Meta-analysis with zero-event studies: a comparative study with application to COVID-19 data

Background Meta-analysis is a statistical method to synthesize evidence from a number of independent studies, including those from clinical studies with binary outcomes. In practice, when there are zero events in one or both groups, it may cause statistical problems in the subsequent analysis. Methods In this paper, by considering the relative risk as the effect size, we conduct a comparative study that consists of four continuity correction methods and another state-of-the-art method without the continuity correction, namely the generalized linear mixed models (GLMMs). To further advance the literature, we also introduce a new method of the continuity correction for estimating the relative risk. Results From the simulation studies, the new method performs well in terms of mean squared error when there are few studies. In contrast, the generalized linear mixed model performs the best when the number of studies is large. In addition, by reanalyzing recent coronavirus disease 2019 (COVID-19) data, it is evident that the double-zero-event studies impact the estimate of the mean effect size. Conclusions We recommend the new method to handle the zero-event studies when there are few studies in a meta-analysis, or instead use the GLMM when the number of studies is large. The double-zero-event studies may be informative, and so we suggest not excluding them.

Meta-analysis With Zero-event Studies: A Comparative Study With Application to COVID-19 Data

Background: Meta-analysis is a statistical method to synthesize evidence from a number of independent studies, including those from clinical studies with binary outcomes. In practice, when there are zero events in one or both groups, it may cause statistical problems in the subsequent analysis. Methods: In this paper, by considering the relative risk as the effect size, we conduct a comparative study that consists of four continuity correction methods and another state-of-the-art method without the continuity correction, namely the generalized linear mixed models. To further advance the literature, we also introduce a new method of the continuity correction for estimating the relative risk. Results: From the simulation studies, the new method performs well in terms of mean squared error when there are few studies. In contrast, the generalized linear mixed model performs the best when the number of studies is large. In addition, by reanalyzing a recent COVID-19 data, it is evident that the double-zero-event studies impact on the estimate of the mean effect size.Conclusion: We recommend the new method to handle the zero-event studies when there are only few studies in the meta-analysis, or instead use the GLMM when the number of studies is large. The double-zero-event study may beinformative, and so we suggest not excluding them.

Double-zero-event studies matter: a re-evaluation of physical distancing, face masks, and eye protection for preventing person-to-person transmission of COVID-19 and its policy impact

Objectives: High-quality meta-analyses on COVID-19 are in urgent demand for evidence-based decision making. However, conventional approaches exclude double-zero-event studies (DZS) from meta-analyses. We assessed whether including such studies impacts the conclusions in a recent systematic urgent review on prevention measures for preventing person-to-person transmission of COVID-19. Study designs and settings: We extracted data for meta-analyses containing DZS from a recent review that assessed the effects of physical distancing, face masks, and eye protection for preventing person-to-person transmission. A bivariate generalized linear mixed model was used to re-do the meta-analyses with DZS included. We compared the synthesized relative risks (RRs) of the three prevention measures, their 95% confidence intervals (CI), and significance tests (at the level of 0.05) including and excluding DZS. Results: The re-analyzed COVID-19 data containing DZS involved a total of 1,784 participants who were not considered in the original review. Including DZS noticeably changed the synthesized RRs and 95% CIs of several interventions. For the meta-analysis of the effect of physical distancing, the RR of COVID-19 decreased from 0.15 (95% CI, 0.03 to 0.73) to 0.07 (95% CI, 0.01 to 0.98). For several meta-analyses, the statistical significance of the synthesized RR was changed. The RR of eye protection with a physical distance of 2 m and the RR of physical distancing when using N95 respirators were no longer statistically significant after including DZS. Conclusions: DZS may contain useful information. Sensitivity analyses that include DZS in meta-analysis are recommended.