Abstract
Background and aims
Internationally, rates of harm from alcohol tend to be higher in lower socio‐economic groups, even while drinking is lower. This is known as the alcohol harm paradox. There are very little Australian data published on socio‐economic disparities in alcohol consumption and harm, and none that has examined changes over time. This paper aimed to describe trends in socio‐economic inequalities in key measures of alcohol consumption and alcohol‐related harm over 21 years in Victoria, Australia.
Design
Trend analysis of population rates of separate data on hospital, emergency department and drinking behaviour.
Setting
Victoria, Australia, between 2000 and 2020.
Participants/cases
Survey data from 37 422 respondents plus 841 792 hospital admissions and 591 824 emergency department presentations.
Measurements
Socio‐economic status was measured using an area‐based index based on postcode of residence, divided into quintiles. Two measures of drinking were assessed based on survey responses: annual volume of drinking and frequency of risky (50 g or more) drinking occasions. Chronic harms were measured via hospital admissions for alcohol‐related liver disease and acute harms via emergency department presentations for alcohol‐related disorders. Differences in drinking and harm rates across quintiles were assessed using negative binomial regression, with interactions to examine whether the social gradients changed over time.
Findings
For men, there were no statistically significant differences in either total volume of drinking or frequency of episodic risky drinking between socio‐economic quintiles. For women, volume of drinking was generally higher for those living in more advantaged neighbourhoods [e.g. Incident Rate Ratio (IRR) = 1.60, 95% confidence interval (CI) = 1.32–1.95 for women in the most advantaged compared with most disadvantaged], while frequency of episodic risky drinking did not differ statistically significantly. Trends in drinking on either measure did not differ by socio‐economic status for men or women. Alcohol‐related harms were higher for people living in disadvantaged neighbourhoods for most outcomes and sub‐groups analysed. For example, male rates of alcohol‐related liver disease were nearly twice as high in the most disadvantaged quintile as in the least disadvantaged quintile (IRR = 0.54, CI = 0.50–0.58). On some measures there was evidence that the gap between socio‐economic groups had widened over time.
Conclusions
Despite similar or lower levels of alcohol consumption, people living in more disadvantaged socio‐economic areas of Victoria, Australia, appear to experience much higher rates of alcohol‐related harm than those in more advantaged areas, with some disparities widening over time.
Keywords: alcohol, alcohol consumption, alcohol harm paradox, Australia, socio‐economic status, trends
INTRODUCTION
Alcohol's impact on population health is well‐established, with over 5% of the global burden and 4.5% of the Australian burden of disease attributable to alcohol [1, 2]. It is also increasingly clear internationally that this burden is not distributed equally across society. Systematic reviews of alcohol‐related harms show that rates for people in low socio‐economic groups are consistently higher than those in high groups, often markedly [3, 4, 5, 6]. For example, Probst et al. [3] show that men in the lowest 20% of the income distribution have relative risks of alcohol‐related mortality of 4 to 6 compared with the highest income men. These inequalities help to drive unequal outcomes in the total population For example, a study of 17 European countries estimated that alcohol was responsible for up to10% of the socio‐economic inequality in mortality [7], while a recent United Kingdom study found alcohol‐specific causes were responsible for between 3% and 5% of socio‐economic inequalities in life expectancy [8]. The role of alcohol‐related morbidity and mortality in health inequalities is particularly striking because, unlike many other health risk factors, there is little evidence that they are explained by socio‐economic disparities in use [6]. A systematic review of studies that incorporated measures of both drinking and mortality found that average drinking levels explained less than 10% of the inequalities in mortality rates, and even measures incorporating drinking pattern explained only approximately one‐quarter [5].
This disconnect: that socio‐economically disadvantaged groups experience greater harm from alcohol despite not drinking more alcohol, has been dubbed the alcohol harm paradox [9] and has been the focus of significant research [10], largely in Europe. In a systematic review Boyd et al. [10] found 79 articles that demonstrated the paradox, reporting rates of harm higher for disadvantaged populations than would be expected based on consumption measures and noted that causal mechanisms for the paradox remained relatively opaque. Despite the significant amount of work on this topic internationally, there has been very little Australian research into alcohol's contribution to health inequalities. The Boyd et al. [10] review identifies only four empirical studies from Australia [11, 12, 13, 14], only one of which (from 1999) relies on population measures of harm [13] and none analysing change over time. In a ground‐breaking early Australian study, Najman et al. [15] showed that between 1981 and 2002, blue collar workers died from liver cirrhosis at over twice the rate of white collar workers and that the mortality gap had widened over the study period. In a cross‐sectional aggregated study, Dietze and colleagues [16] showed that Australian Local Government Areas (LGAs) with higher income inequality but not greater overall disadvantage experienced higher rates of alcohol‐related harm. Since these early articles, there has been very little Australian research on the topic beyond standard risk factor analyses in the Australian Burden of Disease study [1], which showed that rates of disability adjusted life years (DALYs) lost because of alcohol were nearly twice as high in the most disadvantaged fifth of the population compared to the least disadvantaged group. Australia provides a useful comparison point to the predominantly European literature, with a distinctive drinking culture forged from successive waves of migration [17] with relatively high consumption and disease burden [1, 18]. Given broader health disparities identified in Australia [19] and the striking consistency of cross‐national studies of the alcohol harm paradox [10], we expect to find disparities in harm here that are not easily explained by consumption patterns.
Similarly, data on Australian drinking levels by socio‐economic status are relatively sparse. National survey data shows that rates of risky drinking, both long‐ and short‐term, are lower among unemployed than employed Australians and lowest for people living in the most disadvantaged quintile of Australian neighbourhoods [20]. These findings have been replicated in other Australian studies among older people [21] and young adults [22], although some studies have reported less consistent associations [23]. This is consistent with a broad international literature highlighting higher levels of drinking and risky‐drinking among socio‐economically advantaged populations [6], although it is worth noting that rates of heavy episodic drinking have been found to be higher in low socio‐economic status groups in some settings [24, 25].
