Socioeconomic and Geographic Disparities in Emergency Department Visits due to Alcohol in Ontario: A Retrospective Population-level Study from 2003 to 2017

Abstract

Objective:

While the overall health system burden of alcohol is large and increasing in Canada, little is known about how this burden differs by sociodemographic factors. The objectives of this study were to assess sociodemographic patterns and temporal trends in emergency department (ED) visits due to alcohol to identify emerging and at-risk subgroups.

Methods:

We conducted a retrospective population-level cohort study of all individuals aged 10 to 105 living in Ontario, Canada. We identified ED visits due to alcohol between 2003 and 2017 using defined International Classification of Diseases, 10th edition, codes from a pre-existing indicator. We calculated annual age- and sex-standardized, and age- and sex-specific rates of ED visits and compared overall patterns and changes over time between urban and rural settings and income quintiles.

Results:

There were 829,662 ED visits due to alcohol over 15 years. Rates of ED visits due to alcohol were greater for individual living in the lowest- compared to the highest-income quintile neighbourhoods, and disparities (rate ratio lowest to highest quintile) increased with age from 1.22 (95% CI, 1.19 to 1.25) in 15- to 18-year-olds to 4.17 (95% CI, 4.07 to 4.28) in 55- to 59-year-olds. Rates of ED visits due to alcohol were significantly greater in rural settings (56.0 per 10,000 individuals, 95% CI, 55.7 to 56.4) compared to urban settings (44.8 per 10,000 individuals, 95% CI, 44.7 to 44.9), particularly for young adults. Increases in rates of visits between 2003 and 2017 were greater in rural versus urban settings (82 vs. 68% increase in age- and sex-standardized rates) and varied across sociodemographic subgroups with the largest annual increases in rates of visits in young (15 to 29) low-income women (6.9%, 95%CI, 6.7 to 7.3) and the smallest increase in older (45 to 59) high-income men (2.7, 95%CI, 2.4 to 3.0).

Conclusion:

Alcohol harms display unique patterns with the highest burden in rural and lower-income populations. Rural–urban and income-based disparities differ by age and sex and have increased over time, which offers an imperative and opportunity for further interventions by clinicians and policy makers.

Introduction

Alcohol use has negative impacts on the health of Canadians and is a leading cause of death and disability in Canada and globally. ¹ A growing body of evidence has documented that morbidity and mortality from alcohol have increased in high-income countries, particularly in Canada and the United States. ^{2 –5} Several recent studies have found that alcohol-related health-care visits have been disproportionately increasing in younger adults and women. ^2,4,6 Evidence from the United States has also highlighted how deaths from drug and alcohol poisonings and alcoholic liver disease have been increasing faster in rural compared to urban America particularly among younger (aged 35 to 44) Caucasian individuals. ⁷ In Canada, there is a general understanding that certain sociodemographic groups experience a higher burden of alcohol harms (e.g., low-income individuals and middle-aged men) and larger increases in harms over time (e.g., women and young adults). However, there is a limited knowledge about alcohol-related harms in rural populations in Canada as well as less understanding of trends in socioeconomic and geographic disparities over time. In light of a growing epidemic of alcohol-related harms, such data are urgently needed to identify gaps in the delivery of clinical services and better target alcohol control policies based on evidence.

Multiple studies and reports have documented that alcohol-related harms are generally greater for lower-income individuals in Canada and internationally. ^8,9 Problematic alcohol use and hospitalizations due to alcohol are also higher in rural areas of Canada. ⁸ However, almost no data are available on how these geographic and socioeconomic disparities differ by sex and age. ^8,10 In addition, contemporary data are lacking whether income and geographic disparities are widening over time, as seen in the United States, and their potential contribution to the observed increases in alcohol harms in Canada. ¹⁰ In this study, we examined rates of emergency department (ED) visits due to alcohol by sex, age, neighbourhood income quintiles, and rurality of individuals. We described how the rates of ED visits due to alcohol in subpopulations changed over a 15-year time period by examining trends in rates of visits in Ontario, Canada, between 2003 and 2017. Finally, we examined the interactions among these sociodemographic factors (e.g., young women living in low-income and rural regions) to identify populations with the greatest rates of ED visits due to alcohol. We chose to examine ED visits due to alcohol as both a measure of the burden of alcohol on our health systems and as repeat visits have recently been found to be strongly associated with increased risk of mortality. ¹¹

