Gender-based Analysis of the Correlates of Hazardous and Binge Drinking among Canadian Active Duty Military Personnel

Abstract

Alcohol misuse is prevalent among military personnel and has been associated with increased risk of psychosocial problems and physical health issues. Research suggests that military men and women exhibit distinct drinking patterns and differ in terms of alcohol-related problems. However, very little information exits on the gender-specific factors related to alcohol misuse. Information on the environment, such as rurality of serving areas, related to risky drinking among military populations is also scarce. This study examined demographic, military, psychosocial, geographical, and health indicators related to alcohol misuse among a representative sample of active-duty Canadian military members who reported drinking alcohol. Logistic regressions were used to assess the correlates of hazardous drinking and binge drinking among women and men separately. Among women, being younger, single, and experiencing mental health issues were related to risky drinking. Among men, being younger, in the Navy, self-reporting symptoms of psychological distress, and having difficulty coping with anger were related to hazardous drinking, whereas, binge drinking was related to younger age, being single, rurality of serving area, anger management, and poor overall health. Results provide support for gender-specific prevention and intervention strategies and highlight the importance of tailoring programs and policies to the needs of at-risk geographical areas.

Keywords:

• Military personnel
• alcohol misuse
• mental health
• rurality
• gender-based analysis
• logistic regression
• binge drinking
• hazardous drinking

Background

Alcohol misuse among military personnel is a key public health concern (Ames & Cunradi, 2004; Henderson et al., 2009; Institute of Medicine, 2015) and an important contributory factor to morbidity and mortality (Ferrari et al., 2014; Smyth et al., 2015; Wilsnack et al., 2018; Wood et al., 2018; Xi et al., 2017), leading to organizational and occupational productivity issues (Bray et al., 2014; Fisher et al., 2000; Thørrisen et al., 2019).

Alcohol misuse has been associated with numerous individual-level factors including mental health disorders (Ames & Cunradi, 2004; Kline et al., 2014; Marshall et al., 2012; Richer et al., 2016), poor physical health (Bray et al., 2009; Waller et al., 2015), and injuries (Howland et al., 2007). Additionally, alcohol misuse has been associated with socio-demographic and military characteristics, namely, age, marital status, education, service element, rank, greater cumulative time spent on deployment, as well as having been exposed to combat while on deployment (Ferrier-Auerbach et al., 2009; Henderson et al., 2009; Jacobson et al., 2008; Richer et al., 2016; Spera et al., 2011; Stahre et al., 2009; Waller et al., 2015).

A characteristic that has consistently demonstrated differences in alcohol-related consumption and related-effects is sex and gender. In general, compared to men, more women abstain from alcohol, and those that do drink, tend to consume less alcohol and demonstrate lower prevalence of alcohol use disorder (Erol & Karpyak, 2015; World Health Organization, 2018). While male drinkers tend to have more alcohol-related problems than their female counterparts (World Health Organization, 2018), women who drink excessively have a higher risk of developing alcohol-related morbidity, mortality, and social problems (Erol & Karpyak, 2015). Similar results were found among the military population where men were more likely than women to drink heavily and to report alcohol-related problems; however, despite reporting lower rates of alcohol usage, women drinkers showed equivalent or higher rates of deleterious effects such as productivity loss and dependence symptoms (Brown et al., 2010). Additionally, a study examining motives underlying alcohol consumption among military women suggests that their primary motive is to forget problems and improve mood (Jeffery & Mattiko, 2016), implying that military women may be using alcohol as a negative coping mechanism.

Overall, previous research highlights the need to better understand military women’s patterns of alcohol use and related factors to design gender-tailored prevention strategies and effective interventions for alcohol misuse. To our knowledge, very little research has focused on identifying patterns of alcohol consumption and the factors related to alcohol misuse among military women. Further, when examining alcohol-related behaviors, correlates, and consequences among military members, results showing the magnitude of relationships are not readily available for women. Also, women tend to comprise a smaller proportion of military research samples, thus, it is not clear how aggregated results apply to them. Given the previously identified differences among men and women in their patterns of drinking, alcohol-related effects, and psychosocial and physiological outcomes, analyses should be conducted separately for men and women (Clayton & Tannenbaum, 2016).

When examining alcohol misuse among military personnel, a better understanding of the influence of the environment (Sudhinaraset et al., 2016), such as rurality of serving area, is critical. As a result of operational and organizational requirements, military members tend to relocate frequently during their military careers and may be posted to isolated, remote, and rural areas. Several characteristics of the rural environment, such as lack of recreational activities, social contexts and cultural attitudes toward excessive alcohol use (Vander Weg & Cai, 2012), and barriers to accessing health care (Brown et al., 2015), may increase the risk of alcohol misuse among military personnel and impede service provision for problem drinkers.

