Abstract
BACKGROUND: Blood phosphatidylethanol (PEth) is a promising biomarker of alcohol consumption. This study was conducted to evaluate its performance in patients with liver disease. METHODS: This study included 222 patients with liver disease. Patient-reported alcohol use was obtained as a reference standard, and PEth was measured by tandem mass spectrometry. Receiver operating characteristic (ROC) and contingency table analyses were used to assess the performance of PEth in detecting any drinking and averaging 4 or more drinks daily in the past 30 days. RESULTS: At the limit of quantitation (20 ng/ml), PEth was 73% sensitive (95% confidence interval [CI] 65 to 80) and 96% specific (95% CI 92 to 100) for any drinking in the past month. Subjects who drank but had a negative PEth result were mainly light drinkers. Subjects who reported 30-day abstinence but with quantifiable PEth either reported heavy drinking within the past 6 weeks or had data that suggested underreported drinking. At the optimal cutoff concentration of 80 ng/ml, PEth was 91% sensitive (95% CI 82 to 100) and 77% specific (95% CI 70 to 83) for averaging at least 4 drinks daily. CONCLUSIONS: PEth is a useful test for detecting alcohol use in patients with liver disease, but cutoff concentrations for heavy drinking will result in misclassification of some moderate to heavy drinkers.