There remain significant gaps in our knowledge of alcohol's role in health inequalities in Australia, including whether and how it varies by age and sex, or by type of health outcome and whether inequalities are widening or narrowing. Alcohol consumption in Australia has changed markedly in the past two decades, with per‐capita consumption increasing between 2000 and 2008 before a subsequent decline [18]. There have been marked variations in drinking by age group too, with notable declines in risky drinking for teenagers and young adults [26]. Over the same time period, health inequalities in Australia overall appear to have grown [27], although there remains very limited systematic reporting of data to track inequalities [28].
In this study, we use 20 years of survey and health system data from the state of Victoria to examine socio‐economic inequalities in alcohol consumption and in chronic (hospital admissions alcohol‐related liver disease) and acute (emergency presentations for alcohol intoxication) alcohol‐related harms. We assess how these socio‐economic disparities have changed over time.
METHODS
This study examines trends in two repeated cross‐sectional measures of alcohol consumption and two repeated cross‐sectional measures of alcohol‐related harm by socio‐economic status in Victoria between the year 2000 and 2020.
Data sources
Survey data come from the National Drug Strategy Household Survey (NDSHS), conducted by the Australian Institute of Health and Welfare every 3 years. The NDSHS uses stratified cluster samples to draw a representative national sample (excluding households where English is not spoken and people in institutional accommodation), with in‐person recruitment and a ‘drop and collect’ survey form, which is completed by the respondent in their own time and then collected by the data collection agency at a subsequent visit. Other modes have been used alongside ‘drop and collect’ across the waves analysed (in‐person, telephone and online), but sampling mode differences are low on key measures [29] and NDSHS estimates of consumption have been shown to track objective measures of population consumption relatively well [30]. We use data from seven waves (2001–2019) and limit our analyses to the Victorian sample for comparability with the harm measures. This leaves a total sample of 37 422 respondents, varying between 4734 (in 2007) and 6130 (in 2004) per year. Response rates for the NDSHS varied between 46% and 51% with no clear trend between 2001 and 2019.
Harms data were provided by the Victorian Agency for Health Information. We collated annual counts of hospital admissions from the Victorian Admitted Episode Database (VAED), which captures key information on all hospital admissions in Victoria across both public and private hospitals [31]. Annual counts of emergency department presentations were extracted from the Victorian Emergency Minimum Dataset (VEMD) [32], which includes data on all presentations to Victorian public hospitals with designated emergency departments (private hospital emergency departments represent a very small proportion of emergency presentations in Australia) [33].
Measures
Socio‐economic status
Socio‐economic status was operationalised using the postcode of residence reported by the survey respondents (for consumption) and presenting patients (for harms). Postcodes were assigned a quintile of socio‐economic disadvantage according to the Index of Relative Socio‐Economic Disadvantage (IRSD) from the Australian Bureau of Statistics' (ABS) Socio‐Economic Index for Areas (SEIFA). We used the SEIFA indices developed from the 2011 Census, as this represented the mid‐point of the study period and the ABS explicitly notes that combining indices from different census years is problematic [34]. Therefore, we assume that the relative socio‐economic disadvantage of areas has not changed between 2000 and 2020, which smooths over complex patterns of urban renewal, gentrification and development [35]. The SEIFA quintiles break the Australian population up into five roughly equal groups, from the most socio‐economically disadvantaged (quintile 1) to the least disadvantaged (quintile 5).
Alcohol consumption
We examined two different measures of drinking: annual drinking volume (in standard drinks, 10 g of pure alcohol) and frequency of drinking occasions involving five or more standard drinks (hereafter referred to as risky single occasion drinking). Drinking was measured using a graduated quantity‐frequency scale that asked respondents to report how often they drank at various levels (more than 20 standard drinks, 11–19, 7–10, 5–6, 3–4, 1–2, <1). Mid‐points were assigned for volume and frequency [e.g. someone drinking 7–10 drinks 2–3 times per month was assigned drinking occasions of 8.5 standard drinks 30 (2.5 × 12) times in the past year]. These calculations were made for each drinking threshold and the values summed to provide an estimate of total drinking volume in the past year. Respondents who provided more than 365 drinking occasions were capped to their heaviest 365.
Frequency of 5+ drinking occasions was estimated using the same items, summing respondents' frequencies of drinking 5–6, 7–10, 11–19 and 20+ and capping at 365 days. For both consumption measures, we include abstainers coded as drinking 0 g (for the volume measure) and 0 risky occasions (for the risky single occasion measure). Given our harm outcomes are analysed as rates in the total population, this seemed like the most reasonable approach.
Alcohol harms
We looked at two outcomes covering chronic and acute harms related to alcohol consumption. First, alcohol‐related liver disease hospital admissions from the VAED, defined as any admission where the primary diagnosis had an International Classification of Diseases 10th revision code between K70.0 and K70.9. This includes alcoholic fatty liver, alcoholic hepatitis, alcoholic fibrosis and sclerosis of the liver, alcoholic cirrhosis of the liver, alcoholic hepatic failure and unspecified alcoholic liver disease. We focussed on primary diagnoses only to ensure comparability over time, but this means the rates presented here likely underestimate the real burden of alcohol‐related liver disease in Victorian hospitals.
Second, we examined presentations to emergency departments (EDs) in Victorian public hospitals with primary diagnosis codes for alcohol‐related disorders (F10.1–F10.9) or for toxic effects of alcohol (T51.0–T51.9) in the VEMD. These capture a range of behaviours, but are predominantly linked to acute intoxication events. A total of 210 448 hospital admissions and 147 970 ED presentations were included in the final analyses.
Both of these measures are simply aggregated counts of admissions/presentations—the same individual can appear multiple times in the data. This adds the potential for biases if, for example, people from lower socio‐economic groups are more likely to make repeat admissions to hospital, but linked data allowing us to focus on person‐level measures were not available.
Populations
Census data by postal area from 2001, 2006, 2011, 2016 and 2021 were used to develop the underlying populations for the harm rates. Populations for the intervening years were interpolated, assuming linear changes over time (e.g. if a postcode had a population increase of 500 people over 5 years, we assumed an annual increase of 100). Populations were aggregated to quintile (based on the assignment of postcodes to quintiles in the 2011 SEIFA indices). Annual rates per 10 000 population were then estimated by sex, age‐group and quintile for each year of the study. Analyses were conducted for men and women overall, and stratified by age (15–29, 30–49, 50–69, 70+).