In Australia, studies examining the link between alcohol use and geographic location have consistently shown higher rates of alcohol consumption, high-risk drinking, and alcohol-related injuries within rural areas (Coomber et al., 2013; Miller et al., 2010). While some studies examining patterns of alcohol use across geographic regions in the United States (U.S.) have also shown higher rates of alcohol use and high-risk drinking among rural residents compared to their metropolitan counterparts (Borders & Booth, 2007; Lambert et al., 2008), a study has also shown lower rates of heavy drinking and binge drinking in rural areas (Jackson et al., 2006). One study of U.S. veterans showed lower rates of drinking among those living in rural areas compared to those in urban and suburban areas; however, those rural dwellers who did drink showed higher rates of risky alcohol use (Vander Weg & Cai, 2012). These conflicting results may be due to oversimplified dichotomizations (rural versus urban) and the lack of a standardized definition of the continuum of rurality-urbanity (Dixon & Chartier, 2016). As of now, little information is available on the influence of rurality of serving area on variations in patterns of alcohol use among military personnel. Given the large proportion of Canadian military members serving in rural or remote locations at various times throughout their military careers, it is critical to better understand the contribution and effect of rurality on alcohol misuse among active-duty military personnel.

This study aimed to determine the gender-specific individual correlates of alcohol misuse and the relationship between rurality of serving area and alcohol-related behaviors among a representative sample of active-duty military personnel. As previously mentioned, men and women exhibit different drinking patterns which are related to distinct problems and potentially unique support and treatment needs. Thus, the correlates of alcohol misuse were examined separately for men and women. Additionally, previous research suggests that different drinking patterns are associated with different psychosocial profiles (Richer et al., 2016) and alcohol-related consequences (Kraus et al., 2009), with hazardous drinking leading to chronic health conditions and binge drinking increasing risk of accidents and injuries (Ridolfo & Stevenson, 2001). As such, the current study includes two outcomes of alcohol misuse – an indicator of binge drinking and a sex-specific indicator of hazardous drinking – to examine differences in the relationships between correlates across drinking patterns.

Methods

Data for the study were collected through the 2013/2014 Health and Lifestyle Information Survey (HLIS; Thériault et al., 2016), a cross-sectional population-based health survey. The questionnaire was sent to a random sample of 4,314 Canadian Armed Forces (CAF) Regular Force personnel – stratified by gender, age, and rank – selected from the 2013 population of 56,574 eligible CAF personnel. The response rate was 60%, and a total of 2,499 observations were available for the analyses. The survey was approved by an independent human research ethics review board.

Measures and variables

Alcohol misuse

Hazardous drinking was assessed with the 10-item Alcohol Use Disorders Identification Test (AUDIT; Babor et al., 1992; Bohn et al., 1995). The scale examines the frequency and amount of alcohol use, dependence symptoms, and alcohol-related problems in the past 12 months. Each item is rated from zero to four and is summed in a total score (maximum score of 40) with higher values indicating more problematic alcohol consumption. A cut-off score of 8, based on the total score, is recommended to identify hazardous or harmful drinking, which is indicative of chronic alcohol misuse. A slightly lower cut-off point (cut-off score of 7) is recommended for specific populations, namely women (Babor et al., 2001). This scale has demonstrated good validity and efficiency at identifying problematic alcohol use in multiple populations (de Menses-Gaya et al., 2009) and research supports the use of the AUDIT cut-off score of 8 in the military population (Searle et al., 2015). The sex-specific cut-off score was used for the purpose of the study analysis.

Binge drinking or episodic heavy drinking, was measured with the AUDIT item “How often do you have six or more drinks on one occasion?” on a 5-point scale including 0 (never), 1 (less than monthly), 2 (monthly), 3 (weekly), and 4 (daily or almost daily). Participants were considered binge drinkers if they had reported binge drinking monthly or more.

Mental health

Depression symptoms were assessed with the Patient Health Questionnaire-2 (PHQ-2; Kroenke et al., 2003) which includes the core symptoms of major depression disorders, namely anhedonia and feeling down, depressed, or hopeless. Respondents reported how frequently they experienced each of these two symptoms in the previous two weeks. To better describe mental health issues in the study sample, participants were categorised as “depressed” or “not depressed” using the validated cut-off score of 3 (Kroenke et al., 2003). The total score was used in the current analyses with higher scores indicating greater depression symptoms.

Past-month symptoms of posttraumatic stress disorder (PTSD) were assessed using the primary care posttraumatic stress disorder screen (PC-PTSD), which consists of a list of four symptoms with yes (1) or no (0) response options. A validated cut-off score of 3 indicates that respondents need further PTSD evaluation (Prins et al., 2003). The scale total score was used for the current analyses with higher scores indicating greater PTSD symptoms.

Psychological distress was examined using the Kessler Screening Scale for Psychological Distress (K6; Kessler et al., 2003). The K6 assesses the frequency of six psychological symptoms in the past 30 days: nervousness; hopelessness; restlessness; depressive mood; worthlessness; and perceived effort of daily activities on a 5-point Likert scale with response options varying from “none of the time” to “all the time” Participants were classified as either “probably” or “probably not” suffering from a serious mental health issue using validated cut-off scores of 13 or greater. The cut-off has a good specificity (96%), but a low sensitivity (36%), which may lead to respondents experiencing moderate mental health issues being misclassified (Kessler et al., 2003). Given that the purpose of the proposed analysis was to verify the relationship between psychological distress and alcohol misuse, the K6 total score was used with higher total scores indicating greater levels of psychological distress. The total score was previously used as an indicator of mood and anxiety issues (Furukawa et al., 2003).

