Problematic Alcohol Use in Debre Berhan, Ethiopia: A Community-Based Cross-Sectional Study

Abstract

Background

Problematic alcohol use (PAU) negatively affects physical, psychological, and social well-being. This study aims to estimate the community-level prevalence of PAU and identify factors associated with PAU in Central Ethiopia.

Methods

A community-based cross-sectional study was conducted in Debre Berhan, Ethiopia, in January and February 2022. We recruited 1,400 adults aged 18 years and older using multistage sampling, and 1,380 completed the study interview. PAU was assessed with the alcohol use disorder identification test (positive AUDIT score). We conducted multivariable logistic regression to identify factors associated with PAU, with predictors selected using least absolute shrinkage and selection operator (LASSO) regression, in the overall sample and in sex-stratified analyses.

Results

The prevalence of PAU was 15.5% (95% CI: 13.7% –17.5%) overall, and 26.3% (95% CI: 23.2%–29.6%) and 3.4% (95% CI: 2.2%–5.1%) for men and women, respectively. For the overall sample, factors associated with PAU included being male (AOR = 7.83; 95% CI: 4.59–13.36), unemployment (AOR = 2.73; 95% CI: 1.29–5.78) or engaged in income-generating roles (AOR = 2.38; 95% CI: (1.23–4.58) compared to those in non-income generating roles like housewives and students, smoking (AOR = 5.24; 95% CI: 1.73–15.91), high blood pressure (AOR = 2.04; 95% CI: 1.30–3.21), only one or ≥2 stressful life events (AORs = 1.60 and 1.65; 95% CIs: 1.03–2.47 and 1.05–2.61, respectively), and family history of PAU (AOR = 4.04; 95% CI: 2.09–7.81). For men, smoking (AOR = 5.29; 95% CI: 1.66–16.81), high blood pressure (AOR = 2.33; 95% CI: 1.39–3.90), ≥2 stressful events (AOR = 1.71; 95% CI: 1.04–2.82), and family history of PAU (AOR = 4.10; 95% CI: 1.90–8.84) were associated with greater odds of PAU. For women, being 25-34 years compared to >35 years old (AOR = 8.88; 95% CI: 1.56–50.63), being divorced (compared to married) (AOR = 5.72 95% CI: 1.27– 25.72) were associated with greater odds of PAU, as were only one stressful life event (AOR = 4.05; 95% CI: 1.14–14.36), and family history of PAU (AOR = 10.98; 95% CI: 1.69–71.14).

Conclusions

PAU prevalence was high in this sample, significantly higher among men compared to women. Our study highlights the need to implement public health interventions that target risk factors for PAU in Ethiopia. Findings also highlight the need for selective PAU prevention interventions that target individuals with a family history of PAU or exposure to stressful life events.

Introduction

Alcohol consumption is a contributing factor to more than 200 illnesses, including alcohol dependence, liver cirrhosis, cancers, and physical injuries. ¹ The alcohol per capita consumption (APC) for adults was estimated to be 6.4 liters of pure alcohol per person per year worldwide, 9.8 liters in the WHO European Region, 8.0 liters in the Americas, ² 6.0 liters in the WHO African Region, ¹ and 3.2 liters in Ethiopia. ³

The relatively low APC figure for Ethiopia possibly underestimates the true alcohol consumption, given that unrecorded alcohol use, consisting of home-brewed traditional alcoholic drinks such as tella, tej, and araki that are widely consumed in Ethiopia, contributes to nearly 25.0% of total alcohol consumption worldwide and nearly 30% in Africa. ¹ These estimates suggest that Ethiopia’s APC figure underrepresents the actual level of alcohol consumption, and that PAU in Ethiopia may be more prevalent than usually reported—emphasizing the importance of community-based epidemiological surveys like the present one.

Problematic alcohol use (PAU) defines a spectrum of problematic and excessive patterns of alcohol use, ranging from alcohol use at hazardous levels like heavy episodic drinking (HED), which puts users at risk for adverse health and social outcomes, to harmful use of alcohol and more serious disorders–alcohol use disorder (AUD) that have severe health consequences.^3,4 Harmful alcohol use is one of the leading risk factors for death and disability, contributing to 5.1% of all disability-adjusted life years (DALYs) and 5.3% of all deaths globally. ³ Around 0.5%–5.0% of the global population fulfills AUD diagnostic criteria, with the highest prevalence reported in people aged 15–49. ⁵

There is inconsistent evidence of the prevalence of AUD in Africa ⁴ ; however, estimates suggest a 12-month prevalence of nearly 3.7% in the WHO African Region, with notable variation between males (6.3%) and females (0.7%). ³

In Ethiopia, alcohol consumption is an increasing public health concern, partly driven by the expansion of the alcohol industry, with increasing production, marketing, and accessibility of alcoholic drinks. ⁶ Ethiopia has a long history of unrecorded alcohol use, in the form of home-brewed beverages such as “tella” or “talla” (local beer), “tejj” (Indigenous honey wine or mead), and “areki/araki,” also spelled as “areqie/araqe” or “katikala” (traditional liquor akin to vodka), that has long been an integral part of community life, especially among the highlanders.^7,8

Yet epidemiological evidence on PAU prevalence is limited. ⁶ A review from Ethiopia reported a pooled prevalence of PAU of 8.9% (11.6% for men and 1.2% for women). ⁹ However, a spatial analysis reported regional variations in drinking, highlighting the need for representative and localized PAU prevalence data. ¹⁰ For example, Debre Berhan is known for its large production of home-brewed alcoholic drinks, yet there is a dearth of community-level epidemiological alcohol studies. Previous studies have focused on specific segments of Debre Berhan’s population, including prisoners, ¹¹ pregnant women, ¹² and students. ¹³ Community-based alcohol studies provide an opportunity to capture a representative population sample, enabling more accurate estimates of PAU prevalence and potential predictors of PAU. ¹⁴

A comprehensive assessment of alcohol use and its associated factors in a localized setting, such as Debre Berhan, is needed to inform local public health policies, guide contextually relevant prevention and treatment strategies, and facilitate future epidemiological studies. ¹⁵ This localized evidence could feed and inform the development of a national alcohol policy in Ethiopia and support the updating of laws, regulations, proclamations, and restrictions that currently are not adequate to reduce the multifaceted harmful use of alcohol and related health conditions.^3,8

This study addresses this gap and the need for localized information on PAU prevalence in Debre Behran. More specifically, it aims to estimate the prevalence of PAU and identify factors associated with PAU among residents of Debre Berhan town.

Methods

Study Design and Setting

This community-based cross-sectional survey was conducted in Debre Berhan town, located in central Ethiopia,130 km Northeast of Addis Ababa, the capital of Ethiopia. During data collection (10 January to 13 February 2022), there were nine ‘kebeles’ in the town–'kebele’ is the smallest administrative unit of local government next to ‘woreda’ (districts). According to the Ethiopian Statistical Service (ESS) projection for 2023, the total population of Debre Berhan was estimated at 167,946. ¹⁶

Sampling Strategy and Participant Recruitment

The source population was all adults aged 18 years and older residing in Debre Berhan. We used a three-stage random sampling to select the study population. First, four kebeles were selected from nine using simple random sampling.