Methods

Study Design and Population

We conducted a retrospective population-level study in Ontario, Canada. We used linked administrative data collected from the province’s universal health-care system. We created 15 consecutive and contiguous annual cohorts between the years 2003 and 2017. Each year, all Ontario residents between the ages of 10 and 105 (as of July 1 of the calendar year) who were eligible for provincial health insurance were included in the study. Individuals were assessed each year and excluded from that year if they were missing data on age, sex, neighbourhood income, or rurality. Overall, 182,345 individuals (1.18% of eligible Ontarians) were excluded in one or more study years due to missing data. Furthermore, 5,060 individuals (0.03%) were missing one or more covariates in all years of eligibility and, therefore, entirely excluded from the study. In total, our study included 15,444,489 unique individuals.

Data Sources and Covariates

We used the following datasets: (1) the National Ambulatory Care Reporting System database, which contains complete records from all ED visits in Ontario ¹² ; (2) the Ontario Registered Persons Database, a registry of all individuals eligible for the provincial health insurance plan containing individual demographic information; (3) the Postal Code Conversion File Plus (PCCF+), which contains information on individuals’ neighbourhood income and the rurality of their home address. We used the Statistics Canada census definition for rurality, which defines a rural residence as living in a town or municipality outside of the commuting zone of census metropolitan areas (CMA, population of 100,000 or more) or census agglomeration (CA, population of 10,000 or more). ¹³ We characterized neighbourhood income using the Quintile of Annual Income Per Person Equivalent (QAIPPE) from the PCCF+. QAIPPE divides neighbourhoods into quintiles based on the average before-tax income, which is adjusted for household size and is relative to other neighbourhoods either within the same CMA or CA or in rural Ontario. ¹⁴

Outcomes

Our primary outcome was an ED visit due to alcohol, defined using the Canadian Institute for Health Information’s indicator on “hospitalizations entirely due to alcohol.” ¹⁵ ED visits due to alcohol were identified using diagnostic codes from the 10th edition of the International Classification of Diseases (ICD-10); see Online Appendix A for the list of ICD-10 codes. Visits were included when an ICD-10 code from the indicator was listed as the main reason for the visit or having contributed to the visit. Our secondary outcome was “all-cause” or ED visit for any reason.

Statistical Analyses

We present descriptive statistics to characterize and compare the rates of ED visits due to alcohol across age groups, sex, and neighbourhood income quintiles. We calculated event rates as the number of events in the population divided by the number of individuals in the population on July 1 of that year. Rates were directly standardized by age and sex using the latest Census (2016 Census) population as the reference. We present rate ratios (RRs) with 95% confidence intervals to compare crude rates of ED visits due to alcohol between the lowest- and highest-neighbourhood-income quintiles for different subgroups (e.g., men and women, urban and rural and by age).

We examined temporal trends using age- and sex-stratified Poisson regression models. We categorized age into 3 groups (ages 15 to 29, 30 to 44, and 45 to 59). These age groups were selected to allow examination of differences in the burden of visits at different life stages of adulthood and to allow sufficient visits per subgroup for stable rates. We assessed trends in each subgroup by running stratified models examining the count of ED visits, the years of observation (specified as a continuous variable), and the population of the subgroup (e.g., number of highest-neighbourhood-income quintile urban men). We interpreted the incident RRs obtained from the time covariate in our Poisson models as the average annual change in the rate of hospitalizations. All tests were 2-sided, and P values less than 0.05 were considered significant. Statistical analyses were performed using SAS version 9.4 statistical software.

Results

There were 829,662 ED visits due to alcohol over the 15-year study period. Visits due to alcohol were greater for individuals in the lowest-neighbourhood-income quintile (35% of visits, n = 293,233), who experienced 2.5 times the number of visits due to alcohol than individuals in the highest-neighbourhood-income quintile (13.5% of visits, n = 112,336). The disparity by neighbourhood income quintiles was greater for men (2.9 times) than women (2.1 times; see Table 1).

