Skip to main content
NCBI home page
JAMA Network logoLink to JAMA Network
. 2026 Jul 31;7(7):e262348. doi: 10.1001/jamahealthforum.2026.2348

Expanded Alcohol Screening and Brief Intervention to Address Premature Mortality

Julia M Lemp 1,2, Carolin Kilian 3,4,5, Xinyi Kou 6, Charlotte Buckley 6,7, Laura Llamosas-Falcón 1,8, Yachen Zhu 9, Jürgen Rehm 1,5,8,10,11,12,13, Nina Mulia 9, Robin Purshouse 6, Charlotte Probst 1,8,10,11,12,✉
PMCID: PMC13428287  PMID: 42536386

Key Points

Question

What is the potential impact of expanding alcohol screening and brief interventions in primary care on premature mortality from major alcohol-related causes of death in US adults?

Findings

In this decision analytical model and microsimulation study, expanding delivery of alcohol brief interventions to an additional 8 million interventions per year was projected to reduce annual alcohol-related potential years of life lost per 100 000 adults by −51.3 (95% credible interval [CrI], −73.2 to −32.8) in men and −34.1 (CrI, −54.5 to −16.1) in women by 2030 compared with the status quo.

Meaning

These findings suggest that substantially expanding alcohol screening and brief interventions in primary care may reduce premature mortality and help narrow the socioeconomic gap in alcohol-related mortality in the US.

Abstract

Importance

Alcohol use is a leading cause of premature mortality in the US. Despite public health recommendations and insurance coverage mandates, alcohol screening and brief intervention (ASBI) in primary care remains underused.

Objective

To estimate the population-level impact of expanded ASBI delivery on alcohol use and potential years of life lost (YLL) from major alcohol-related causes of death in US adults through 2030, by sex, race and ethnicity, and educational attainment as a proxy for socioeconomic status.

Design, Setting, and Participants

This was a decision analytical study that used a dynamic individual-level microsimulation model of nationally representative of US adults (age ≥18 years) from 2000 to 2030 and considered sociodemographic variables, alcohol use, and mortality risks.

Interventions

Four expansion scenarios increasing the annual number of eligible individuals receiving ASBI from 2025 to 2030: scenario 1, expanding alcohol screening (AS) only; scenario 2, expanding brief intervention (BI) by 4 million; scenario 3, expanding BI by 8 million; and scenario 4, universal ASBI delivery. Scenarios were compared with a no-expansion counterfactual assuming historical trends.

Main Outcomes and Measures

Prevalence of hazardous alcohol use (>20 or >40 g of alcohol per day for women and men) and YLL per 100 000 before age 75 years from 5 key alcohol-related causes of death (alcohol use disorder, liver disease and cirrhosis, motor vehicle injuries, other unintentional injuries, and suicide) in 2030.

Results

The analysis of the microsimulation model found that scenario 3 (additional 8 million BI) and scenario 4 (universal ASBI) had the strongest impact on alcohol use and mortality, and were projected to reduce annual alcohol-related YLL per 100 000 by −51.3 (95% credible interval [CrI], −73.2 to −32.8) and −68.9 (CrI, −102.2 to −44.8) among men, respectively, and −34.1 (CrI, −54.5 to −16.1) and −52.0 (CrI, −75.0 to −30.2) among women, respectively, by 2030. In scenario 4, the gap in YLL from causes associated with alcohol use between low and high education narrowed by −3.5% (−64.5 YLL per 100 000; 95% CrI, −106.3 to −35.9) among men and −4.9% (−31.4 YLL per 100 000; 95% CrI, −56.2 to −12.4) among women.

Conclusions and Relevance

In decision analytical model expanding ASBI delivery in primary care was projected to reduce premature mortality and may represent an important strategy for improving population health. Substantial expansion had measurable but modest potential to narrow socioeconomic inequalities in premature mortality from causes of death associated with alcohol use.

This decision analytical model estimates the population-level impact of expanded alcohol screening and brief intervention delivery on alcohol use and potential years of life lost from major alcohol-related causes of death among US adults.

Introduction

Screening and brief intervention for alcohol use is a core component of preventive care. The US Preventive Services Task Force recommends that all adults in primary care settings be screened for alcohol use and that those engaging in hazardous or risky drinking receive brief behavioral counseling.1 Under the Affordable Care Act, commercial health insurers are required to cover these services at no additional cost to patients, and most Medicare & Medicaid plans cover at least annual provision of alcohol screening and brief intervention (ASBI).2,3

Despite these mandates, implementation of ASBI remains limited, with access differing across population groups.4,5 Routine health care visits constitute a major entry point to ASBI, yet 15% of US adults had no health care visit in the past 12 months.6 Even among those who did, survey data suggest that 4 in 5 adults who attended a “health checkup” were asked about their alcohol use, while only 1 in 5 who engaged in risky drinking received advice to reduce drinking.4 This is especially concerning given that alcohol use is among the most important risk factors for premature mortality in the US. In 2021, alcohol was estimated to account for more than 111 500 deaths across all ages, the highest number recorded since 1990.7 In addition, deaths from causes associated with alcohol use are concentrated among population groups with lower educational attainment, making alcohol-attributable mortality a major contributor to the widening socioeconomic gap in life expectancy in the US.8

Low ASBI rates among individuals with hazardous alcohol use may represent a missed opportunity to mitigate the growing alcohol-related burden. While brief interventions (BIs) are effective at reducing alcohol use at the individual level,9 their potential to counteract national trends in premature mortality and variation across sociodemographic groups in the US has received little attention. Our study used an advanced microsimulation to model how expanding ASBI at different stages of the care cascade may affect alcohol use and potential years of life lost (YLL) from key alcohol-related causes of death through 2030 in the adult US population, by sex, race and ethnicity, and educational attainment.

Methods

This study is part of the Simulation of Alcohol Control Policies for Health Equity (SIMAH) project,10 which has been reviewed and approved by the Centre for Addiction and Mental Health Research Ethics Board (#115/2020). The underlying computer model is documented using the Overview, Design Concepts and Details (ODD) protocol available in the eMethods 1 in Supplement 1, a recognized best practice for reporting simulation models.11 We followed the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guideline.

Model Overview

A dynamic, individual-level microsimulation model was developed, calibrated, and validated to simulate alcohol use and deaths due to alcohol-related causes over 30 years from 2000 through 2030, progressing in 1-year time steps. Individuals in the microsimulation constituted a synthetic population representative of the actual US population aged 18 to 79 years. Individuals were characterized by educational attainment (a proxy for socioeconomic status, denoted “education”), race and ethnicity, age, and sex. Education was categorized as high school degree or less (≤ high school), some college, and college degree or more (≥ college) and synthetic individuals could progress in their education until the age of 34 years. Race and ethnicity were categorized as non-Hispanic White (hereafter, White), non-Hispanic Black (Black), Hispanic, and other (combining American Indian or Alaska Native, Asian, and Native Hawaiian or Other Pacific Islander due to small sample sizes; additional details in eMethods, sections 1.3.3, in Supplement 1). Alcohol use was modeled based on each individual’s characteristics, where individuals had the opportunity to change their alcohol consumption level each year (eMethods, sections 1.1.2 and 1.3.5, in Supplement 1). Four expansion scenarios for ASBI delivery were modeled and compared to a reference scenario. The reference scenario assumed continuation of historical trends in ASBI rates (ie, no expansion beyond observed trends; eMethods, sections 1.3.5, in Supplement 1). Each expansion scenario increased the number of individuals receiving ASBI starting in the year 2025. For each comparison, annual subgroup-specific mean effect estimates and 95% credible intervals (CrIs) were quantified to 2030, reflecting the impact of expansion over a 6-year period.