Second, systematic random sampling was applied to select households (HHs) within each selected kebele to participate in the survey. We used government data to estimate population and household size (HHs) to prepare a sampling frame and enumeration areas (EAs) within each kebele. Data from the Central Statistical Agency (CSA) of Ethiopia, Debre Berhan city administration sector offices, and HH registration folder records of health extension workers (HEWs) were used for HH selection. The HEWs’ records, which included routinely updated HHs, were mainly used to map existing HHs and randomly select the required number of HHs within each EA. Each kebele’s respective sample size (ni) was determined using a proportional allocation formula based on the total sample size (N) and the number of HHs in each selected kebele. The sampling fraction (K) was determined by dividing the total study HHs by N, giving a value of nearly 5, and every fifth consecutive HH was approached to identify an eligible participant until the required sample size was reached. In cases where an individual in a selected HH was unavailable after three visits or refused to consent, the HH was recorded as a non-respondent. Recruitment continued with the next scheduled HH until the target number of respondents was enrolled to maintain the representativeness and randomization of the systematic sampling strategy.

Third, from each selected household that agreed to participate, we randomly selected one individual who fulfilled the eligibility criteria. All adult HH members aged 18 years and older were screened for age, residence duration, and capacity to consent to identify eligible participants. To be eligible, HH members had to be at least 18 years old and have resided in the area for at least six months–an adequate period for community integration and exposure to local contexts. Individuals with impaired communication or severe physical or mental disabilities that limitted their ability to understand or answer the questions, as well as those who refused to provide consent for participation, were excluded. Eligibility was assessed through observation or self-report during the screening process. If two or more people from the HHs met eligibility criteria, a Kish Method ¹⁷ was employed to select one individual for the interview (Appendix S1, Supplementary File 1).

Local authorities were informed before the initiation of the study to ensure their support for the research team. A team of 24 people (17 female and seven male), consisting of four team leaders, ten interviewers, and ten assistant data collectors, was involved in data collection. The data collectors administered an Amharic-language semi-structured face-to-face survey interview orally, recording participant responses using an Open Data Kit (ODK) electronic data collection application software. This approach continued until the sample of total enrolled participants was recruited.

We used a single population proportion formula to estimate the minimum required sample size to estimate the prevalence of PAU. Taking a 21% prevalence (P) of hazardous alcohol use in Ethiopia ¹⁸ and assuming this prevalence to obtain the maximum sample size at 95% certainty with a maximum discrepancy of 4.0% (d) between the sample and the underlying population, a calculation was done with Epi-Info software for a population survey. To ensure an adequate sample for less common outcomes and allow for sex-stratified analyses, the sample was increased with a conservative design effect of 3.0, resulting in a recruitment target of 1194 participants. With a 15% increase for possible non-response, the final target sample size was enhanced to 1373. To facilitate data collection and ensure a manageable workload for the 10 data collection groups, 140 respondents were assigned to each team for recruitment, resulting in 1,400 respondents being approached for screening.

Ethics Approval

This study was approved by the Institutional Review Board (IRB) of the College of Health Sciences (CHS), Addis Ababa University (AAU) (Protocol No. 119/20/Psy; Date of ethical approval: 30 December 2020). All study participants were provided with detailed information about the study’s aims before starting the interview, and informed written consent was obtained. The study was conducted in accordance with the Declaration of Helsinki. Participant confidentiality was ensured by using coded identifiers in place of personal identifying information.

Measures

Alcohol Outcomes

Problematic alcohol use (PAU) was the primary outcome variable, with PAU defined as a total score of ≥8 on the 10-item Alcohol Use Disorder Identification Test (AUDIT), which is the recommended cut-off indicating hazardous, harmful alcohol use and possible alcohol dependence. A positive AUDIT score (AUDIT ≥8) is a public health screening threshold, not a diagnostic criterion.

The AUDIT assesses risk of alcohol-related harms in the past 12 months, with items assessing alcohol consumption patterns, drinking behaviours, and alcohol-related harms. ¹⁹ The AUDIT has been widely used in Ethiopia but has not been formally validated for use within the general population.^20,21 In this study, we developed a showcard depicting common traditional and commercially produced alcoholic drinks and their approximate alcohol content.^18,20,22 Detailed methods for the measures of alcohol outcomes and showcards are provided in (Appendix S2, Supplementary File 1).

Independent Variables

The selection of independent variables for examining their association with the primary outcome was informed by existing literature and their availability in the baseline dataset.

Demographics and Living Situation

Sociodemographic factors such as age, sex, marital status, relationship in the family, length of stay, education, employment, socioeconomic status, religion, and household size were assessed using a self-report questionnaire.

Physical Health and Other Health Risk Factors

Self-report questions were used to assess chronic medical conditions, defined as the presence of at least one medically confirmed chronic illness, including hypertension, diabetes mellitus, HIV, other related medical problems, and current medication use– defined as the use of one or more prescription medications for medical conditions in the previous 12 months.

Blood pressure was measured using the Boso Medicus-X digital upper-arm blood pressure monitor, which met the International Validation Protocol standards of the European Society of Hypertension. ²³ Blood pressure was classified as normal (<120 and < 80), pre-hypertension (120-139 or 80–89), stage 1 hypertension (140-159 or 90–99), stage 2 hypertension (≥160 or ≥100), and stage 3 hypertension (≥180 mmHg or ≥110). ²⁴ This study defined high blood pressure (HBP) as the presence of any of the following diagnostic categories: stage 1, stage 2, or stage 3 hypertension.

Physical activity was assessed based on the WHO recommendations, including leisure time, physical activity, transportation, occupational activities, household chores, play, games, and sports or planned exercise. ²⁵ Participants were asked to estimate their average daily physical activity as less than 20 minutes or 20 minutes or more.

Injuries or admissions were assessed using compiled items from previous studies to detect participants’ medical conditions and experiences. ²⁶ It was defined as the presence of either a self-report of physical traumas, injuries, and accidents or a history of hospitalizations, admissions, or healthcare visits in the last 12 months. ²⁷

Other Substance Use

The use of tobacco, khat, and other substances was measured using the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST). ²⁸ Based on the ASSIST scoring guidelines, each substance was initially scored using the standard scoring categories–low, moderate, and high risk. However, because of the low frequency of responses in those categories and difficulties in maintaining sufficient statistical power for the analysis, commonly used substances in our study setting–including tobacco, khat, and injectable drugs–were recoded into binary categories (use vs. non-use). ASSIST was also adapted to the local context and used in Ethiopia previously. ²⁹

Mental Health and Psychosocial Factors

Depressive symptoms were assessed using the 9-item Patient Health Questionnaire (PHQ-9). ³⁰ Probable major depressive disorder (MDD) was defined as a PHQ-9 cut-off point of ≥10 based on recommended use in Ethiopia. ³¹ Suicidal behavior was screened using a Mini International Neuropsychiatric Interview (MINI Plus 5.0.0). ³² The risk of suicidal behavior was categorized as ‘no risk’ (scores <5) and ‘has risk’ (low, moderate, and high-risk: scores of 1-5, 6-9, and ≥10, respectively). Anxiety symptoms were assessed using a 7-item Generalized Anxiety Disorder checklist (GAD-7), which has good reliability. ³³ Probable anxiety symptoms were defined using GAD-7 scores of 7-10, based on pooled estimates from a review ³⁴ and previous use in Ethiopia. ³⁵

