatment two-period crossover design, and develop sample size procedures for continuous and binary outcomes corresponding to a population-averaged model estimated by generalized estimating equations, accounting for both within-period and interperiod correlations. In particular, we show that the required sample size depends on the correlation parameters through an eigenvalue of the within-cluster correlation matrix for continuous outcomes and through two distinct eigenvalues of the correlation matrix for binary outcomes. We demonstrate that the empirical power corresponds well with the predicted power by the proposed formulae for as few as eight clusters, when outcomes are analyzed using the matrix-adjusted estimating equations for the correlation parameters concurrently with a suitable bias-corrected sandwich variance estimator.",
keywords = "cluster randomized trials, crossover, finite-sample correction, generalized estimating equations (GEE), matrix-adjusted estimating equations (MAEEs), sandwich variance estimator",
author = "Fan Li and Forbes, {Andrew B.} and Turner, {Elizabeth L.} and Preisser, {John S.}",
year = "2019",
month = "2",
day = "20",
doi = "10.1002/sim.7995",
language = "English",
volume = "38",
pages = "636--649",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley & Sons",
number = "4",

}

Power and sample size requirements for GEE analyses of cluster randomized crossover trials. / Li, Fan; Forbes, Andrew B.; Turner, Elizabeth L.; Preisser, John S.

In: Statistics in Medicine, Vol. 38, No. 4, 20.02.2019, p. 636-649.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Power and sample size requirements for GEE analyses of cluster randomized crossover trials

AU - Li, Fan

AU - Forbes, Andrew B.

AU - Turner, Elizabeth L.

AU - Preisser, John S.

PY - 2019/2/20

Y1 - 2019/2/20

N2 - The cluster randomized crossover design has been proposed to improve efficiency over the traditional parallel cluster randomized design, which often involves a limited number of clusters. In recent years, the cluster randomized crossover design has been increasingly used to evaluate the effectiveness of health care policy or programs, and the interest often lies in quantifying the population-averaged intervention effect. In this paper, we consider the two-treatment two-period crossover design, and develop sample size procedures for continuous and binary outcomes corresponding to a population-averaged model estimated by generalized estimating equations, accounting for both within-period and interperiod correlations. In particular, we show that the required sample size depends on the correlation parameters through an eigenvalue of the within-cluster correlation matrix for continuous outcomes and through two distinct eigenvalues of the correlation matrix for binary outcomes. We demonstrate that the empirical power corresponds well with the predicted power by the proposed formulae for as few as eight clusters, when outcomes are analyzed using the matrix-adjusted estimating equations for the correlation parameters concurrently with a suitable bias-corrected sandwich variance estimator.

AB - The cluster randomized crossover design has been proposed to improve efficiency over the traditional parallel cluster randomized design, which often involves a limited number of clusters. In recent years, the cluster randomized crossover design has been increasingly used to evaluate the effectiveness of health care policy or programs, and the interest often lies in quantifying the population-averaged intervention effect. In this paper, we consider the two-treatment two-period crossover design, and develop sample size procedures for continuous and binary outcomes corresponding to a population-averaged model estimated by generalized estimating equations, accounting for both within-period and interperiod correlations. In particular, we show that the required sample size depends on the correlation parameters through an eigenvalue of the within-cluster correlation matrix for continuous outcomes and through two distinct eigenvalues of the correlation matrix for binary outcomes. We demonstrate that the empirical power corresponds well with the predicted power by the proposed formulae for as few as eight clusters, when outcomes are analyzed using the matrix-adjusted estimating equations for the correlation parameters concurrently with a suitable bias-corrected sandwich variance estimator.

KW - cluster randomized trials

KW - crossover

KW - finite-sample correction

KW - generalized estimating equations (GEE)

KW - matrix-adjusted estimating equations (MAEEs)

KW - sandwich variance estimator

UR - http://www.scopus.com/inward/record.url?scp=85054559720&partnerID=8YFLogxK

U2 - 10.1002/sim.7995

DO - 10.1002/sim.7995

M3 - Article

VL - 38

SP - 636

EP - 649

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 4

ER -