Analysis
Data are initially presented descriptively. To test whether there were significant differences in consumption by socio‐economic quintile, a series of negative binomial models were estimated with volume of consumption or frequency of 5+ occasions as outcome variables and IRSD quintile of disadvantage as the key independent variable. All survey analyses adjusted for the complex sampling of the NDSHS, using Stata's ‘svy’ commands [36]. For the harms analyses, the number of admissions (or presentations) was the outcome variable, the underlying population was the exposure variable and year and IRSD quintile were included. For all models, initial main effects models were estimated and then interactions between quintile and year were also included to assess whether any socio‐economic disparities changed over time. Where the interaction models provided a better fit to the data (assessed via Bayesian information criteria) they are presented, otherwise main effects models are presented. Results for the two consumption outcomes and ED presentations are provided overall and for all age groups. For the alcohol‐related liver disease hospital admissions, the youngest and oldest age group are not presented separately as numbers were too low. These analyses were not pre‐registered and should be considered exploratory.
RESULTS
Consumption
Figure 1 shows the overall trends in total volume of consumption and frequency of risky single occasion drinking for men and women, by socio‐economic status.
FIGURE 1.
Mean yearly volume of drinking (in 10 g standard drinks) and frequency of 5+ drinking occasions for men and women, 2001–2019, National Drug Strategy Household Survey, by index of relative socio‐economic disadvantage quintile.
Negative binomial regressions for men and women overall and by age group for the two drinking outcomes are presented in Table 1. For men, consumption levels appear broadly similar across socio‐economic groups, while there is some evidence that women's drinking has a positive social gradient, at least for annual volume, where women in the most advantaged quintile reported significantly higher consumption than those in lower quintiles. Men in the most disadvantaged quintile drank significantly less than those in the least disadvantaged quintile, both overall and for the youngest age group, while those in the fourth quintile reported significantly lower drinking overall and for 30 to 49 year olds. Notwithstanding these differences, incident rate ratios were all relatively close to 1.0, suggesting a broadly flat socio‐economic distribution of drinking behaviour for men in Victoria. There were no significant interactions between socio‐economic status and survey year, meaning these patterns were stable over time—the models including the interaction terms did not improve the model fit, so the models with main effects only are reported here.
TABLE 1.
Negative binomial regression models for overall drinking volume and frequency of 5+ drinking occasions, overall and by age group.
| IRR (95% CI) | |||||
|---|---|---|---|---|---|
| Total volume | All men a | Men 15–29 | Men 30–49 | Men 50–69 | Men 70+ |
| Q1 (most disadvantaged) | 0.87 (0.79–0.96) | 0.78 (0.62–0.97) | 0.92 (0.78–1.07) | 0.92 (0.78–1.08) | 0.81 (0.64–1.01) |
| Q2 | 0.95 (0.86–1.04) | 0.94 (0.74–1.18) | 0.90 (0.78–1.04) | 1.03 (0.89–1.19) | 0.86 (0.68–1.08) |
| Q3 | 0.94 (0.87–1.02) | 0.85 (0.69–1.04) | 0.98 (0.87–1.11) | 0.97 (0.85, 1.11) | 0.80 (0.66–0.98) |
| Q4 | 0.88 (0.81–0.95) | 0.85 (0.70–1.03) | 0.84 (0.76–0.94) | 0.94 (0.82, 1.07) | 0.88 (0.71–1.09) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 0.99 (0.98–1.00) | 0.97 (0.96–0.98) | 0.99 (0.98–1.00) | 1.00 (0.99–1.01) | 0.99 (0.98–1.01) |
| All womena | Women 15–29 | Women 30–49 | Women 50–69 | Women 70+ | |
|---|---|---|---|---|---|
| Q1 (most disadvantaged) | 0.70 (0.62–0.80) | 0.68 (0.52–0.88) | 0.82 (0.66–1.02) | 0.64 (0.53–0.78) | 0.51 (0.39–0.67) |
| Q2 | 0.80 (0.72, 0.90) | 0.62 (0.50, 0.76) | 0.91 (0.76, 1.09) | 0.89 (0.72, 1.09) | 0.74 (0.52, 1.05) |
| Q3 | 0.77 (0.71–0.84) | 0.81 (0.67–0.99) | 0.73 (0.64–0.83) | 0.79 (0.69–0.91) | 0.62 (0.49–0.80) |
| Q4 | 0.83 (0.77–0.90) | 0.82 (0.68–0.98) | 0.82 (0.73–0.92) | 0.90 (0.78–1.04) | 0.69 (0.55–0.86) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 0.99 (0.98–0.99) | 0.97 (0.96–0.98) | 1.00 (0.99–1.01) | 1.01 (1.00–1.02) | 1.00 (0.98–1.02) |
| Frequency of risky single occasion drinking | All men a | Men 15–29 | Men 30–49 | Men 50–69 | Men 70+ |
|---|---|---|---|---|---|
| Q1 (most disadvantaged) | 0.91 (0.81–1.03) | 0.73 (0.58–0.93) | 0.93 (0.76–1.14) | 1.12 (0.89–1.41) | 0.87 (0.54–1.43) |
| Q2 | 1.01 (0.90–1.13) | 0.94 (0.74–1.19) | 0.89 (0.75–1.07) | 1.34 (1.08–1.65) | 0.74 (0.42–1.30) |
| Q3 | 0.97 (0.87–1.09) | 0.79 (0.63–0.99) | 1.03 (0.88–1.22) | 1.07 (0.86–1.32) | 0.95 (0.59–1.53) |
| Q4 | 0.88 (0.79–0.97) | 0.78 (0.64–0.96) | 0.82 (0.70–0.95) | 1.09 (0.88–1.34) | 0.77 (0.48–1.23) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 0.99 (0.98–1.00) | 0.96 (0.95–0.98) | 1.00 (0.99–1.01) | 1.00 (0.99–1.01) | 0.98 (0.95–1.02) |
| All women a | Women 15–29 | Women 30–49 | Women 50–69 | Women 70+ | |
|---|---|---|---|---|---|
| Q1 (most disadvantaged) | 0.89 (0.74–1.07) | 0.71 (0.54–0.94) | 1.16 (0.86–1.58) | 0.87 (0.56–1.36) | 0.58 (0.22–1.53) |
| Q2 | 0.90 (0.75–1.07) | 0.60 (0.47–0.77) | 1.16 (0.87–1.54) | 1.22 (0.82–1.83) | 0.91 (0.32–2.60) |
| Q3 | 0.85 (0.72–0.99) | 0.83 (0.66–1.04) | 0.80 (0.61–1.02) | 0.95 (0.63–1.44) | 0.63 (0.24–1.61) |
| Q4 | 0.89 (0.77–1.04) | 0.86 (0.68–1.07) | 0.88 (0.70–1.11) | 1.02 (0.69–1.50) | 0.50 (0.20–1.26) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 0.99 (0.98–1.01) | 0.96 (0.95–0.98) | 1.01 (1.00–1.03) | 1.05 (1.02–1.07) | 0.99 (0.93–1.05) |
Note: Bold figures represent statistically significant effects.
