Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Feb 21.
Published in final edited form as: Nutr Metab Cardiovasc Dis. 2026 Jan 28;36(6):104580. doi: 10.1016/j.numecd.2026.104580

Does Educational Attainment Modify the Relationship Between Alcohol Use and Ischemic Heart Disease Mortality? Evidence from the UK Biobank

Tara Martin 1,2, Laura Llamosas-Falcón 1,6, Yachen Zhu 3, Jürgen Rehm 1,2,4,5,6,7,8, Charlotte Probst 1,4,5,6,*, Carolin Kilian 1,7,9,10,*
PMCID: PMC12922503  NIHMSID: NIHMS2142742  PMID: 41708426

Abstract

Background and Aims:

Alcohol use is associated with ischemic heart disease (IHD) mortality. The significant health burden of IHD is not distributed equally across socioeconomic strata and individuals with low socioeconomic status (SES) may be more vulnerable to the health effects of alcohol use. We aim to explore the effect modification of educational attainment on the association between alcohol use and IHD mortality in the United Kingdom (UK).

Methods and Results:

This cohort study utilized UK Biobank data from 387,914 participants (203,081 females and 184,833 males) aged 40–69 years, who reported being current drinkers at baseline. IHD deaths were identified through national death registries. Individuals with low education had a higher risk of IHD mortality compared to those with high education. Among males, alcohol use of >1 to 30 g/day (without heavy episodic drinking) was associated with 39% lower risk of IHD mortality compared with occasional alcohol use. In the Cox Proportional Hazard model, a differential association was observed among males with low (interaction term HR: 1.29, 95% CI 1.01, 1.66) and medium education (interaction term HR: 1.36, 95% CI:1.03, 1.80), compared to highly educated occasional drinkers. No differential association of alcohol use with IHD mortality by education was observed among females.

Conclusions:

These findings suggest that the association between alcohol use and IHD mortality may vary across educational attainment in males. Future research needs to strengthen our understanding of the risk relationship between alcohol use and IHD mortality, while accounting for potential differences by sex and educational attainment.

Keywords: alcohol use, ischemic heart disease, mortality, educational attainment, socioeconomic status, effect modification, UK Biobank

INTRODUCTION

Ischemic heart disease (IHD) is a leading cause of death in the United Kingdom (UK), with 59,356 deaths (over 10% of all deaths) in England and Wales in 2022.1 There are several factors that contribute to the high IHD burden, with alcohol use being one such factor.2 The association between alcohol use and IHD is complex. The risk relationship has been characterized by a ‘J-shape’, with lifetime abstainers showing a higher risk of IHD compared to women drinking up to 12.5 g and men up to 25 g of pure alcohol per day.3,4 The risk of IHD increases with higher consumption levels, particularly for individuals drinking more than 60 g/day.5 However, this risk is significantly affected by drinking patterns, as heavy episodic drinking (HED) has been shown to increase cardiovascular risk, offsetting protective effects observed at moderate intake levels in the absence of HED.5

The significant health burden of IHD is not distributed equally across socioeconomic strata and individuals with low socioeconomic status (SES) may be more vulnerable to the harmful effects of alcohol use.6 For example, a meta-analysis of retrospective and prospective cohort studies reported that individuals with low education (a key facet of SES)7 bear 1.39 times the risk of cardiovascular (CV) death and 1.50 times the risk of CV events when compared to individuals with high education, based on events observed over the follow-up periods of the included studies.8 Moreover, it has been observed that among participants in Norwegian health surveys, drinking 2–3 times/week was associated with a lower risk of CV disease, especially among those with high SES (as determined by life course household conditions, income, and education).9 However, drinking 4–7 times per week and engaging in HED were both associated with higher risks of CV disease, particularly among individuals with low SES. One study conducted in the United States (US) found a stronger protective association between consuming <20 g/day and IHD mortality among individuals with high SES compared to those with low SES in both sexes, adjusting for risk factors (including smoking status, body mass index [BMI], and physical activity), showing an effect modification of SES in the relationship between alcohol use and IHD mortality.10 While the exact biological mechanism may not be fully understood, the observed differences could be due to various factors, including differences in drinking patterns, access to healthcare, and overall lifestyle behaviours associated with different SES groups.10–13 However, similar effect modification has not been explored in other samples, including the UK.

Given the high incidence of IHD mortality in the UK, it is crucial to delve deeper into the intricate relationship between alcohol consumption and IHD, while exploring how SES may influence this dynamic. Although SES is known to be associated with both alcohol consumption patterns and CV outcomes, few studies have explicitly examined its potential effect modification in the relationship between alcohol use and IHD.13 Based on previous research that found individuals with higher SES (indicated by educational attainment) experiencing more favourable cardiovascular outcomes at similar levels of alcohol use,10 we hypothesize that the inverse association between low alcohol intake and IHD mortality is more pronounced among individuals with high-education individuals compared to those with low education (i.e., multiplicative interaction effect). Furthermore, we hypothesize that the harmful association between heavy alcohol intake and IHD mortality is stronger among individuals with lower education compared to those with higher education. Importantly, we have incorporated HED into our alcohol consumption categories, allowing for a more detailed assessment of drinking patterns beyond average intake levels. Our main analysis focuses exclusively on current drinkers, excluding non-drinkers, to minimize bias from individuals who may declare themselves as non-drinkers but are occasional drinkers. This approach helps ensure the accuracy of our findings by reducing the potential for misclassification. Moreover, the exclusion of non-drinkers in such analyses has been rarely explored in existing research, highlighting the novelty of our study.