Anger and stress management were assessed with two items - developed specifically for the HLIS “In general, how would you rate your ability to manage your anger/stress?” Respondents were asked to answer on a 5-point scale ranging from: excellent to poor. To ensure adequate distribution, the scales were categorized as excellent/very good, good, fair/poor.

Overall health

Self-rated health is considered an excellent proxy of overall health and well-being and is one of the best indicators of population health (Sargent-Cox et al., 2010). Self-rated health was measured with the item “In general, would you say your health is (excellent, very good, good, fair, or poor). By health, we mean not only the absence of disease or injury but also physical, mental and social well-being.” The scale was categorised as excellent/very good, good, fair/poor. Recent acute injuries were measured with a dichotomous item “In the past 12 months, have you had an acute injury serious enough to limit your normal activities?”

Rurality of serving area

The geographic location and level of rurality-urbanity of the serving area was used as a proxy to categorize participants’ rurality of community context. A Geographic Information System (GIS) mapping was used to identify the geographic location of bases/wings/units. Rurality was determined using the size and the density of population centers (Statistics Canada, 2016a) and the census subdivisions indicating the degree of metropolitan influence on a region (Statistics Canada, 2016b). Area of serving bases/wings/units were divided into three categories: 1) urban; 2) suburban; and 3) semi-rural, rural and remote.

Demographics and military characteristics

Several demographic and military characteristics were included in the analyses: age (categorised as: 18-29; 30-44; 45-60), level of education (categorised as: completed some or completed secondary/high school education; completed some or completed college or university courses; completed university degree or graduate degree), first official language (French or English), marital status (married/common law relationship; divorce/separated/widow; single), rank (junior non-commissioned member [NCM]; senior NCM; officer), element (Air Force; Navy; Army), and having been deployed in the past two years (yes/no). Respondents were asked to report their sex (male vs female). This study examined the association between psychosocial factors and alcohol misuse, which may be influenced by gender identity and norms. As suggested by Clayton and Tannenbaum (2016), we used the terminology related to gender identity (men vs women) to report our findings, but we recognize the limitations of this approach.

Data analysis strategy

Analyses were conducted on a sub-sample of alcohol drinkers. Evidence suggests that alcohol abstainers present a different profile compared to light or moderate alcohol users including previous heavy drinking, higher psychological distress, and lower social well-being (Goodwin et al., 2017; Lucas et al., 2010). Based on evidence from the literature, purposeful bivariate logistic regression analyses were used to explore the relationship between alcohol misuse and each of the following variables: indicators of overall self-rated health, acute injuries, mental health (i.e., posttraumatic stress symptoms, depression, psychological distress, ability to manage stress and anger), military characteristics (i.e., service element, rank, and having been deployed in the past two years), sociodemohic characteristics (i.e., age, marital status, education level, and first official language) and rurality of the serving base or wing. The variable selection was based on a combination of p-value, clinical importance, and support from the literature. The variables which were associated with hazardous drinking and binge drinking in the unadjusted models (p < .25) were then included in the initial multivariable logistic regression models. Weighted prevalence estimates with 95% confidence intervals were calculated for correlates and outcomes. Analyses were conducted using the svyset package in Stata (StatsCorp, 2015) to account for the complex sampling design.

Multivariable logistic regressions with the Newton-Raphson optimization technique and first-order Taylor series linear method were used to assess the independent association of each correlate with alcohol misuse (i.e., hazardous drinking and binge drinking). Two separate sets of multivariable models were run for the two alcohol use variables, one for men and one for women. Multicollinearity was tested using the variance inflation factor and tolerance values and there was no indication of multicollinearity among the study variables. The final logistic regression models were derived using backward elimination process until the final model included only factors that were significantly associated with risky drinking behaviors (p ≤ .05). The linearity assumption between continuous variables and the log of the outcome were examined through the Tidwell test (Box & Tidwell, 1962), and corresponding transformations were conducted when needed. Data analyses were conducted using IBM SPSS Statistics (Version 25; IBM Corp, 2017) and Stata (Version 14; StataCorp, 2015).

Results

The estimated proportion of CAF members who drank at least once in the past 12 months was 94.3% (95% CI [93.4, 95.1]). More than half of respondents were middle-aged, Anglophone, married men, residing in an urban area. Population estimates of drinkers’ characteristics are presented in Tables 1 and 2.

Table 1. Socio-demographic and military characteristics of Canadian Armed Forces drinkers.

