Methodological quality of systematic reviews used in clinical practice guidelines: focus on clinical imaging

Purpose To summarise the methodological quality of systematic reviews used by clinical practice guidelines, focusing on the field of clinical imaging. Methods We searched PubMed (up to January 2021), EMBASE, and Cochrane Methodology Register (both up to April 2021) for studies quantitatively assessing the methodological quality of systematic reviews referenced in clinical practice guidelines. We summarised the overall quality, methodological areas with low quality, and commonly used assessment tools with their applicability to systematic reviews of diagnostic accuracy studies. Results 2433 records were retrieved and 12 studies were included, two of which investigated imaging guidelines. The overall methodological quality of systematic reviews was moderate in eight studies (67%), high in one study (8%), and low in one study (8%). One study (8%) assessed merely the risk of bias and found unclear or high risk in general. The Assessment of Multiple Systematic reviews (AMSTAR) tool was used in ten studies, and systematic reviews in imaging guidelines received the lowest AMSTAR scores (mean 4/11, range 0/11-9/11 in an interventional radiology guideline; mean 3/11, range 0/11-11/11 in a mammography screening guideline), correlating to low and borderline moderate quality. Methodological areas frequently found with low quality were financial disclosure, publication bias assessment, and review of grey literature. All commonly used tools are applicable, with or without minor modifications, to assessing diagnostic systematic reviews. Conclusion The methodological quality of systematic reviews in guidelines for intervention radiology and mammography screening was lower than that in guidelines for other topics. Diagnostic imaging guidelines should also be assessed for the quality of systematic review evidence underlying their recommendations.