Perceived social support was measured using the Oslo-3 item Social Support Scale (OSS-3), which is widely used in Ethiopia. ³⁶ This scale was scored from 3 to 14, with scores 3-8 indicating poor support, 9-11 indicating moderate support, and 12-14 indicating strong support. Stressful life events were measured using the List of Threatening Experiences (LTE), ³⁷ a 12-item instrument that assesses main life events related to long-term threats. This instrument has good test-retest reliability and validity. ³⁸ The LTE was translated into Amharic and culturally adapted for use in Ethiopia. The number of stressful events was categorized into no events, one event, or two or more events. ³⁹

Quality of Life, Functioning, Insomnia, and Sexual Health

Quality of Life was assessed using the WHO Quality of Life scale-Brief version (WHOQOL-BREF), ⁴⁰ which has been semantically translated into Amharic for use in the Ethiopian context. ⁴¹ It contains 26 questions across four domains (physical, psychological, social relations, and environment), with each item rated on a 5-point Likert scale (1 = very dissatisfied to 5 = very satisfied), yielding a total score of 26-130. The four domains were then transformed to a 0 to 100 scale using the SPSS syntax conversion method that helps interpret scores, with 100 indicating the best possible quality of life. ⁴²

Impairment of functioning was assessed using the 12-item WHO Disability Assessment Schedule-2.0 (WHODAS-2.0), ⁴³ which was validated in Ethiopia. ⁴⁴ The WHO simple scoring method was employed, rating each of the 12 items on a 5-point scale,^1-5 yielding a total score of 12-60, with higher scores indicating greater disability. This method is easy to interpret and strongly aligns with complex scoring. ⁴⁵ Sleep and sexual problems were measured using the 7-item Insomnia Severity Index (ISI) ⁴⁶ and the 5-item Arizona Sexual Experiences Scale (ASEX), ⁴⁷ respectively, with higher scores reflecting greater severity.

Data Collectors and Data Quality Control

Data collection tools that had not been previously validated in Ethiopia were semantically translated into Amharic and adapted to the local context. Training was provided to interviewers, assistants, and team leaders on the use of both paper-based and electronic versions of the questionnaire. A pre-test was conducted on 5.0% of the total sample size prior to the commencement of actual data collection. Based on findings from the pre-test, the questionnaire and data collection techniques were revised accordingly. Specifically, the skip patterns for related questions were modified to enhance logical flow, and the ODK form’s required-response function was enabled to prevent missing data. Items related to consumption in the AUDIT were discussed during the debriefing, and additional training was provided to data enumerators on the proper administration of the AUDIT. To avoid confusion caused by redundant IDs during the survey, an interval-based ID assignment system was followed for each enumerator.

Data collected electronically using KoboToolbox and saved simultaneously on Android devices via ODK Briefcase as offline backup. The principal investigator daily checked data completeness, consistency, and appropriate entry, and participant confidentiality was maintained by deleting data from interviewers’ devices immediately after successful online submission and backup.

Data Processing and Statistical Analyses

Data were collected using the ODK electronic tool programmed on Android phones, with pencil-and-paper for documentation and written consent (Appendix S3, Supplementary File 1).

The least absolute shrinkage and selection operator (LASSO) regularization technique, which applies a penalty to prevent overfitting and enhance the accuracy of statistical models incorporating an adaptive selection method, was used to identify variables for building the best-fitting predictive model of factors associated with PAU in our study. ⁴⁸ Using logistic regression with adaptive LASSO regularization, we developed three separate models to identify factors associated with PAU. Model 1 (overall sample, n=1380) fits 15 independent variables (variables of interest): age group, marital status, educational status, employment status, family history of alcohol-related problems, high blood pressure, tobacco, and khat use in the last three months, PHQ-9 total scores, risk of suicidal behavior, GAD-7 total scores, social support, stressful life events, WHODAS-12 total scores, and ISI total scores and sex to control confounding. Model 2 (men’s sample, n = 730) fits all 15 independent variables in Model 1, and Model 3 (women’s sample, n = 630) fits only 13 independent variables, avoiding tobacco and khat use due to inadequate sample observations for these two variables (Table S1, Supplementary File 2).

All models included 11 pre-specified control variables (monthly income, relative wealth, number of family members who use alcohol, exposure to alcohol ads, known chronic medical illness, medically confirmed known hypertension, medications prescribed in the last 12 months, and the physical, psychological, social relations, and environment domains of WHOQOL-BREF tool for quality of life) for LASSO regularization to minimize potential overfitting while preserving model parsimony ⁴⁸ which were selected a priori according to their plausible relevance to the outcome. All independent variables were treated as categorical except PHQ-9, GAD-7, WHODAS-12, and ISI total scores, which were treated as continuous.

Model fitness was checked using the Hosmer-Lemeshow goodness-of-fit test (GOF) and area under the receiver operating characteristic (ROC) curve (AUC) to evaluate predictive power. ⁴⁹ The p-values of the GOF test for models 1-3 were 0.0987, 0.2686, and 0.9763, respectively, showing a good fit. The AUC for all three models was greater than 0.7, indicating acceptable predictive performance.

This study was reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. ⁵⁰ The completed STROBE checklist is provided in (Supplementary File 3).

Results

Of the 1,400 individuals approached and planned for recruitment, 1,380 (98.6%) met the eligibility criteria, provided consent to participate, and completed the interview. This slightly increased the estimated sample size (1,373), allowing for better data capture and statistical power. Of those approached, 20 individuals were non-respondents (12 declined participation, and 8 were not found at home after 3 visits). The number of participants interviewed daily, and the time needed for a session were estimated at 5-7 people and 19-132 minutes, averaging 45.9 minutes.

Sociodemographic and Economic Characteristics of the Sample

The mean age of the respondents was 35.6±13.5 years, ranging from 18 to 95. Of the total, 730 (52.9%) were males, 759 (55.0%) were married, and 1,327(96.2%) were Orthodox Christians. Close to 62.0% of the participants were employed, 77.3% earned ≤5000 Ethiopian Birr (37.5 USD) per month, and only 0.8% reported experiencing hunger in the month before the interview (Table 1).

Table 1.