	Men (Unweighted n = 1,334; Weighted n = 45,907)		Women (Unweighted n = 980; Weighted n = 6,928)		Overall (Unweighted n = 2,314; Weighted n = 52,835)
Variables	Weighted n (%)	95% CI	Weighted n (%)	95% CI	Weighted n (%)	95% CI
Demographic characteristics
Age
18–29	13,494 (29.4)	29.4-29.4	1,720 (24.8)	24.8–24.8	15,213 (28.8)	28.8–28.8
30–44	22,696 (49.4)	47.9–50.9	3,809 (55.0)	52.9–57.1	26,505 (50.2)	48.8–51.5
45–60	9,717 (21.2)	19.7–22.7	1,399 (20.2)	18.2–22.4	11,117 (21.0)	19.7–22.4
First official language
French	12,977 (28.3)	25.5–31.3	1,889 (27.3)	24.1–30.6	14,866 (28.2)	25.7–30.8
English	32,808 (71.7)	68.7–74.5	5,036 (72.7)	69.4–75.8	37,844 (71.8)	69.2–74.3
Marital status
Married or common law relationship	31,958 (69.8)	66.9–72.7	4,267 (61.8)	58.3–65.2	36,225 (68.8)	66.2–71.3
Divorced/separated/widowed	3,855 (8.4)	6.8–10.4	1,010 (14.6)	12.3–17.3	4,865 (9.2)	7.8–11.0
Single	9,945 (21.7)	19.3–24.3	1,626 (23.6)	20.8–26.5	11,572 (22.0)	19.8–24.2
Highest level of education
Completed or completed some secondary education	16,155 (35.8)	32.8–38.9	1,433 (21.0)	18.2–24.2	17,588 (33.8)	31.2–36.6
Completed College or completed some university education	18,160 (40.2)	37.1–43.3	2,919 (42.9)	39.7–46.1	21,079 (40.6)	37.9–43.3
Completed post-secondary education	10,846 (24.0)	22.8–25.3	2,455 (36.1)	34.2–37.9	13,301 (25.6)	24.5–26.7
Geographical location
Remote/rural/semi-rural	12,691 (29.5)	26.5–32.7	1,564 (24.0)	20.9–27.4	14,255 (28.8)	26.2–31.6
Suburban	4,895 (11.4)	9.4–13.7	1,101 (16.9)	14.2–20.0	5,995 (12.1)	10.4–14.1
Urban	25,400 (59.1)	55.8–62.3	3,844 (59.1)	55.3–62.7	29,243 (59.1)	56.2–61.9
Military characteristics
Rank
Junior NCM	23,262 (50.7)	48.4–53.0	3,249 (46.9)	44.1–49.8	26,511 (50.2)	48.1–52.2
Senior NCM	11,429 (24.9)	22.7–27.3	1,585 (22.9)	20.1–25.9	13,014 (24.6)	22.6–26.7
Officer	11,216 (24.4)	24.4–24.4	2,094 (30.2)	30.2–30.2	13,310 (25.2)	25.2–25.2
Element
Air force	13,708 (30.1)	27.3–33.1	2,729 (39.6)	36.2–43.2	16,437 (31.4)	28.9–34.0
Navy	8,594 (18.9)	16.5–21.5	1,127 (16.4)	13.9–19.2	9,722 (18.6)	16.4–20.9
Army	23,200 (51.0)	47.8–54.2	3,026 (44.0)	40.4–47.6	26,226 (50.1)	47.3–52.9
Has been deployed in the past 2 years	6,983 (15.3)	13.1–17.7	1,043 (15.1)	12.6–17.9	8,025 (15.2)	13.3–17.4

Table 2. Mental and physical health of Canadian Armed Forces drinkers.

Variables	Men (Unweighted n = 1,334; Weighted n = 45,907)		Women (Unweighted n = 980; Weighted n = 6,928)		Overall (Unweighted n = 2,314; Weighted n=52,835)
	Weighted n (%)	95% CI	Weighted n (%)	95% CI	Weighted n (%)	95% CI
Mental health
Psychological distress (Kessler 6) probable serious mental illness	1,040 (2.3)	1.5–3.6	331.6 (4.9)	3.5–6.8	1,372 (2.7)	1.9–3.8
Posttraumatic stress Symptoms (further PTSD evaluation required)	4,751 (10.6)	8.7–12.9	792 (11.7)	9.5–14.4	5,542 (10.7)	9.0–12.7
Depressed (PHQ-2)	3,320 (7.4)	5.7–9.4	600 (8.9)	7.01–11.3	3,920 (7.6)	6.1–9.3
Ability to manage anger
Excellent or very good	27,693 (60.7)	57.4–63.8	4,571 (66.5)	62.9–69.8	32,254 (61.4)	58.6–64.2
Good	11,470 (25.1)	22.4–28.1	1,645 (23.9)	20.9–27.2	13,115 (25.0)	22.6–27.5
Fair or poor	6,484 (14.2)	12.0–16.7	661 (9.6)	7.7–11.9	7,145 (13.6)	11.7–15.8
Ability to manage stress
Excellent or very good	26,331 (57.6)	54.3–60.7	3,529 (51.3)	47.7–54.9	29,860 (56.7)	53.9–59.5
Good	13,616 (29.8)	26.9–32.8	2,323 (33.8)	30.4–37.3	15,939 (30.3)	27.7–32.9
Fair or poor	5,802 (12.7)	10.6–15.1	1,028 (14.9)	12.6–17.7	6,830 (13.0)	11.1–15.1
Physical health
Serious injury in the past year	8,874 (19.7)	17.2–22.4	1,330 (19.5)	16.8–22.6	10,204 (19.7)	17.5–22.0
Self-rated health
Excellent or very good	29,646 (65.0)	61.9–68.0	4,821 (69.8)	66.3–73.0	29,646 (65.0)	61.9–68.0
Good	13,249 (29.1)	26.2–32.1	1,549 (22.4)	19.5–25.6	13,249 (29.1)	26.2–32.0
Fair or poor	2,678 (5.9)	4.5–7.6	540 (7.8)	6.0–10.1	2,678 (5.9)	4.5–7.6
Alcohol misuse
Hazardous drinking^1 (AUDIT)	9,445 (22.4)	19.7–25.4	860 (13.5)	11.2–16.1	10,305 (21.2)	18.9–23.8
Monthly binge drinking^2	10,273 (24.2)	21.5–27.2	691 (10.8)	8.7–13.4	10,964 (22.5)	20.1–25.1
Mental health indicators (continuous)	Range	Mean (SE)	Range	Mean (SE)	Range	Mean (SE)
Posttraumatic stress symptoms total (PC–PTSD)	0–4	0.58 (0.04)	0–4	0.69 (0.05)	0–4	0.59 (0.03)
Depression (PHQ-2) score	0–6	0.78 (0.04)	0–6	0.80 (0.05)	0–6	0.78 (0.03)
Psychological distress (K6) score	0–22	2.9 (0.12)	0–24	3.38 (0.16)	0–24	2.98 (0.10)