METHODS

Data source

The UKB is a large database comprising demographics, lifestyle, and health information from participants recruited between 2006 and 2010 at 22 assessment centres, targeting individuals aged 40 to 69 years, with a participation rate of 5.45%.14,15 Baseline assessments included a self-completed touchscreen questionnaire, interview, and physical measurements. Mental health follow-up assessments, including questions on alcohol use and HED, were self-completed by touchscreen questionnaires in 2016 by 155,264 participants. The prospective design allows for the assessment of a diverse range of exposures, and the large sample size makes it well-suited for studying the independent and combined effects of multiple risk factors on health outcomes.16

Study population

Data from a total of 502,244 participants was available. However, we excluded 1,297 (0.3%) individuals who were lost to follow-up, 82,419 (16.4%) participants who had missing data for one or more key covariates, participants who were below 40 (n = 5, 0.001%) or above 70 years (n = 7, 0.002%) of age at baseline, participants who withdrew from the study (n=94, 0.022%), and participants who reported being non-drinkers (i.e. lifetime abstainers and former drinkers) at baseline (n=30,508, 7.29%). Participants with missing data on covariates were excluded. Descriptive comparison shows that those excluded due to missing data had lower educational attainment (higher proportion with low education and lower proportion with high education) and lower median daily alcohol consumption (Supplementary Materials Table 1) compared to our included sample. The decision to exclude non-drinkers was made to minimize bias due to misclassification. Specifically, we observed that, of those who completed the mental health follow-up questionnaire in 2016, 13,672 participants reported being lifetime abstainers. However, 24.6% (n=3,355) of self-reported lifetime abstainers in 2016 reported being former drinkers at baseline, and 45.2% (n=6,116) reported being current drinkers at baseline. Thus, the final sample in our main analysis was 387,914 participants, of which 203,081 were females and 184,833 were males (Figure 1).

Figure 1.

Figure 1.

Open in a new tab

Study participant’s selection

Exposure assessment

SES was operationalized via educational attainment (denoted education). As early-life education has been shown to significantly delay the onset of poor health, surpassing the influence of income, education was selected as indicator of SES.17,18 Previous research has also identified education as the most relevant factor for alcohol-attributable burden and socioeconomic disparities in mortality risk.19 It was assessed by self-reported highest achieved qualification. We re-categorized the six categories provided by UKB into three groups using the International Standard Classification for Education (ISCED) 2011 coding:20 low education (completed a Certificate of Secondary Education or an equivalent qualification, or less as their highest educational qualification), medium education (completed O levels/General Certificate of Secondary Education or an equivalent, or A levels/AS levels or an equivalent), and high education (reference category, completed a college or university degree, National Vocational Qualification, Higher National Diploma, Higher National Certificate or an equivalent, or another professional qualification such as nursing or teaching).

Alcohol use was defined based on participants self-reported consumption. Non-weekly drinkers were asked about the number of glasses of any alcoholic beverages consumed each month, on average. Participants who reported more frequent alcohol intake were asked to report the number of drinks of each type consumed each week, on average. For each participant, the average grams (g) of pure alcohol consumed daily was calculated by dividing the number of g of alcohol in each alcohol type reported (assuming 8 g per UK standard drink) by 30.4 for monthly reports and by 7 for weekly reports to obtain the average g/day.21 As the association between IHD and alcohol use is decisively impacted by the presence of HED in addition to average consumption, this was accounted for.22 HED was only assessed in a subset of participants (31.6%) at the mental health follow-up in 2016 and was defined as consuming 6 or more units of alcohol (48 g) within a single day at a frequency of monthly or more often. We conducted systematic, descriptive comparisons of complete cases (baseline and mental health follow-up) and missing cases (baseline but without mental health follow-up) to identify the type of missingness, suggesting that data were missing at random (Supplementary Table 2). Using data from both baseline and mental health follow-up, HED at baseline (2006–2010) was estimated by implementing a regression-based multiple imputation model (for detailed approach, see Supplementary Material Methods). Our alcohol use categories for the main analysis were occasional drinkers (past year daily average of (0, 1] g/day pure alcohol, reference), Category I (past year daily average of (1, 30] g/day) without HED, Category I with HED, and Category II (past-year daily average of over 30 g/day). HED was incorporated into the definition of Category I alcohol use, as we anticipated that it would have a major impact on the risk on IHD mortality only at lower levels of alcohol intake.5

Outcome assessment

IHD death or last presumed alive was linked to the UKB using the NHS England death registry for participants in England and Wales, and the NHS Central Register for participants in Scotland. Data related to mortality outcomes were extracted as of November 30th, 2022 (end date). IHD mortality was operationalized using the International Classification of Diseases, 10th revision (ICD-10) codes I20-I25 as the underlying (primary) cause of death.14,16 The time to death, censoring (died of another cause), or last presumed alive was calculated as the difference between the participant’s age at baseline and their age at the respective mortality end date if deceased. For participants still alive at censoring, the difference was determined by subtracting the date of baseline assessment from the end date.

Covariates

Covariates were selected a priori based on the literature: self-reported ethnicity (White [reference category] vs non-White or Mixed participants), smoking status (former smoker, current someday smoker, current every day smoker, or never smoker [reference category]),23,24 BMI (kg/m2) (underweight: <18.5, healthy weight: 18.5–24.99 [reference category], overweight: 25–29.99, and obese: >=30),25 physical activity levels (sedentary: 0 min/week, somewhat active: <150 min/week, and active: >=150 min/week [reference category]),26 and region of assessment centre at recruitment.