Abbreviations: IRR, incidence rate ratios, Q, quintile.
Adjusted for age group.
In contrast to the results for men, there were quite clear socio‐economic differences across measures and age groups for women. In general, women living in the least disadvantaged neighbourhoods reported significantly higher overall volumes of drinking (with the exception of 30–49 year olds) than those in other neighbourhoods. For risky single occasion drinking, a similar social gradient existed for women age 15 to 29, with the most disadvantaged two quintiles reporting less frequent episodes of risky single occasion drinking than the least disadvantaged. Again, interaction terms worsened the model fit and were non‐significant, so the main effects models are presented here.
Alcohol‐related harms
Figure 2 shows the overall trends in alcohol‐related liver disease hospital admissions and alcohol‐related ED presentations for men and women, by socio‐economic status.
FIGURE 2.
Alcohol‐related liver disease hospital admissions and alcohol‐related emergency department presentations by index of relative socio‐economic disadvantage quintile, Victoria, 2000–2020, men and women.
For all outcomes, a social gradient is evident, with attendance/presentation rates generally higher in more disadvantaged postcodes than more advantaged ones. Acute harms, in terms of presentations to ED for alcohol‐related issues have increased over the study period, while overall rates of alcohol‐related liver disease appear broadly stable. Negative binomial regression model results for alcohol‐related liver disease hospital admissions (chronic harms) are presented in Table 2.
TABLE 2.
Negative binomial regression models, alcohol‐related liver disease hospital admissions—men.
| IRR (95% CI) | |||
|---|---|---|---|
| All men | Men 30–49 | Men 50–69 | |
| Q1 (most disadvantaged) | 1.82 (1.41–2.35) | 2.18 (1.60–3.00) | 1.73 (1.31–2.29) |
| Q2 | 1.33 (1.03–1.73) | 1.60 (1.15–2.22) | 1.43 (1.07–1.89) |
| Q3 | 1.03 (0.79–1.33) | 1.35 (0.98–1.86) | 0.99 (0.75–1.32) |
| Q4 | 1.02 (0.79–1.33) | 0.87 (0.63–1.19) | 1.23 (0.93–1.63) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 0.98 (0.96–0.99) | 0.97 (0.95–0.99) | 0.98 (0.96–0.99) |
| P‐value for overall interaction | 0.33 | 0.43 | 0.05 |
| Q1 × year | 1.01 (0.99–1.04) | 1.00 (0.98–1.03) | 1.02 (0.99–1.05) |
| Q2 × year | 1.02 (0.99–1.04) | 1.02 (0.99–1.05) | 1.01 (0.99–1.04) |
| Q3 × year | 1.02 (0.99–1.05) | 1.01 (0.98–1.04) | 1.03 (1.00–1.05) |
| Q4 × year | 1.01 (0.99–1.03) | 1.02 (0.99–1.05) | 1.00 (0.97–1.02) |
| All women | Women 30–49 | Women 50–69 | |
|---|---|---|---|
| Q1 (most disadvantaged) | 1.20 (0.91–1.58) | 1.47 (0.98–2.20) | 0.96 (0.67–1.36) |
| Q2 | 1.52 (1.15–2.00) | 2.82 (1.91–4.12) | 0.98 (0.69–1.40) |
| Q3 | 1.18 (0.89–1.55) | 1.79 (1.21–2.64) | 0.95 (0.66–1.35) |
| Q4 | 0.92 (0.70–1.21) | 1.37 (0.93–2.01) | 0.64 (0.45–0.90) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 1.02 (0.99–1.03) | 1.03 (1.00–1.05) | 1.00 (0.98–1.02) |
| P‐value for overall interaction | 0.08 | <0.01 | 0.21 |
| Q1 × year | 1.02 (0.99–1.04) | 1.01 (0.98–1.05) | 1.03 (1.00–1.06) |
| Q2 × year | 0.99 (0.96–1.01) | 0.95 (0.91–0.98) | 1.02 (0.99–1.05) |
| Q3 × year | 1.01 (0.99–1.02) | 0.99 (0.96–1.03) | 1.02 (0.99–1.05) |
| Q4 × year | 1.00 (0.98–1.03) | 0.98 (0.95–1.02) | 1.03 (1.00–1.06) |
Note: Rates of liver disease were too low for the youngest and oldest age groups considered in this study and models for those groups are therefore, not presented. Bold figures represent statistically significant effects.
Abbreviations: IRR, incidence rate ratios; Q, quintile.
For men, there was clear evidence for a social gradient overall, with the two lowest quintiles experiencing significantly higher rates of alcohol‐related liver disease than the least disadvantaged quintile (the reference category), both overall and for the two specific age groups examined. Overall rates and rates for each age group declined significantly over the study period. Only one significant interaction effect was found, reflecting limited changes in the social gradient of male alcohol‐related liver disease over the study period. The social gradients were less steep for women than for men, but women in the least disadvantaged quintile still recorded lower rates of alcohol‐related liver disease than in more disadvantaged quintiles, both in total and for the 30 to 49 age group. Rates in the second least disadvantaged quintile were significantly lower than in the least disadvantaged quintile for 50 to 69 year olds. Rates for women were increasing in the 30 to 49 year old age group. Interactions were generally non‐significant, with the few significant effects not strongly suggesting a widening or narrowing of the social gradient of hospital admissions. The regression model estimates for ED presentations (acute harms) are provided in Table 3.