Note: ¹Hazardous drinking refers to the AUDIT sex-specific cut-off (8 for male and 7 for female). ²Consuming six drinks or more in the same occasion; SE = standard error.

The first outcome examined was hazardous drinking among women. Non-linearity was found for the variable of psychological distress, and this continuous variable was transformed as its inverse (1/(total score +1)) which fixed the linearity issue. Results of the multivariable logistic regression indicated that the odds of hazardous drinking are greater for younger women aged 18 to 29 (adjusted odds ratio [AOR] = 3.92; 95% CI [1.70, 9.04]) compared to older women aged 45 to 60, and the odds were higher for single women (AOR = 1.73; 95% CI [1.05, 2.85]) versus married women or women in common law relationships. Additionally, the odds of hazardous drinking decreased significantly with lower scores of self-rated psychological distress (AOR = 0.16; 95% CI [0.07, 0.39]). With respect to binge drinking among women the final multivariable model suggests that the odds of binge drinking were greater for younger women (AOR = 3.35; 95% CI [1.39, 8.08]) and single women (AOR = 1.78; 95% CI [1.02, 3.08]). Results also indicated a relationship between binge drinking and mental health. Women self-reporting less psychological distress were less likely to binge drink (AOR = 0.34; 95% CI [0.14, 0.84]). See Table 3 for the details of the analysis.

Table 3. Logistic regression analysis of hazardous and binge drinking among Canadian Armed Forces women (Unweighted n = 980; Weighted n = 6,928)

Variables	Hazardous drinking		Monthly binge drinking
	Crude OR [95% CI]	Adjusted OR [95% CI]	Crude OR [95% CI]	Adjusted OR[95% CI]
Age
18–29	5.29** [2.50, 11.20]	3.92** [1.70, 9.04]	4.22** [1.83, 9.73]	3.35** [1.39, 8.08]
30–44	1.84 [0.83, 4.07]	1.74 [0.75, 4.02]	1.76 [0.73, 4.27]	1.66 [0.66, 4.17]
45 to 60^a
Marital status
Divorced/separated/widowed	1.49 [0.76, 2.91]	1.55 [0.75, 3.20]	1.35 [0.62, 2.94]	1.47 [0.65, 3.29]
Single	2.52** [1.59, 4.00]	1.73*[1.05, 2.85]	2.32** [1.38, 3.89]	1.78* [1.02, 3.08]
Married or common law relationship^a
Element
Navy	1.76† [0.95, 3.26]	1.70 [0.89, 3.26]	1.46 [0.70, 3,02]
Army	1.76* [1.08, 2.85]	1.63† [0.98, 2.71]	1.82* [1.06, 3.13]
Air Force^a
Deployed overseas in the past 2 years
Yes			1.72† [0.93, 3.20]
No^a
Posttraumatic stress symptoms (PC-PTSD)	1.22* [1.03, 1.44]
Psychological distress (K6) inverse score	0.16** [0.07, 0.35]	0.16** [0.07, 0.39]	0.30** [0.13, 0.72]	0.34* [0.14, 0.84]
Depression (PHQ-2)	1.27** [1.10, 1.47]		1.14 [0.95, 1.38]
Ability to manage stress
Excellent or very good	0.46** [0.26, 0.79]
Good	0.55† [0.30, 1.00]
Fair or poor^a
Having sustained a serious injury in the past 12 months	1.39 [0.84, 2.32]

Note: ^aReference category. OR = odds ratio. CI = confidence interval. † p ≤ .10 * p ≤ .05; ** p ≤ .01. The initial multivariable model examining hazardous drinking included the following covariates: age, marital status, element, symptoms of posttraumatic stress, psychological distress, depression, ability to manage stress, and past-year serious acute injuries. The initial multivariable model examining monthly binge drinking included the following covariates: age, marital status, element, deployment status, psychological distress, and depression score.

The third model examined hazardous drinking among men. The final multivariable regression model suggested that the odds of hazardous drinking are greater for younger men (aged 18 to 29; AOR = 3.30; 95% CI [1.95, 5.58]) compared to older men aged 45 to 60. Serving in the Navy (AOR = 1.94; 95% CI [1.07, 3.52]) was also associated with greater odds of hazardous drinking than serving in the Air Force. Rurality status of serving area was marginally significant. Men who reported serving in rural or remote areas tended to be at greater odds of hazardous drinking (AOR = 1.50; 95% CI [0.96, 2.35]) compared to those serving in urban areas. Regarding mental health, men who self-rated their ability to manage anger as excellent or very good were at lower odds (AOR = 0.50, 95% CI [0.29, 0.87]) of hazardous drinking compared to fair or poor anger management. Finally, the odds of hazardous drinking increased with greater psychological distress measured by the K6 scale (AOR = 1.06; 95% CI [1.00, 1.12]).

