Article type
Year
Abstract
Background: Observational studies, where there is no randomized assignment of subjects into a treated group versus a control group, focus on the change and difference before and after exposure or intervention. The importance of observational studies and when to include them has been detailed in the Cochrane Handbook; GRADE is required to rate the quality of evidence, especially considering the three upgrade factors (large effect, all plausible biases, dose-response relation) with so-called 'low-quality' observational studies. No study, however, has examined the use of GRADE in Cochrane Reviews (CRs) for observational studies.
Objectives: Investigate GRADE evidence assessment with observational studies in CRs.
Methods: We conducted a full text search in the Cochrane Library using 'GRADE' or 'Grading of Recommendations Assessment, Development and Evaluation', and screened the records for CRs using GRADE for observational studies.
Results: We included 21 CRs, published from 2008 to 2014. Among these, evidence from observational studies was graded in 19 (90%) of the CRs, and the EP and summary of findings were reported in 18 (86%) and 17 (82%) respectively. Fifteen (71%) of the included studies reported factors that affected the quality of evidence. Details of factors under consideration included two (10%) large effect, and none for all plausible biases or dose-response relation. Whereas the downgrade factors were applied for observational studies as follows: 11 (52%) risk of bias, 12 (57%) imprecision, nine (43%) inconsistency, two (10%) indirectness, and none for publication bias. In addition, 13 (62%) CRs included randomized control studies along with observational studies, four (19%) of which assessed the evidence all mixed; six (29%) rated the evidence of these two types of study separately; and three (14%) were unclear.
Conclusions: A low proportion of CRs for observational studies were found and the use and reporting of the evidence assessment was irregular. Considering the existing problems, we suggest reviewers who include observational studies in their CR should learn how to use GRADE systematically, to achieve appropriate use and reporting.
Objectives: Investigate GRADE evidence assessment with observational studies in CRs.
Methods: We conducted a full text search in the Cochrane Library using 'GRADE' or 'Grading of Recommendations Assessment, Development and Evaluation', and screened the records for CRs using GRADE for observational studies.
Results: We included 21 CRs, published from 2008 to 2014. Among these, evidence from observational studies was graded in 19 (90%) of the CRs, and the EP and summary of findings were reported in 18 (86%) and 17 (82%) respectively. Fifteen (71%) of the included studies reported factors that affected the quality of evidence. Details of factors under consideration included two (10%) large effect, and none for all plausible biases or dose-response relation. Whereas the downgrade factors were applied for observational studies as follows: 11 (52%) risk of bias, 12 (57%) imprecision, nine (43%) inconsistency, two (10%) indirectness, and none for publication bias. In addition, 13 (62%) CRs included randomized control studies along with observational studies, four (19%) of which assessed the evidence all mixed; six (29%) rated the evidence of these two types of study separately; and three (14%) were unclear.
Conclusions: A low proportion of CRs for observational studies were found and the use and reporting of the evidence assessment was irregular. Considering the existing problems, we suggest reviewers who include observational studies in their CR should learn how to use GRADE systematically, to achieve appropriate use and reporting.