Alcohol Use Disorder During Pregnancy: Harmonizing Multiple Datasets

To the Editor:

Alcohol use disorder (AUD) during pregnancy receives relatively little attention compared to other substance use disorders (ie, opioids).¹ Recognizing that insurance disparities may impact access to medications for AUD (MAUD),^2,3 we replicated the 2024 work by Roberts et al.⁴ within an electronic health records (EHR) data source (TriNetX Analytics) encompassing people from > 100 health care organizations. We secondarily explored the prevalence of AUD diagnoses over time in this same data source, including through the year 2024.

To identify people with > 1 delivery-related encounter for AUD, we queried the TriNetX data from 2006 to 2019, the same observation window used by Roberts et al.⁴ TriNetX is refreshed as frequently as weekly and is a component of the FDA’s efforts for drug safety surveillance (Sentinel Initiative).⁵ Details on TriNetX query methods done by our group have been previously reported.⁶ Whereas Roberts et al.⁴ identified people with > 1 claim for alcohol-related diagnoses, encompassing both AUD and other alcohol-related medical problems (ie, alcoholic liver disease), we focused on claims for specifically AUD. We evaluated the number of individuals with probable AUD (ICD-10 = F10.xx and ICD-9 equivalents) in the year before through after delivery who received > 1 fill for acamprosate, disulfiram, or naltrexone during that timeframe. Estimates of MAUD prescribing during pregnancy in the TriNetX data were compared to the estimates of Roberts et al. in the MarketScan commercial databases.⁴

From 2006 to 2019, 0.5% of 649,033 birthing people in the TriNetX database had a diagnosis of AUD during pregnancy (Fig. 1). While our estimate of AUD prevalence ostensibly mirrors the estimate of Roberts et al. of 0.2%–0.5%, we suspect that the estimated prevalence of probable AUD would be even higher if we expanded our codes to encompass alcohol-related diagnoses beyond strictly AUD codes (ie, F10.xx).

FIGURE 1.

Prescribing of medication for alcohol use disorder across multiple datasets.

Across both the analysis of Roberts et al.⁴ and our data, few people with AUD received MAUD in the year before through after delivery (TriNetX = 1.4% naltrexone, 0.7% disulfiram, 0.6% acamprosate; Market-Scan commercial = 1.9% naltrexone, 1.4% disulfiram, 0.6% acamprosate). Approximately 40% had diagnoses for mood (ie, major depression) disorders in the year postpartum, reflecting the high degree of comorbidity between AUD and other psychiatric disorders. Interestingly, the annual prevalence of AUD diagnoses in birthing individuals increased in the TriNetX data from ~0.2% in 2006–2007 to 0.4% by 2018–2019, and 0.6% when we extended the observation period to 2024.

The lack of continuous insurance enrollment requirements is an important strength, permitting us to capture subpopulations that may be underrepresented in commercial insurance databases. However, individuals may have sought MAUD treatment in institutions outside the database before and after pregnancy-related insurance coverage. We do not have access to detailed socioeconomic data (eg, the precise percentage of the sample that is uninsured and Medicaid-insured in our sample). Nonetheless, replicating the published analysis of Roberts et al. in another data source with different strengths and limitations is a critical part of establishing generalizability. Even as AUD’s prevalence may be rising in pregnancy, we found that filled MAUD prescriptions are uncommon in birthing people,⁴ a finding that warrants further investigation.