The fourth model examined binge drinking among men. In the final multivariable model, the variables that remained significantly associated with binge drinking were younger age (aged 18 to 29, AOR = 3.10; 95% CI [1.94, 5.31] and aged 30 to 44, AOR = 1.75; 95% CI [1.07, 2.84]) and being single (AOR = 2.01, 95% CI [1.32, 3.08]). Rurality of serving area was associated with greater odds of binge drinking (AOR = 1.65; 95% CI [1.11, 2.44]) compared to serving in an urban area. Self-rated excellent or very good ability to manage anger was associated with lower odds of binge drinking (AOR = 0.53; 95% CI [0.32, 0.87]) compared to poorer anger management abilities. Finally, poorer overall health was associated with greater odds of binge drinking (AOR = 2.43, 95% CI [1.09, 5.43]). See Table 4 for the details of the analysis.

Table 4. Logistic regression analysis of hazardous and binge drinking among Canadian Armed Forces men (Unweighted n = 1,334; Weighted n = 45,907)

Variable	Hazardous drinking		Monthly binge drinking
	Crude OR [95% CI]	Adjusted OR [95% CI]	Crude OR [95% CI]	Adjusted OR [95% CI]
Age
18 to 29	3.92** [2.47, 6.23]	3.30** [1.95, 5.58]	4.74** [3.02, 7.45]	3.10**[1.81, 5.31]
30 to 44	2.06** [1.31, 3.25]	1.61† [0.96, 2.72]	2.18**[1.40, 3.41]	1.75* [1.07, 2.84]
45 to 60^a
First official language
French	0.73 [0.49, 1.08]		0.64* [0.43, 0.94]
English^a
Marital status
Divorced/separated/widowed	0.98 [0.53, 1.80]		1.16 [0.64, 2.10]	1.20 [0.62, 2.33]
Single	1.81** [1.23, 2.65]		2.92** [2.02, 4.21]	2.01** [1.32, 3.08]
Married or common law relationship^a
Highest level of education
Completed some or completed secondary education	1.55* [1.07, 2.26]		1.50* [1.04, 2.15]
Completed College or completed some university education	1.38† [0.96, 1.98]		1.41† [0.99, 2.00]
Completed post-secondary education^a
Geographical location
Remote/rural/semi-rural	1.52* [1.04, 2.21]	1.50† [0.96, 2.35]	1.89** [1.31, 2.72]	1.65** [1.11, 2.44]
Suburban	1.22 [0.70, 2.11]	1.57 [0.85, 2.91]	1.58† [0.93, 2.67]	1.70† [0.96, 3.00]
Urban^a
Element
Navy	1.48† [0.91, 2.38]	1.94* [1.07, 3.52]	1.19 [0.75, 1.87]
Army	1.27 [0.86, 1.87]	1.33 [0.84, 2.10]	1.05 [0.73, 1.52]
Air Force^a
Rank
Junior NCM	1.79** [1.30, 2.48]		1.83** [1.34, 2.50]
Senior NCM	0.95 [0.61, 1.46]		1.07 [0.71, 1.60]
Officer^a
Depression score (PHQ-2)	1.27** [1.11, 1.45]		1.12† [0.98, 1.30]
Posttraumatic stress symptoms (PC-PTSD)	1.12† [0.97, 1.29]
Psychological distress (K6 score)	1.11** [1.06, 1.16]	1.06* [1.00, 1.12]	1.05*[1.01, 1.10]
Ability to manage anger
Excellent or very good	0.35** [0.22, 0.55]	0.50* [0.29, 0.87]	0.54**[0.33, 0.86]	0.53** [0.32, 0.87]
Good	0.66† [0.40, 1.10]	0.91 [0.52, 1.62]	0.85 [0.51, 1.42]	0.89 [0.51, 1.54]
Fair or poor^a
Ability to manage stress
Excellent or very good	0.38** [0.23, 0.62]		0.45** [0.28, 0.74]
Good	0.70 [0.42, 1.17]		0.68 [0.41, 1.15]
Fair or poor^a
Overall health
Fair or poor	2.16* [1.14, 4.10]		2.23** [1.21, 4.09]	2.43* [1.09, 5.43]
Good	0.95 [0.66, 1.40]		0.71† [0.49, 1.04]	0.73 [0.48, 1.11]
Excellent or very good^a
Having sustained a serious injury in the past 12 months			0.76 [0.50, 1.14]

Note: ^aReference category. OR = odds ratio. CI = confidence interval. †p ≤ .10 * p ≤ .05. ** p ≤ .01. The initial multivariable model examining hazardous drinking included the following covariates: age, first official language, marital status, education, geographical location, element, rank, depression, posttraumatic stress symptoms, psychological distress, self-rated ability to manage stress and anger, and overall health. The initial multivariable model examining monthly binge drinking included the following covariates were: age, first official language, marital status, education, geographical location of the serving base or wing, rank, depression, psychological distress, ability to manage anger and stress, overall health, and injuries.

