Military veterans’ drinking behaviour: how segmentation can inform intervention development

Military veterans are often regarded as being at high risk of the deleterious effects of alcohol consumption (Schumm and Chard, 2012). In particular, the culture of using alcohol as a means of coping with trauma in the military is well known (Jones and Fear, 2011) and has been shown to have a negative impact on veterans’ alcohol use after leaving the military (Head et al., 2016). What is less well known is how experiences in the military, attitudes towards support services offered, as well as emotional regulation strategies employed may impact veterans’ harmful use of consumption. This research draws on a cohort of military veterans from Aotearoa New Zealand to identify statistically discrete segments based on alcohol consumption patterns. The aim of this research is to better understand what impact, if any, experience in the military (including drinking culture), attitudes towards support services offered, and emotional regulation strategies have on harmful alcohol consumption behaviours after leaving the military. From here, development of targeted interventions based on the needs of military personnel can be undertaken with a focus on the key needs for each group to mitigate alcohol related harm.

Alcohol misuse among military personnel and veterans is a significant public health concern with far-reaching consequences. The unique stressors associated with military service, including exposure to traumatic events, frequent deployments and the challenges of transition into non-military life, can contribute to increased alcohol consumption and the development of alcohol use disorders in this population (Capone et al., 2013; Jacobson et al., 2008; Kehle et al., 2012; Kelley et al., 2013). Excessive alcohol use not only jeopardises the health and well-being of service members and veterans but also has broader implications for their families, communities and the military as an institution (Hitch et al., 2023; Tinney and Gerlock, 2014). We look to explore the issue of health impacts from alcohol harm further by exploring what factors are specific to the military cohort being studied here. A social marketing led segmentation approach achieves this by firstly assessing whether different groups within a sample behave differently and then what other factors are statistically associated with these differences in behaviour. The outcome of this process is to ascertain what specific needs are evident for different groups within a sample set and lead more effective intervention development for them in the future.

Numerous studies have documented elevated rates of alcohol misuse among military personnel and veterans compared to their civilian counterparts. For instance, Jacobson et al. (2008) found that the prevalence of binge drinking increased significantly after military deployment, with 25.6% of active-duty personnel engaging in binge drinking after returning from combat zones, compared to 11.8% before deployment. Similarly, Cucciare et al. (2011) reported that nearly half of the veterans receiving a brief alcohol intervention met criteria for binge drinking.

Several factors have been identified as contributing to the elevated risk of alcohol misuse among military personnel and veterans. Exposure to combat trauma and the associated risk of developing post traumatic stress disorder (PTSD) is a well-documented risk factor (Jacobson et al., 2008; Kehle et al., 2012). In addition, the military culture, which often normalises and even encourages heavy drinking as a means of bonding and coping, can perpetuate alcohol misuse (Meadows et al., 2023). Other risk factors include younger age, male gender, lower educational attainment and a lack of social support (Capone et al., 2013; Cucciare et al., 2011; Kelley et al., 2013).

Conversely, certain protective factors have been identified that may mitigate the risk of alcohol misuse in this population. Access to mental health support services (Williamson et al., 2019) and strong social support networks, particularly those involving family and non-military peers, have been associated with lower rates of alcohol misuse (Burnett-Zeigler et al., 2011). In addition, positive coping strategies, such as seeking professional help or engaging in physical activity, can serve as protective factors against excessive alcohol consumption (Kehle et al., 2012). Understanding attitudes towards mental health services as well as understanding what cognitive regulation strategies veterans employ may help to understand if these practices correlate with harmful alcohol consumption (Bystritsky et al., 2005; Garnefski and Kraaij, 2006; Vogt et al., 2014). The following section outlines the methodology employed to collect data from the former New Zealand military personnel and carry out the segmentation analysis.