Sociodemographic and Clinical Characteristics of Respondents by Problematic Alcohol Use Status in Debre Berhan, Central Ethiopia, 2026: Stratified by sex (n = 1380; Men = 730, Women = 650)

Characteristics	Population n (%)	Overall PAU (n = 214; 15.51%)		PAU in Men (n = 192; 26.30%)		PAU in women (n = 22; 3.38%)
		Yes (%)	p-value	Yes (%)	p-value	Yes (%)	p-value
Sociodemographic and economic factors
Sex	​	​	0.001	​	​	​	​
Male	730 (52.9)	192 (26.3)	​
Female	650 (47.1)	22 (3.4)	​
Age: mean±SD (range); 35.55±13.47 (18-95)
Age group (years)	​	​	0.218	​	0.142	​	0.009
≤24	271 (19.6)	37 (13.7)	​	33 (20.5)	​	04 (3.6)	​
25-34	521 (37.8)	92 (17.7)	​	77 (29.1)	​	15 (5.9)	​
≥35	588 (42.6)	85 (14.5)	​	82 (27.0)	​	03 (1.1)	​
Marital status	​	​	0.000	​	0.426	​	0.407
Single	447 (32.4)	94 (21.0)	​	87 (28.4)	​	07 (5.0)	​
Married	759 (55.0)	107 (14.1)	​	97 (25.3)	​	10 (2.7)	​
Divorced	100 (7.3)	10 (10.0)	​	06 (26.1)	​	04 (5.2)	​
Widowed/separated/cohabited	74 (5.4)	03 (4.1)	​	02 (11.8)	​	01 (1.8)	​
Relationships in the family	​	​	0.000	​	0.731	​	0.320
Husband	386 (28.0)	96 (24.9)	​	96 (25.0)	​	0 (0.0)	​
Wife	445 (32.3)	14 (3.2)	​	03 (23.1)	​	11 (2.6)	​
Child in the family	320 (23.2)	64 (20.0)	​	59 (29.2)	​	05 (4.2)	​
Others*	229 (16.6)	40 (17.5)	​	34 (26.0)	​	06 (6.1)	​
Length of stay in the town	​	​	0.051	​	0.036	​	0.646
≤ 5 years	256 (18.6)	36 (14.1)	​	34 (22.5)	​	02 (1.9)	​
5-10 years	230 (16.7)	25 (10.9)	​	21 (18.8)	​	04 (3.4)	​
> 10 years	894 (64.8)	153 (17.1)	​	137 (29.3)	​	16 (3.8)	​
Educational status	​	​	0.043	​	0.117	​	0.350
No formal education	172 (12.5)	14 (8.1)	​	13 (18.3)	​	01 (1.0)	​
Only primary education	297 (21.5)	50 (16.8)	​	42 (29.8)	​	08 (5.1)	​
Only secondary education	443 (32.1)	73 (16.5)	​	66 (30.1)	​	07 (3.1)	​
College/university education	468 (33.9)	77 (16.5)	​	71 (23.8)	​	06 (3.6)	​
Employment status	​	​	0.000	​	0.008	​	0.084
Productive a	382 (27.7)	16 (4.2)	​	11 (13.1)	​	05 (1.7)	​
Income generating b	860 (62.3)	168 (19.5)	​	154 (27.3)	​	14 (4.7)	​
Unemployed	138 (10.0)	30 (21.7)	​	27 (32.9)	​	03 (5.4)	​
Monthly income (mean±SD): 4099.8±8605	​	​	0.001	​	0.826	​	0.002
≤ 5000 ETB	1,067 (77.3)	145 (13.6)	​	130 (25.7)	​	15 (2.7)	​
5001-10000 ETB	213 (15.4)	44 (20.7)	​	41 (27.2)	​	03 (4.8)	​
> 1000 ETB	100 (7.3)	25 (25.0)	​	21 (28.8)	​	04 (14.8)	​
Relative wealth	​	​	0.775	​	0.674	​	0.076
Lower	608 (44.1)	97 (16.0)	​	88 (27.9)	​	09 (3.1)	​
Moderate	710 (51.5)	106 (14.9)	​	96 (25.3)	​	10 (3.0)	​
Better	62 (4.5)	11 (17.7)	​	08 (22.9)	​	03 (11.1)	​
Hunger in households in the last month	​	​	0.555	​	0.468	​	0.707
No	1,369 (99.2)	213 (15.6)	​	191 (26.4)	​	22 (3.1)	​
Yes	11 (0.8)	01 (9.1)	​	01 (14.3)	​	0 (0.0)	​
Religion	​	​	0.001	​	0.005	​	0.278
Orthodox Christian	1,327 (96.2)	214 (16.1)	​	192 (27.1)	​	22 (3.6)	​
Others C	53 (3.8)	0 (0.0)	​	0 (0.0)	​	0 (0.0)	​
Household size	​	​	0.855	​	0.630	​	0.683
1-2	297 (21.5)	49 (16.5)	​	44 (23.7)	​	05 (4.5)	​
3-5	841 (60.9)	129 (15.3)	​	115 (27.1)	​	14 (3.4)	​
6 or more	242 (17.5)	36 (14.9)	​	33 (27.7)	​	03 (2.4)	​
Alcohol use-related characteristics
Ever use of alcohol drinks	​	​	0.000	​	0.002	​	0.173
No	75 (5.4)	0 (0.0)	​	0 (0.0)	​	0 (0.0)	​
Yes	1,305 (94.6)	214 (16.4)	​	192 (27.3)	​	22 (3.7)	​
Current use of alcohol (last 12 months)	​	​	0.000	​	0.000	​	0.074
No	132 (9.6)	0 (0.0)	​	0 (0.0)	​	0 (0.0)	​
Yes	1,248 (90.4)	214 (17.2)	​	192 (28.3)	​	22 (3.9)	​
Current use of alcohol (last 03 months)	​	​	0.000	​	0.000	​	0.057
No	146 (10.6)	0 (0.0)	​	0 (0.0)	​	0 (0.0)	​
Yes	1,234 (89.4)	214 (17.3)	​	192 (28.5)	​	22 (3.9)	​
Drinking initiated age (n=1305), Mean/SD: 15 ±5.8		​	0.552	​	0.489	​	0.060
≤ 15 years	695 (53.3)	110 (15.8)	​	102 (28.4)	​	08 (2.4)	​
> 15 years	610 (46.7)	104 (17.1)	​	90 (26.1)	​	14 (5.3)	​
Number of family members who use alcohol	​	​	0.000	​	0.013	​	0.579
No alcohol use in the family	125 (9.1)	13 (10.4)	​	12 (19.1)	​	01 (1.6)	​
One family member who uses alcohol	339 (24.6)	75 (22.1)	​	69 (33.7)	​	06 (4.5)	​
≥2 family members who use alcohol	916 (66.4)	126 (13.8)	​	111 (24.0)	​	15 (3.3)	​
Family history of alcohol-related problem	​	​	0.000	​	0.000	​	0.055
No	1,304 (94.5)	182 (14.0)	​	163 (23.8)	​	19 (3.1)	​
Yes	76 (5.5)	32 (42.1)	​	29 (65.9)	​	03 (9.4)	​
Drinking frequency with friends (n = 1305)	​	​	0.000	​	0.000	​	0.000
Never	637 (48.8)	15 (2.4)	​	11 (5.1)	​	04 (1.0)	​
Rarely	558 (42.8)	136 (24.4)	​	121 (31.0)	​	15 (9.0)	​
Sometimes/often/always	110 (8.4)	63 (57.3)	​	60 (60.6)	​	03 (27.3)	​
Friends’ drinking frequency when socializing	​	​	0.000	​	0.000	​	0.000
Never	686 (49.7)	14 (2.0)	​	11 (4.9)	​	03 (0.7)	​
Rarely	547 (39.6)	130 (23.8)	​	115 (29.9)	​	15 (9.3)	​
Sometimes/often/always	147 (10.7)	70 (47.6)	​	66 (54.6)	​	04 (15.4)	​
Normative misconception (n = 1248)	​	​	0.000	​	0.000	​	0.000
Higher than me	427 (34.2)	114 (26.7)	​	105 (34.1)	​	09 (7.6)	​
Similar as much to me	338 (27.1)	74 (21.9)	​	66 (31.7)	​	08 (6.2)	​
Lower than me	38 (3.0)	12 (31.6)	​	09 (40.9)	​	03 (18.8)	​
I do not know	445 (35.7)	14 (3.2)	​	12 (8.6)	​	02 (0.7)	​
Exposure to alcohol ads in the last 12 months	​	​	0.000	​	0.009	​	0.172
Never exposed to ads	1,162 (84.2)	162 (13.9)	​	145 (24.3)	​	17 (3.0)	​
Exposed for ads	218 (15.8)	52 (23.9)	​	47 (35.3)	​	05 (5.9)	​
Physical health/illness characteristics
Known chronic medical illness	​	​	0.186	​	0.277	​	0.321
No	1,228 (89.0)	196 (16.0)	​	175 (26.9)	​	21 (3.6)	​
Yes	152 (11.0)	18 (11.8)	​	17 (21.3)	​	01 (1.4)	​
Medically confirmed hypertensive person	​	​	0.713	​	0.783	​	0.340
No	1,322 (95.8)	206 (15.6)	​	184 (26.4)	​	22 (3.5)	​
Yes	58 (4.2)	08 (13.8)	​	08 (24.2)	​	0 (0.0)	​
High Blood Pressure (HBP)	​	​	0.011	​	0.000	​	0.249
No	1,135 (82.3)	163 (14.4)	​	143 (23.6)	​	20 (3.8)	​
Yes	245 (17.8)	51 (20.8)	​	49 (39.2)	​	02 (1.7)	​
Prescription medications in the last 12 months	​	​	0.209	​	0.302	​	0.715
No	1,230 (89.1)	196 (15.9)	​	176 (26.9)	​	20 (3.5)	​
Yes	150 (10.9)	18 (12.0)	​	16 (21.3)	​	02 (2.7)	​
Average physical activity per day	​	​	0.738	​	0.649	​	0.645
<20 minutes/day	16 (1.2)	2 (12.5)	​	02 (20.0)	​	0 (0.0)	​
≥20 minutes/day	1,364 (98.8)	212 (15.5)	​	190 (26.4)	​	22 (3.4)	​
Number of injuries/admissions in 12 months	​	​	0.524	​	0.065	​	0.586
No injuries/admissions	1,174 (85.1)	179 (15.3)	​	162 (25.2)	​	17 (3.2)	​
≥1 injury/admissions	206 (14.9)	35 (17.0)	​	30 (34.5)	​	05 (4.2)	​
Other substance uses d
Tobacco use in the last three months	​	​	0.000	​	0.000	​	ϕ
No	1,361 (98.6)	200 (14.7)	​	178 (25.0)	​	22 (3.4)	​
Yes	19 (1.4)	14 (73.7)	​	14 (73.7)	​	0 (0.0)	​
Khat use in the last three months	​	​	0.000	​	0.000	​	ϕ
No	1,345 (97.5)	195 (14.5)	​	173 (24.8)	​	22 (3.4)	​
Yes	35 (2.5)	19 (54.3)	​	19 (57.6)	​	0 (0.0)	​
Ever drug injection history	​	​	0.227	​	0.300	​	0.000
No	1,374 (99.6)	212 (15.4)	​	192 (26.4)	​	20 (3.1)	​
Yes	06 (0.4)	02 (33.3)	​	0 (0.0)	​	02 (66.7)	​
Mental health and psychosocial factors
Probable major depressive disorder (MDD)	​	​	0.010	​	0.000	​	0.551
No	1,361 (98.6)	207 (15.2)	​	185 (25.7)	​	22 (3.4)	​
Yes	19 (1.4)	07 (36.8)	​	07 (77.8)	​	0 (0.0)	​
Risk of suicidal behavior	​	​	0.312	​	0.076	​	0.375
No suicidal risk	1,286 (93.2)	196 (15.2)	​	177 (25.6)	​	19 (3.2)	​
Suicidal risk (low to high risk)	94 (6.8)	18 (19.2)	​	15 (38.5)	​	03 (5.5)	​
Probable generalized anxiety disorder (GAD)	​	​	0.092	​	0.015	​	0.395
No	1,340 (97.1)	204 (15.2)	​	182 (25.6)	​	22 (3.5)	​
Yes	40 (2.9)	10 (25.0)	​	10 (50.0)	​	0 (0.0)	​
Social support	​	​	0.281	​	0.126	​	0.523
Poor support	212 (15.4)	30 (14.2)	​	28 (28.6)	​	02 (1.8)	​
Moderate	744 (53.9)	126 (16.9)	​	112 (28.6)	​	14 (4.1)	​
Strong	424 (30.7)	58 (13.7)	​	52 (21.6)	​	06 (3.3)	​
Stressful life events	​	​	​	​	0.000	​	0.471
No stressful events	761 (55.1)	88 (11.6)	0.000	78 (19.7)	​	10 (2.7)	​
One stressful event	313 (22.7)	56 (17.9)	​	49 2(8.7)	​	07 (4.9)	​
Two or more stressful events	306 (22.2)	70 (22.9)	​	65 (39.9)	​	05 (3.5)	​