Discussion

The focus of the current study was to examine the individual factors related to alcohol misuse and the contribution of rurality of the serving area separately for military men and women. The results can inform alcohol-related policies and prevention and intervention strategies by identifying at-risk individuals and environments. Alcohol misuse is prevalent among military personnel (Richer et al., 2016; Whitehead & Hawes, 2010) and can be costly to both the organization, through loss of productivity (Bray et al., 2014; Thørrisen et al., 2019), and the members themselves as it can contribute to morbidity and mortality (Ferrari et al., 2014; Smyth et al., 2015; Wilsnack et al., 2018; Wood et al., 2018; Xi et al., 2017). In this study, 13.5% of women reported hazardous drinking and 10.8% reported monthly binge drinking (six or more drinks in the same occasion). Among men, the prevalence of hazardous drinking was 22.4%, with binge drinking reported by 24.2%. Different definitions of heavy drinking and binge drinking exist, making it difficult to compare prevalence estimates across populations. As a point of reference, 30.0% of U.S. active-duty service members reported binge drinking defined as drinking five or more drinks for men and four or more drinks for women in the past month (Meadows et al., 2018). A recent study conducted among the overall Canadian population, including drinkers and non-drinkers, reported binge drinking (defined as five or more drinks on one occasion at least one time per month for men and four or more for women) by 13.2% of women and 23.1% of men (Pham et al., 2020) respectively. In the current study, the binge drinking criteria was six or more drinks in a single occasion. The proportion of binge drinkers would likely increase with lower cut-offs.

The findings from the current study add to the literature examining the contribution of age and marital status to the likelihood of problematic drinking. Previous research conducted among military personnel found that younger and single military members are more likely to misuse alcohol (Stahre et al., 2009; Waller et al., 2015). Importantly, this study highlights similar odds of problem drinking among young and single military men and women. The protective effect of marriage (or committed relationship) on risky alcohol use, as seen in the current findings, has also been observed and discussed in previous studies (Jeffery & Mattiko, 2016; Poehlman et al., 2011; Waller et al., 2015). Consistent with previous research, military members in the Navy were more likely to report hazardous drinking compared to members in the Air Force possibly due to occupational restrictions (abstinence requirements before a flight), operational demands, and the social context and culture related to alcohol use (Bray & Hourani, 2007; Waller et al., 2015). Since the data were collected, multiple policies and orders have been implemented to encourage responsible provision, serving, and consumption of alcohol on ships and in garrison (Baines, 2014). Additionally, measures have been put in place to enforce policies and standards related to the serving of alcohol in the Canadian Navy. Future analyses will provide more insights on the effectiveness of these policies and measures in reducing problematic drinking among CAF Navy personnel.

Findings suggest that rurality of the serving area is associated with binge drinking among men, but not women. These findings are partly in line with studies conducted among veterans and civilians across numerous countries (Borders & Booth, 2007; Lambert et al., 2008; Miller et al., 2010; Vander Weg & Cai, 2012). The differences observed between men and women may be explained by the motivations underlying heavy drinking. The motives for binge drinking in rural locations may include boredom and lack of recreational activities. Our study results may reflect gender differences in the motivations underlying problem drinking. Men may be more prone to risk-taking and sensation seeking (Breivik et al., 2019) and turning to alcohol to socialize, increase cohesion and camaraderie, and alleviate boredom, whereas women may be more likely on average to use alcohol to cope with stress and negative emotions (Jeffery & Mattiko, 2016). Community-based factors can also lead to differing patterns of alcohol use, including availability of alcohol, enforcement of laws related to alcohol consumption, socioeconomic status of the area, and community social capital (Dixon & Chartier, 2016). Regardless of the underlying cause, the current results suggest that consideration of the rurality of military members’ postings is important when developing prevention and intervention tools. Importantly, mental health and addiction services may be less accessible in these locations (Brown et al., 2015). This, and research suggesting that military personnel engaging in riskier drinking tend to underestimate the need for treatment and are less likely to seek mental health care (Fink et al., 2015), suggest it is imperative to enhance and tailor addiction programs in these specific locations and ensure that they are readily accessible.

Findings also support the comorbidity between alcohol misuse and mental health symptoms, a finding that is consistent with previous research examining civilian and military populations (Richer et al., 2016; Tailieu et al., 2020). Authors have proposed the self-medication hypothesis (Khantzian, 1999) to explain the relationship between problem drinking and mental health symptoms. They suggest that alcohol is used to cope with negative emotions, military stress, and trauma (Schumm & Chard, 2012). Furthermore, in a study conducted only among active-duty women, it was found that the primary motivations for using alcohol were to forget problems and improve mood (Jeffery & Mattiko, 2016). The current study also suggests that, among men, self-rated excellent or very good ability to manage anger was inversely associated with risky drinking, consistent with previous research showing a strong relationship between alcohol misuse and anger (Bray & Hourani, 2007; Coccaro et al., 2016), and suggesting that alcohol may be used as a maladaptive coping strategy to manage negative emotions among men.

