Assessing the sensitivity and specificity of phosphatidylethanol (PEth) cutoffs to identify alcohol exposed pregnancies

In the literature on alcohol use biomarkers, there has been debate as to what a valid and/or utilitarian cut off level should be for various research applications. In this manuscript, we assessed the sensitivity and specificity of multiple cutoff values for phosphatidylethanol (PEth) from bloodspots relative to self-report, the Alcohol Use Disorder Identification Test (AUDIT) scores, and another alcohol use biomarker ethyl glucuronide (EtG) from fingernails in a sample of 222 pregnant women in the Western Cape Province of South Africa. Receiver operating characteristic (ROC) curves were used to assess the area under the curve (AUC) and assess PEth cutoff values of ≥2, ≥4, ≥8, ≥14, and ≥20 nanograms per milliliter (ng/ml). The highest AUC value was attained when PEth was compared to an AUDIT score of 1 or more. Depending on the cutoff used to determine alcohol consumption, PEth identified 47%–70% of the individuals as alcohol-consuming while 62.6%–75.2% were identified by self-reported measures, and 35.6% were identified by EtG. In this sample, sensitivity and accuracy were highest at less stringent PEth cutoffs when compared to self-report, AUDIT score of 1 or more, 5 or more, 8 or more, and EtG ≥ 8 picograms per milligram (pg/mg). For research purposes, less stringent cutoffs, such as PEth ≥ 8 ng/ml, may be considered a valid, positive cutoff for identifying women who consume alcohol during pregnancy in this population. A cutoff of PEth ≥ 20 ng/ml may miss individuals who reported consuming alcohol (false negatives).