TABLE 3.
Regression models, alcohol‐related emergency department presentations.
| IRR (95% CI) | |||||
|---|---|---|---|---|---|
| All men | Men 15–29 | Men 30–49 | Men 50–69 | Men 70+ | |
| Q1 (most disadvantaged) | 1.51 (1.35–1.70) | 1.34 (1.09–1.64) | 1.86 (1.64–2.11) | 1.53 (1.32–1.77) | 0.85 (0.65–1.11) |
| Q2 | 1.66 (1.48–1.86) | 1.24 (1.01–1.52) | 2.12 (1.88–2.41) | 1.78 (1.54–2.05) | 0.82 (0.62–1.09) |
| Q3 | 1.01 (0.90–1.13) | 1.01 (0.83–1.25) | 1.07 (0.94–1.22) | 1.02 (0.87–1.19) | 0.55 (0.41–0.74) |
| Q4 | 0.99 (0.88–1.10) | 1.07 (0.87–1.30) | 0.99 (0.88–1.12) | 0.85 (0.73–0.98) | 0.90 (0.70–1.15) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 1.03 (1.02–1.04) | 1.02 (1.01–1.04) | 1.04 (1.03–1.05) | 1.05 (1.04–1.06) | 1.02 (1.00–1.03) |
| P‐value for overall interaction | <0.01 | 0.52 | <0.01 | <0.01 | 0.01 |
| Q1 × year | 1.01 (0.99–1.02) | 1.01 (0.99–1.02) | 1.00 (0.99–1.01) | 1.01 (0.99–1.02) | 1.03 (1.00–1.05) |
| Q2 × year | 0.99 (0.98–1.00) | 1.00 (0.98–1.02) | 0.99 (0.98–0.99) | 0.99 (0.98–1.01) | 1.03 (1.01–1.05) |
| Q3 × year | 1.01 (1.00–1.02) | 1.01 (0.99–1.02) | 1.01 (0.99–1.02) | 1.01 (0.99–1.02) | 1.04 (1.02–1.06) |
| Q4 × year | 1.00 (0.99–1.02) | 1.00 (0.98–1.02) | 1.01 (0.99–1.01) | 1.02 (1.00–1.03) | 1.01 (0.99–1.03) |
| All women | Women 15–29 | Women 30–49 | Women 50–69 | Women 70+ | |
|---|---|---|---|---|---|
| Q1 (most disadvantaged) | 1.19 (1.01–1.40) | 1.18 (0.93–1.49) | 1.42 (1.23–1.64) | 1.23 (0.97–1.57) | 0.65 (0.44–0.98) |
| Q2 | 1.02 (0.86–1.21) | 0.98 (0.77–1.24) | 1.23 (1.06–1.43) | 0.89 (0.69–1.15) | 0.54 (0.35–0.83) |
| Q3 | 1.07 (0.91–1.27) | 1.06 (0.84–1.34) | 1.22 (1.05–1.41) | 0.89 (0.70–1.15) | 0.46 (0.29–0.74) |
| Q4 | 1.05 (0.89–1.23) | 1.02 (0.81–1.29) | 1.10 (0.96–1.26) | 0.92 (0.73–1.16) | 0.90 (0.64–1.27) |
| Q5 (least disadvantaged) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) | 1 (Ref ) |
| Year/trend | 1.04 (1.03–1.05) | 1.03 (1.02–1.05) | 1.05 (1.04–1.06) | 1.07 (1.05–1.08) | 1.03 (1.01–1.05) |
| P‐value for overall interaction | 0.71 | 0.79 | 0.81 | 0.26 | 0.01 |
| Q1 × year | 1.00 (0.99–1.01) | 0.99 (0.97–1.01) | 0.99 (0.98–1.01) | 1.01 (0.99–1.03) | 1.02 (0.99–1.05) |
| Q2 × year | 1.01 (0.99–1.02) | 1.01 (0.98–1.03) | 1.00 (0.99–1.01) | 1.02 (1.00–1.04) | 1.04 (1.01–1.07) |
| Q3 × year | 1.00 (0.99–1.02) | 1.00 (0.98–1.02) | 1.00 (0.98–1.01) | 1.01 (0.99–1.03) | 1.04 (1.01–1.07) |
| Q4 × year | 1.00 (0.99–1.01 | 1.00 (0.98–1.02) | 1.00 (0.99–1.01) | 1.02 (0.99–1.04) | 0.99 (0.97–1.02) |
Note: Bold figures represent statistically significant effects.
Abbreviations: IRR, incidence rate ratios; Q, quintile.
For men, sharp socio‐economic gradients were evident for overall ED presentations and for all age groups except men age 70 years and over. For women, the rates of alcohol‐related ED presentations were consistently significantly higher in the most disadvantaged group than in all other quintiles, with the exception of the 70 + age group. Rates increased overall and for every age group. There were minimal significant interactions, and all related to slightly faster increases in quintile 2 than in quintile 1.
DISCUSSION
Our results clearly demonstrate that the alcohol harm paradox is evident in Victoria, with relatively minimal socio‐economic differences in drinking contrasting with marked socio‐economic inequalities for both acute‐ and chronic‐alcohol related harms over two decades. These patterns were broadly consistent across outcome, age group and sex and point to fundamental health inequalities in Victoria linked to alcohol harms, even where (as for women) drinking rates were higher in less socio‐economically disadvantaged groups. This is consistent with national estimates of the disease burden linked to alcohol, which estimate markedly higher impacts in disadvantaged neighbourhoods [1]. We found some limited evidence that socio‐economic disparities in these outcomes had widened over the study period, which is consistent with broader analyses highlighting increasing inequalities in premature mortality between 2006 and 2016 [37, 38]. This is in spite of the fact that our results show no significant changes in socio‐economic patterns of alcohol consumption over the same period [39] and suggests that, if anything, the alcohol harms paradox is strengthening in Victoria. In our data, there were minimal differences in consumption levels for men, but stark differences in harm rates over socio‐economic quintiles, while women's consumption was more strongly skewed toward socio‐economically advantaged groups, but harms remained more prevalent in disadvantaged neighbourhoods.
