Article type
Year
Abstract
Background:
Incorporating cluster randomized trials into meta-analyses is challenging because appropriate standard errors accounting for clustering are not always reported. Systematic reviews of such trials often use a single external estimate of the intraclass correlation coefficient (ICC) to adjust standard errors and facilitate meta-analyses.
Objectives:
To investigate whether developing a database of ICCs and using posterior predictive distributions to impute ICCs can improve meta-analyses by accounting for ICC uncertainty using a large systematic review of the effects of diabetes quality improvement interventions.
Methods:
We combined internal ICC estimates from studies included in our review with those obtained from external sources. For outcomes with two or more available ICC estimates, we constructed posterior predictive ICC distributions in a Bayesian framework. For a selected continuous outcome, glycated hemoglobin (HbA1c), we compared the impact of incorporating a single ICC versus imputing ICCs drawn from the posterior predictive distribution when estimating the effect of intervention components on post treatment mean in a case study of diabetes quality improvement trials.
Results:
Using internal and external ICC estimates, we were able to construct a database of 59 ICCs for 12 of the 13 review outcomes (range 1-10 per outcome) and estimate the posterior predictive ICC distribution for 11 review outcomes. Synthesized results of continuous outcome HbA1c were not markedly changed by our approach.
Conclusions:
Building posterior predictive distributions to impute missing ICCs is a feasible approach to facilitate meta-analyses of cluster randomized trials. Further work is needed to establish whether the application of these methods leads to improved review inferences.
Patient or healthcare consumer involvement:
None
Incorporating cluster randomized trials into meta-analyses is challenging because appropriate standard errors accounting for clustering are not always reported. Systematic reviews of such trials often use a single external estimate of the intraclass correlation coefficient (ICC) to adjust standard errors and facilitate meta-analyses.
Objectives:
To investigate whether developing a database of ICCs and using posterior predictive distributions to impute ICCs can improve meta-analyses by accounting for ICC uncertainty using a large systematic review of the effects of diabetes quality improvement interventions.
Methods:
We combined internal ICC estimates from studies included in our review with those obtained from external sources. For outcomes with two or more available ICC estimates, we constructed posterior predictive ICC distributions in a Bayesian framework. For a selected continuous outcome, glycated hemoglobin (HbA1c), we compared the impact of incorporating a single ICC versus imputing ICCs drawn from the posterior predictive distribution when estimating the effect of intervention components on post treatment mean in a case study of diabetes quality improvement trials.
Results:
Using internal and external ICC estimates, we were able to construct a database of 59 ICCs for 12 of the 13 review outcomes (range 1-10 per outcome) and estimate the posterior predictive ICC distribution for 11 review outcomes. Synthesized results of continuous outcome HbA1c were not markedly changed by our approach.
Conclusions:
Building posterior predictive distributions to impute missing ICCs is a feasible approach to facilitate meta-analyses of cluster randomized trials. Further work is needed to establish whether the application of these methods leads to improved review inferences.
Patient or healthcare consumer involvement:
None