Outcomes

Primary outcomes were (1) prevalence of hazardous alcohol use and (2) combined YLL per 100 000 adults from alcohol use disorder (AUD, including alcohol poisonings), liver disease and cirrhosis (including hepatitis C−related liver cirrhosis), motor vehicle injuries, other unintentional injuries, and suicide (eMethods, section 1.1.2, in Supplement 1 for codes from the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision). Each of these causes has an alcohol-attributable fraction of 15% or more in the US.8,12 YLL was calculated as the difference between a benchmark age of 75 years (following the general practice of most federal and state agencies’ YLL calculations13,14,15) and the age at death and standardized per 100 000 adults, and were assessed in absolute and relative terms compared to the reference scenario. Hazardous alcohol use was defined as average consumption exceeding 20 g of pure alcohol per day (GPD) for women or 40 GPD for men, consistent with the World Health Organization’s classification of drinking levels based on sex-specific daily alcohol consumption and associated alcohol-related harm,16 thereby defining the population eligible for BI in the model. A secondary outcome was heavy episodic drinking prevalence (HED; defined as 60 g or more of pure alcohol on at least 1 occasion in the past 30 days), which is updated annually and modifies the cause-specific risk functions for injuries (eMethods, section 1.3.5, in Supplement 1).

Data Input and Model Validation

Data Input

Microsimulation parameters were informed by US data sources and existing studies (eMethods, section 1.3, in Supplement 1). In brief, population estimates were based on decennial US Census Bureau data17 and the annual American Community Survey (ACS)18; transitions between levels of education (by race and ethnicity, age, and sex) were informed by data from the Panel Study of Income Dynamics.19 Individual-level data to inform alcohol exposure and changes in alcohol use over time were retrieved from the Behavioral Risk Factor Surveillance System (BRFSS)20 and adjusted for underreporting.21 Mortality was based on individual death records from the US National Vital Statistics System.22 Relative risk functions linking alcohol use to cause-specific mortality risks were informed by published meta-analyses and secondary data analyses. Two distinct cause-specific relative risk functions were applied within the liver disease and cirrhosis category: one for chronic liver disease and cirrhosis, and a second for attenuated function for hepatitis C virus−related liver cirrhosis, reflecting the indirect alcohol-attributable pathway through disease progression (details available in eMethods, section 1.3.6, and eTable 6 in Supplement 1).

Data inputs for projecting forward are detailed in eMethods, section 1.3.4, in Supplement 1. In brief, total population growth and birth rates (from 2022 onward) were informed by the US Census Bureau’s population projections that uses census estimates of the resident population on July 1, 2022, as the base for modeling the US population forward.23 Migration rates (from 2022 onward) were projected forward accounting for year, sex, age, and race and ethnicity. Mortality (from 2024 onward) was projected forward using the US Census Bureau’s death projections.23

To account for disruptions during the COVID-19 pandemic, education transitions in 2020 to 2022 were parameterized using Panel Study of Income Dynamics 2019 to 2021 data (eMethods, section 1.3.5, in Supplement 1). We applied the same ordinal model to represent alcohol use transitions for all years given that secondary analyses of BRFSS data from 2020 onward did not reveal consistent directional changes in annual consumption. Progression through the ASBI care cascade—routine health care access, AS, and BI for hazardous drinkers—was informed by regression models built using individual-level data from the US National Survey on Drug Use and Health (NSDUH) from 2013 to 2019, accounting for heterogeneity across population subgroups (eMethods, section 1.3.5, in Supplement 1).24 Effect estimates are expected to be insensitive to pandemic-era adjustments, as both expansion and reference scenarios share identical adjustment assumptions, such that the projected differences are largely driven by the modeled expansion rather than the changes to baseline trends due to the pandemic.

Model Calibration and Validation

We used state-of-the-art calibration and validation techniques to refine and test the modeling of education and alcohol transitions, following a Bayesian probabilistic framework (eMethods, section 1.3.5, in Supplement 1). Calibration used ACS data (2000-2010) for education and BRFSS data (2000-2015) for alcohol use. Model performance was validated against later ACS (2011-2019) and BRFSS (2016-2019) data. Mortality risks are not calibrated; instead, reference mortality rates are adjusted to align mortality outcomes from the model with the US National Vital Statistics System data and US Census Bureau projections.

The validated model was used to estimate changes in hazardous alcohol use and YLL by sex, education, and race and ethnicity, accounting for uncertainty in model parameters. The final simulation comprised 70 parameters sets with 20 stochastic replications each (1400 runs per scenario). Outcomes are summarized as means with 95% CrI. The 95% CrI represents the 2.5th and 97.5th percentile of outcomes across the 70 parameter sets and reflects uncertainty in model inputs following calibration (eMethods, section 1.3.6, in Supplement 1).

Expansion Scenarios

Modeling Flow

BI rates among past-year alcohol users were increased starting in 2025. For each step in the care cascade, individuals who would have received the service according to historical rates were first sampled using empirically informed regression models (refer to the Data Input section) before additional recipients were selected among the remaining individuals using the same model and according to the respective increase in each expansion scenario. Only individuals newly receiving BI due to expanded ASBI were assigned intervention effects because the effects of existing interventions are assumed to be already captured in historical alcohol use data. Both intervention effects are based on the most recent Cochrane review9: all newly intervened individuals received a reduction in GPD, sampled from the trial-level effect distribution (mean [SE] reduction: −2.86 [0.58] GPD) and scaled using each individual’s current level of consumption (eMethods, section 1.3.5, in Supplement 1). Among newly intervened individuals classified as heavy episodic drinkers, a proportion corresponding to the pooled risk difference (RD, −0.07; 95% CI, −0.02 to −0.12) were randomly selected to transition to non-HED status. GPD reduction and HED transition were sampled separately, such that HED status change was not determined by the magnitude of GPD reduction. Sampling of individuals who progress through the cascade was independent of the previous year’s sampling process, allowing individuals to be resampled in any year during the expansion period.

Expansion Strategies

Four expansion scenarios were modeled, each defined by annual increases in ASBI delivery. Scenario 1 increased screening among all past-year alcohol users by 20 million annually compared with historical rates (ie, an increase in the mean screened proportion from 55.7% to 66.9%). Scenario 1 was included to represent settings in which screening uptake is expanded without a commensurate expansion of BI delivery resources. In this scenario, total BI delivery still slightly increased as a larger pool of screened individuals means more individuals were identified as eligible for BI under existing delivery rates. Scenarios 2 and 3 increased screening by 20 million while simultaneously increasing BIs among screened individuals with hazardous alcohol use by 4 million and 8 million, respectively. Scenario 4 modeled the upper limit of ASBI’s impact by providing universal ASBI to all individuals accessing health care. All scenarios assumed continuation of historical trends of health care use (eMethods, section 1.3.5, in Supplement 1).