Table 1.

Characteristics of Ontario Residents Who Visited the Emergency Department because of Alcohol Use (2003 to 2017).^a

Characteristic	No. (%) of All Patients			No. (%) of All Patients
	Women			Men
	Q1 (Lowest), n = 34,596	Q5 (Highest), n = 23,073	Standardized Difference	Q1 (Lowest), n = 61,906	Q5 (Highest), n = 34,320	Standardized Difference
No. of visits due to alcohol use (% of total visits due to alcohol over study)	88,879 (10.7)	41,983 (5.0)	—	203,543 (24.5)	69,614 (8.4)	—
Age at first visit, yr
Age, mean ± SD	34.9 ± 17.5	32.7 ± 18.5	.12	39.9 ± 17.5	36.6 ± 18.7	.18
10 to 18b	6,269 (18.1)	6,488 (28.1)	.24	5,948 (9.6)	6,223 (18.1)	.25
19 to 29	10,812 (31.3)	6,930 (30.0)	.03	16,153 (26.1)	10,234 (29.8)	.08
30 to 39	4,715 (13.6)	2,150 (9.3)	.14	9,349 (15.1)	3,845 (11.2)	.12
40 to 49	5,279 (15.3)	2,624 (11.4)	.11	11,203 (18.1)	4,664 (13.6)	.12
50 to 59	3,886 (11.2)	2,379 (10.3)	.03	9,946 (16.1)	4,435 (12.9)	.09
60+	3,635 (10.5)	2,502 (10.8)	.01	9,307 (15.0)	4,919 (14.3)	.02
No. of all-cause ED visits in year of initial visit, mean ± SD	3.3 ± 4.9	2.3 ± 3.2	.23	3.0 ± 4.9	2.3 ± 3.2	.16
No. of ED visits due to alcohol in the year of initial visit, mean ± SD	1.2 ± 0.9	1.1 ± 0.6	.07	1.2 ± 1.2	1.2 ± 0.8	.08
1 visit	31,063 (89.8)	21,249 (92.1)	.08	54,305 (87.7)	31,219 (91.0)	.11
2 visits	2,463 (7.1)	1,334 (5.8)	.05	5,111 (8.3)	2,193 (6.4)	.07
3 visits	581 (1.7)	290 (1.3)	.04	1,220 (2.0)	483 (1.4)	.04
≥4 visits	489 (1.4)	200 (0.9)	.05	1,270 (2.1)	425 (1.2)	.06
No. of visits due to alcohol in rural settings (% of visits in column)	5,246 (15.2)	3,061 (13.3)	.05	8,620 (13.9)	4,909 (14.3)	.01

^a The characteristics (age, sex, income quintile, rurality, and number of ED visits) of individuals with 1 ED visit due to alcohol during the study were determined at the time of the visit. Characteristics of individuals who had multiple visits (multiple visits in the same year or visits over multiple years) were counted 1 time at the time of the first visit in the study.

^b19 is the legal drinking age in Ontario.

Age and Sex Differences in Geographic and Neighbourhood Income Disparities

Figure 1 depicts the influence of age and sex on the relationship between neighbourhood income and rates of ED visits due to alcohol. For both men and women, neighbourhood income–based disparities in rates of ED visits due to alcohol increased with age. The RR between rates of ED visits due to alcohol for individuals living in the lowest- compared to the highest-neighbourhood-income quintiles was 1.22 (95% CI, 1.19 to 1.25) for those aged 15 to 18, 2.45 (95% CI, 2.38 to 2.51) for those aged 30 to 34, 4.17 (95% CI, 4.07 to 4.28) for those aged 55 to 59, and 2.51 (95% CI, 2.45 to 2.57) for those aged 65+.

Figure 1.

Rates of emergency department (ED) due to alcohol across 5 income quintiles (Q1 = poorest quintile and Q5 = wealthiest quintile), overall for all years (2003 to 2017) in 5-year age increments for men and women.