P-values are derived from Pearson chi-square tests examining whether PAU prevalence differs significantly across categories of each sociodemographic and other variable in the overall sample, among men, and among women independently. Statistical significance was set at p < 0.05. PAU = Problematic Alcohol Use; SD = Standard Deviation.

^aEmployment status: Housewife and student.

^bEmployment status: All actively working and retired.

^CReligion: Muslim, Protestant, Catholic, Jehovah’s Witnesses, or Hawariyat.

^dOther substance uses: Using the ASSIST screening tool.

*Others (Relationships in the family): living alone, a tenant, grandchild, a child-in-law, sibling, cousins, friends, and other relatives.

^ϕFor the women sample, x² for tobacco and khat use is not calculated due to the small sample size.

Prevalence of PAU and Harmful Patterns of Alcohol Use

Among the 1,380 participants, lifetime alcohol use was 94.6% (95% CI: 93.2%–95.7%), and current alcohol use as defined by use in the past 12 months prior to the interview was 90.4% (95% CI: 88.8%–91.9%). Among all participants, 15.5% (95% CI: 13.7%–17.5%) met the criteria for PAU, defined as positive AUDIT score (AUDIT score ≥8). The prevalence was significantly higher among men (26.3%, 95% CI: 23.2%–29.6%) compared to women (3.4%, 95% CI: 2.2%–5.1%) (Table 1).

