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




Poster session 4


Tuesday 25 October 2016 - 15:30 to 16:00


All authors in correct order:

Wu XY1, Lee CC1, Ho RS1, Du XJ1, Wong CH1, Yip BH1, Wong LK1, Lau AY1, Wong SY1, Chung VC1
1 The Chinese University of Hong Kong, China
Presenting author and contact person

Presenting author:

Xin Yin Wu

Contact person:

Abstract text
Background: Methodological limitations among meta-analyses (MAs) can impact trustworthiness of conclusions, and subsequently affect the quality of treatment decision making.

Objectives: The aim of this study was to evaluate the methodological rigor of MAs on stroke treatments.

Methods: A cross-sectional study was conducted. MAs on stoke treatments published between 2000-2014 were retrieved from databases. The methodological quality of the included MAs was assessed using the validated AMSTAR (A MeaSurement Tool to Assess systematic Reviews) instrument. Association between characteristics of MAs and their individual score in each AMSTAR item was analyzed with regression analysis.

Results: We included 179 MAs, of which 83 (46.4%) were Cochrane Reviews. Most of them (65.4%) focused on non-pharmacological treatments. The included MAs generally did well in providing information on conflict of interests (84.4%); performing a comprehensive literature search (79.9%); using appropriate methods to combine results (74.9%); and assessing and documenting the scientific quality of primary studies (77.1%). However, only 39.7% assessed publication bias; 47.5% conducted duplicate study selection and data extraction. About half (49.2%) provided characteristics of the primary studies, or mentioned harms of the treatment (45.8%). Cochrane Reviews generally had good methodological quality. Results from regression analyses showed that more recently published MAs, those produced by Asian authors, and those focused on non-pharmacological treatments, were associated with better methodological quality.

Conclusions: Overall, the methodological quality of included MAs was mediocre. Improvements are needed in assessing publication bias, conducting duplicate literature selection and data extraction, providing characteristics of included studies, and providing information on harmful effects of treatment.