Are meta-analyses of Chinese herbal medicine trials trustworthy and clinically applicable? A cross-sectional study

Abstract

Ethnopharmacological relevanceMeta-analysis (MA) on Chinese herbal medicine (CHM) trials is increasingly published and indexed in major international databases but their trustworthiness and clinical applicability is uncertain. We aimed to assess the characteristics and methodological quality of MA on CHM. Materials and MethodsCross-sectional study. MA published during 1993–2013 was sampled from MEDLINE, EMBASE, Cochrane Database of Systematic Reviews and Database of Abstracts of Reviews of Effect. Bibliographical characteristics were abstracted and methodological quality was assessed using the validated AMSTAR tool by two independent reviewers. ResultsTotal of 201 MA were included and half were published in or after 2009. Only 7.5% being updates of previous reviews. Majority are published in journals with low or no impact factor, with a median of 1.5. These MA demonstrated methodological strengths in ensuring comprehensive literature search, providing characteristics of the included studies, assessing the scientific quality of included studies and appropriately using the scientific quality of included studies in formulating conclusions. Nevertheless, weaknesses in protocol provision, listing of included and excluded studies, inclusion of grey literature, use of appropriate meta-analytic technique as well as reporting of funding sources were prevalent. CHM and control interventions pooled in majority of MA are found to have substantial clinical heterogeneity in terms of composition, dosage form and route of administration. ConclusionsThere are rooms for improvement in methodological rigor, and in choosing clinically homogenous interventions and control for statistical pooling. These shortcomings limit the trustworthiness and clinical applicability of existing MA on CHM trials. To overcome the limitations of pair-wise meta-analysis in synthesizing trials comparing different CHM and control interventions, the potential of network meta-analysis should be explored.