Hazardous drinking (AUDIT 8-15), harmful drinking (AUDIT 16-19), and possible alcohol dependence (AUDIT ≥20) were reported to be 11.6% (95% CI: 10.0%–13.4%), 1.9% (95% CI: 1.3%–2.8%), and 2.0% (95% CI: 1.4%–2.9%), respectively. Based on item 3 of AUDIT (drinking six or more drinks per occasion), HED was 16.2% (95% CI: 14.4%–18.3%). Participants with AUDIT scores ≥8 were considered screening-positive for PAU and referred for further mental health professional (MHP) evaluation; though, AUDIT is a screening tool and does not replace a formal clinical diagnosis of Alcohol Use Disorder (AUD). The internal consistency of AUDIT in this study (Cronbach’s α) was 0.79.

Early initiation of alcohol consumption was common; 53.3% (95% CI: 50.5%–56.0%) started drinking before 15 years of age. Among individuals who currently drink alcohol, those who believed others drank less than they did (“lower than me” normative misconception) had a higher prevalence of PAU, 31.6% (95% CI: 17.5%–48.7%). Those who were exposed to alcohol advertisements in the last 12 months had a higher prevalence of PAU, 23.9% (95% CI: 18.7%–30.0%), compared to those never exposed, 13.9% (95% CI: 12.1%–16.1%). Concerning other substances, PAU was higher for people with concurrent tobacco use, 73.7% (95% CI: 50.2%–88.6%), and khat use, 54.3% (95% CI: 37.9%–69.8%) (Table 1).

Among those who had ever used alcohol, the most commonly reported reason for drinking was to celebrate social occasions (95.5%), followed by enhancing social situations (47.1%). Regarding preferred drinking places, most participants reported drinking at home (82.1%), followed by social gatherings like ekub (traditional savings associations), religious ceremonies, and feasts (39.1%). Other reported places included public pubs or bars (31.2%), friends’ homes (20.7%), and sellers’ homes (11.1%). Participants reported two or more responses for reasons and places of drinking.

Physical Health, Mental Health, and Psychosocial Factors

Individuals with HBP had a higher prevalence of PAU, 20.8% (95% CI: 16.2%–26.4%) than those without HBP, 14.4% (95% CI: 12.4%–16.5%). PAU prevalence was higher among those who had encountered one or more physical injuries or had admissions in the past 12 months, at 17.0% (95% CI: 12.5%–22.8%), compared to those with no injuries, at 15.2% (95% CI: 13.3%–17.4%). PAU prevalence was higher among individuals with probable major depressive disorder, 36.8% (95% CI: 18.7%–59.7%) and generalized anxiety disorder, 25.0% (95% CI: 14.0%–40.6%) (Table 1).

For participants in the age group ≤24 years, females with PAU have a higher mean AUDIT total score than males with PAU in the same age group. Across most age groups, individuals with PAU had higher mean AUDIT total scores than those without PAU. There are potential differences in AUDIT scores between males and females across age groups, particularly noticeable in the younger cohorts (Figure 1).

Figure 1.

Bar chart illustrating the mean AUDIT total scores stratified by sex and age, comparing individuals with and without PAU in Debre Berhan, Central Ethiopia, 2026

In males, the mean AUDIT total scores in the 25-44 year age group were high. The scores decreased significantly after age 44, with the lowest scores observed in individuals aged 55 and above. In females, higher mean AUDIT total scores were observed in younger age groups, particularly those below 34 years. There were age- and sex-related differences in AUDIT total scores, indicating that younger individuals, particularly males aged 25-44, had higher scores, which decreased with advancing age. Both males and females exhibited a decline in mean AUDIT total scores with increasing age (Figure 2).

Figure 2.

Bar chart displaying the mean AUDIT total scores for males and females across different age groups in Debre Berhan, Central Ethiopia, 2026

Factors Associated With Problematic Alcohol Use (PAU)

In model 1 (overall sample), factors statistically associated with PAU were being male (AOR= 7.83; 95% CI: 4.59–13.36), having an income-generating job (AOR = 2.38; 95% CI: 1.23–4.58) or being unemployed (AOR = 2.73; 95% CI: 1.29–5.78) compared to being in a productive, non-income generating job (housewife and student), a family history of alcohol-related problem (AOR = 4.04; 95% CI: 2.09–7.81), HBP (AOR = 2.04; 95% CI: 1.30–3.21), current smoking (AOR = 5.24; 95% CI: 1.73–15.91), and exposure to one (AOR = 1.60; 95% CI: 1.03–2.47) or ≥2 stressful life events (AOR = 1.65; 95% CI: 1.05–2.61) compared to no events (Table 2).

Table 2.

Multivariable Logistic Regression Analysis With LASSO Regularization for Factors Associated With Problematic Alcohol Use in Debre Berhan, Central Ethiopia, 2026: Overall Sample (n = 1380)

Characteristics	n	PAU Yes (%)	Crude OR (95% CI)	Adjusted OR (95% CI)
Sociodemographic and economic factors
Sex
Male	730	192 (26.3)	10.19 (6.46, 16.07)	7.83 (4.59, 13.36)***
Female	650	22 (3.4)	1.00	1.00
Age: mean±SD (range); 35.55±13.47 (18-95)
Age group (years)
≤24	271	37 (13.7)	​	0.80 (0.41, 1.57)
25-34	521	92 (17.7)	1.27 (0.92, 1.76)	1.08 (0.68, 1.71)
≥35	588	85 (14.5)	1.00	1.00
Marital status
Single	447	94 (21.0)	1.62 (1.20, 2.20)	1.59 (0.98, 2.58)
Married	759	107 (14.1)	1.00	1.00
Divorced	100	10 (10.0)	0.68 (0.34, 1.34)	1.37 (0.60, 3.13)
Widowed/separated/cohabited	74	03 (4.1)	0.26 (0.08, 0.83)	0.37 (0.10, 1.37)
Educational status
No formal education	172	14 (8.1)	1.00	1.00
Only primary education	297	50 (16.8)	2.84 (1.22, 4.27)	1.67 (0.84, 3.31)
Only secondary education	443	73 (16.5)	2.23 (1.22, 4.06)	1.52 (0.77, 3.03)
College/university education	468	77 (16.5)	2.22 (1.22, 4.04)	1.02 (0.50, 2.09)
Employment status
Productive a	382	16 (4.2)	1.00	1.00
Income generating b	860	168 (19.5)	5.55 (3.27, 9.42)	2.38 (1.23, 4.58)*
Unemployed	138	30 (21.7)	6.35 (3.34, 12.09)	2.73 (1.29, 5.78)**
Alcohol use-related characteristics
Family history of alcohol-related problem
No	1,304	182 (14.0)	1.00	1.00
Yes	76	32 (42.1)	4.48 (2.77, 7.26)	4.04 (2.09, 7.81)***
Physical health/illness characteristics
High Blood Pressure (HBP)
No	1,135	163 (14.4)	1.00	1.00
Yes	245	51 (20.8)	1.57 (1.10, 2.23)	2.04 (1.30, 3.21)**
Other substance uses
Tobacco use in the last three months
No	1,361	200 (14.7)	1.00	1.00
Yes	19	14 (73.7)	16.25 (5.79, 45.62)	5.24 (1.73, 15.91)**
Khat use in the last three months
No	1,345	195 (14.5)	1.00	1.00
Yes	35	19 (54.3)	7.00 (3.54, 13.85)	1.63 (0.73, 3.65)
Mental health and psychosocial factors
PHQ-9 total scores (depressive symptoms)	​	​	1.11 (1.05, 1.18)	1.06 (0.96, 1.17)
Risk of suicidal behavior
No suicidal risk	1,286	196 (15.2)	1.00	1.00
Suicidal risk (low to high risk)	94	18 (19.2)	1.32 (0.77, 2.25)	1.23 (0.61, 2.50)
GAD-7 total scores (anxiety symptoms)	​	​	1.07 (1.01, 1.14)	1.00 (0.91, 1.11)
Social support
Poor support	212	30 (14.2)	1.04 (0.65, 1.67)	1.22 (0.69, 2.17)
Moderate	744	126 (16.9)	1.29 (0.92, 1.80)	1.44 (0.97, 2.14)
Strong	424	58 (13.7)	1.00	1.00
Stressful life events
No stressful events	761	88 (11.6)	1.00	1.00
One stressful event	313	56 (17.9)	1.67 (1.16, 2.40)	1.60 (1.03, 2.47)*
Two or more stressful events	306	70 (22.9)	2.27 (1.60, 3.21)	1.65 (1.05, 2.61)*
Disability (WHODAS-12) total scores	​	​	0.99 (0.95, 1.03)	0.97 (0.91, 1.04)
Insomnia Severity Index (ISI) total scores	​	​	1.06 (1.02, 1.10)	1.05 (0.99, 1.12)