The lack of consistent reporting on inequalities in alcohol‐related harm in Australia is striking, with only the periodic updates to the Australian Burden of Disease (e.g. Australian Institute of Health and Welfare) [1] providing data, often in ways that are incomparable over time and of limited use for monitoring. This study provides some of the longest‐running data on disparities in alcohol harms in Australia and demonstrates clearly that no progress has been made at reducing the social gradient of morbidity because of alcohol here. Standardised reporting and monitoring of these disparities are a key first step in responding to them, and future research should explore ways to collate national data on the topic.
Socio‐economic disparities in alcohol‐related outcomes remain a key puzzle for alcohol epidemiology [10, 40]. While some of the differences in outcomes appear to relate to drinking patterns [5, 41], there remain marked gaps in morbidity and mortality that appear to require explanations beyond alcohol consumption [42]. Previous analyses focussing on disparities in other risk behaviours [12] have not produced striking results, suggesting more fundamental structural issues may be driving these inequalities. Evidence from the broader research field examining inequities in health outcomes is likely to be informative here, and the links between alcohol‐related harm and issues like social support, housing, financial stress and primary healthcare utilisation may play key roles. The recent call from Boyd et al. [10] to empirically examine a wider range of mechanisms for the alcohol harm paradox is important and timely, especially given the substantial impact of alcohol harms on health inequalities more broadly [7].
This study has a number of important limitations. First, we are presenting basic trends in drinking and harm rates from separate data sources. The alcohol harm paradox is better addressed with data that links individuals' consumption to harm outcomes, but these data are not readily available in Australia. Further, as with most studies of alcohol consumption, we rely on survey estimates that systematically under‐represent marginalised heavy drinkers [43] and consistently underestimate actual alcohol consumption in the population [44]. Still, previous work from Finland has demonstrated that even with more robust sampling frames and response rates the disparities in socio‐economic patterns for consumption and harm are maintained [45]. Further, our study focusses on primary diagnoses only for both the acute and chronic harms examined, which likely underestimates the true burden of alcohol in Victorian health systems. Further, it is worth noting that our measure for chronic harms—hospital admissions for alcohol‐related liver disease—typically takes many years to develop within individuals, but changes in population level consumption are reflected relatively quickly in population rates of disease [46]. More importantly, our measure of socio‐economic status has some important flaws. First, place‐based measures like the IRSD and the aggregate analyses we have conducted here are open to aggregation bias, and other measures of socio‐economic status (especially at the individual level) would improve our confidence in these findings. Having said that, the broad consistency of the global literature provides some reassurance that the socio‐economic disparities identified are real and not artefacts of measurement. Our use of the 2011 IRSD, while examining data between 2000 and 2020 assumes fixed neighbourhood disadvantage levels, which may mean our analyses of trends over time are flawed. This potential misclassification could serve to either underestimate or overestimate social gradients, although our use of the mid‐point in time means that, for example, a suburb that has steadily gentrified would be misclassified in opposite directions either side of 2011, potentially smoothing out overall biases. Still, more specific measurement of socio‐economic status is crucial for future work in this space. Our use of broad age groups potentially disguises heterogeneity in trends within them (e.g. 18–24 year olds may have different trends to 25–39 year olds), but we were constrained by the number of events in our harm measures to these broad categories.
Finally, our outcome measures rely on episode‐ not individual‐level data, meaning repeated admissions for the same condition are all counted separately. It is unclear how this might affect estimates of health inequalities, but future work should make use of increasingly available linked data (e.g. [47]) to ensure these patterns are reflected at the person‐level, not just the episode‐level. Finally, for the sake of simplicity we have assumed linear time trends in our models, which likely smooths over some important variations over the entire period studied. Our data focuses on Victoria because hospital and ED data in Australia are captured at the state, not national level. There are substantial differences in population demographics and drinking rates across Australian states and so more work is required to understand how our findings are reflected nationally.
Despite these limitations, this article provides the most comprehensive and up‐to‐date examination of socio‐economic inequalities in alcohol consumption and related harm in Australia. It provides further evidence that alcohol is a key driver of health inequality even while survey data suggests consumption levels do not have a strong social gradient. Our findings point clearly toward an urgent need for better understanding of the mechanisms driving these disparities so that they can be reduced.
AUTHOR CONTRIBUTIONS
Michael Livingston: Conceptualization (equal); data curation (lead); formal analysis (lead); methodology (lead); writing—original draft (lead); writing—review and editing (equal). Nic Taylor: Data curation (supporting); formal analysis (supporting); methodology (supporting); writing—review and editing (equal). Sarah Callinan: Conceptualization (supporting); formal analysis (supporting); funding acquisition (supporting); writing—review and editing (equal). Yvette Mojica‐Perez: Formal analysis (supporting); methodology (supporting); project administration (supporting); writing—review and editing (equal). Alexandra Torney: Data curation (equal); formal analysis (equal); methodology (supporting); writing—review and editing (equal). Gabriel Caluzzi: Conceptualization (supporting); writing—review and editing (equal). Klaudia Kepa: Conceptualization (supporting); project administration (equal); writing—review and editing (equal). Amy Pennay: Conceptualization (equal); funding acquisition (lead); project administration (lead); writing—review and editing (equal).
DECLARATION OF INTERESTS
None.
ACKNOWLEDGEMENTS
This article was substantially improved thanks to feedback from Inge Kersbergen at the 2024 Kettil Bruun Society Conference. The Australian Institute of Health and Welfare provided access to the National Drug Strategy Household Survey and the Victorian Agency for Health Information provided access to the hospital and emergency department data. Open access publishing facilitated by Curtin University, as part of the Wiley ‐ Curtin University agreement via the Council of Australian University Librarians.