Statistical analyses

A longitudinal study design with one baseline assessment, one follow-up assessment, and a mortality follow-up through record linkage were used to perform the survival analyses using Cox proportional hazards (PH) models to examine the effect modification of education on the association between alcohol use and IHD mortality. Age was chosen as the time scale for the survival analyses.27–29 The PH assumption that the ratio of hazard rates between exposure groups remains constant over time was assessed using Schoenfeld residual plots to ensure the validity of the models. Hazard ratios (HRs) were calculated to quantify the associations between alcohol use, education, and IHD mortality, including main effects of alcohol use and education, as well as interactions between alcohol use and education. We first checked the main effects of alcohol use by running a model without the interaction term of alcohol use with education. Then, in a second model, we included the multiplicative interaction term between alcohol use and education to examine potential effect modification of education in the relationship between alcohol use and IHD mortality. All analyses were stratified by sex, based on previous literature showing differences between sexes in the relationship between alcohol use and IHD.2,30 Additionally, a priori analysis exploring a model with a multiplicative interaction term between alcohol use (as continuous) and sex showed statistical significant interaction effects (results not shown).

Sensitivity analyses

To test the robustness of our findings, and following recommendations of a recent systematic review,31 we performed the following sensitivity analyses. First, we employed Fine-Gray sub distribution models to account for competing risks (i.e., causes of death other than IHD), which can influence the likelihood of IHD mortality. For this model, we also adjusted for age as continuous variable.32 Second, the primary analysis was repeated including participants younger than 60 at baseline only to account for potential health effects due to older age. Third, we excluded current everyday smokers to test the effect of smoking on the model. Fourth, we included non-drinkers (i.e., lifetime abstainers and former drinkers as two separate categories), with lifetime abstainers as our reference category and accounting for occasional drinkers within category 1. Finally, we repeated the main analysis including participants lost to follow-up, treating them as censored. All sensitivity analyses included models with and without the interaction term between alcohol use and education.

This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.33 All statistical analyses were performed using R, version 4.4.1.34 We considered statistical significance at a level of two-sided p < 0.05.

RESULTS

Participants characteristics can be found in Table 1. The average follow-up time was 13.4 years (Standard Deviation [SD] = 2.0; females: 13.5 years, SD = 1.7; males: 13.2 years, SD = 2.2). Median daily alcohol consumption was 17.1 g/day in males (interquartile range [IQR] 24) and 8 g/day in females (IQR 14.4). Overall, there were a total of 3,217 IHD deaths during follow-up, with 2,676 deaths occurring in males and 541 in females, with an overall mortality rate of 6.2 (95% CI 6.0 – 6.4) per 10,000 person years (for stratified deaths by education and by sex see Table 2).

Table 1.

Participant characteristics, overall and stratified by sex.

Overall Males Females
Sample Size, n (%) 387,914 184,833 203,081
Follow-up, mean years (SD) 13.4 (2.0) 13.2 (2.2) 13.5 (1.7)
Age at baseline, mean years (SD) 56.3 (8.1) 56.6 (8.2) 55.9 (8.0)
Daily alcohol use (g), median (IQR) 11.4 (19.4) 17.1 (24.0) 8.0 (14.4)
Alcohol use, n (%)
 Occasional drinker 73,867 (19.0) 23,658 (12.8) 50,209 (24.7)
 Category I without HED 200,769 (51.8) 83,849 (45.4) 116,920 (57.6)
 Category I with HED 52,016 (13.4) 30,138 (16.3) 21,878 (10.8)
 Category II 61,262 (15.8) 47,188 (25.5) 14,074 (6.9)
Education, n (%)
 High 182,350 (47.0) 92,105 (49.8) 90,245 (44.4)
 Low 75,758 (19.5) 37,378 (20.2) 38,380 (18.9)
 Medium 129,806 (33.5) 55,350 (29.9) 74,456 (36.7)
Smoking, n (%)
 Never smoker 209,865 (54.1) 90,937 (49.2) 118,928 (58.6)
 Former smoker 138,743 (35.8) 72,011 (39.0) 66,732 (32.9)
 Current some day smoker 11,075 (2.9) 6,571 (3.6) 4,504 (2.2)
 Current everyday smoker 28,231 (7.3) 15,314 (8.3) 12,917 (6.4)
Body Mass Index, n (%)
 Healthy weight 129,850 (33.5) 46,615 (25.2) 83,235 (41.0)
 Underweight 1,834 (0.5) 386 (0.2) 1,448 (0.7)
 Overweight 167,185 (43.1) 92,446 (50.0) 74,739 (36.8)
 Obese 89,045 (23.0) 45,386 (24.6) 43,659 (21.5)
Physical activity, n (%)
 Active 244,610 (63.1) 117,698 (63.7) 126,912 (62.5)
 Sedentary 45,865 (11.8) 21,617 (11.7) 24,248 (11.9)
 Somewhat active 97,439 (25.1) 45,518 (24.6) 51,921 (25.6)
Ethnicity, n (%)
 White 373,400 (96.3) 177,917 (96.3) 195,483 (96.3)
 Non-White or Mixed 14,514 (3.7) 6,916 (3.7) 7,598 (3.7)
Open in a new tab

Abbreviations: SD, standard deviation. IQR, interquartile range, HED, heavy episodic drinking

Table 2.