A segmentation approach was undertaken in a similar fashion to that outlined by Slater (1996) with refinements for available technology and appropriate approaches for the cohort under investigation. Recruitment was undertaken by posting an online questionnaire on social media sites popular with former military personnel and sent out via newsletters of various veteran support associations. All distribution was done in accordance with ethical standards governed by the University of Canterbury Human Ethics Committee (HREC 2023/102) and with the permission of social media administrators and association leadership. All participation was voluntary but an inducement of winning one of two iPads was offered to those who completed the questionnaire. All identifiable information was held in a separate questionnaire and never matched with participant responses. Participants were screened based on having served in the New Zealand Military in an official capacity. It was not a requirement for participants to be deployed into any active combat zone to participate in the study. Active and Reserve personnel were not eligible to participate in the study.

Harmful Alcohol Use was measure by the AUDIT-C scale which measures how often people drink, how many drinks they have in a typical day, and how often they have consumer six or more drinks in a single drinking session in the previous year (Babor et al., 2001). The test provides a score from 0 to 12 based on these questions, with a participant scoring greater than 4 for men and greater than 3 for women seen as being a positive indication for problematic alcohol use. In general, the higher a participant’s score, the more likely the participant’s alcohol use is affecting their health and safety (Babor et al., 2001).

Cultural Factors measured drinking culture in the military was measured using Meadows et al.’s (2023) 4-point scale. To understand participants’ perceptions of stigma associated with mental health services offered whilst in the military, Vogt et al.’s (2014) Endorsed and Anticipated Stigma Inventory (EASI) was employed.

The two main structural factors of interest in this study were perceived difficulties in accessing support services whilst in the military and perceived readiness for transition out of the military into non-military life. Access to support services used Williamson et al.’s (2019) 4-point Access to Mental Health Services subscale. Perceived preparation for transition was measured using Pritchard’s (2023) 5-point transition preparation scale.

Personal Factors were based on the literature and included participants’ childhood trauma Bernstein et al.’s (2003) childhood trauma questionnaire [Short-Form] (CTQ-SF). Attitudes towards support services in general was measured using Williamson et al.’s (2019) perceived stigma of mental health care/providers 9-point subscale. Finally, Garnefski and Kraaij’s (2006) 18-point cognitive emotion regulation questionnaire (CERQ-Short) was used to understand how participants are able to respond to, reframe, or reflect upon distressing or stressful life events.

Demographic information was also collected relating to age, education level, current occupation, highest military rank, gender and time spent on military deployments, if any.

All data analysis was completed in SPSS 29.0.0. A total of 387 participants began the study with 53 leaving their answers incomplete and subsequently removed. A further 6 participants completed the questionnaire, but their answers intimated an unnatural approach to completion (e.g. scoring all answers a 5/5, even when reverse coding is employed and/or completing the questionnaire in an unusually fast time). These six responses were also removed as part of sample cleaning. The final participant sample was 328. Given the relatively small size of the New Zealand Defence Force [nearly 15,000 active, reserve and civilian personnel as of June 2023 (New Zealand Defence Force, 2023)], this was deemed to be a suitable sample size to begin understanding the cohort. It should be noted that generalisation to all military or former military personnel is not validated in this study. The purpose of this study is to aid in the understanding of behaviour so that future interventions can be made more effective.

Cluster analysis was conducted to create statistically discrete segments from the data. Cluster analysis began with hierarchical cluster analysis based on participants total AUDIT score. Cluster membership was confirmed and assigned through k-means clustering. As only one variable was used to create the clusters, z-score standardisation was not required, as it would not affect the relative distances between observations or the resulting cluster solution. A Kruskal–Wallis H test was conducted to examine whether AUDIT total scores differed significantly across the three identified clusters (Under Control, Problematic and Harmful Coping). The results indicated a statistically significant difference in AUDIT scores between clusters, H(2) = 197.36, p < 0.001. The ANOVA table from the k-means clustering further showed a significant difference between the final three clusters.

