Evaluation of Systematic Reviews of Interventions for Retina and Vitreous Conditions.

Abstract

Importance\nPatient care and clinical practice guidelines should be informed by evidence from reliable systematic reviews. The reliability of systematic reviews related to forthcoming guidelines for retina and vitreous conditions is unknown.\n\n\nObjectives\nTo summarize the reliability of systematic reviews on interventions for 7 retina and vitreous conditions, describe characteristics of reliable and unreliable systematic reviews, and examine the primary area in which they appeared to be lacking.\n\n\nDesign, Setting, and Participants\nA cross-sectional study of systematic reviews was conducted. Systematic reviews of interventions for retina- and vitreous-related conditions in a database maintained by the Cochrane Eyes and Vision United States Satellite were identified. Databases that the reviewers searched, whether any date or language restrictions were applied, and bibliographic information, such as year and journal of publication, were documented. The initial search was conducted in March 2007, and the final update was performed in July 2018. The conditions of interest were age-related macular degeneration; diabetic retinopathy; idiopathic epiretinal membrane and vitreomacular traction; idiopathic macular hole; posterior vitreous detachment, retinal breaks, and lattice degeneration; retinal and ophthalmic artery occlusions; and retinal vein occlusions. The reliability of each review was evaluated using prespecified criteria. Data were extracted by 2 research assistants working independently, with disagreements resolved through discussion or by 1 research assistant with verification by a senior team member.\n\n\nMain Outcomes and Measures\nProportion of reviews that meet all of the following criteria: (1) defined eligibility criteria for study selection, (2) described conducting a comprehensive literature search, (3) reported assessing risk of bias in included studies, (4) described using appropriate methods for any meta-analysis performed, and (5) provided conclusions consistent with review findings.\n\n\nResults\nA total of 327 systematic reviews that addressed retina and vitreous conditions were identified; of these, 131 reviews (40.1%) were classified as reliable and 196 reviews (59.9%) were classified as not reliable. At least 1 reliable review was found for each of the 7 retina and vitreous conditions. The most common reason that a review was classified as not reliable was lack of evidence that a comprehensive literature search for relevant studies had been conducted (149 of 196 reviews [76.0%]).\n\n\nConclusion and Relevance\nThe findings of this study suggest that most systematic reviews that addressed interventions for retina and vitreous conditions were not reliable. Systematic review teams and guideline developers should work with information professionals who can help navigate sophisticated and varied syntaxes required to search different resources.