Characteristics and methodological quality of meta-analyses on hypertension treatments: a cross-sectional study

Article type
Year
Authors
Chung VC1, Du XJ1, Ho RS1, Lee CC1, Yip BH1, Wong MC1, Wong SY1, Wu XY1
1The Chinese University of Hong Kong, China
Abstract
Background: Hypertension is one of the top contributors to the global disease burden. Identifying effective interventions for hypertension is a major global public health challenge. Evidence from systematic reviews (SR) is of great importance for the management of hypertension. Methodological quality of meta-analysis on hypertension treatments can affect treatment decisions.

Objectives: To investigate the methodological quality of meta-analyses of hypertension treatments.

Methods: We searched the Cochrane Database of Systematic Reviews and the Database of Abstracts of Reviews of Effect. SRs with at least one meta-analysis on hypertension treatment effect were considered eligible. We assessed methodological quality with the validated Assessing the Methodological Quality of Systematic Reviews (AMSTAR) tool.

Results: We identified 158 meta-analyses on hypertension treatments, with 32 (20.3%) being Cochrane meta-analyses and 16 (10.1%) being an update of a previous meta-analysis. Overall, methodological quality was unsatisfactory in the following aspects: comprehensive reporting of the sources of support (1.9%), provision of included and excluded list of studies (22.8%), inclusion of grey literature (27.2%), and inclusion of protocols (32.9%). The 126 non-Cochrane meta-analyses had poor performance on almost all the methodological items except for providing characteristics (63.5%) and assessing the scientific quality (61.1%) of included studies. Among non-Cochrane meta-analyses, those that focused on non-pharmacological treatments were more likely to consider the scientific quality of included studies when drawing conclusions; meta-analyses published recently were better at using appropriate statistical methods and assessing publication bias. The 32 Cochrane meta-analyses generally had good methodological quality except for comprehensive reporting of the sources of support.

Conclusions: Our results highlight the need for cautious interpretation of these meta-analyses, especially among physicians and policy-makers when guidelines are formulated. Future meta-analyses should address the shortcomings in these methodological items.