OR = Odds Ratio, 95% CI = 95% confidence interval, 1.00 = reference for category.

Statistical significance was set at p < 0.05; * p<.05, ** p<.01, *** p<.001.

p-value of Hosmer-Lemeshow goodness-of-fit test for the model of the ‘overall model sample’ = 0.0987.

Crude odds ratios (CORs) are presented alongside adjusted odds ratios (AORs) to enable assessment of confounding and to facilitate independent verification and critical appraisal of the associations before and after covariate adjustment.

^aEmployment status: Housewife and student.

^bEmployment status: All actively working and retired.

In model 2 (men sample), factors statistically associated with PAU were a family history of alcohol-related problems (AOR = 4.10; 95% CI: 1.90–8.84), HBP (AOR = 2.33; 95% CI: 1.39–3.90), current smoking (AOR = 5.29; 95% CI: 1.66–16.81) and exposure to ≥2 stressful events (AOR = 1.71; 95% CI: 1.04–2.82) compared to no stressful event (Additional File 4).

In model 3 (women sample), factors statistically associated with PAU include being in the age group of 25-34 years (AOR = 8.88; 95% CI: 1.56–50.63) compared to 35 years and above, being divorced (AOR = 5.72 95% CI: 1.27–25.72) compared to being married, a family history of the alcohol-related problem (AOR = 10.98; 95% CI: 1.69–71.14), and exposure to one stressful event (AOR = 4.05 95% CI: 1.14–14.36) compared to no stressful event (Additional File 4).

Discussion

This study estimated the population-level prevalence of PAU and the associated factors in Central Ethiopia. The overall prevalence of PAU was 15.5%. In the overall sample, PAU was significantly associated with being male, employment status, having a family history of problem drinking, hypertension, smoking, and experiencing stressful life events. These factors were commonly identified among men, while for women, younger age, being divorced, and experiencing a single stressful life event were associated with PAU.

In this study, AUDIT-specific classifications of PAU, including hazardous drinking (11.6%), harmful drinking (1.9%), and possible alcohol dependence (2.0%), give notable insights for understanding the spectrum of problem drinking and the importance of targeting interventions at different levels of drinking. ³ Similarly, HED was reported to be 16.2%, comparable to the 2015 Ethiopian STEPS survey, which reported a 12.4% prevalence of HED. ⁶ HED is an important pattern of alcohol consumption because of its relationship with acute alcohol-related harms like injuries, accidents, and violence. ⁵¹ It is also an important predictor for the emergence of PAU and diverse chronic health outcomes. ⁵²

The overall prevalence of PAU in this study aligns with findings from other parts of Ethiopia that used the AUDIT screening tool. A prevalence of 13.9% was reported in the rural Sodo district of Ethiopia. ²⁰ Other studies from Ethiopia have reported higher prevalences: 21.0% among adults ≥18 years old in the same setting using the FAST screening tool, ¹⁸ 30.6% among a predominantly male sample of Dilla town, ⁵³ and 23.7% in a majority of male participants too in South Gondar Zone. ⁵⁴ A 27.5% prevalence of PAU was reported among adults aged ≥60 years in Southwest Ethiopia. ⁵⁵ These variations may be attributed to differences in sample composition, urban-rural distinctions, and age group demographics, as well as the assessment tools employed. Despite numerous studies conducted in Ethiopia, the remarkable differences reported may be influenced by local contexts, including sociocultural norms, gender roles, and religion, which can impact drinking habits. Specifically, sample characteristics that primarily encompass males, urban residents, and older age groups, as observed in particular studies, may contribute to the higher prevalences of PAU reported by previous studies.

There were inconsistent prevalence estimates of PAU in other studies from Africa or LMICs. For instance, the prevalence of PAU in South Africa–a national survey, showed 9.0% ⁵⁶ ; in Suriname, AUDIT-based reports ranged from 5.8% in rural to 6.4% in urban settings, ⁵⁷ in Thailand 12.2%, ⁵⁸ in rural India 6.0%, ⁵⁹ in Bengal, India 34.5%, ⁶⁰ in Nepal 23.8%, ⁶¹ and in Thailand, a suburban setting 24.3%. ⁶² In contrast, PAU prevalence using AUDIT from HICs in New Zealand was reported to be 17.7%. ⁶³ More generally, the global 12-month prevalence of PAU in an adult population from WHO regions has been estimated at nearly 5.1%, with regional differences from 0.8%-8.8%, ³ and the lifetime prevalence of PAU is approximately 8.0%-11.0%. ⁶⁴ Such differences in the prevalence of PAU across studies and regions can be due to inconsistencies in screening tools, sample composition, variation in social or cultural norms, economic disparities, different study methods or designs, temporal and seasonal changes, contextual influences like access to healthcare services, local alcohol-related strategies or policies, and stigma.

Nevertheless, the gender variation in PAU prevalence in our study (26.3% in men vs. 3.4% in women) aligns with existing reports showing excessive alcohol use and higher PAU rates among men. ⁶⁵ The prevalence of PAU across genders also varies notably–male PAU prevalence ranges from 4.1% to 36.0% in studies from rural Uganda, India, and Switzerland, ⁶⁶ while female PAU prevalence was reported as 8.0% in Mozambique ⁶⁷ and 14.3% in Sweden. ⁶⁸ The male sex was significantly associated with PAU, which is consistent with various existing studies from Ethiopia^6,18,20,69 and other studies from both LMICs ⁷⁰ and high-income countries (HICs).^71,72 Global trends consistently highlight that being male is more likely to lead to problem drinking than being female.^3,4 The reported variability of PAU between men and women might be attributed to sociocultural norms across settings, biological differences, and gender-specific alcohol use patterns, strong societal and cultural norms that frequently normalize excessive drinking in men while stigmatizing drinking in women.^65,73 It is worth noting that variations in assessment tools and study settings may have limited the comparability of findings with our study.