Livingston M, Taylor N, Callinan S, Mojica‐Perez Y, Torney A, Caluzzi G, et al. Describing the alcohol harm paradox: 20 years of data from Victoria, Australia. Addiction. 2026;121(2):349–359. 10.1111/add.70194
Funding information This study was funded by a Victorian Health Promotion Foundation grant. M.L. is supported by an Australian Research Council Future Fellowship (FT210100656).
DATA AVAILABILITY STATEMENT
Access to both the NDSHS and Victorian Health System data were provided on the undertaking that data confidentiality was maintained. Data are available from the original data custodians on request, but we cannot make them freely available with this article.
REFERENCES
- 1. Australian Institute of Health and Welfare . Australian burden of disease study 2018: interactive data on risk factor burden Canberra: AIHW; 2021. [Google Scholar]
- 2. Vos T, Lim SS, Abbafati C, Abbas KM, Abbasi M, Abbasifard M, et al. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the global burden of disease study 2019. Lancet. 2020;396(10258):1204–1222. 10.1016/S0140-6736(20)30925-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Probst C, Roerecke M, Behrendt S, Rehm J. Gender differences in socioeconomic inequality of alcohol‐attributable mortality: a systematic review and meta‐analysis. Drug Alcohol Rev. 2015;34(3):267–277. 10.1111/dar.12184 [DOI] [PubMed] [Google Scholar]
- 4. Probst C, Roerecke M, Behrendt S, Rehm J. Socioeconomic differences in alcohol‐attributable mortality compared with all‐cause mortality: a systematic review and meta‐analysis. Int J Epidemiol. 2014;43(4):1314–1327. 10.1093/ije/dyu043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Probst C, Kilian C, Sanchez S, Lange S, Rehm J. The role of alcohol use and drinking patterns in socioeconomic inequalities in mortality: a systematic review. Lancet Public Health. 2020;5(6):e324–e332. 10.1016/S2468-2667(20)30052-9 [DOI] [PubMed] [Google Scholar]
- 6. Collins SE. Associations between socioeconomic factors and alcohol outcomes. Alcohol Res. 2016;38(1):83–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Mackenbach JP, Kulhánová I, Bopp M, Borrell C, Deboosere P, Kovács K, et al. Inequalities in alcohol‐related mortality in 17 European countries: a retrospective analysis of mortality registers. PLoS Med. 2015;12(12):e1001909. 10.1371/journal.pmed.1001909 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Angus C, Pryce R, Holmes J, de Vocht F, Hickman M, Meier P, et al. Assessing the contribution of alcohol‐specific causes to socio‐economic inequalities in mortality in England and Wales 2001–16. Addiction. 2020;115(12):2268–2279. 10.1111/add.15037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Smith K, Foster J. Alcohol, health inequalities and the harm paradox: why some groups face greater problems despite consuming less alcohol Institute of Alcohol Studies; 2014. [Google Scholar]
- 10. 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. 10.1111/add.15567 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Salom CL, Williams GM, Najman JM, Alati R. Does early socio‐economic disadvantage predict comorbid alcohol and mental health disorders? Drug Alcohol Depend. 2014;142:146–153. 10.1016/j.drugalcdep.2014.06.011 [DOI] [PubMed] [Google Scholar]
- 12. Livingston M. Socio‐economic differences in alcohol‐related risk‐taking behaviours. Drug Alcohol Rev. 2014;33(6):588–595. 10.1111/dar.12202 [DOI] [PubMed] [Google Scholar]
- 13. Jonas H, Dietze P, Rumbold G, Hanlin K, Cvetkovski S, Laslett AM. Associations between alcohol related hospital admissions and alcohol consumption in Victoria: Influence of socio‐demographic factors. Aust N Z J Public Health. 1999;23:272–279. [DOI] [PubMed] [Google Scholar]
- 14. Boyle T, Fritschi L, Tabatabaei SM, Ringwald K, Heyworth JS. Smoking, alcohol, diabetes, obesity, socioeconomic status, and the risk of colorectal cancer in a population‐based case–control study. Cancer Causes Contr. 2014;25(12):1659–1668. 10.1007/s10552-014-0470-7 [DOI] [PubMed] [Google Scholar]
- 15. Najman JM, Williams GM, Room R. Increasing socioeconomic inequalities in male cirrhosis of the liver mortality: Australia 1981‐2002. Drug Alcohol Rev. 2007;26(3):273–278. 10.1080/09595230701247699 [DOI] [PubMed] [Google Scholar]
- 16. Dietze P, Jolley D, Chikritzhs T, Clemens S, Catalano P, Stockwell T. Income inequality and alcohol attributable harm in Australia. BMC Public Health. 2009;9:70. 10.1186/1471-2458-9-70 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Room R. The dialectic of drinking in Australian life: From the rum corps to the wine column. Austr Drug Alcohol Rev. 1988;7(4):413–437. 10.1080/09595238880000741 [DOI] [Google Scholar]
- 18. Australian Institute of Health and Welfare . Apparent consumption of alcohol in Australia, Canberra Australia: AIHW; 2023. [Google Scholar]
- 19. Turrell G, Mathers CD. Socioeconomic status and health in Australia. Med J Aust. 2000;172(9):434–438. 10.5694/j.1326-5377.2000.tb124041.x [DOI] [PubMed] [Google Scholar]
- 20. Australian Institute of Health and Welfare . National Drug Strategy Household Survey ‐ 2022‐2023 Canberra: AIHW; 2024. [Google Scholar]
- 21. Bonevski B, Regan T, Paul C, Baker AL, Bisquera A. Associations between alcohol, smoking, socioeconomic status and comorbidities: evidence from the 45 and up study. Drug Alcohol Rev. 2014;33(2):169–176. 10.1111/dar.12104 [DOI] [PubMed] [Google Scholar]
- 22. Livingston M, Laslett AM, Dietze P. Individual and community correlates of young people's high‐risk drinking in Victoria, Australia. Drug Alcohol Depend. 2008;98(3):241–248. 10.1016/j.drugalcdep.2008.06.002 [DOI] [PubMed] [Google Scholar]
