An evaluation of constrained randomization for the design and analysis of group‐randomized trials

Li, Fan; Lokhnygina, Yuliya; Murray, David M.; Heagerty, Patrick J.; DeLong, Elizabeth R.

doi:10.1002/sim.6813

Cited by 72 publications

(161 citation statements)

References 24 publications

Supporting

Mentioning

154

Contrasting

Order By: Relevance

“…This l 2 metric accommodates both continuous and categorical variables (dummy variables are used to represent a multicategory factor) and is invariant to linear transformations of covariates with the default choice of ω k . Two alternative balance metrics were developed with constrained randomization, but they were specifically designed to balance group‐level factors and may not accommodate continuous variables. To compare with the l 2 metric, we propose an l 1 analog

B_{false(l 1 false)} = \sum_{k} {\overset{ω}{true}}_{k} false| {\overset{x}{true}}_{T k} - {\overset{x}{true}}_{C k} false|,

where

{\overset{x}{true}}_{T k}, {\overset{x}{true}}_{C k}

and

{\overset{ω}{true}}_{k}

are defined similarly, and by default,

{\overset{ω}{true}}_{k}

is taken to be the inverse standard deviation of the group means.…”

Section: Motivating Examplesmentioning

confidence: 99%

An evaluation of constrained randomization for the design and analysis of group‐randomized trials with binary outcomes

Turner

Heagerty

et al. 2017

Statistics in Medicine

Self Cite

143

View full text Add to dashboard Cite

Group-randomized trials (GRTs) are randomized studies that allocate intact groups of individuals to different comparison arms. A frequent practical limitation to adopting such research designs is that only a limited number of groups may be available, and therefore, simple randomization is unable to adequately balance multiple group-level covariates between arms. Therefore, covariate-based constrained randomization was proposed as an allocation technique to achieve balance. Constrained randomization involves generating a large number of possible allocation schemes, calculating a balance score that assesses covariate imbalance, limiting the randomization space to a pre-specified percentage of candidate allocations and randomly selecting one scheme to implement. When the outcome is binary, a number of statistical issues arise regarding the potential advantages of such designs in making inference. In particular, properties found for continuous outcomes may not directly apply, and additional variations on statistical tests are available. Motivated by two recent trials, we conduct a series of Monte Carlo simulations to evaluate the statistical properties of model-based and randomization-based tests under both simple and constrained randomization designs, with varying degrees of analysis-based covariate adjustment. Our results indicate that constrained randomization improves the power of the linearization F-test, the KC-corrected GEE t-test (Kauermann and Carroll, 2001, Journal of the American Statistical Association 96, 1387–1396) and two permutation tests when the prognostic group-level variables are controlled for in the analysis and the size of randomization space is reasonably small. We also demonstrate that constrained randomization reduces power loss from redundant analysis-based adjustment for non-prognostic covariates. Design considerations such as the choice of the balance metric and the size of randomization space are discussed.

show abstract

B_{false(l 1 false)} = \sum_{k} {\overset{ω}{true}}_{k} false| {\overset{x}{true}}_{T k} - {\overset{x}{true}}_{C k} false|,

where

{\overset{x}{true}}_{T k}, {\overset{x}{true}}_{C k}

and

{\overset{ω}{true}}_{k}

are defined similarly, and by default,

{\overset{ω}{true}}_{k}

is taken to be the inverse standard deviation of the group means.…”

Section: Motivating Examplesmentioning

confidence: 99%

An evaluation of constrained randomization for the design and analysis of group‐randomized trials with binary outcomes

Turner

Heagerty

et al. 2017

Statistics in Medicine

Self Cite

143

View full text Add to dashboard Cite

show abstract

“…16,17 Recently, Donner et al reported that ignoring matching can adversely affect other analyses, such as analyses that examine the relationship between a risk factor and an outcome; 18 for this reason, investigators considering pair-matching should consider small strata instead (e.g., strata of 4). Li et al 19 compared model-based and permutation methods in the context of constrained randomization adjusting for group-level covariates. They found that both the adjusted F-test and permutation test maintained the nominal size and had improved power under constrained randomization compared to simple randomization.…”

Section: Developments In the Analysis Of Parallel Group-randomized Trmentioning

confidence: 99%

“…They found that both the adjusted F-test and permutation test maintained the nominal size and had similar power, but cautioned that the randomization distribution must be calculated within the constrained randomization space to prevent inflating the type I error rate. 19 …”

