Availability of evidence and comparative effectiveness for surgical versus drug interventions: an overview of systematic reviews and meta-analyses

Abstract

ObjectivesThis study aims to examine the prevalence of comparisons of surgery to drug regimens, the strength of evidence of such comparisons and whether surgery or the drug intervention was favoured.DesignSystematic review of systematic reviews (umbrella review).Data sourcesCochrane Database of Systematic Reviews.Eligibility criteriaSystematic reviews attempt to compare surgical to drug interventions.Data extractionWe extracted whether the review found any randomised controlled trials (RCTs) for eligible comparisons. Individual trial results were extracted directly from the systematic review.SynthesisThe outcomes of each meta-analysis were resynthesised into random-effects meta-analyses. Egger’s test and excess significance were assessed.ResultsOverall, 188 systematic reviews intended to compare surgery versus drugs. Only 41 included data from at least one RCT (total, 165 RCTs) and covered a total of 103 different outcomes of various comparisons of surgery versus drugs. A GRADE assessment was performed by the Cochrane reviewers for 87 (83%) outcomes in the reviews, indicating the strength of evidence was high in 4 outcomes (4%), moderate in 22 (21%), low in 27 (26%) and very low in 33 (32%). Based on 95% CIs, the surgical intervention was favoured in 38/103 (37%), and the drugs were favoured in 13/103 (13%) outcomes. Of the outcomes with high GRADE rating, only one showed conclusive superiority in our reanalysis (sphincterotomy was better than medical therapy for anal fissure). Of the 22 outcomes with moderate GRADE rating, 6 (27%) were inconclusive, 14 (64%) were in favour of surgery and 2 (9%) were in favour of drugs. There was no evidence of excess significance.ConclusionsThough the relative merits of surgical versus drug interventions are important to know for many diseases, high strength randomised evidence is rare. More randomised trials comparing surgery to drug interventions are needed.