Sensitivity Analyses

We investigated alternative parameter settings in 2 sensitivity analyses. First, we examined the impact of a lower (−1.71 GPD) or higher (−4.01) mean BI effect9 to gauge uncertainty in the treatment effect separately from other parameter variation. Second, in scenario 4, we extended eligibility for BI to include alcohol use category I (up to 20 GPD for women and 40 GPD for men, generally not considered hazardous use) to test whether results were sensitive to broader intervention coverage. Third, we conducted a sensitivity analysis to quantify the mortality cost of delaying expansion by comparing the total cumulative YLL averted over 2025 to 2030 under 4 implementation start dates: 2025 (original), 2026, 2027, and 2028.

Results

Screening and Brief Intervention Cascade by Subgroup

In the reference scenario, 18.0% (CrI, 10.4%-26.4%; 2.3 M; CrI, 1.3-3.3) of men and 12.2% (CrI, 5.8%-19.7%; 1.2 M; CrI, 0.5-1.9) of women with hazardous alcohol use were projected to receive a BI annually by 2030. Expanding the annual number of ASBI in primary care was simulated to have differential effects on the share of adults receiving a BI across population subgroups (Figure 1 and eFigure 1 in Supplement 1), consistent with the observed heterogeneity in the ASBI care cascade (eMethods, section 1.3.5, in Supplement 1).

Figure 1. Line Graph of Simulated Mean Percentage of Adults With Hazardous Alcohol Use Who Received a Brief Intervention Each Year, by Sex and Education for Each Modeled Scenario.

Three-panel line charts of percent vs year by education, sex, and scenario. Three side-by-side panels titled, from left to right, less than or equal to High school degree, Some college, and greater than or equal to College degree. Each panel contains multiple colored lines with circular markers for years 2025 through 2030 on the horizontal axis labeled Year, with tick labels 2025, 2026, 2027, 2028, 2029, 2030. The vertical axis in each panel is labeled Adults with hazardous alcohol use who received brief intervention, percent, ranging from 0 to 100 with labeled ticks at 0, 25, 50, 75, and 100. A legend centered below maps Scenario colors: Reference: no expansion in dark teal; Scenario 1: screening expansion only plus 20 million screenings in bright blue; Scenario 2: plus 4 million interventions in tan; Scenario 3: plus 8 million interventions in orange; Scenario 4: Universal A S B I in green. The legend also maps Sex line style: Men as solid lines and Women as dashed lines. In each panel, green lines are highest, with a solid green line near the upper 60s to around 80 depending on panel and a dashed green line higher near the mid 80s to low 90s. Orange lines rise over time in all panels, with dashed orange above solid orange; in the left panel orange increases from roughly high 30s to low 50s, in the middle panel from roughly high 40s to mid 60s, and in the right panel from roughly high 50s to mid 70s by 2030. Tan lines sit midrange and increase slightly, from about mid 20s to around 30 in the left panel, about upper 20s to low 30s in the middle, and about low 30s to mid 30s in the right. Blue and dark teal lines are lowest; blue sits slightly above dark teal in each panel, both increasing modestly from roughly around 10 to low teens in the left panel, around low teens to mid teens in the middle, and around mid teens to upper teens in the right. No panel letter labels, arrows, or scale bars are present.

Open in a new tab

Hazardous use was defined as more than 20 GPD for women and more than 40 GPD for men. Results by sex and race and ethnicity are presented in eFigure 1 in Supplement 1. ASBI indicates alcohol screening and brief intervention; GPD, grams of pure alcohol per day.

Changes in Hazardous Alcohol Use

Figure 2 shows simulated changes in the prevalence of hazardous alcohol use from 2022 to 2030 under the different expansion scenarios, by sex and education (see also eTable 1 in Supplement 1). Results by sex and race and ethnicity are presented in eFigure 2 and eTable 2 in Supplement 1. In the reference scenario, hazardous alcohol use is projected to remain relatively stable, with only slight decreases among men with less than or equal to high school or some college education.

Figure 2. Line Graph of Simulated Mean Prevalence of Hazardous Alcohol Use, by Sex and Education for Each Modeled Scenario.

Six-panel line graphs of hazardous alcohol use percent by year and education. Two-row, three-column figure with six line charts. Panel labels at far left: A Men on the top row and B Women on the bottom row. Column titles above each chart: less than or equal to High school degree, Some college, and greater than or equal to College degree. Each chart has a horizontal axis labeled Year with ticks at 2022, 2024, 2026, 2028, and 2030, and a vertical axis labeled Hazardous alcohol use, percent with a scale from 0 to 12 point 5. A legend in the lower left of the top-left chart lists Scenario with five entries and matching markers and colors: Reference No expansion as a dark teal circle; Scenario 1 Screening expansion only plus 20 million screenings as a bright blue square; Scenario 2 plus 4 million interventions as a tan triangle; Scenario 3 plus 8 million interventions as an orange square; Scenario 4 Universal A S B I as a light green diamond. In every panel, lines begin near 2022, converge around 2024, then diverge through about 2029, with a separate cluster of large markers at 2030 indicating the end values for each scenario. Men, less than or equal to High school degree starts near 10 percent and ends around 9 point 5 for reference and scenario 1, about 8 point 8 for scenario 2, about 7 point 6 for scenario 3, and about 6 point 4 for scenario 4. Men, Some college starts near 9 percent and ends around 8 point 4 for reference and scenario 1, about 7 point 3 for scenario 2, about 6 point 1 for scenario 3, and about 5 point 5 for scenario 4. Men, greater than or equal to College degree starts near 11 point 5 percent and ends around 11 point 4 for reference and scenario 1, about 9 point 7 for scenario 2, about 8 point 0 for scenario 3, and about 7 point 4 for scenario 4. Women, less than or equal to High school degree starts near 6 point 8 percent and ends around 6 point 7 for reference and scenario 1, about 6 point 1 for scenario 2, about 5 point 0 for scenario 3, and about 4 point 0 for scenario 4. Women, Some college starts near 6 point 4 percent and ends around 6 point 0 for reference and scenario 1, about 5 point 2 for scenario 2, about 4 point 2 for scenario 3, and about 3 point 2 for scenario 4. Women, greater than or equal to College degree starts near 9 point 0 percent and ends around 8 point 9 for reference and scenario 1, about 7 point 6 for scenario 2, about 5 point 8 for scenario 3, and about 4 point 8 for scenario 4.

Open in a new tab

Hazardous use was defined as 20 GPD for women and 40 GPD for men. Results by sex and race and ethnicity are presented in eFigure 1, and for heavy episodic drinking prevalence, in eFigure 3 in Supplement 1. ASBI indicates alcohol screening and brief intervention; GPD, grams of pure alcohol per day.

Hazardous alcohol use was simulated to be most prevalent among those with college or higher degree (men, 11.6% [CrI, 11.5%-11.7%]; women, 8.9% [CrI, 8.8%-8.9%]) and White adults (men, 12.2% [CrI, 12.1%-12.4%]; women, 9.2% [CrI, 9.1%-9.4%]), and lowest among those with some college (men, 8.5% [CrI, 8.4%-8.6%]; women, 5.9% [CrI, 5.8%-6.0%]) and Black adults (men, 4.4% [CrI, 4.4%-4.5%]; women, 3.2% [CrI, 3.1%-3.3%]).