To evaluate how these clusters differed in terms of the factors measures in this study, a MANCOVA was used to compare what significant differences may exist with the cluster membership as independent variable and demographics used as covariates. MANCOVA results showed the full model was significant {Wilks’ Lambda = 0.086, [F(32, 610) = 46.076, p < 0.001]}. The only covariates that showed a significant impact on the model were Age {Wilks’ Lambda = 0.700 [F(16, 305) = 8.178, p < 0.001]}, and how many months a participant was deployed {Wilks’ Lambda = 0.874, [F(16, 305) = 2.750, p < 0.001]}. Table 1 shows the between subjects’ effects for the key dependent variables in this study. Descriptive statistics reveal the key differences between each group and are also presented in Table 1. The three groups and their interpretations are described further in the following sections.

This group reported the highest AUDIT score with an average of a 11.365 / 12. This group reported drinking alcohol heavily and regularly. They would be categorised as exhibiting extremely harmful drinking behaviours. They represented the youngest group out of all three with a mean age of 40–44 years old. They also reported the highest likelihood to be deployed whilst in the military. This group is typified as believing that the drinking culture whilst in the military was very much geared towards intoxication, they ruminate negatively about stressful life events, they catastrophise stressful events, and report the highest disdain for mental health providers and support services. They also felt that these services were inaccessible. Access was measured both in terms of physically accessible (ability to gain adequate transport or time off to meet a service provider) as well as perceptions of psychological inaccessibility (perceptions that accessing these services would negatively impact their standing amongst peers). This group struggles to engage in positive reappraisal of stressful life events and are more likely to blame others for the negative impacts on their life. This group reported the highest score for having a traumatic childhood. This score could be elevated because of experiencing significant negative life events. An alternative explanation could be that their propensity to ruminate and catastrophise life events more than others may exacerbate their childhood experiences, potentially making them feel their negative life events are more prominent than the same experiences for other groups. This would need to be explored more in any interventions developed. The Problematic Drinkers group has all the hallmarks of a cohort that is struggling with their well-being, and this correlates strongly with their harmful drinking behaviour.

With a mean score for the AUDIT measure of 7.402/12 this group’s drinking was significantly lower than the Problematic cluster; however, their behaviour can still be classified as being harmful to their health and well-being. The mean age of this group was 45–49 years old and did not report being deployed as often as the Problematic Group. This group did not feel the drinking culture was prominent whilst in the military. Interestingly, their score for this factor was the lowest of all groups and perhaps indicates that the culture in the military was either perceived as being not negative or perhaps very similar to the drinking culture they experienced either before or after their time in the military. What differentiates this group from the Problematic cohort is their approach to cognitive emotion regulation appears healthier. This group is less likely to ruminate and catastrophise negative life events compared to the Problematic Drinkers group. They are also the best group at Positive Refocusing (concentrating on the positives, rather than negatives of a life event), and Positive Reappraisal (able to see how negative life events have benefitted them). They also do not perceive support services to be as inaccessible or stigmatised as the Problematic Drinkers group. They also showed a significantly more positive perception of mental health support services in general, not just the support services offered during their time in the military. This group, although engaging in potentially harmful drinking behaviours, do have a healthier outlook on their experience in the military and a more positive attitude towards help-seeking behaviours.

This group represented the largest group in the analysis and reported a mean AUDIT score of 2.662, which is significantly lower than both other groups. This represents a group that perceive their drinking to be under control and not harmful to their well-being. This group represents the oldest of the three cohorts with the average age of 60–64 years old. They also reported the lowest likelihood of being deployed whilst in the military. The size of this group in relation to other segments is in line with previous studies exploring alcohol abuse in the military (Polich, 1981). However, it should also be noted that underreporting of alcohol use by military personnel has been shown to be prevalent (Mattiko et al., 2011). This group reported the lowest perception of a drinking culture whilst in the military. This group showed the most positive attitudes towards accessing support services, attitudes towards mental health support, and the lowest levels of stigmatisation of mental health support in the military. Cognitive emotional regulation strategies were more varied and often not as positive as those in the Harmful but Coping group. That is, this group, although reporting the healthiest drinking behaviour, did not necessarily report the most positive emotional regulation techniques. This could be because of the lack of focus on negative life events or the more positive outlook on life events not necessitating the need to regulate their emotions. The Garnesfski and Kraaij (2006) scale does focus heavily on being able to recall negative life events and responses to these events. This is discussed when further comparisons between groups is made in the discussion section.

