The perils of conducting meta-analyses of observational data.

Guy Harling,Sv Subramanian

doi:10.7448/ias.17.1.19112

Abstract

Dear Editors, We read with interest Li and colleagues’ recent article in this journal, which conducted a systematic review and meta-analysis of the literature on associations between intimate partner violence (IPV) and HIV infection in women. The authors found a significant positive association between various types of IPV and HIV, concluding that “physical violence, sexual violence, a combination of physical and sexual violence, and any type of IPV were associated with HIV infection in women (p. 7)”. (Published: 9 June 2014) Citation: Harling G and Subramanian SV. Journal of the International AIDS Society 2014, 17 :19112 http://www.jiasociety.org/index.php/jias/article/view/19112 | http://dx.doi.org/10.7448/IAS.17.1.19112

Highlights

Dear Editors, We read with interest Li and colleagues’ recent article in this journal [1], which conducted a systematic review and metaanalysis of the literature on associations between intimate partner violence (IPV) and HIV infection in women
The authors found a significant positive association between various types of IPV and HIV, concluding that ‘‘physical violence, sexual violence, a combination of physical and sexual violence, and any type of IPV were associated with HIV infection in women (p. 7)’’ [1]
We were surprised to see that, in conducting their meta-analysis, the authors report that our data show significant positive relationships between HIV and several types of IPV (Figures 2 and 3 in [1])

Summary

Introduction

Dear Editors, We read with interest Li and colleagues’ recent article in this journal [1], which conducted a systematic review and metaanalysis of the literature on associations between intimate partner violence (IPV) and HIV infection in women. We were surprised to see that, in conducting their meta-analysis, the authors report that our data show significant positive relationships between HIV and several types of IPV (Figures 2 and 3 in [1]).

Results

Conclusion