Sex also modified the relationship between age, neighbourhood income, and ED visits due to alcohol. Among individuals aged 15 to 29, the RRs for the lowest- versus highest-neighbourhood-income quintiles was similar for men (1.56, 95% CI, 1.54 to 1.58) and women (1.47, 95% CI, 1.41 to 1.45). In comparison, among individuals aged 45 to 59, the RR for lowest- versus highest-neighbourhood-income quintiles was greater in men (4.35, 95% CI, 4.29 to 4.41) than women (3.05, 95% CI, 2.99 to 3.11). Age and sex did not modify disparities between lowest- and highest-income neighbourhoods for all-cause ED visits (see Online Appendix B Figure 1). Much of the disparities in ED visits related to neighbourhood income was related to being in the lowest-income quintile. For example, among men aged 45 to 59, each quintile decrease in neighbourhood income from Q5 (highest) to Q2 (second lowest) was on average associated with a 1.30 times (95% CI, 1.27 to 1.33) increase in rates of ED visits due to alcohol. In contrast, the decrease in neighbourhood income from the second lowest to lowest quintile was associated with a 1.98 (95% CI, 1.94 to 2.02) increase in the rate of ED visits.

Figure 2 depicts the influence of age and sex on the geographic disparities in rates of ED visits due to alcohol. Overall, the age- and sex-standardized rate of ED visits due to alcohol per 10,000 individuals was greater in rural (56.0, 95% CI, 55.7 to 56.4) than in urban (44.8, 95% CI, 44.7 to 44.9) regions. The rate of ED visits due to alcohol per 10,000 individuals for those aged 10 to 29 was 42.3% higher in rural (62.0, 95% CI, 61.4 to 62.7) compared to urban regions (43.9, 95% CI, 43.7 to 44.1). In contrast, the rate among those aged 40 or older in rural (46.1, 95% CI, 45.7 to 46.5) and urban regions (46.2, 95% CI, 46.1 to 46.3) was comparable.

Figure 2.

Rates of emergency department (ED) visits due to alcohol depicting the unadjusted rates of visits in urban and rural settings, overall for all years (2003 to 2017) in 5-year age increments for men and women.

Interaction of Risk Factors and High-risk Groups

Figure 3 shows the interaction between age, sex, neighbourhood income, and geography on rates of ED visits due to alcohol in the year 2017. We found that among youth and young adults (aged 15 to 29), the neighbourhood income gradient (i.e., RR comparing the lowest to the highest quintile) was greater in rural (RR: 2.83, 95% CI, 2.53 to 3.17) compared to urban regions (RR: 1.54, 95% CI, 1.47 to 1.61). Among those aged 30 to 44, the neighbourhood income gradient was greater in rural compared to urban regions, particularly in women (RR: 3.17, 95% CI, 2.90 to 3.45 for urban women vs. 8.03 95% CI, 6.24 to 10.35 for rural women in this age group only). Finally, among middle-aged adults (aged 45 to 59), we observed a larger neighbourhood income gradient for men in urban (RR: 5.57, 95% CI, 5.25 to 5.91) compared to rural settings (RR: 3.27, 95% CI, 2.83 to 3.77).

Figure 3.

Rates of emergency department (ED) visits due to alcohol for men and women aged 15 to 29, 30 to 44, and 45 to 54, living in urban or rural settings in 2017. Rate ratios with 95 percent confidence intervals between the lowest (Q1)-neighbourhood-income quintile and highest (Q5)-neighbourhood-income quintile are presented above the bars.

Temporal Changes for Lowest and Highest Neighbourhood Income and Urban and Rural Regions

Figure 4 shows how rates of ED visits due to alcohol have changed over time. Annual rates increased significantly in all 24 subgroups over our study period. The increases in visits, between 2003 and 2017, were greater in rural compared to urban settings (82% vs. 68% increase in age- and sex-standardized rates); this may be attributable to the relatively larger increase in age-standardized rates among women (114% rural vs. 85% urban) compared to men (60% rural vs. 63% urban). The greatest increase in rates of annual visits was women aged 15 to 29 living in the lowest-neighbourhood-income quintile (6.9% average increase per year, 95% CI, 6.7 to 7.3), while the smallest relative increase in rates was in men aged 45 to 59 living in the highest-neighbourhood-income quintile (2.7% average increase per year, 95% CI, 2.4 to 3.0).

