Assessing Bias in Documentation of Alcohol Use Disorders in Primary Care Settings Across Intersections of Race or Ethnicity, Sex, and Socioeconomic Status: The Role of Symptoms and Stigma

Abstract

Alcohol use disorder (AUD) is harmful, with an 11.3% prevalence among adults living in the United States (U.S.), and is growing among key vulnerable demographics. Although there is effective AUD treatment available, only 4.6% of adults with AUD receive any alcohol-related treatment. Primary care has the potential to deliver evidence-based treatment and is accessible to most adults, making it an ideal setting for systematically improving both the diagnosis and treatment of AUD. However, in most clinical settings, including primary care, AUD is frequently underdiagnosed and undertreated. Part of this is likely due to a heavy reliance on clinicians’ subjective assessments of AUD symptoms (i.e., typically without standardized assessment measures of AUD symptoms) and which likely results in inconsistent and incomplete information about AUD symptoms. Additionally, the disparities in clinically-documented AUD across race, ethnicity, and sex shown in prior research suggest that there are likely potential biases in diagnosing practices in clinical settings. Further, the stigmatizing language used to describe mental health conditions, such as AUD, may reflect underlying biased beliefs held by clinicians making diagnoses, while also perpetuating stigma. These limitations imply that there is a need to develop, test, and adopt innovative approaches to assist clinicians in recognizing AUD symptoms in a systematic and standardized manner, which may help them accurately diagnose AUD in an unbiased manner. This dissertation describes the distribution of clinically-documented AUD and the proportion of AUD diagnoses reported in electronic health records (EHRs) using stigmatized language across intersecting identities, therefore providing foundational knowledge for whether bias occurs in the diagnosis of AUD in primary care settings. Specific aims of this dissertation were to: 1) describe patterns in the prevalence of clinically-documented AUD in EHRs in primary care patients across intersections of race or ethnicity, sex, and socioeconomic status (SES); 2) describe, among primary care patients who reported high-risk drinking, patterns in the prevalence of clinically-documented AUD in EHRs in primary care across intersections of race or ethnicity and sex, adjusting for patient-reported alcohol consumption and AUD symptoms; and 3) describe the proportion of AUD diagnoses documented in EHRs with stigmatized descriptors among primary care patients across intersections of race or ethnicity and sex. Aim 1 used Kaiser Permanente Washington (KPWA) EHR data linked to a census tract neighborhood deprivation index, representing community-level SES, from 439,375 adult primary care patients who completed alcohol screenings from 3/1/2015-9/30/2020. We described the prevalence of clinically-documented AUD based across 36 subgroups defined by intersections of race or ethnicity, sex, and terciles of community-level SES based on documented EHR diagnostic codes for AUD. Among women, the prevalence was highest for AI/AN women with middle SES, 1.5% (95% CI:1.0-2.3), and lowest for Asian women with middle SES, 0.1% (95% CI:0.1-0.2). Among men, the prevalence was highest for AI/AN men with high and middle SES, 2.0% (95% CI:1.1-3.4) and 2.0% (95% CI:1.2-3.2), respectively, and lowest for Asian men with high SES, 0.5% (95% CI:0.3-0.7). Black and Latine patients tended to have a lower prevalence of AUD than White patients, across all intersections of sex and SES except for Black women with high SES. The overall prevalence of clinically-documented AUD was 1.0% and varied across race or ethnicity and sex, but no consistent pattern emerged for SES. Aim 2 used KPWA EHR data from 14,442 adult patients who self-reported high-risk drinking (AUDIT-C score 7-12), had primary care encounters from 3/1/2015-2/28/2022, were Asian, Black, Latine, or White, and completed a DSM-5 Alcohol Symptom Checklist (0-11). We described the prevalence of clinically-documented AUD across 8 intersections of race or ethnicity and sex. Rates of clinically-documented AUD diagnoses increased as alcohol consumption and AUD symptoms increased. The prevalence of clinically-documented AUD diagnoses differed across the 8 intersectional subgroups differed in unadjusted analyses ranging from 12.2% (95% CI: 9.8-15.0) to 21.7% (95% CI: 17.7-26.3) but did not differ after adjustment for both AUDIT-C score and AUD symptoms ranging from 11.0% (95% CI: 8.7-13.8) to 15.1% (95% CI: 14.3-16.0). This suggests that observed differences in the prevalence of clinically-documented AUD diagnoses across intersectional subgroups likely were due to differences in alcohol consumption and AUD symptom burden across subgroups. Aim 3 used AUD text descriptors from adult primary care patients who were Asian, Black, Latine, and White and had AUD diagnoses documented in EHRs during primary care visits from 3/1/2015-5/31/2023. We described the proportion of AUD diagnoses documented with stigmatized descriptors across intersections of race or ethnicity and sex. The overall proportion of stigmatized AUD descriptors in the EHR was 18.6% or 88.5%, depending on whether stigmatized AUD descriptors excluded or included stigmatizing language from valid DSM-IV diagnoses, respectively (i.e., alcohol abuse, alcohol dependence). There appeared to be no meaningful variation across race or ethnicity, or intersections based on race or ethnicity and sex, but men had AUD diagnoses documented with a significantly higher proportion of stigmatized text descriptors compared to women. Findings suggest that further efforts are needed to reduce the amount of stigmatized language associated with alcohol use disorder in the EHR. Overall, the findings of this dissertation support the use of intersectional approaches to describing disparities in AUD, the use of standardized tools for AUD symptom assessment alongside routine population-based alcohol screenings, and efforts to reduce stigmatized language associated with alcohol use.