Across expansion scenarios, the magnitude of reductions in the prevalence of hazardous alcohol use by 2030 increased with the intensity of expansion. Screening expansion only (scenario 1) was simulated to have very small effects, whereas scenarios that include additional BIs (scenarios 2-4) show progressive reductions in hazardous alcohol use. Reductions were somewhat larger among men than women, and among White adults and those with higher education. Heavy episodic drinking followed a similar pattern, with absolute reductions by 2030 ranging from 0.4 (CrI, 0.3-0.5) percentage points among men with some college education to 1.4 (CrI, 1.3-1.5) percentage points among women with a college degree or more (eFigure 3, eTable 3-4 in Supplement 1).

Changes in Years of Life Lost

In 2030, the annual mortality burden from key alcohol-related causes in the reference scenario was projected to be highest among those with high school or less education (men, 2642 YLL per 100 000 [CrI, 2539-2742]; women, 1068 [CrI, 1035-1112]) and Black adults (men, 2048 [CrI, 1979-2,103]; women, 853 [CrI, 828-876]), and lowest among those with college or more education (men, 825 YLL per 100 000 [CrI, 804-855]; women, 430 YLL per 100 000 [CrI, 414-447]), White men (1786 [CrI, 1755-1808]), and Hispanic women (659 [CrI, 632-693]).

Changes in annual YLL per 100 000 from alcohol-related causes under the different expansion scenarios by subgroup are shown in eTable 5 to 7 in Supplement 1. In men, the decrease compared with the reference scenario in 2030 was projected to be −8.0 YLL (CrI, −19.6 to 2.4) per 100 000 in scenario 1, −29.0 (CrI, −42.8 to −14.0) in scenario 2, −51.3 (CrI, −73.2 to −32.8) in scenario 3, and −68.9 (CrI, −102.2 to −44.8) in scenario 4. In women, the decrease is projected to be −1.9 YLL per 100 000 (CrI, −8.8 to 4.9) in scenario 1, −15.1 (CrI, −22.9 to −5.8) in scenario 2, −34.1 (CrI, −54.5 to −16.1) in scenario 3, and −52.0 (CrI, −75.0 to −30.2) in scenario 4 (eFigure 4 in Supplement 1). Figure 3 shows the simulated differences in 2030 across scenarios. In all expansion scenarios, those with a high school degree or less education benefit more than those with higher education in absolute terms, whereas the reverse applies in relative terms (eTable 6 in Supplement 1). White adults benefit more than Black and Hispanic adults, among both sexes in absolute and relative terms (eTable 7 in Supplement 1).

Figure 3. Bar and Whisker Chart of Simulated Mean Change in Annual Potential Years of Life Lost (YLL) per 100 000 From Key Alcohol-Related Causes for Each Modeled Expansion Scenario Compared to the Reference Scenario in 2030, Sex, Education, and Race and Ethnicity.

Four-panel boxplot figure of YLL reduction per 100000 by scenario, sex, education, and race. Four panels arranged in a two by two grid, with panel letters A on the upper left and B on the lower left. All panels use box and whisker plots with a dashed horizontal reference line at zero. The vertical axis in every panel reads Reduction in YLL per 100000, with tick marks at 50, zero, minus 50, minus 100, and minus 150. The horizontal axis in every panel reads Scenario with categories 1, 2, 3, and 4. Top row, panel A titled YLL by education: left subplot labeled Men and right subplot labeled Women. In each scenario group, three adjacent boxplots represent education levels, keyed by a legend labeled Education with three gray swatches: less than or equal to High school degree, Some college, and greater than or equal to College degree. Scenario 1 boxes cluster near zero for both sexes, with medians slightly below zero. Scenario 2 boxes extend further below zero, roughly from about minus 10 to minus 40 depending on education and sex. Scenario 3 boxes extend lower, with medians roughly around minus 25 to minus 70. Scenario 4 boxes are lowest overall; for men, the lightest education group reaches near minus 100 with whiskers extending to about minus 145, and the other education groups center around roughly minus 40 to minus 60; for women, scenario 4 centers around roughly minus 35 to minus 70 with whiskers extending to about minus 95. A second legend in the upper right panel area labels Scenario with colored squares: scenario 1 in blue, scenario 2 in tan, scenario 3 in orange, and scenario 4 in green, with accompanying text describing screening expansion, added interventions, and universal A S B I. Bottom row, panel B titled YLL by race and ethnicity: left subplot Men and right subplot Women. Each scenario group contains three adjacent boxplots keyed by a legend labeled Race with gray swatches: Non-Hispanic Black, Hispanic, and Non-Hispanic White. Scenario 1 boxes lie close to zero for both sexes. Scenario 2 and 3 shift downward, with women generally less negative than men. Scenario 4 shows the largest downward shifts; for men, Non-Hispanic White centers near about minus 85 with whiskers to about minus 120, while the other two groups center near about minus 35 to minus 55; for women, Non-Hispanic White centers near about minus 70 with whiskers to about minus 100, while the other two groups center near about minus 20 to minus 25.

Open in a new tab

Errors bars represent 95% credible intervals across 70 unique parameter combinations. Full results are presented in eTables 6 and 7 in Supplement 1. Results are not reported for the other race and ethnicity category because model parameters could not be reliably informed by empirical data. Key alcohol-related causes of death are alcohol use disorder, liver disease and cirrhosis (including hepatitis C−related liver cirrhosis), motor-vehicle injuries, other unintentional injuries, and suicide. Results considering additional causes of deaths are presented in eTables 8 to 10 in Supplement 1. ASBI indicates alcohol screening and brief intervention.

Potential Impact on the Socioeconomic Gap

The decrease in YLL per 100 000 was projected to be most pronounced for causes with the highest alcohol-attributable fractions, that is, AUD and liver disease and cirrhosis (Figure 4; eTable 8 in Supplement 1). Simulation results further indicate that those with high school or less education will experience the largest declines: In scenario 4, annual YLL per 100 000 attributable to liver disease and cirrhosis in 2030 decreased by −38.5 (CrI, −81.4 to −11.8) for men and by −33.3 (CrI, −53.4 to −16.8) for women with high school or less, but only by −11.9 (CrI, −26.8 to 0.5) for men and by −15.1 (CrI, −25.9 to −6.0) for women with college or more (eTable 10 in Supplement 1). Figure 5 illustrates the potential narrowing of the alcohol-related mortality gap in 2030 between those with low (≤ high school) and high (≥ college) education across all modeled causes. Among men, the gap consistently narrowed in scenario 3 by −2.0% (−36.7 YLL per 100 000 [CrI, −63.8 to −11.9]) and in scenario 4 by 3.5% (−64.5 YLL per 100 000 [CrI, −106.4 to −35.9]). Among women, a consistent narrowing of the gap was only observed in scenario 4 by −4.9% (−31.4 YLL per 100 000 [CrI, −56.2 to −12.4]).

Figure 4. Line Graph of Simulated Mean Potential Years of Life Lost (YLL) per 100 000 by Sex and Education .