Number of deaths caused by ischemic heart disease, stratified by sex, by education and by alcohol use categories.

Males Females
Sample size Deaths Sample size Deaths
N Rate, per 10,000 py (95% CI) N Rate, per 10,000 py (95% CI)
Education
Low 37,378 891 18.4 (17.2, 19.6) 38,380 188 3.7 (3.2, 4.2)
Medium 55,350 737 10 (9.3, 10.8) 74,456 174 1.7 (1.5, 2.0)
High 92,105 1,048 8.6 (8, 9.1) 90,245 179 1.5 (1.3, 1.7)
Alcohol use
Occasional drinkers 23,658 478 15.1 (13.8, 16.6) 50,209 175 2.6 (2.2, 3)
Category I with HED 83,849 1,131 10.2 (9.6, 10.9) 116,920 293 1.9 (1.7, 2.1)
Category I without HED 30,138 305 7.6 (6.7, 8.5) 21,878 39 1.3 (0.9, 1.8)
Category II 47,188 762 12.3 (11.4, 13.2) 14,074 34 1.8 (1.2, 2.5)
Open in a new tab

Abbreviations: N, counts. py, person-year. CI, confidence interval. HED, heavy episodic drinking

Among males, our model without interaction term (Table 3) indicated a higher risk of IHD mortality with low education (HR: 1.45, 95% CI 1.32, 1.59) and medium education (HR: 1.13, 95% CI 1.02, 1.24) compared to those with high education. Also, consistently across our alcohol categories, protective association effects with IHD mortality were found compared to occasional drinkers. We observed differential association of drinking more than 1 and up to 30 g per day without HED with IHD mortality by education. Specifically, the protective association of Category I without HED and IHD mortality was attenuated among individuals with low (interaction term: HR: 1.29, 95% CI 1.01, 1.66) and medium education (interaction term: HR: 1.36, 95% CI:1.03, 1.80), compared to occasional drinkers with high education. Among females, in our model without interaction term, we found a higher risk of IHD mortality with low education (HR: 1.36, 95% CI 1.11, 1.68). When including the interaction term in the model, there was no indication of a differential association of alcohol use with IHD mortality by education (Table 3). For example, the interaction term for drinking more than 1 and up to 30 g/day without HED and low education yielded an HR of 1.31 (95% CI 0.84–2.06), and for medium education an HR of 1.37 (95% CI 0.85, 2.20).

Table 3.

Adjusted hazard ratios with 95% CIs for the association of alcohol use on IHD mortality, based on Cox regression models, and the interaction effect between alcohol use and educational attainment on IHD mortality, stratified by sex.

Males Females
Model with interaction term Model without interaction term Model with interaction term Model without interaction term
High Education (Reference) 1.00 1.00 1.00 1.00
Low Education 1.20 (0.97, 1.47) 1.45 (1.32, 1.59) * 1.14 (0.80, 1.63) 1.36 (1.11, 1.68) *
Medium Education 0.89 (0.70, 1.12) 1.13 (1.02, 1.24) * 0.87 (0.59, 1.27) 1.05 (0.85, 1.30)
Occasional drinkers (Reference) 1.00 1.00 1.00 1.00
Category I without HED 0.61 (0.51, 0.73) * 0.73 (0.65, 0.81) * 0.72 (0.51, 1.00) 0.88 (0.73, 1.06)
Category I with HED 0.58 (0.46, 0.73) * 0.69 (0.60, 0.80) * 0.84 (0.47, 1.52) 0.95 (0.67, 1.36)
Category II 0.71 (0.59, 0.85) * 0.80 (0.71, 0.90) * 0.75 (0.41, 1.39) 0.84 (0.58, 1.22)
Interaction: Alcohol use and Education
High education:Occasional drinkers (Reference) 1.00 1.00
Low education:Category I without HED 1.29 (1.01, 1.66) * 1.31 (0.84, 2.06)
Low education:Category I with HED 1.28 (0.91, 1.80) 1.50 (0.67, 3.38)
Low education:Category II 1.19 (0.92, 1.56) 0.90 (0.34, 2.42)
Medium education:Category I without HED 1.36 (1.03, 1.80) * 1.37 (0.85, 2.20)
Medium education:Category I with HED 1.42 (0.99, 2.03) 0.91 (0.37, 2.24)
Medium education:Category II 1.24 (0.93, 1.67) 1.37 (0.59, 3.20)
Open in a new tab

Abbreviations: CI, confidence interval. IHD, ischemic heart disease. HED, heavy episodic drinking.

Models adjusted for smoking status, BMI, physical activity, ethnicity and assessment centre region

*

Indicates a p-value of < 0.05.

