Random effects meta-analysis of event outcome in the framework of the generalized linear mixed model with applications in sparse data

Abstract

We consider random effects meta-analysis where the outcome variable is the occurrence of some event of interest. The data structures handled are where one has one or more groups in each study, and in each group either the number of subjects with and without the event, or the number of events and the total duration of follow-up is available. Traditionally, the meta-analysis follows the summary measures approach based on the estimates of the outcome measure(s) and the corresponding standard error(s). This approach assumes an approximate normal within-study likelihood and treats the standard errors as known. This approach has several potential disadvantages, such as not accounting for the standard errors being estimated, not accounting for correlation between the estimate and the standard error, the use of an (arbitrary) continuity correction in case of zero events, and the normal approximation being bad in studies with few events. We show that these problems can be overcome in most cases occurring in practice by replacing the approximate normal within-study likelihood by the appropriate exact likelihood. This leads to a generalized linear mixed model that can be fitted in standard statistical software. For instance, in the case of odds ratio meta-analysis, one can use the non-central hypergeometric distribution likelihood leading to mixed-effects conditional logistic regression. For incidence rate ratio meta-analysis, it leads to random effects logistic regression with an offset variable. We also present bivariate and multivariate extensions. We present a number of examples, especially with rare events, among which an example of network meta-analysis.

Meta-analysis using individual participant data: one-stage and two-stage approaches, and why they may differ

“…One‐stage and two‐stage methods may yield different summary results when the second stage of the two‐stage method assumes that study treatment effect estimates ( ${\hat{\theta }}_i$) have a normal sampling distribution and that their variances ( $Var\left({\hat{\theta }}_i\right)true)$ are known 3, 11, 56, 57. This first assumption is based on the central limit theorem, and the second assumes that the variance is estimated with reasonable accuracy.…”

Single incision versus standard multiport laparoscopic cholecystectomy: Up-dated systematic review and meta-analysis of randomized trials

“…Conventional meta-analytical methods do not perform well with sparse binary data (procedure failure, complications). 23,29,30 For convenience, they were summarized as Mantel-Haenszel (M-H) odds ratio (OR). However, zero frequency cell correction and omission of zero-event studies (inherent to the M-H method) can introduce bias.…”

“…However, zero frequency cell correction and omission of zero-event studies (inherent to the M-H method) can introduce bias. 23,29,30 Hence, we implemented methods that use all trials and do not employ corrections: a) a randomeffects method for sparse dichotomous data by Shuster et al 29 and b) random-effects analysis within the bivariate binomial-normal model (BN). 30 The latter provides estimates of event incidence for each treatment and of treatment difference (OR).…”

Preventive Pharmacologic Treatments for Episodic Migraine in Adults

“…We used meta-analysis of RCTs for evaluating drug safety based on all available trials. 46 We analyzed sparse adverse effects data with various statistical methods 43,[47][48][49][50][51] for robustness by comparing statistical significance and magnitude of the harms. In cases of multi-arm trials, we created a single pair-wise comparison.…”