Is there scope to improve the selection of patients with alcohol-related liver disease for referral to secondary care? A retrospective analysis of primary care referrals to a UK liver centre, incorporating simple blood tests.

OBJECTIVES Twenty per cent of people with alcohol use disorders develop advanced fibrosis and warrant referral to secondary care. Improving outcomes in alcohol-related liver disease (ArLD) relies on its earlier detection in primary care with non-invasive tests (NIT). We aimed to determine the proportion of alcohol-related referrals who were diagnosed with advanced fibrosis in secondary care, the prevalence of both alcohol and fatty liver disease ('BAFLD') and the potential impact of NIT on referral stratification. DESIGN/SETTING Retrospective analysis of all general practitioner-referrals with suspected ArLD/non-alcoholic fatty liver disease (NAFLD) to a UK hepatology-centre between January 2015 and January 2018. PARTICIPANTS Of 2944 new referrals, 762 (mean age 55.5±13.53 years) met inclusion criteria: 531 NAFLD and 231 ArLD, of which 147 (64%) could be reclassified as 'BAFLD'. PRIMARY OUTCOME MEASURE Proportion of referrals with suspected ArLD/NAFLD with advanced fibrosis as assessed by tertiary centre hepatologists using combinations of FibroScan, imaging, examination and blood tests and liver histology, where indicated. SECONDARY OUTCOME MEASURES Included impact of body mass index/alcohol consumption on the odds of a diagnosis of advanced fibrosis, and performance of NIT in predicting advanced fibrosis in planned post-hoc analysis of referrals. RESULTS Among ArLD referrals 147/229 (64.2%) had no evidence of advanced fibrosis and were judged 'unnecessary'. Advanced fibrosis was observed in men drinking ≥50 units per week (U/w) (OR 2.74, 95% CI 1.51 to 5, p=0.001) and ≥35 U/w in women (OR 5.11, 95% CI 1.31 to 20.03, p=0.019). Drinking >14 U/w doubled the likelihood of advanced fibrosis in overweight/obesity (OR 2.11; 95% CI 1.44 to 3.09; p<0.001). Use of fibrosis 4 score could halve unnecessary referrals (OR 0.50; 95% CI 0.32 to 0.79, p=0.003) with false-negative rate of 22%, but was rarely used. CONCLUSIONS The majority of referrals with suspected ArLD were deemed unnecessary. NIT could improve identification of liver damage in ArLD, BAFLD and NAFLD in primary care. Anecdotal thresholds for harmful drinking (35 U/w in women and 50 U/w in men) were validated. The impact of alcohol on NAFLD highlights the importance of multi-causality in chronic liver disease.