Allowing for informative missingness in aggregate data meta-analysis with continuous or binary outcomes: Extensions to metamiss

Abstract

Missing outcome data can invalidate the results of randomized trials and their meta-analysis. However, addressing missing data is often a challenging issue because it requires untestable assumptions. The impact of missing outcome data on the meta-analysis summary effect can be explored by assuming a relationship between the outcome in the observed and the missing participants via an informative missingness parameter. The informative missingness parameters cannot be estimated from the observed data, but they can be specified, with associated uncertainty, using evidence external to the meta-analysis, such as expert opinion. The use of informative missingness parameters in pairwise meta-analysis of aggregate data with binary outcomes has been previously implemented in Stata by the metamiss command. In this article, we present the new command metamiss2, which is an extension of metamiss for binary or continuous data in pairwise or network meta-analysis. The command can be used to explore the robustness of results to different assumptions about the missing data via sensitivity analysis.

Keywords: informative missingness; meta-analysis; metamiss2; mixed treatment comparison; sensitivity analysis; st0540.

Figure 1

Plot of the summary mean difference of mirtazapine versus placebo and the respective 95% confidence interval (random-effects meta-analysis) for various values of sd(logimrom) under the imrom = 1 assumption