This study also identified other factors associated with PAU that are consistent with or different from previous studies. Compared with individuals in productive job status, those who were employed or unemployed was more likely to have PAU. This finding is consistent with previous studies reporting that employment or higher income was associated with PAU in Ethiopia^20,74 and other LMIC studies.^57,70 In contrast, being unemployed or experiencing economic instability was reported to be associated with problem drinking in studies from Ethiopia,^75,76 other LMICs, ⁷⁷ and HICs. ⁷⁸

Study participants who reported alcohol-related problems in the family were more likely to have PAU across all three models, which is consistent with established psychosocial and genetic studies that show a strong association between having problem drinking tendencies and familial factors. ⁷⁹ According to a report from a meta-analysis of 26 studies from Ethiopia, family history of alcohol consumption was reported to be significantly associated with alcohol consumption in adolescents and young people with more than three-fold increased risk. ⁸⁰

Having HBP was associated with PAU, which is in line with studies in China ⁸¹ and other global reports indicating a link between excessive drinking and hypertension. ⁸² This finding is currently important in Ethiopia, where there is an increasing morbidity and mortality from HBP and other noncommunicable diseases. ⁸³

The association of current smoking among men is similar across many studies, including Ethiopia and other LMIC or HIC studies.^6,81 The relationship between experiencing adverse life events and PAU aligns with other global studies that indicate stressful events as an important trigger for risky drinking, commonly in vulnerable groups. ⁸⁴

A key finding in our study is that PAU was associated with exposure to two or more stressful events among men, but with only a single event among women. This indicates that women might be more vulnerable to risky drinking in response to adversity. However, it may not be the amount of stressors alone that matters. Women may be more likely to experience specific adverse events, the severity and timing of which can be problematic, such as sexual violence, which is strongly associated with PAU. Existing studies have also reported inconsistencies in the relationship between PAU and levels of stress across genders. ⁸⁴ These findings suggest that gender-focused interventions should consider not only the number but also the timing and nature of stressors.

Strengths and Limitations

The main strength of this study lies in the use of probability sampling methods, with a rigorous Kish Method for participant selection, drawing the sample from the general population, and including the entire age range of adults aged 18 years and older, regardless of gender, thereby increasing the study’s representativeness in Ethiopia. Furthermore, the 98.6% response rate in this study warrants an excellent internal validity of the findings.

Though this study broadened our understanding of problematic alcohol use, there were some limitations. First, the study’s cross-sectional design cannot establish causal associations. Second, the absence of an established sampling frame in the study setting could introduce selection bias, potentially affecting the reliability, validity, and generalizability of our findings. Third, the self-report nature of data collection could introduce biases, such as social desirability bias or information bias, leading to underestimation or overestimation of drinking behavior. However, evidence shows that there is good reliability in self-reported information about alcohol use behavior. ⁸⁵

Fourth, since the study was a community-based household survey, special populations like military personnel, institutionalized persons in nursing homes or prisons, and homeless individuals were excluded, which limits the generalizability of the findings. Fifth, although AUDIT is widely used in Ethiopia, it has not been validated, and its use could introduce measurement bias or not capture respondents’ reports of drinking accurately and possibly affect the actual prevalence of PAU. Nevertheless, given the good internal consistency of the AUDIT in our study, along with the prior adaptation of its specific items^20,21 and further adaptation in this study, the findings are unlikely to be notably affected. Sixth, the sample size was determined to estimate the prevalence in the overall sample. Thus, post hoc stratified analyses by sex may have reduced statistical power; therefore, findings should be interpreted with caution. Likewise, due to small sample sizes in specific categories (mainly among women), confidence intervals are wide for some associations, suggesting that important findings should be interpreted cautiously.

Findings from this study are most directly generalizable to Ethiopian urban settings sharing comparable socioeconomic, cultural, and ecological characteristics with Debre Berhan, particularly highland cities in the Amhara, Oromia, and Tigray regional states of Ethiopia, where traditional alcoholic beverages are commonly consumed in cultural and social contexts. Future multi-site studies across different Ethiopian settings would further increase the national evidence base and strengthen broader generalizability.

Conclusions

The prevalence of problematic alcohol use (PAU) among adults aged 18 years and older in Central Ethiopia was notably high, with a significant gender disparity, as PAU was remarkably higher in men than women. Male gender, employment status, family history of problem drinking, hypertension, smoking, and stressful life events were significantly associated with PAU.

The findings of our study have important implications for public mental health policy and intervention programming. Gender-targeted interventions are needed that male-focused programs should focus on risky drinking patterns, while women-focused programs should target stressful life events contributing to harmful alcohol drinking. Economic interventions emphasizing on unemployment, family education programs aiming at intergenerational transmission of problem drinking, and routine use of alcohol use screening tools such as AUDIT in community and clinical settings for early detection are strongly recommended, mainly in resource-limited settings like Ethiopia.

Findings of this study may serve as an important reference for developing evidence-based policies and interventions in similar Ethiopian urban settings. Future studies should prioritize longitudinal designs to better identify factors predisposing individuals to PAU and to strengthen the national evidence base.

Finally, as AUDIT is screening instead of a diagnostic tool, the prevalence estimates reported in this study reflect screening-positive PAU rather than clinically confirmed AUD diagnoses. Clinical assessment using standardized diagnostic criteria is advised to confirm the diagnosis of AUD.

Appendix.

List of Abbreviations

AOR

Adjusted Odds Ratio

AUC

Area Under the Curve

AUD

Alcohol Use Disorder

AUDIT

The Alcohol Use Disorder Identification Test

CI

Confidence Interval

GAD

Generalized Anxiety Disorder

HBP

High Blood Pressure

HED

Heavy Episodic Drinking

HH

Household

HICs

High-Income Countries

ISI

Insomnia Severity Index

LASSO

Least Absolute Shrinkage and Selection Operator

LMICs

Low- and Middle- Income Countries

ODK

Open Data Kit

PAU

Problematic Alcohol Use

PHQ-9

Patient Health Questionnaire

SDs

Standard Drinks

WHODAS

World Health Organization Disability Assessment Schedule.

ORCID iD

Kefyalew Dagne https://orcid.org/0000-0001-7272-6351

Ethical Considerations

Ethical clearance was obtained from the Institutional Review Board (IRB) of the College of Health Sciences (CHS), Addis Ababa University (AAU), with Protocol Number (119/20/Psy) on 30 December 2020.

Consent to Participate

Individuals were fully informed about the study’s objectives and procedures before participation, and written informed consent was obtained from all participants before data collection. An information sheet containing the study objectives and procedures was read aloud in Amharic to all participants, and a copy was given to each individual for their records. Participants who could read and write signed the consent form independently. For respondents who could not read or write, a neutral witness was present during the consent process and signed a witness statement verifying that the study and information sheet had been explained to and understood by the participant, and that consent was provided voluntarily. The data collected for this study did not include identifying information of the participant; instead, coded numbers were used to ensure confidentiality.

Data Availability Statement

All relevant materials and data supporting the findings of this study are contained within the manuscript, and the relevant document will be available upon request.*