Figure 4.

Changes in rates of emergency department visits due to alcohol for men and women in 3 age groups (15 to 29, 30 to 44, and 45 to 59) between urban and rural settings and for the lowest-income quintile (Q1) and the highest-income quintile (Q5) across 4 cross-sectional time periods (2003 to 2005, 2007 to 2009, 2011 to 2013, and 2015 to 2017). A test of linear trend (year as a continuous variable) for change in rates of visits for each group was significant.

We note 2 key trends; first, by 2017, neighbourhood income became a stronger indicator of rates of visits due to alcohol than sex. The rates of visits in lowest-neighbourhood-income women were 1.54 (95% CI, 1.48 to 1.59) times greater than in highest-neighbourhood-income men. Second, increases in ED visits were evenly distributed across age groups in lowest-neighbourhood-income individuals but concentrated in young adults in highest-neighbourhood-income individuals. For example, between 2003 and 2017, men aged 15 to 29 were responsible for 39% (1,099 visits) of the total increase in ED visits due to alcohol that occurred in the highest-neighbourhood-income quintile (2,823 visits) compared to 21% (2,129) of the total increase that occurred in lowest-neighbourhood-income quintile (10,208 visits; see Online Appendix B Figure 2).

Interpretation

We found that individuals living in the lowest-income neighbourhoods and rural individuals experience far greater burden from ED visits due to alcohol than urban or highest-income-neighbourhood individuals. Importantly, these disparities were modified by age and sex. Neighbourhood income disparities were minimal in younger adults while in older age groups, especially men, those living in lowest-income neighbourhoods had much higher rates than those in highest-income neighbourhoods. In contrast, urban–rural disparities were greater in youth compared to older adults. The demographics of individuals experiencing an ED visit due to alcohol have changed over time, with the greatest increases occurring in rural compared to urban settings for young individuals living in the lowest-income neighbourhoods and the smallest increases occurring in older adults living in highest-income neighbourhoods.

Our findings of neighbourhood income–based disparities are consistent with a large body of international work showing that individuals of lower socioeconomic status experience greater harms from alcohol. ^16,17 Adding to the literature, we found that alcohol-related disparities in ED visits increase with age, particularly in men. A potential explanation for these findings may be a greater accumulation of comorbid conditions or social circumstances in lower-income individuals over time, which may decrease their resiliency to or increase the consequences of heavy drinking. ¹⁸ This possibility is supported by multiple studies showing that despite experiencing greater alcohol-related harms, alcohol consumption is generally lower or comparable in lower socioeconomic groups than individuals with higher socioeconomic status. ¹⁹ This finding may also partially be the result of reverse causation, where adults who display harmful patterns of alcohol consumption gravitate to low-income areas over time—a pattern that could worsen with age. ¹⁷ Individuals of lower socioeconomic status have almost universally been found to have worse health outcomes; our results also indicate larger income-based disparities in ED visits due to alcohol than in rates of all-cause ED visits, especially in older adults—a finding that requires further investigation.

We found that rates of ED visits due to alcohol were higher in rural compared to urban regions of Ontario. These observations are consistent with surveys of drinking behaviour, which have found that high-risk drinking is more prevalent in residents of rural areas, particularly youth, see Online Appendix C for further details. ^19,20 While there is little research into potential contributing factors, they are likely influenced by differences in demographic makeup and socioeconomic factors. There is also evidence demonstrating distinct drinking cultures across urban and rural Canada, with rural school-aged youth reporting greater alcohol-impaired driving and heavy drinking than their urban counterparts. ^21,22 Further, important considerations are how government policies, such as alcohol pricing and the availability of alcohol stores, may promote harmful drinking in rural settings. ^23,24 Policies that increase the price of alcohol have been shown to reduce alcohol use and harms, particularly for high-risk drinkers. ²³ Alcohol policies in Ontario maintain the same alcohol pricing in rural and urban regions ²⁵ while other household commodities, which are typically not price-controlled, are more expensive in rural and northern areas of Ontario compared to urban regions. ²⁶ These policies may both make alcohol more affordable (e.g., less expensive) in rural regions than it would be in their absence and create a relative price difference between alcohol and other goods making alcohol purchases feel more affordable and appealing. Greater access to off-premise stores (i.e. liquor stores) has previously been shown to increase alcohol use and consequent harms. ²⁴ Public health units (PHUs) in rural Ontario have over 2.5 times more off-premise alcohol stores per capita than urban and mainly urban PHUs (4.00 vs. 1.50 stores per 10,000 individuals in 2014), which may be contributing to urban–rural disparities. ²⁶

