View Article

Reviews provide summaries of evidence to answer important practice and 
						policy questions without readers having to spend the time and effort to 
						summarise for themselves. Reviews vary greatly, ranging from unstructured 
						reviews 
				
		
		
				
						(e.g., editorials or journalistic reviews) 
				
				to systematic reviews 
		
		
				(e.g., those produced 
				by the Cochrane Collaboration). 
		
		Unstructured reviews, often produced by experts in 
		the subject area, are usually based on unscientific, subjective, and nonreplicable 
		methods of information collection and interpretation. In contrast, 
		systematic reviews have explicit, scientific, and comprehensive descriptions of 
		their objectives and methods. It is important that those using reviews to 
		inform practice and policy are able to judge what makes a good quality review. 
		
				• 
				Explicit methods limit bias in identifying and rejecting studies 
		
		
				• 
				Conclusions are more reliable and accurate depending on methods 
		
		used. Lots of information can be assimilated quickly by providers, 
		researchers, and policymakers, possibly reducing delay between 
		discovery and implementing effective diagnostic and therapeutic
		strategies and enables formal comparison of different study results to 
		
				establish generalisability of findings and consistency 
				
						(lack of heterogeneity)
				
				.
		
		
				Many checklists have been published the assess quality of systematic reviews. 
				
						They all have strengths and limitations. Eight questions 
						‐ 
						derived from various 
				
				
						checklists 
						‐ follow as one option to assess the quality of a review. 
				
				
						• 
						Did the review address an appropriate and clearly focused question? 
				
				Systematic reviews should address precise questions, focused in terms of the 
				intervention under consideration (for example, medical antishock trousers), 
				the group of patients ("population") being studied (for example, patients after 
				trauma) and the outcomes assessed (for example, death, length of stay in 
				hospital, and so on). 
				
						• 
						Did the authors of the review make efforts to search for the 
				
				appropriate studies to answer the question? 
				Searches for relevant primary studies should try to find as comprehensive a list 
				as possible, whether published or unpublished, limiting the opportunity for 
				publication bias. Search strategies should be clearly defined, with any 
				restrictions placed on the search made explicit. The primary sources for 
				searching are bibliographic databases (e.g. Medline, Cochrane Library). These 
				may miss important papers and so should be supplemented by follow up of 
				reference lists in relevant studies, contact with researchers in the field, hand
				
						searching subject specific journals, and exploring grey 
				
				
						
								(unpublished) 
						
						literature. 
				
				
						• 
						3: Were criteria for deciding studies to include in review clearly stated? 
				
				
						Studies identified by the search should be sorted, using pre
						‐determined criteria 
				
				
						(to avoid selection bias), 
				
				into those to be included or excluded from the review. 
				
						• 
						4: Did authors adequately assess the quality of studies included ? 
				
				Authors should assess methodological quality of studies included in review, as 
				this may affect the reliability of results, so the appropriate weight can be given 
				to the different studies in summarising results and in drawing conclusions. 
				Such criteria assess the opportunity for bias in the study, giving greatest 
				weight to studies with the highest quality and least opportunity for bias. 
				
						• 
						5: If included studies were combined, was this appropriate and were 
				
				the methods used suitable? 
				Where appropriate, the results of studies are combined through narrative 
				
						(qualitative) 
				
				
						assessment of evidence. This may be accompanied by meta‐analysis 
				
				
						(statistical analysis), 
				
				the intention being to provide a summary estimate of effect 
				size with an investigation of whether it holds in different studies, settings and 
				participants, 
				
						but studies, should only be combined where the included studies are similar in 
						terms of patients, settings, and outcomes used
				
				. 
				
						• 
						6: What were the results of the review? 
				
				Results of the review should be presented clearly using appropriate measures, 
				which may be odds ratios, relative risks, or numbers needed to treat. Also, 
				
						there should be an indication of the precision of these 
						‐ 
						usually with 95% 
				
				confidence intervals. 
				
						• 
						7: Have the results been interpreted sensibly, and how generalisable 
				
				are the results of the review? 
				Interpretation of results should accurately reflect the results themselves, as 
				well as being placed in the context of the clinical question posed. The 
				
						generalisability (
						
								external validity) 
						
						of the results should be assessed also. 
				
				
						• 
						8: Have the methods of the review been applied appropriately? 
				
				Where authors applied specific methods in the review, like inclusion and quality 
				criteria or data extraction, they should state how these were undertaken. e.g., 
				authors should indicate whether criteria were applied by one reviewer or two 
				
						reviewers independently. This allows an assessment of reviewer Bias 
						‐ 
						that is, 
				
				whether reviewers have been influenced by factors such as the results rather 
				than the relevance or quality.
		
		
				
						Source 
						BMJ