Comparison of recommendations from cirrhosis Clinical Practice Guidelines and from Cochrane systematic reviews

Abstract Background: Clinical practice guidelines (CPGs) contain recommendations intended to aid clinicians in providing optimal care for patients and should be derived from the best quality evidence available. The highest quality evidence is obtained through systematic reviews using Cochrane methodology. This exploratory study aimed to determine the proportion of guideline statements from major liver organisations in USA and Europe that were consistent with the conclusions of recent systematic reviews using Cochrane methodology. Methods: We selected four topics related to decompensated cirrhosis on which our group has recently published Cochrane systematic reviews (variceal bleeding; spontaneous bacterial peritonitis; ascites; hepatorenal syndrome). We performed a purposive sample of CPGs from leading organisations covering these topics: American Association for the Study of Liver Diseases (AASLD); British Society of Gastroenterology (BSG); European Association for the Study of the Liver (EASL); and National Institute for Health and Care Excellence (NICE) until January 2022. For each guideline recommendation statement, two researchers independently assessed whether there was agreement in the statement and the strength of recommendation given by guideline authors when compared to those derived from the Cochrane reviews. We meta-analysed the proportion of agreement using ‘meta’ package for R utilising the ‘Generalised Linear Mixed Model’ method for meta-analysis of proportions. Results: We identified 9 CPGs containing 62 guideline statements that were also covered in 7 Cochrane reviews. Agreement between the guideline authors and independent assessment was 0.145 (95% CI: 0.077 to 0.256), i.e., there was disagreement in 85.5% of recommendation statements. There was disagreement on four of the 62 guideline statements about the use or consideration of an intervention when there was no evidence to support their use or consideration. The remaining disagreements related to the disagreement in the strength of recommendation: in each of these disagreements, the guideline author had provided a strong recommendation for or against an intervention when there was only weak evidence for or against the intervention. For each disagreement in the strength of recommendation, the strength of recommendation was overstated rather than understated by the guideline authors. The disagreement could not be explained by the difference in the search dates between the guidelines and Cochrane reviews. Conclusion: The strength of recommendation given in the majority of CPGs from leading liver organisations on decompensated liver cirrhosis did not match those derived from Cochrane reviews. Our findings suggest a systematic issue with guideline developers overstating the strength of recommendations, which has the potential for direct impact on patient care.