Ten line charts of Y L L per 100000 by year, sex, education, and cause. Two-row, five-column multi-panel line figure with panel A labeled Men across the top row and panel B labeled Women across the bottom row. Columns, left to right, titled Alcohol-use disorder, Liver cirrhosis, Motor vehicle injuries, Unintentional injuries, and Suicide. Each subplot has horizontal axis labeled Year with ticks from 2010 through 2030, and vertical axis labeled Y L L per 100000; vertical ranges differ by subplot: men alcohol-use disorder 0 to 400, men liver cirrhosis 0 to 800, men motor vehicle injuries 0 to 1500, men unintentional injuries 100 to 700, men suicide 250 to 1250; women alcohol-use disorder 0 to 120, women liver cirrhosis 0 to 500, women motor vehicle injuries 0 to 500, women unintentional injuries 50 to 250, women suicide 100 to 300. A black dashed vertical reference line at year 2025 appears in every subplot. Three education strata appear as dark teal lines with distinct markers: circle for less than or equal to high school degree, triangle for some college, and square for greater than or equal to college degree. Two scenarios appear: a dark teal reference trajectory and a light green scenario trajectory beginning at 2025 and extending to 2030, with a filled light green marker at 2030 for each education group. Men: alcohol-use disorder peaks near 2021 to 2022 around 390 for less than or equal to high school, around 230 for some college, and around 125 for greater than or equal to college; by 2030 light green endpoints near 235, 175, and 110. Men liver cirrhosis peaks near 2021 around 780, 430, and 200; by 2030 light green endpoints near 460, 320, and 170. Men motor vehicle injuries peak near 2021 around 1300, 650, and 170; by 2030 light green endpoints near 720, 360, and 90. Men unintentional injuries drop sharply at 2025 from about 580 to 430 for less than or equal to high school, from about 320 to 250 for some college, and from about 180 to 140 for greater than or equal to college; by 2030 light green endpoints near 360, 220, and 120. Men suicide rises to about 1180 before 2025 for less than or equal to high school and about 860 for some college, then drops at 2025; by 2030 light green endpoints near 750, 580, and 270. Women: alcohol-use disorder rises to about 115 before 2025 for less than or equal to high school and about 90 for some college; by 2030 light green endpoints near 95 for less than or equal to high school, about 92 for some college, and about 55 for greater than or equal to college. Women liver cirrhosis peaks near 2021 around 430, 290, and 125; by 2030 light green endpoints near 295, 245, and 110. Women motor vehicle injuries peak near 2021 around 420, 280, and 90; by 2030 light green endpoints near 275, 170, and 55. Women unintentional injuries peak near 2021 around 215, 140, and 80, then decline; by 2030 light green endpoints near 152, 105, and 60. Women suicide fluctuates near 240 to 270 before 2025 for less than or equal to high school and some college, then declines; by 2030 light green endpoints near 180 for less than or equal to high school, about 190 for some college, and about 110 for greater than or equal to college. A combined legend at lower right labels Education with the three marker shapes and labels Scenario with dark teal Reference: no expansion and light green Scenario 4: universal A S B I.

Open in a new tab

Ranges vary by group and cause of death. Dashed vertical line at 2025 indicates the start of ASBI expansion. Results considering additional causes of deaths are presented in eTables 8 to 12 in Supplement 1. ASBI indicates alcohol screening and brief intervention.

Figure 5. Bar and Whisker Chart of Simulated Mean Reduction in the Gap in Combined Potential Years of Life Lost (YLL) per 100 000 From Key Alcohol-Related Causes Between Adults With High School Degree or Less Compared With Those With a College Degree or More in 2030, Stratified by Sex.

Two-panel bar charts of education gap reduction by scenario, sex, and units. Two side-by-side panels labeled A and B, each with horizontal bars and thin horizontal whiskers. Panel A title at upper left: Reduction in education gap by YLL. Vertical axis labeled Scenario with tick labels 1, 2, 3, 4. Horizontal axis labeled Reduction in education gap, YLL per 100000, with ticks from minus 125 to 25. A vertical dashed reference line at zero. For each scenario row, two adjacent bars appear: a darker saturated color and a lighter, more transparent version of the same color, each with a central bar and a thin whisker line extending left and right. Colors correspond to a legend box in the lower left labeled Scenario: bright blue for Scenario 1: Screening expansion only plus 20 million screenings; tan for Scenario 2: plus 4 million interventions; orange for Scenario 3: plus 8 million interventions; green for Scenario 4: Universal A S B I. Panel B title at upper left: Reduction in education gap by percentage. Vertical axis labeled Scenario with tick labels 1 through 4. Horizontal axis labeled Reduction in education gap, percent, with ticks from minus 9 to 3. A vertical dashed reference line at zero. The same bar-and-whisker layout repeats by scenario, with darker and lighter bars per scenario. A legend box at lower left labeled Sex with two gray swatches: lighter gray labeled Women and darker gray labeled Men. Across both panels, most bars extend left of zero into negative values, with whiskers spanning wider ranges; some whiskers extend slightly to the right of zero.

Open in a new tab

Errors bars represent 95% credible intervals across 70 unique parameter combinations. ASBI indicates alcohol screening and brief intervention.

Sensitivity Analyses

First, varying the mean treatment effect of additional BIs considerably affected the projected impact of expansion. Under scenario 4, the decrease among men with a high school or less ranged between −132.8 YLL per 100 000 (CrI, −194.7 to −87.1) with a higher mean effect and −64.8 YLL per 100 000 (CrI, −91.6 to −35.6) with a lower mean effect (eFigure 5 and eTable 11-14 in Supplement 1). Second, extending universal ASBI in scenario 4 to include alcohol use category I had very little impact on the projected population-level effects (eTable 5 and eTable 15 in Supplement 1). Third, total cumulative YLL averted over 2025 to 2030 decreased with each year of delayed expansion (eFigure 6 and eTable 16 in Supplement 1). Under scenario 3, delaying the start of expansion from 2025 to 2026 (1-year delay), 2027 (2-year delay), or 2028 (3-year delay) was associated with a cumulative loss of 47.0 YLL per 100 000 (CrI, 18.3-75.3), 92.1 (CrI, 55.3-134.4), and 116.8 (CrI, 80.2-178.2) in men and 30.9 (CrI, 5.7-54.0), 63.5 (CrI, 32.6-100.6), and 82.5 (CrI, 41.6-122.8) in women by 2030, which quantifies the mortality cost of deferred action.