To our knowledge, this is the first study to examine changes over time in health-care visits due to alcohol in rural regions of Canada. We observed a greater increase in alcohol harms in rural compared to urban Ontario, which is consistent with data showing that high-risk drinking in rural regions has increased significantly over time while remaining steady in urban regions. ^19,20 The cause of this increasing disparity in Canada is unknown but mirrors a number of studies from the United States showing a widening urban–rural gap in mortality related to suicides, drug poisoning, and alcohol liver disease collectively known as “deaths of despair.” ²⁷ Increasing deaths of despair in the rural United States have been attributed to a growing socioeconomic gulf, including growing poverty and unemployment and lack of social opportunity. ⁷ Available information supports that similar patterns may be occurring in rural Ontario where rural men have experienced almost no increase in wages between 2001 and 2015 and rural regions have seen greater declines in labour force participation than urban regions. ²⁸

Our findings highlight that further health service and policy interventions are required to reduce rising alcohol-related harms in Ontario. Screening for problematic alcohol use and offering counselling and referral for treatment in high-risk patients is an effective but underutilized intervention to reduce alcohol harms. ^29,30,31 Our study has identified several groups that may benefit from improved access to screening and addiction service, including women living in low-income neighbourhoods who now experience higher rates of harms than men living in high-income neighbourhoods; younger individuals in rural and remote settings, where harms are rapidly increasing; and older men in lower-income neighbourhoods who have the greatest overall burden of harms. Importantly, concerted effort should be made to increase access to these screening interventions specifically for high-risk populations to avoid inadvertently exacerbating health inequities. Additionally, a number of alcohol control policies including minimum unit pricing, restricting the availability of alcohol sales, and limiting alcohol promotion have been found to be highly effective at reducing problematic alcohol use and consideration should be given to strengthening these policies. ²⁴

Limitations

Our study has several limitations. First, we used an area-based measure to characterize income. Consequently, we cannot determine whether our observations are due to (a) living in a low-income neighbourhood, (b) individuals themselves having a low income, or (c) a combination or interaction between these 2 factors. Second, our study uses an ecological design that limits inferences about individual-level risk factors. Third, our inclusion criteria (valid Ontario Health Insurance Plan [OHIP] and home address) likely excluded a portion of individual experiencing homelessness or precarious housing. Research has found that individuals experiencing homelessness are more likely to live in urban settings and have problematic alcohol use; exclusion of these individuals may have underestimated the burden of ED visits in urban settings. ³² Fourth, the diagnoses included in our main outcome measure likely represented the “tip of the iceberg” in alcohol harms, with other types of direct harms (e.g., cancers related to alcohol) and indirect alcohol harms (e.g., violence) not being captured. Finally, greater rates of ED visits due to alcohol in rural compared to urban regions may be partially explained by different patterns of health-care utilization across geographic settings. ³³ However, these patterns would be unlikely to change over time and would not explain the observed widening disparity of ED visits due to alcohol.

Conclusion

We found that that current patterns and temporal trends in ED visits due to alcohol show large disparities between urban and rural regions of Canada and by socioeconomic status. These disparities also vary considerably in different age groups and between men and women. Overall, the burden of alcohol on the health system in Canada appears to be increasing particularly in rural regions. Future studies examining potential drivers of these trends are warranted to better inform clinical services and policy interventions. In the interim, clinicians should consider increasing screening and counselling for patients who have been identified as part of the high-risk groups. In addition, policy makers should increase evidence-based interventions aimed at reducing alcohol use and harms at the population level.