Fine-Gray sub-distribution models accounting for competing risks confirmed the results of the main analyses (Supplementary Table 3). We found a similar effects than the ones observed in our main model between drinking more than 1 and up to 30 g per day without HED and low (interaction HR: 1.28, 95% CI 1.00, 1.64) and medium (interaction HR: 1.35, 95% CI 1.03, 1.79) education among males. When restricting the sample to participants younger than 60 years, we identified differential associations among women between drinking more than 1 and up to 30 g per day without HED and low education (HR 2.84, 95% CI 1.01, 7.99; Supplementary Table 4), similar to the effect observed among males in the main analysis. However, among males, the effects were weaker when excluding those aged 60 years and older (except for Category II alcohol use and medium education, HR: 1.64, 95% CI 1.01, 2.65). When excluding current everyday smokers from our sample, differential associations were found among men with medium education and drinking more than 1 and up to 30 g per day with and without HED (Supplementary Table 5). In these models, no differential association of alcohol use with IHD mortality by education was observed among females. When including non-drinkers in our sample and using lifetime abstainers as the reference category (Supplementary Table 6), in the model without interaction term, we found a similar higher risk of IHD mortality among males with low and medium education compared to high education. A higher risk among former drinkers compared to lifetime abstainers was found (HR: 1.40, 95% CI 1.07, 1.82), as well as a protective association between drinking more than 0 and up to 30 g per day with HED (HR: 0.78, 95% CI 0.61, 0.99) compared to lifetime abstainers. Among females, a higher risk of IHD mortality was found in individuals with low education compared to high education. A protective association was found among females who drink more than 0 and up to 30 g per day without HED and who drink more than 30 g per day compared to lifetime abstainers. Using lifetime abstainers as reference category, no differential association of alcohol use with IHD mortality by education was observed for both males and females. Finally, including participants lost to follow-up as censored (n = 1,003) yielded results consistent with the main analysis (Supplementary Table 7).

DISCUSSION

In our analysis, we found a greater protective association between drinking more than 1 and up to 30 g per day (vs. occasional drinking) and IHD mortality in the high-education group compared with the low- and medium-education group among males, suggesting an effect modification of educational attainment in the alcohol-IHD mortality association. The corresponding test among females did not indicate such differential associations.

Consistent with our main analysis, previous survey-based research has shown similar effects of alcohol consumption on IHD mortality.9,13 When re-evaluating the alcohol consumption and IHD risk relationship using burden of proof methodology, Carr et al.35 found an inverse association between average alcohol intake and IHD risk in conventional observational studies, aligning with prior meta-analyses. However, this conflicts with studies using methods genetically predicting alcohol consumption, which do not show any significant association with IHD risk. Findings from a Mendelian randomization study using the UKB found that human genetic data indicate causal associations between any level of alcohol intake and increased risk of hypertension and coronary artery disease.36 The authors suggest that the cardioprotective effects of light to moderate alcohol consumption found in observational studies may be influenced by confounding lifestyle factors. However, there is still a comprehensive body of research showing potential protective effects of low to moderate alcohol consumption, including on all-cause mortality,37–39 and there continues to be a scientific debate on the robustness and potential explanations of these findings.40–42 In contrast, the protective effect found in the highest alcohol category in males is not consistent with the literature, as several meta-analyses (e.g., Roerecke & Rehm5) as well as biological reasoning (e.g. Roerecke et al.3) suggest higher IHD mortality risks. To the best of our knowledge, there are no biological pathways that can explain this finding. Therefore, this observation may arise from sample idiosyncrasies, or methodological limitations of our study.31

The differential associations found among males between low alcohol consumption without HED and low and medium education indicate a complex interplay between SES, alcohol use, and IHD mortality. This aligns with the results found by Zhu et al. in a US sample, where they identified an interaction between low education and drinking more than 0 and up to 20 g per day in both men and women compared to lifetime abstention.13 However, our study differs in that we used occasional drinkers as the reference category to minimize bias from misclassification. Despite this difference in reference categories, our findings remain robust across several sensitivity analyses. Our results therefore suggest that protective associations of low alcohol consumption against IHD mortality may be reduced among males with low and medium SES, indicated by educational attainment. Various pathways may contribute to the observed effect modification: For example, high SES individuals tend to access healthcare at earlier stages of disease progression and to achieve better treatment outcomes,43 resulting in fewer comorbidities that may interact negatively with alcohol use.44 There are also differences in alcohol use patterns, drinking contexts, and beverage preferences, with more irregular heavy and spirits drinking among individuals with low and medium SES,45–47 and other health behaviours, such as smoking, that may diminish protective effects of low alcohol intake.37,47–49

The absence of a similar differential associations among females may be better explained by limited statistical power, as the number of IHD deaths in women was substantially lower and the corresponding confidence intervals were wider. However, there are also known sex differences that may have influenced this observation. For example, irregular heavy and spirits drinking are generally lower among women, resulting in lower variations across socioeconomic groups in women compared to men.50 More sex-specific research with larger sample sizes is needed to increase precision of potential effect modification effects.

There are several limitations that should be discussed. While the UKB provides comprehensive individual-level data on various risk factors and outcomes, it is important to note that the sample lacks representativeness of the general adult UK population.14 As the UKB only recruited participants aged 40 to 69 years at baseline, our findings may not be generalizable to younger or older populations with different alcohol metabolism and cardiovascular risk profiles. The cohort consists of underrepresented important groups. Specifically, UKB participants tend to reside in more affluent areas and demonstrate lower rates of obesity, smoking, and daily alcohol consumption compared to the general population, in addition to reporting fewer health issues. Notably, all-cause mortality among UKB participants is about half that of the overall UK population.15 The protective effect of alcohol use on IHD mortality may have been observed due to the potential underrepresentation of alcohol users in the analyzed sample. The protective effect could also be attributed to the healthier behavioral and lifestyle risk factor distributions among UKB participants. Reverse-causality bias cannot be entirely excluded, as individuals with pre-existing health conditions may change their drinking behavior, potentially influencing observed associations.51 However, using occasional drinkers as the reference group may have reduced this risk, since those who abstain due to illness were less likely to be included in that category. Moreover, average daily alcohol consumption was assessed by self-report. This may have led to additional bias due to social desirability and memory bias, misclassification, and the changing nature of drinking patterns. HED status was estimated for a large number of participants (68.4%) based on follow-up data collected 6–10 years post-baseline, introducing a risk of misclassification. While our estimation model was designed to minimize bias, the delayed measurement and incomplete HED data remain important limitations. Another limitation is a high degree of missingness in the physical activity variable, with data missing for 13.6% (n=65,985) from the total final sample, leading to its exclusion in the final analysis. Although we adjusted for several major confounders (e.g., smoking, BMI, and physical activity), other relevant variables such as family history of CVD, diet, lipid levels, medication use, and a history of other chronic conditions were not included. Further investigation is required to elucidate mechanisms through which alcohol use impact IHD mortality and how these may differ by educational attainment.