Discussion

This microsimulation study estimated that expanding ASBI in primary care through 2030 could lower hazardous drinking and thereby annually save up to 68.9 (CrI, 44.8-102.2) and 52.0 (CrI, 30.2-75.0) YLL per 100 000 in men and women, respectively, due to key causes of deaths associated with alcohol use. Reductions in hazardous alcohol use were projected to be more pronounced among adults with a college degree or more (than those with lower educational attainment), and among White adults (than Black and Hispanic adults), likely driven by their higher baseline consumption and more regular health care visits.25

Overall, the simulations indicated that only substantial investments in BI expansion (eg, 8 million additional annual BI in scenario 3) would yield meaningful declines in YLL. This would correspond to approximately 3.6 times as many BIs as simulated under the reference scenario in 2025, with uncertainty driven by baseline coverage estimates derived from self-report in NSDUH data (reference, 3.1 million [CrI, 1.8-4.8] vs scenario 3: 11.1 million [CrI, 9.8-12.8]; ratio 3.6 [CrI, 2.7-5.4]). However, community implementation of ASBI remains limited due to persistent barriers, including time constraints and competing priorities in primary care, limited training and confidence among clinicians, inadequate reimbursement, and fragmented workflows, as documented by a comprehensive body of existing literature.26,27,28 These challenges, while beyond the scope of this analysis, highlight the importance of implementation strategies that integrate ASBI into standard primary care processes. Recent evidence from a multisite Agency for Healthcare Research and Quality initiative29,30,31 provides important context for the feasibility of our expansion scenarios, showing that structured practice support increased the proportion of patients screened for unhealthy alcohol use from 23.2% to 35.8% and those receiving brief counseling from 20.1% to 27.6% across project sites.32 While these gains are encouraging, the scale of expansion modeled in our scenario 3 would demand broader system-level implementation than has yet been demonstrated in any single initiative.

A sensitivity analysis further illustrates the urgency of action: delaying roll-out of scenario 3 by 3 years (to 2028) was projected to result in 116.8 and 82.5 YLL foregone per 100 000 men and women by 2030, suggesting that the value of timely action is sizeable even within a relatively short implementation window. Another practically relevant finding is that expanding screening alone under stagnant BI delivery rates (scenario 1) produced YLL reductions that were statistically indistinguishable from zero in both men and women. This finding suggests that screening rates alone are an inadequate proxy for the quality and impact of ASBI programs.

With respect to education-based socioeconomic status and mortality, the model projected the largest absolute decreases in YLL per 100 000 among those with high school or less education, a group with a disproportionately high alcohol-attributable burden. This pattern—comparatively larger gains in YLL despite smaller reductions in hazardous drinking—is consistent with the alcohol harm paradox: individuals with lower socioeconomic status experience greater adverse health consequences at comparable or even lower levels of consumption.33 This paradox is thought to result from compounding structural and contextual factors including co-occurring risk factors such as smoking or illicit drug use, greater exposure to social and environmental stressors, higher prevalence of comorbid conditions, and lower health care access.34

Despite these larger absolute gains, ASBI expansion had measurable but modest equity-correcting potential: under universal ASBI delivery, the YLL gap between those with high school or less and college or more education narrowed by only 3.5% in men and 4.9% in women. ASBI scale-up alone is therefore unlikely to eliminate the socioeconomic inequalities in key causes associated with alcohol use. Structural barriers to health care access and broader social determinants of health remain important constraints, even under ambitious ASBI expansion. Complementary policies that improve primary care access and affordability for disadvantaged populations and address the broader social determinants of health will be needed to substantially narrow these inequalities.

Limitations

Our modeling is based on several key assumptions; as such, the following limitations need to be acknowledged. First, the ASBI care cascade was informed by regression models built on cross-sectional NSDUH data, where alcohol use and ASBI receipt are self-reported and subject to biases, most importantly underreporting of level and patterns of use. ASBI receipt was operationalized using the survey item on self-reported receipt of advice from a physician to cut down on drinking, which is subject to recall bias and may capture heterogeneous clinical interactions, ranging from brief, directive advice to reduce alcohol use to more structured BIs that incorporate motivational techniques and patient-led goal setting.

Second, socioeconomic status was represented only by educational attainment without considering employment or income due to high missingness in input data. Third, race and ethnicity analyses exclude the other category, which combines American Indian and Alaska Native, Asian, and Native Hawaiian or Other Pacific Islander individuals, due to small sample sizes, high heterogeneity, and high racial misclassification rates for American Indian and Alaska Native on death certificates.35 This aggregation can obscure important disparities, particularly given that American Indian and Alaska Native communities experience the highest alcohol-attributable mortality rates of any US population group.36 The potential equity implications of ASBI expansion for American Indian and Alaska Native communities therefore could not be examined.

Fourth, modeled causes of death do not include alcohol-related cancers, which contribute substantially to alcohol-attributable mortality. Because these cancers are characterized by substantial latency between exposure and disease manifestation,37 they cannot be adequately represented within the present modeling framework based on annual drinking transitions and comparatively short-term follow-up; our estimates therefore capture only short- and intermediate-term impacts of ASBI expansion.

Fifth, projections into the future inherently involve uncertainty. Although we used US Census Bureau population and mortality projections as benchmarks, these do not probabilistically quantify uncertainty, and all projections assume that no external events or interventions alter model parameters. Long-term effects from the COVID-19 pandemic may also not be fully captured.

Sixth, we assumed a homogenous BI effect scaled only by baseline consumption level because evidence was insufficient to parameterize subgroup-specific effects.9 While BI effectiveness may vary across demographic groups,5,38 differential access to ASBI—which is empirically substantiated and modeled across subgroups—may be at least as important for differential population-level impact as treatment effect heterogeneity.

Lastly, we did not model referral to specialized alcohol treatment or pharmacotherapy for AUD (eg, naltrexone, acamprosate),39 both of which address individuals with the heaviest alcohol use (and likely addiction) and are, therefore, expected to have a meaningful impact on mortality. However, evidence that ASBI reliably increases specialized alcohol care is inconsistent,40,41 which is why modeling these pathways was beyond the scope of this study and remain important avenues for future work.

Conclusions

This decision analytical study shows that expanding AS and, particularly, BI in primary care has the potential to decrease mortality from key causes of death associated with alcohol use. Although the largest absolute YLL reductions were projected among lower−socioeconomic status groups, ASBI expansion alone is unlikely to eliminate socioeconomic inequalities in alcohol-attributable mortality. Concerted investment and effort by payers and health care professionals will be needed to routinely integrate evidence-based ASBI into standard primary care workflows. If combined with policies that increase primary care access generally, expanded ASBI would reach even more US residents and likely achieve greater reductions in mortality across diverse population subgroups.

Supplement 1.

eMethods 1. ODD Protocol

eTable 1. Simulated prevalence of alcohol use categories by sex and education (standard effect)

eTable 2. Simulated prevalence of alcohol use categories by sex and race and ethnicity (standard effect)

eTable 3. Simulated prevalence of heavy episodic drinking by sex and education (standard effect)

eTable 4. Simulated prevalence of heavy episodic drinking by sex and race and ethnicity (standard effect)

eTable 5. Simulated combined YLL per 100,000 by sex

eTable 6. Simulated combined YLL per 100,000 by sex and education (standard effect)

eTable 7. Simulated combined YLL per 100,000 by sex and race and ethnicity (standard effect)

eTable 8. Simulated YLL per 100,000 by sex and cause (standard effect)

eTable 9. Simulated YLL per 100,000 by sex comparing different liver disease aetiologies (standard effect)

eTable 10. Simulated YLL per 100,000 by sex, education, and cause (standard effect)

eTable 11. Simulated combined YLL per 100,000 by sex and education (sensitivity analysis 1, maximum effect)

eTable 12. Simulated combined YLL per 100,000 by sex and education (sensitivity analysis 1, minimum effect)

eTable 13. Simulated combined YLL per 100,000 by sex and race and ethnicity (sensitivity analysis 1, maximum effect)

eTable 14. Simulated combined YLL per 100,000 by sex and race and ethnicity (sensitivity analysis 1, minimum effect)

eTable 15. Simulated combined YLL per 100,000 (sensitivity analysis 2)

eTable 16. Cumulative YLL reductions per 100,000 across the 2025–2030 period (sensitivity analysis 3)

eFigure 1. Simulated percentage of adults with hazardous alcohol use who received a brief intervention by sex and race and ethnicity

eFigure 2. Simulated mean prevalence of hazardous alcohol use by sex and race and ethnicity

eFigure 3. Simulated mean change in heavy episodic drinking prevalence by subgroup

eFigure 4. Simulated mean change in YLL per 100,000 by sex.

eFigure 5. Simulated mean change in combined YLL per 100,000 by sex, education and average brief intervention effect (sensitivity analysis 1)

eFigure 6. Cumulative changes in YLL per 100,000 by sex across the 2025–2030 period (sensitivity analysis 3)

jamahealthforum-e262348-s001.pdf (4.9MB, pdf)
Supplement 2.