- 23. Giskes K, Turrell G, Bentley RJ, Kavanagh A. Individual and household‐level socioeconomic position is associated with harmful alcohol consumption behaviours among adults. Aust N Z J Public Health. 2011;35(3):270–277. 10.1111/j.1753-6405.2011.00683.x [DOI] [PubMed] [Google Scholar]
- 24. Jefferis BJ, Manor O, Power C. Social gradients in binge drinking and abstaining: Trends in a cohort of British adults. J Epidemiol Community Health. 2007;61(2):150–153. 10.1136/jech.2006.049304 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Grittner U, Kuntsche S, Gmel G, Bloomfield K. Alcohol consumption and social inequality at the individual and country levels—Results from an international study. Eur J Public Health. 2013;23(2):332–339. 10.1093/eurpub/cks044 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Livingston M, Callinan S, Vashishtha R, Yuen WS, Dietze P. Tracking the decline in Australian adolescent drinking into adulthood. Addiction. 2022;117(5):1273–1281. 10.1111/add.15720 [DOI] [PubMed] [Google Scholar]
- 27. Xu KQ, Payne CF. A growing divide: Trends in social inequalities in healthy longevity in Australia, 2001–20. Popul Stud. 78(2):1–250. 10.1080/00324728.2023.2241429 [DOI] [PubMed] [Google Scholar]
- 28. Flavel J, McKee M, Freeman T, Musolino C, van Eyk H, Tesfay FH, et al. The need for improved Australian data on social determinants of health inequities. Med J Aust. 2022;216(8):388–391. 10.5694/mja2.51495 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Roy Morgan Research . 2005. National Drug Strategy Household Survey 2004 ‐ technical report. Melbourne: Roy Morgan Research. [Google Scholar]
- 30. Livingston M, Dietze P. National survey data can be used to measure trends in population alcohol consumption in Australia. Aust N Z J Public Health. 2016;40(3):233–235. 10.1111/1753-6405.12511 [DOI] [PubMed] [Google Scholar]
- 31. Health Data Standards and Systems Unit . VAED 24th edition user manual 2024–25 Melbourne, Australia: Victorian Department of Health; 2024. [Google Scholar]
- 32. Health Data Standards and Systems Unit . The Victorian emergency minimum dataset (VEMD) user manual, 29th edition, 2024–25, Melbourne: Victorian Department of Health; 2024.
- 33. FitzGerald G, Toloo G, He J, Doig G, Rosengren D, Rothwell S, et al. Private hospital emergency departments in Australia: challenges and opportunities. Emerg Med Australas. 2013;25(3):233–240. 10.1111/1742-6723.12082 [DOI] [PubMed] [Google Scholar]
- 34. Australian Bureau of Statistics . Census of population and housing: socio‐economic indexes for areas (SEIFA), Australia, 2011 Canberra: ABS; 2013. [Google Scholar]
- 35. Randolph B, Tice A. Suburbanizing disadvantage in Australian cities: Sociospatial change in an era of neoliberalism. J Urban Aff. 2014;36(sup1):384–399. 10.1111/juaf.12108 [DOI] [Google Scholar]
- 36. Kreuter F, Valliant R. A survey on survey statistics: what is done and can be done in Stata. Stata Journal. 2007;7(1):1–21. 10.1177/1536867X0700700101 [DOI] [Google Scholar]
- 37. Adair T, Lopez AD. Widening inequalities in premature mortality in Australia, 2006‐16. Aust Popul Stud. 2020;4(1):37–56. 10.37970/aps.v4i1.62 [DOI] [Google Scholar]
- 38. Adair T, Lopez AD. An egalitarian society? Widening inequalities in premature mortality from non‐communicable diseases in Australia, 2006–16. Int J Epidemiol. 2021;50(3):783–796. 10.1093/ije/dyaa226 [DOI] [PubMed] [Google Scholar]
- 39. Ding D, Do A, Schmidt H‐M, Bauman AE. A widening gap? Changes in multiple lifestyle risk Behaviours by socioeconomic status in New South Wales, Australia, 2002–2012. PLoS ONE. 2015;10(8):e0135338. 10.1371/journal.pone.0135338 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Bellis MA, Hughes K, Nicholls J, Sheron N, Gilmore I, Jones L. The alcohol harm paradox: using a national survey to explore how alcohol may disproportionately impact health in deprived individuals. BMC Public Health. 2016;16(1):111. 10.1186/s12889-016-2766-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Lewer D, Meier P, Beard E, Boniface S, Kaner E. Unravelling the alcohol harm paradox: a population‐based study of social gradients across very heavy drinking thresholds. BMC Public Health. 2016;16(1):1–11. 10.1186/s12889-016-3265-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Bloomfield K. Understanding the alcohol‐harm paradox: what next? Lancet Public Health. 2020;5(6):e300–e301. 10.1016/S2468-2667(20)30119-5 [DOI] [PubMed] [Google Scholar]
- 43. Rehm J, Kilian C, Rovira P, Shield KD, Manthey J. The elusiveness of representativeness in general population surveys for alcohol. Drug Alcohol Rev. 2021;40(2):161–165. 10.1111/dar.13148 [DOI] [PubMed] [Google Scholar]
- 44. Livingston M, Callinan S. Under‐reporting in alcohol surveys: whose drinking is under‐estimated? J Stud Alcohol Drugs. 2015;76(1):158–164. 10.15288/jsad.2015.76.158 [DOI] [PubMed] [Google Scholar]
- 45. Mäkelä P, Huhtanen P. The effect of survey sampling frame on coverage: the level of and changes in alcohol‐related mortality in Finland as a test case. Addiction. 2010;105(11):1935–1941. 10.1111/j.1360-0443.2010.03069.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Holmes J, Meier PS, Booth A, Guo Y, Brennan A. The temporal relationship between per capita alcohol consumption and harm: a systematic review of time lag specifications in aggregate time series analyses. Drug Alcohol Depend. 2012;123(1‐3):7–14. 10.1016/j.drugalcdep.2011.12.005 [DOI] [PubMed] [Google Scholar]
- 47. Australian Institute of Health and Welfare . National Integrated Health Services Information: AIHW 2023.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Access to both the NDSHS and Victorian Health System data were provided on the undertaking that data confidentiality was maintained. Data are available from the original data custodians on request, but we cannot make them freely available with this article.