Despite our analysis suggesting a protective association of alcohol use with IHD mortality, potential benefits are outweighed by well-documented negative health effects of alcohol consumption on conditions including cancers, non-communicable diseases, and more.52,53 As such, the potential effects of alcohol on IHD mortality must be weighed against its well-established harms, as no level of consumption is without risk and the damage increases with the amount consumed.54 Given our findings that males with low educational attainment face higher IHD mortality risk, future research is needed to elucidate the mechanisms by which alcohol influences cardiovascular health in socioeconomically disadvantaged populations, and to investigate the factors underlying the observed sex differences.

Supplementary Material

Supplementary Materials

HIGHLIGHTS.

  • Moderate alcohol use without HED was associated with a lower risk of IHD mortality in males.

  • The association between alcohol and IHD mortality differed by education in males.

  • No such differential association was observed in females.

Acknowledgements

This research has been conducted using the UK Biobank Resource under Application Number 103421.

Funding source

Research reported in this publication was supported by the National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award Number R01AA028009. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

Declaration of interest

Authors declare no conflict of interest

Data statement

Access to the UK Biobank data is restricted to bona fide researchers who must apply and be approved to use the resource for health-related research that is in the public interest. Due to these restrictions, the research data is confidential and cannot be made publicly available. Researchers interested in accessing the UK Biobank data can apply through the UK Biobank’s access procedures.