Data Sharing Statement

jamahealthforum-e262348-s002.pdf (18.1KB, pdf)

References

  • 1.Curry SJ, Krist AH, Owens DK, et al. ; US Preventive Services Task Force . Screening and behavioral counseling interventions to reduce unhealthy alcohol use in adolescents and adults: US Preventive Services Task Force Recommendation Statement. JAMA. 2018;320(18):1899-1909. doi: 10.1001/jama.2018.16789 [DOI] [PubMed] [Google Scholar]
  • 2.Curfman G, Boumil M, Bibbins-Domingo K. Insurance coverage without cost sharing for preventive health care restored by the Supreme Court. JAMA. 2025;334(7):573-575. doi: 10.1001/jama.2025.12149 [DOI] [PubMed] [Google Scholar]
  • 3.Centers for Medicare & Medicaid Services . Decision Memo for Screening and Behavioral Counseling Interventions in Primary Care to Reduce Alcohol Misuse. US Department of Health and Human Services, 2011. Accessed July 9, 2025. https://www.cms.gov/medicare-coverage-database/view/ncacal-decision-memo.aspx?NCAId=249
  • 4.McKnight-Eily LR, Okoro CA, Turay K, Acero C, Hungerford D. Screening for alcohol use and brief counseling of adults−13 states and the District of Columbia, 2017. MMWR Morb Mortal Wkly Rep. 2020;69(10):265-270. doi: 10.15585/mmwr.mm6910a3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Kilian C, Lemp JM, Probst C. Who benefits from alcohol screening and brief intervention? A mini-review on socioeconomic inequalities with a focus on evidence from the United States. Addict Behav. 2023;145:107765. doi: 10.1016/j.addbeh.2023.107765 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.National Center for Health Statistics . Percentage of doctor visit in the past 12 months for adults aged 18 and over, United States, 2023. National Health Interview Survey. Accessed August 19, 2025. https://wwwn.cdc.gov/NHISDataQueryTool/ER_Quarterly/index_quarterly.html
  • 7.Institute for Health Metrics and Evaluation . Global Burden of Disease. Accessed July 9, 2025. 2023. http://vizhub.healthdata.org/gbd-compare
  • 8.Probst C, Könen M, Rehm J, Sudharsanan N. Alcohol-attributable deaths help drive growing socioeconomic inequalities in US life expectancy, 2000-18. Health Aff (Millwood). 2022;41(8):1160-1168. doi: 10.1377/hlthaff.2021.01905 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Kaner EF, Beyer FR, Muirhead C, et al. Effectiveness of brief alcohol interventions in primary care populations. Cochrane Database Syst Rev. 2018;2(2):CD004148. doi: 10.1002/14651858.CD004148.pub4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Probst C, Buckley C, Lasserre AM, et al. Simulation of alcohol control policies for health equity (SIMAH) project: study design and first results. Am J Epidemiol. 2023;192(5):690-702. doi: 10.1093/aje/kwad018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Grimm V, Railsback SF, Vincenot CE, et al. The ODD protocol for describing agent-based and other simulation models: a second update to improve clarity, replication, and structural realism. J Artif Soc Soc Simul. 2020;23:7. doi: 10.18564/jasss.425933204215 [DOI] [Google Scholar]
  • 12.Shield K, Manthey J, Rylett M, et al. National, regional, and global burdens of disease from 2000 to 2016 attributable to alcohol use: a comparative risk assessment study. Lancet Public Health. 2020;5(1):e51-e61. doi: 10.1016/S2468-2667(19)30231-2 [DOI] [PubMed] [Google Scholar]
  • 13.Missouri Department of Health and Senior Services . Years of Potential Life Lost (YPLL). Accessed May 11, 2026. https://health.mo.gov/data/ypll/
  • 14.State Health Access Data Assistance Center . Premature Deaths: Understanding Years of Potential Life Lost as a Measure of Public Health. Accessed May 11, 2026. https://www.shadac.org/news/premature-deaths-measure-years-potential-life-lost-public-health
  • 15.Khosla S, Del Rios M, Kotini-Shah P, Weber J, Vanden Hoek T; Illinois Heart Rescue . Years of potential life lost and mean age of adults experiencing nontraumatic, out-of-hospital cardiac arrests-Chicago, 2014-2021. MMWR Morb Mortal Wkly Rep. 2024;73(9):199-203. doi: 10.15585/mmwr.mm7309a2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.World Health Organization . International Guide for Monitoring Alcohol Consumption and Related Harm. World Health Organization; 2001. [Google Scholar]
  • 17.US Census Bureau . US Census. Accessed June 30, 2023. https://www.census.gov/programs-surveys/decennial-census.html
  • 18.Ruggles S, Flood S, Goeken R, Schouweiler M, Sobek M. IPUMS USA, version 12.0. 2022. Accessed June 25, 2026. https://www.ipums.org/projects/ipums-usa/d010.v12.0
  • 19.Institute for Social Research, University of Michigan, Ann Arbor Survey Research Center . Panel Study of Income Dynamics, public use dataset. Accessed June 30, 2023. https://psidonline.isr.umich.edu
  • 20.US Centers for Disease Control and Prevention . Behavioral Risk Factor Surveillance System. Accessed December 28, 2022. https://www.cdc.gov/brfss/
  • 21.Buckley C, Brennan A, Kerr WC, et al. Improved estimates for individual and population-level alcohol use in the United States, 1984-2020. Int J Alcohol Drug Res. 2022;10(1):24-33. doi: 10.7895/ijadr.383 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.National Center for Health Statistics . National Vital Statistics System. Accessed June 30, 2023. https://www.cdc.gov/nchs/nvss/index.htm
  • 23.US Census Bureau . 2023. National Population Projections. Accessed October 28, 2024. https://www.census.gov/data/datasets/2023/demo/popproj/2023-popproj.html
  • 24.Substance Abuse and Mental Health Services Administration . National Survey on Drug Use and Health (NSDUH). Accessed October 9, 2024. https://www.samhsa.gov/data/data-we-collect/nsduh-national-survey-drug-use-and-health
  • 25.Mulia N, Schmidt LA, Ye Y, Greenfield TK. Preventing disparities in alcohol screening and brief intervention: the need to move beyond primary care. Alcohol Clin Exp Res. 2011;35(9):1557-1560. doi: 10.1111/j.1530-0277.2011.01501.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Vendetti J, Gmyrek A, Damon D, Singh M, McRee B, Del Boca F. Screening, brief intervention and referral to treatment (SBIRT): implementation barriers, facilitators and model migration. Addiction. 2017;112(suppl 2):23-33. doi: 10.1111/add.13652 [DOI] [PubMed] [Google Scholar]
  • 27.Rosário F, Santos MI, Angus K, Pas L, Ribeiro C, Fitzgerald N. Factors influencing the implementation of screening and brief interventions for alcohol use in primary care practices: a systematic review using the COM-B system and Theoretical Domains Framework. Implement Sci. 2021;16(1):6. doi: 10.1186/s13012-020-01073-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Chan PS, Fang Y, Wong MC, Huang J, Wang Z, Yeoh EK. Using Consolidated Framework for Implementation Research to investigate facilitators and barriers of implementing alcohol screening and brief intervention among primary care health professionals: a systematic review. Implement Sci. 2021;16(1):99. doi: 10.1186/s13012-021-01170-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Huffstetler AN, Villalobos G, Webel B, et al. Practice facilitation to address unhealthy alcohol use in primary care: a cluster randomized clinical trial. JAMA Health Forum. 2024;5(8):e242371. doi: 10.1001/jamahealthforum.2024.2371 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Jortberg BT, Dickinson LM, Fernald DH, et al. Improving care for unhealthy alcohol use: results from the Facilitating Alcohol Screening and Treatment (FAST) Colorado Study. J Am Board Fam Med. 2024;37(6):1027-1037. doi: 10.3122/jabfm.2024.240048R1 [DOI] [PubMed] [Google Scholar]
  • 31.Jonas DE, Riley SR, Brouwer L, et al. Screening and counseling for unhealthy alcohol use in primary care practices. JAMA Netw Open. 2026;9(1):e2553518. doi: 10.1001/jamanetworkopen.2025.53518 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.AHRQ Primary Care . EvidenceNOW Webinar: Findings From the AHRQ Unhealthy Alcohol Use in Primary Care Initiative. 2024. Accessed May 14, 2026. https://www.youtube.com/watch?v=UxCEtUlPlbQ
  • 33.Bellis MA, Hughes K, Nicholls J, Sheron N, Gilmore I, Jones L. The alcohol harm paradox: using a national survey to explore how alcohol may disproportionately impact health in deprived individuals. BMC Public Health. 2016;16:111. doi: 10.1186/s12889-016-2766-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Boyd J, Sexton O, Angus C, Meier P, Purshouse RC, Holmes J. Causal mechanisms proposed for the alcohol harm paradox-a systematic review. Addiction. 2022;117(1):33-56. doi: 10.1111/add.15567 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Arias E, Schauman WS, Eschbach K, Sorlie PD, Backlund E. The validity of race and Hispanic origin reporting on death certificates in the United States. Vital Health Stat 2. 2008;(148):1-23. [PubMed] [Google Scholar]
  • 36.Spillane S, Shiels MS, Best AF, et al. Trends in alcohol-induced deaths in the United States, 2000-2016. JAMA Netw Open. 2020;3(2):e1921451. doi: 10.1001/jamanetworkopen.2019.21451 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Holmes J, Meier PS, Booth A, Guo Y, Brennan A. The temporal relationship between per capita alcohol consumption and harm: a systematic review of time lag specifications in aggregate time series analyses. Drug Alcohol Depend. 2012;123(1-3):7-14. doi: 10.1016/j.drugalcdep.2011.12.005 [DOI] [PubMed] [Google Scholar]
  • 38.Roheger M, Gürtler D, Eidswick J, Ittermann T, Freyer-Adam J, Baumann S. Socio-economic equity in the impact of population-based interventions to reduce alcohol consumption. Cochrane Database Syst Rev. 2025;5(5):CD016015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.McPheeters M, O’Connor EA, Riley S, et al. Pharmacotherapy for alcohol use disorder: a systematic review and meta-analysis. JAMA. 2023;330(17):1653-1665. doi: 10.1001/jama.2023.19761 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Karno MP, Rawson R, Rogers B, et al. Effect of screening, brief intervention and referral to treatment for unhealthy alcohol and other drug use in mental health treatment settings: a randomized controlled trial. Addiction. 2021;116(1):159-169. doi: 10.1111/add.15114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Glass JE, Hamilton AM, Powell BJ, Perron BE, Brown RT, Ilgen MA. Specialty substance use disorder services following brief alcohol intervention: a meta-analysis of randomized controlled trials. Addiction. 2015;110(9):1404-1415. doi: 10.1111/add.12950 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1.

