Quality of Meta-Analyses for Randomized Trials in the Field of Hypertension: an Updated and Improved Systematic Review.

Publications of hypertension-related meta-analyses (MAs) have increased exponentially in the past 25 years and now average 8/month. Theoretically, this is facilitating evidence-based management of patients. However, some practitioners and authors of guidelines have questioned the quality of published MAs. By extending a prior review, we have assessed the quality of 212 hypertension-related meta-analyses over 5 years based on systematically searching three computerized libraries. Seventeen criteria grouped into four domains of quality yielded the following results: (1) Assessment of trial quality was accomplished in 89% of MAs, and 38% analyzed trials in subgroups of trial quality where appropriate. (2) All three measures of heterogeneity (I 2, tau, and P for heterogeneity) were reported in 36%, reflecting the failure to report tau, the standard deviation of the main effect. (3) Publication bias was assessed in 75%, and 43% of MAs used a statistical test for publication bias. (4) Regarding transparency, 9 to 31% of MAs reported problems in the previous three domains in the article's abstract. Journal impact factor reporting the MAs declined significantly over 5 years. The percent with criteria of quality in a MA was modestly correlated with journal impact factor (R 2 = 0.05, P = 0.001). False-positive results from inappropriate application of the DerSimonian-Laird model affected 25% of articles, which reported these false positives in the article's abstract in 72%. No more than 25% of MAs had 67% or more of the criteria of quality. In conclusion, skepticism of hypertension-related MAs is justified, but their quality can be readily corrected.