REFERENCES

  • 1.Office for National Statistics. Deaths registered in England and Wales: 2022. Published online 2023. https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsregistrationsummarytables/2022
  • 2.Zhao J, Stockwell T, Roemer A, Naimi T, Chikritzhs T. Alcohol Consumption and Mortality From Coronary Heart Disease: An Updated Meta-Analysis of Cohort Studies. J Stud Alcohol Drugs. 2017;78(3):375–386. doi: 10.15288/jsad.2017.78.375 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Roerecke M Alcohol’s Impact on the Cardiovascular System. Nutrients. 2021;13(10):3419. doi: 10.3390/nu13103419 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Ronksley PE, Brien SE, Turner BJ, Mukamal KJ, Ghali WA. Association of alcohol consumption with selected cardiovascular disease outcomes: a systematic review and meta-analysis. BMJ. 2011;342(feb22 1):d671–d671. doi: 10.1136/bmj.d671 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Roerecke M, Rehm J. Alcohol consumption, drinking patterns, and ischemic heart disease: a narrative review of meta-analyses and a systematic review and meta-analysis of the impact of heavy drinking occasions on risk for moderate drinkers. BMC Med. 2014;12(1). doi: 10.1186/s12916-014-0182-6 [DOI] [Google Scholar]
  • 6.Diderichsen F, Hallqvist J, Whitehead M. Differential vulnerability and susceptibility: how to make use of recent development in our understanding of mediation and interaction to tackle health inequalities. Int J Epidemiol. 2019;48(1):268–274. doi: 10.1093/ije/dyy167 [DOI] [PubMed] [Google Scholar]
  • 7.Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 1). J Epidemiol Community Health. 2006;60(1):7–12. doi: 10.1136/jech.2004.023531 [DOI] [Google Scholar]
  • 8.Khaing W, Vallibhakara SA, Attia J, McEvoy M, Thakkinstian A. Effects of education and income on cardiovascular outcomes: A systematic review and meta-analysis. Eur J Prev Cardiol. 2017;24(10):1032–1042. doi: 10.1177/2047487317705916 [DOI] [PubMed] [Google Scholar]
  • 9.Degerud E, Ariansen I, Ystrom E, et al. Life course socioeconomic position, alcohol drinking patterns in midlife, and cardiovascular mortality: Analysis of Norwegian population-based health surveys. Rehm J, ed. PLOS Med. 2018;15(1):e1002476. doi: 10.1371/journal.pmed.1002476 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Zhu Y, Llamosas-Falcón L, Kerr WC, Rehm J, Probst C. Behavioral risk factors and socioeconomic inequalities in ischemic heart disease mortality in the United States: A causal mediation analysis using record linkage data. PLOS Med. 2024;21(9):e1004455. doi: 10.1371/journal.pmed.1004455 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Norström T, Landberg J. The association between population drinking and ischemic heart disease mortality in educational groups. Alcohol Alcohol. 2023;58(4):385–392. doi: 10.1093/alcalc/agad033 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Propper C Socio-economic inequality in the distribution of health care in the UK. Oxf Open Econ. 2024;3(Supplement_1):i577–i581. doi: 10.1093/ooec/odad090 [DOI] [Google Scholar]
  • 13.Zhu Y, Llamosas-Falcón L, Kerr W, Puka K, Probst C. Differential Associations of Alcohol Use With Ischemic Heart Disease Mortality by Socioeconomic Status in the US, 1997–2018. JAMA Netw Open. 2024;7(2):e2354270. doi: 10.1001/jamanetworkopen.2023.54270 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Sudlow C, Gallacher J, Allen N, et al. UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old Age. PLOS Med. 2015;12(3):e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Fry A, Littlejohns TJ, Sudlow C, et al. Comparison of Sociodemographic and Health-Related Characteristics of UK Biobank Participants With Those of the General Population. Am J Epidemiol. 2017;186(9):1026–1034. doi: 10.1093/aje/kwx246 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Allen NE, Lacey B, Lawlor DA, et al. Prospective study design and data analysis in UK Biobank. Sci Transl Med. 2024;16(729). doi: 10.1126/scitranslmed.adf4428 [DOI] [Google Scholar]
  • 17.Herd P, Goesling B, House JS. Socioeconomic Position and Health: The Differential Effects of Education versus Income on the Onset versus Progression of Health Problems. J Health Soc Behav. 2007;48(3):223–238. doi: 10.1177/002214650704800302 [DOI] [PubMed] [Google Scholar]
  • 18.Spiers GF, Liddle JE, Stow D, et al. Measuring older people’s socioeconomic position: a scoping review of studies of self-rated health, health service and social care use. J Epidemiol Community Health. 2022;76(6):572–579. doi: 10.1136/jech-2021-218265 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Probst C, Lange S, Kilian C, Saul C, Rehm J. The dose-response relationship between socioeconomic deprivation and alcohol-attributable mortality risk—a systematic review and meta-analysis. BMC Med. 2021;19(1):268. doi: 10.1186/s12916-021-02132-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.OECD, Institute for Statistics, Europäische Kommission, UNESCO, eds. ISCED 2011 Operational Manual: Guidelines for Classifying National Education Programmes and Related Qualifications. OECD; 2015. [Google Scholar]
  • 21.National Health Service. Alcohol units. https://www.nhs.uk/live-well/alcohol-advice/calculating-alcohol-units/
  • 22.Jones L, Bates G, McCoy E, Bellis MA. Relationship between alcohol-attributable disease and socioeconomic status, and the role of alcohol consumption in this relationship: a systematic review and meta-analysis. BMC Public Health. 2015;15(1):400. doi: 10.1186/s12889-015-1720-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Bobo JK, Husten C. Sociocultural influences on smoking and drinking. Alcohol Res Health J Natl Inst Alcohol Abuse Alcohol. 2000;24(4):225–232. [Google Scholar]
  • 24.Hbejan K Smoking effect on ischemic heart disease in young patients. Heart Views. 2011;12(1):1. doi: 10.4103/1995-705x.81547 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.British Health Intelligence Team. BHF UK CVD Factsheet. Published online 2025. https://www.bhf.org.uk/-/media/files/for-professionals/research/heart-statistics/bhf-cvd-statistics-uk-factsheet.pdf
  • 26.Powell L, Feng Y, Duncan MJ, Hamer M, Stamatakis E. Does a physically active lifestyle attenuate the association between alcohol consumption and mortality risk? Findings from the UK biobank. Prev Med. 2020;130:105901. doi: 10.1016/j.ypmed.2019.105901 [DOI] [PubMed] [Google Scholar]
  • 27.Canchola AJ, Stewart SL, Bernstein L, et al. Cox Regression Using Different Time Scales. In: 2003. https://api.semanticscholar.org/CorpusID:13154158 [Google Scholar]
  • 28.Thiébaut ACM, Bénichou J. Choice of time‐scale in Cox’s model analysis of epidemiologic cohort data: a simulation study. Stat Med. 2004;23(24):3803–3820. doi: 10.1002/sim.2098 [DOI] [PubMed] [Google Scholar]