Poor overall health was associated with binge drinking among men, but not women. It is not clear whether this finding reflects the health impact of excessive drinking or usage of alcohol to cope with health-related issues. The gender differences may be explained by diverging subjective assessment of fair/poor overall health or different levels of disability and functioning limitations between men and women.

The relationship between having sustained a serious injury and alcohol misuse was not significant for either men or women. This result may be due to the infrequent occurrence of serious injuries and inability to observe an association with a smaller effect size. Prevention and intervention programs focusing on alcohol misuse should be tailored to young, single military personnel, but should also be developed in such a way to adapt to the different genders in different geographic and social contexts. Early and synchronized interventions should be provided to military women exhibiting mental health symptoms and problematic alcohol use. Early detection and brief interventions in primary care can encourage problematic drinkers to seek treatment and are recommended to prevent severity and chronicity of problematic alcohol use (O’Donnell et al., 2014). Recent research has shown the feasibility of using a brief screening tool, the AUDIT-C, and the alcohol brief intervention in the context of routine dental inspections to detect and intervene upon military members at increased risk of alcohol-related issues (Dermont et al., 2020).

Furthermore, military organizations must ensure additional treatment availability and accessibility in rural and semi-rural areas. Previous research shows that individuals living in rural contexts face more barriers to accessing mental health services (Jackson et al., 2006). The availability of health care provision has been found to be suboptimal in rural and remote areas given geographic, logistic, and structural issues (Brown et al., 2015; Fitzpatrick et al., 2017). The availability of military care targeting addictions with mental health comorbidity may not be dependable in some locations, forcing service members to rely on external community services. Ideally, more military-specific supports and support personnel would be provided to these under-served communities, however the number of these rural locations and their remoteness does not lend itself well to the huge number of resources required to do so. That said, service provision of alcohol-related treatment through Telehealth is a promising alternative to mitigate accessibility issues (Mitchell et al., 2019). Indeed, the recent COVID-19 pandemic has put a spotlight on Telehealth services and availability. A rapid review of the literature examining digital tools and services to deliver healthcare virtually demonstrated the significant advances in this area as a result of the pandemic and the various considerations when adopting a Telehealth system (Bokolo, 2020).

This study was conducted with a representative sample of the 2013-2014 CAF population and the findings can be generalized to the Canadian military population who consume alcohol. Generalizability of findings to other military populations with similar characteristics and culture may be possible. Some limitations must be considered when interpreting the study findings. The study design is cross-sectional; therefore, causality and directionality cannot be inferred. It is not clear whether serving in a rural environment directly impacts binge drinking or whether alcohol misuse leads to mental health issues and vice versa. Further prospective analyses and additional criteria are needed to determine causation. Also, social desirability bias may be a limitation, as the self-reported nature of measures used to assess alcohol consumption and mental health symptoms in the study may have led to underreporting. The rurality of the serving area is a proxy of respondents’ place of residence. It is possible that some participants are commuting for work and the level of rurality of serving area does not correspond to their living area, which may have weakened the strength of the association between alcohol misuse and rurality. Also, the geographical location of serving base or wing was categorised based on three groups: remote/rural/semi-rural, suburban, and urban. This categorization was an attempt to address the oversimplified dichotomy of rural vs urban areas. The remote/semi-rural/rural categories were amalgamated due to the low number of respondents in each category. This limits our capacity to examine more nuanced relationships between rurality of serving area and alcohol misuse and may have impacted findings given the variability between remote and semi-rural areas.

The measure of gender identity was not included in the HLIS 2013/2014 survey design. As the current study examined psychosocial factors – which may be influenced by gender identity and norms – we used the concept of gender to categorize military men and women using information derived from the variable sex (male vs female). This may have led to misclassification of a small proportion of respondents. Future research should include more complex measures of gender identity including binary, transgender, and non-binary gender identities (e.g., genderqueer, gender variant, gender nonconforming). Finally, the data for the current study do not use the latest source of data regarding alcohol use in the CAF which comes from the 2019 CAF Health Survey. Analyses of the newer data will provide opportunity to compare results and verify whether relationships between individual and environmental factors and alcohol misuse maintain over time.

Conclusion

This study makes an important contribution to understanding the individual and environmental factors related to problem drinking among military men and women. The findings suggest the psychosocial development stage of young adulthood and the social contexts related to single life seem to contribute to a lifestyle involving alcohol misuse. For women, experiencing psychological distress seems to be associated with risky alcohol use whereas, for men, alcohol misuse seems is to be associated with the rurality of the serving area, anger management, and poor self-reported overall health. Results provide support for gender-specific prevention and intervention strategies, given the potential differences in underlying motives for misusing alcohol. Findings highlight the need to tailor alcohol-related prevention and intervention strategies and policies to at-risk areas, such as military bases and wings located in rural and semi-rural environments. Future research should examine the impact of rurality on social contexts and norms promoting risky drinking among military personnel more in depth.

Acknowledgments

The authors would like to acknowledge François Thériault and Barbara Strauss from the Epidemiology Section at the Directorate of Force Health Protection, Canadian Forces Health Services Group, who conducted the design, data collection, data cleaning and preparation, and top line analyses of the 2013/14 Health and Lifestyle Information Survey.

Disclosure statement

The authors have no conflicts of interest to disclose.

Additional information

Funding

Funding for this work was provided by the Government of Canada.