eMethods 1. ODD Protocol

eTable 1. Simulated prevalence of alcohol use categories by sex and education (standard effect)

eTable 2. Simulated prevalence of alcohol use categories by sex and race and ethnicity (standard effect)

eTable 3. Simulated prevalence of heavy episodic drinking by sex and education (standard effect)

eTable 4. Simulated prevalence of heavy episodic drinking by sex and race and ethnicity (standard effect)

eTable 5. Simulated combined YLL per 100,000 by sex

eTable 6. Simulated combined YLL per 100,000 by sex and education (standard effect)

eTable 7. Simulated combined YLL per 100,000 by sex and race and ethnicity (standard effect)

eTable 8. Simulated YLL per 100,000 by sex and cause (standard effect)

eTable 9. Simulated YLL per 100,000 by sex comparing different liver disease aetiologies (standard effect)

eTable 10. Simulated YLL per 100,000 by sex, education, and cause (standard effect)

eTable 11. Simulated combined YLL per 100,000 by sex and education (sensitivity analysis 1, maximum effect)

eTable 12. Simulated combined YLL per 100,000 by sex and education (sensitivity analysis 1, minimum effect)

eTable 13. Simulated combined YLL per 100,000 by sex and race and ethnicity (sensitivity analysis 1, maximum effect)

eTable 14. Simulated combined YLL per 100,000 by sex and race and ethnicity (sensitivity analysis 1, minimum effect)

eTable 15. Simulated combined YLL per 100,000 (sensitivity analysis 2)

eTable 16. Cumulative YLL reductions per 100,000 across the 2025–2030 period (sensitivity analysis 3)

eFigure 1. Simulated percentage of adults with hazardous alcohol use who received a brief intervention by sex and race and ethnicity

eFigure 2. Simulated mean prevalence of hazardous alcohol use by sex and race and ethnicity

eFigure 3. Simulated mean change in heavy episodic drinking prevalence by subgroup

eFigure 4. Simulated mean change in YLL per 100,000 by sex.

eFigure 5. Simulated mean change in combined YLL per 100,000 by sex, education and average brief intervention effect (sensitivity analysis 1)

eFigure 6. Cumulative changes in YLL per 100,000 by sex across the 2025–2030 period (sensitivity analysis 3)

jamahealthforum-e262348-s001.pdf (4.9MB, pdf)
Supplement 2.

Data Sharing Statement

jamahealthforum-e262348-s002.pdf (18.1KB, pdf)

Articles from JAMA Health Forum are provided here courtesy of American Medical Association

RESOURCES