Section: Developments In the Analysis Of Parallel Group-randomized Trmentioning

confidence: 99%

Review of Recent Methodological Developments in Group-Randomized Trials: Part 2—Analysis

et al. 2017

Self Cite

View full text Add to dashboard Cite

In 2004, Murray et al. published a review of methodological developments in both the design and analysis of group-randomized trials (GRTs). Over the last 13 years, there have been many developments in both areas. The goal of the current paper is to review developments in analysis, with a companion paper to focus on developments in design. As a pair, these papers update the 2004 review. This analysis paper includes developments in topics included in the earlier review, such as methods for parallel-arm GRTs, inference for conditional and marginal effects, and new topics including methods to account for multiple levels of clustering and alternative estimation methods such as augmented GEE, targeted maximum likelihood and quadratic inference functions. We also examine developments in dealing with missing outcome data, including doubly robust approaches, software available for analysis, and analysis of alternative group designs (including stepped wedge GRTs, network-randomized trials, pseudo-cluster randomized trials and individually-randomized group treatment trials). These alternative designs, like the parallel-arm GRT, require clustering to be accounted for in both their design and analysis.

show abstract

“…These were split into 2 strata for randomization: (1) Front Range and Eastern Plains and (2) mountains and Western Slope. Within each stratum, we used covariate constrained randomization 21,22 to achieve balanced study arms with respect to population and health resource characteristics. Variables included total population of the region, average county population, percentage at or below poverty level, median income, percentage white, percentage Hispanic, number of primary care physicians per 10,000 population, number of physician assistants/nurse practitioners per 10,000 population, unemployment rate, and number of uninsured adults aged 18 to 64 years old.…”

Section: Covariate Constrained Randomizationmentioning

confidence: 99%

A Community Engagement Method to Design Patient Engagement Materials for Cardiovascular Health

et al. 2018

View full text Add to dashboard Cite

PURPOSE EvidenceNOW Southwest is a cluster-randomized trial evaluating the differential impact on cardiovascular disease (CVD) care of engaging patients and communities in practice transformation in addition to standard practice facilitation support. The trial included development of locally tailored CVD patient engagement materials through Boot Camp Translation (BCT), a community engagement process that occurred before practice recruitment but after cluster randomization. METHODSWe introduce a cluster randomization method performed before recruitment of small to medium-size primary care practices in Colorado and New Mexico, which allowed for balanced study arms while minimizing contamination. Engagement materials for the enhanced study arm were developed by means of BCT, which included community members, practice members, and public health professionals from (1) metropolitan Denver, (2) rural northeast Colorado, (3) Albuquerque, and (4) rural southeast New Mexico. Outcome measures were messages and materials from BCTs and population characteristics of study arms after using geographic-based covariate constrained randomization. RESULTSThe 4 BCTs' messages and materials developed by the BCT groups uniquely reflected each community and ranged from family or spiritual values to early prevention or adding relevance to CVD risk. The geographic-based covariate of a cluster randomization method constrained randomization-assigned regions to study arms, allowing BCTs to precede practice recruitment, reduce contamination, and balance populations.CONCLUSIONS Cluster-randomized trials with community-based interventions present study design and implementation challenges. The BCTs elicited unique contextual messages and materials, suggesting that interventions designed to help primary care practices decrease CVD risk may not be one size fits all. 3 The benefits of engaging patients and families in the care of individual patients is well established.4-11 The outcomes of engaging communities, patients, and families at the practice level, however, are less well documented.12,13 The EvidenceNOW: Advancing Heart Health in Primary Care is an Agency for Healthcare Research and Quality grant initiative dedicated to using the latest evidence to improve the heart health of Americans. EvidenceNOW Southwest is a regional cooperative that has sought to test the differential impacts on cardiovascular disease (CVD) of interventions focused on patient, family, and community engagement to standard practice transformation approaches. S59In EvidenceNOW Southwest, enhanced-arm practices received practice facilitation plus additional support to engage patients and families in practice transformation. These practices were able to order free heart health materials that were developed through a community engagement method known as Boot Camp Translation (BCT) in 4 communities within the enhanced-arm geographic regions (2 in Colorado, 2 in New Mexico). Community members and health professionals worked together to learn about complex health to...

show abstract

An evaluation of constrained randomization for the design and analysis of group‐randomized trials

Cited by 72 publications

References 24 publications

An evaluation of constrained randomization for the design and analysis of group‐randomized trials with binary outcomes

An evaluation of constrained randomization for the design and analysis of group‐randomized trials with binary outcomes

Review of Recent Methodological Developments in Group-Randomized Trials: Part 2—Analysis

A Community Engagement Method to Design Patient Engagement Materials for Cardiovascular Health

Contact Info

Product

Resources

About