Benefits and harms of adhesion barriers for abdominal surgery: a systematic review and meta-analysis

Summary Background Formation of adhesions after peritoneal surgery results in high morbidity. Barriers to prevent adhesion are seldom applied, despite their ability to reduce the severity of adhesion formation. We evaluated the benefits and harms of four adhesion barriers that have been approved for clinical use. Methods In this systematic review and meta-analysis, we searched PubMed, CENTRAL, and Embase for randomised clinical trials assessing use of oxidised regenerated cellulose, hyaluronate carboxymethylcellulose, icodextrin, or polyethylene glycol in abdominal surgery. Two researchers independently identified reports and extracted data. We compared use of a barrier with no barrier for nine predefined outcomes, graded for clinical relevance. The primary outcome was reoperation for adhesive small bowel obstruction. We assessed systematic error, random error, and design error with the error matrix approach. This study is registered with PROSPERO, number CRD42012003321. Findings Our search returned 1840 results, from which 28 trials (5191 patients) were included in our meta-analysis. The risks of systematic and random errors were low. No trials reported data for the effect of oxidised regenerated cellulose or polyethylene glycol on reoperations for adhesive small bowel obstruction. Oxidised regenerated cellulose reduced the incidence of adhesions (relative risk [RR] 0·51, 95% CI 0·31–0·86). Some evidence suggests that hyaluronate carboxymethylcellulose reduces the incidence of reoperations for adhesive small bowel obstruction (RR 0·49, 95% CI 0·28–0·88). For icodextrin, reoperation for adhesive small bowel obstruction did not differ significantly between groups (RR 0·33, 95% CI 0·03–3·11). No barriers were associated with an increase in serious adverse events. Interpretation Oxidised regenerated cellulose and hyaluronate carboxymethylcellulose can safely reduce clinically relevant consequences of adhesions. Funding None.