Validity of prenatal AUDIT screening for alcohol disorders - a Nationwide Swedish register study.

Objective: This study aims to assess the external validity of the Alcohol Use Disorders Identification Test (AUDIT) in Swedish prenatal care as an indicator for alcohol-addiction disorders, and to characterize women with mismatched information in healthcare registers.

Design: This study was designed as a National register-based study.

Setting: Sweden.

Participants: The study sample included 739,735 pregnancies over the period 2014-2020.

Methods: Prospectively collected prenatal AUDIT screening in the was linked to national health databases through individual identification number. The AUDIT score was dichotomized into < 6 points (low-risk use) and ≥ 6 points (hazardous use). Alcohol addiction disorders were defined by a diagnostic code in or drugs dispensed for alcohol dependence in the .

Primary Outcome Measures: The diagnostic properties of AUDIT were assessed based on sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR+), negative likelihood ratio (LR-), and accuracy (proportion of true positive and true negative) for an AUDIT score of ≥ 6 points for alcohol disorders. Women with mismatched information in the register were characterized and assessed by multinominal logistic regression, using women with matched information in the registers for reference.

Results: An alcohol-related disorder was recorded in 3.1%, while 25,770 (3.5%) had an AUDIT point ≥ 6. The diagnostic accuracy of the AUDIT ≥ 6 points for detection of an alcohol related disorder during a year prior to pregnancy was 95.7% (95% confidence interval [CI]: 95.7, 95.8), with a positive LR of 8.03 (95% CI: 7.5, 8.6). The sensitivity for detecting a pre-pregnancy alcohol related disorder was 33.0% (95% CI: 30.9, 35.1). Being young, nulliparous, of low education, and of Swedish origin increased the likelihood of being misclassified with the AUDIT. Prior psychiatric care was associated with false negatives, especially for women with neuropsychiatric disorders (odds ratio [OR]: 10.39, 95% CI: 9.89, 10.90).

Conclusions: The accuracy of AUDIT in screening for alcohol disorders at a population-based level was high, but only identified one third of women with alcohol-related disorders when using a cut-off of six points criterion.