  • 29.Therneau TM, Grambsch PM. Modeling Survival Data : Extending the Cox Model. Springer; 2000. [Google Scholar]
  • 30.Roerecke M, Rehm J. The cardioprotective association of average alcohol consumption and ischaemic heart disease: a systematic review and meta‐analysis. Addiction. 2012;107(7):1246–1260. doi: 10.1111/j.1360-0443.2012.03780.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Stockwell T, Zhao J, Clay J, et al. Why do only some cohort studies find health benefits from low volume alcohol use? A systematic review and meta-analysis of study characteristics that may bias mortality risk estimates. J Stud Alcohol Drugs. Published online January 30, 2024. doi: 10.15288/jsad.23-00283 [DOI] [Google Scholar]
  • 32.Fine JP, Gray RJ. A Proportional Hazards Model for the Subdistribution of a Competing Risk. J Am Stat Assoc. 1999;94(446):496–509. doi: 10.1080/01621459.1999.10474144 [DOI] [Google Scholar]
  • 33.Von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
  • 34.R Core Team. A language and environment for statistical computing. R Foundation for Statistical Computing. Version 4.4.1. https://www.R-project.org/ [Google Scholar]
  • 35.Carr S, Bryazka D, McLaughlin SA, et al. A burden of proof study on alcohol consumption and ischemic heart disease. Nat Commun. 2024;15(1). doi: 10.1038/s41467-024-47632-7 [DOI] [Google Scholar]
  • 36.Biddinger KJ, Emdin CA, Haas ME, et al. Association of Habitual Alcohol Intake With Risk of Cardiovascular Disease. JAMA Netw Open. 2022;5(3):e223849. doi: 10.1001/jamanetworkopen.2022.3849 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Di Castelnuovo A, Bonaccio M, Costanzo S, et al. Drinking alcohol in moderation is associated with lower rate of all-cause mortality in individuals with higher rather than lower educational level: findings from the MORGAM project. Eur J Epidemiol. 2023;38(8):869–881. doi: 10.1007/s10654-023-01022-3 [DOI] [PubMed] [Google Scholar]
  • 38.Bassett JK, Peng Y, MacInnis RJ, et al. Alcohol consumption trajectories over the life course and all-cause and disease-specific mortality: the Melbourne Collaborative Cohort Study. Int J Epidemiol. 2025;54(2):dyaf022. doi: 10.1093/ije/dyaf022 [DOI] [PubMed] [Google Scholar]
  • 39.Xi B, Veeranki SP, Zhao M, Ma C, Yan Y, Mi J. Relationship of Alcohol Consumption to All-Cause, Cardiovascular, and Cancer-Related Mortality in U.S. Adults. J Am Coll Cardiol. 2017;70(8):913–922. doi: 10.1016/j.jacc.2017.06.054 [DOI] [PubMed] [Google Scholar]
  • 40.Stockwell T, Zhao J, Panwar S, Roemer A, Naimi T, Chikritzhs T. Do “moderate” drinkers have reduced mortality risk? A systematic review and meta-analysis of alcohol consumption and all-cause mortality. J Stud Alcohol Drugs. 2016;77(2):185–198. doi: 10.15288/jsad.2016.77.185 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Britton A, Bell S. The protective effects of moderate drinking: lies, damned lies, and… selection biases?: commentaries. Addiction. 2017;112(2):218–219. doi: 10.1111/add.13585 [DOI] [PubMed] [Google Scholar]
  • 42.Visontay R, Mewton L, Sunderland M, Chapman C, Slade T. Is low‐level alcohol consumption really health‐protective? A critical review of approaches to promote causal inference and recent applications. Alcohol Clin Exp Res. Published online April 21, 2024:acer.15299. doi: 10.1111/acer.15299 [DOI] [Google Scholar]
  • 43.Propper C Socio-economic inequality in the distribution of health care in the UK. Oxf Open Econ. 2024;3(Supplement_1):i577–i581. doi: 10.1093/ooec/odad090 [DOI] [Google Scholar]
  • 44.Stringhini S, Carmeli C, Jokela M, et al. Socioeconomic status and the 25 × 25 risk factors as determinants of premature mortality: a multicohort study and meta-analysis of 1·7 million men and women. The Lancet. 2017;389(10075):1229–1237. doi: 10.1016/S0140-6736(16)32380-7 [DOI] [Google Scholar]
  • 45.Degerud E, Ariansen I, Ystrom E, et al. Life course socioeconomic position, alcohol drinking patterns in midlife, and cardiovascular mortality: Analysis of Norwegian population-based health surveys. Rehm J, ed. PLOS Med. 2018;15(1):e1002476. doi: 10.1371/journal.pmed.1002476 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Beard E, Brown J, West R, Kaner E, Meier P, Michie S. Associations between socio-economic factors and alcohol consumption: A population survey of adults in England. Palazón-Bru A, ed. PLOS ONE. 2019;14(2):e0209442. doi: 10.1371/journal.pone.0209442 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Boyd J, Hayes K, Green D, Angus C, Holmes J. The contribution of health behaviour to socioeconomic inequalities in alcohol harm: Analysis of the UK biobank, a large cohort study with linked health outcomes. SSM - Popul Health. 2023;23:101443. doi: 10.1016/j.ssmph.2023.101443 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Rehm J, Hasan OSM. Is burden of disease differentially linked to spirits? A systematic scoping review and implications for alcohol policy. Alcohol. 2020;82:1–10. doi: 10.1016/j.alcohol.2019.06.005 [DOI] [PubMed] [Google Scholar]
  • 49.Jani BD, McQueenie R, Nicholl BI, et al. Association between patterns of alcohol consumption (beverage type, frequency and consumption with food) and risk of adverse health outcomes: a prospective cohort study. BMC Med. 2021;19(8):1–14. doi: 10.1186/s12916-020-01878-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Kilian C, Manthey J, Moskalewicz J, et al. The socioeconomic profile of alcohol use in Europe: findings from a cross-sectional survey of 33 European countries. Int J Alcohol Drug Res. 2023;11(1):13–23. doi: 10.7895/ijadr.0407 [DOI] [Google Scholar]
  • 51.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]
  • 52.Rehm J The risks associated with alcohol use and alcoholism. Alcohol Res Health J Natl Inst Alcohol Abuse Alcohol. 2011;34(2):135–143. [Google Scholar]
  • 53.Rehm J, Gmel GE, Gmel G, et al. The relationship between different dimensions of alcohol use and the burden of disease—an update. Addiction. 2017;112(6):968–1001. doi: 10.1111/add.13757 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Anderson BO, Berdzuli N, Ilbawi A, et al. Health and cancer risks associated with low levels of alcohol consumption. Lancet Public Health. 2023;8(1):e6–e7. doi: 10.1016/s2468-2667(22)00317-6 [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

Supplementary Materials
NIHMS2142742-supplement-Supplementary_Materials.docx (48.3KB, docx)

Data Availability Statement

Access to the UK Biobank data is restricted to bona fide researchers who must apply and be approved to use the resource for health-related research that is in the public interest. Due to these restrictions, the research data is confidential and cannot be made publicly available. Researchers interested in accessing the UK Biobank data can apply through the UK Biobank’s access procedures.

RESOURCES