Only length of time deployed and age showed a significant association in the MANCOVA model. Post hoc correlations showed that harmful alcohol use had a significant negative correlation with age, showing that it was very much the younger participants that reported higher alcohol consumption [r(326) = −0.148, p = 0.07]. The number of months deployed showed an insignificant correlation with drinking behaviour [r(326) = 0.087, p = 0.117] but was positively correlated with propensity for Rumination [r(326) = 0.149, p = 0.07], Catastrophising [r(326) = 0.155, p = 0.05] and stigma towards support services in the military [r(326) = 0.113, p = 0.040]. It was also negatively correlated with the emotional regulation strategy of Positive Reinforcement [r(326) = −0.126, p = 0.022]. This does indicate that those who spend more time deployed do not necessarily engage in more harmful drinking behaviour but do engage in more unhealthy emotional regulation strategies and hold more unhelpful attitudes towards support services on offer. This is a key area for further investigation to aid those transitioning out of the military, as those who have experienced deployment may need more targeted support in developing healthy emotional regulation strategies.

The data and analysis suggest that more effective interventions can be developed by focusing on the key differences between the Not Harmful segment and the Harmful but Coping and Problematic segments. Ideally, all personnel that transition out of the military would be in the Not Harmful segment; however, exploring the differences between Problematic and Harmful But Coping groups can help in developing a stepwise plan for healthy alcohol use. The Problematic group have different perceptions towards support services and their time in the military than the other groups. This group felt the stigma associated with mental health services was significantly higher, they felt that access to support services was significantly harder, and that there was a prominent drinking culture during their time in the military. These combined perceptions could understandably hinder a person’s ability to seek support, when needed, or simply feel they can go against social norms and not consume alcohol in a harmful manner. Providing more open discussions regarding the efficacy of support services offered and enabling personnel to feel that these services will not impact their standing or work could also help to bridge the gap felt by the Problematic group.

With regards to emotional regulation strategies the Problematic group are most likely to both ruminate on and catastrophise their past experiences. They are also most likely to blame others for their past experiences, which coincides with this groups’ belief that the drinking culture was very harmful whilst in the military. These negative cognitive emotion regulation strategies could be a key focus for social marketers wishing to create more effective interventions in this space.

This study explores a relatively small number of ex-military personnel and does not propose to form any causal linkages between the factors measured and the reported alcohol consumption by participants. It is still possible to examine the significant differences between different groups to identify gaps that social marketing interventions could help close – please see Table 2 for summary. Firstly, where military personnel present with significantly higher levels of childhood trauma, coupled with high prevalence for rumination and catastrophising particular care should be taken to support their well-being as these are strongly correlated with unhealthy alcohol consumption patterns. These personnel may require specific interventions that break down the perceived stigma associated with support services. Normalisation strategies that allow greater acceptance of support-seeking and championing healthy engagement with support services have been shown to be effective and can be employed here (Fellbaum et al., 2023).

When exploring the key differences between the Problematic and Harmful but Coping groups the focus on positive emotional regulation strategies appears to be a defining factor. Interventions that specifically teach positive reappraisal and perspective setting strategies may be of use.

This study is based on a sample of New Zealand veterans and relies upon self-report surveys and although the survey was completely anonymous there may be the potential for underreporting of alcohol consumption (Mattiko et al., 2011) and other social desirability biases (Hair et al., 1998). Finally, as a segmentation study there are no implications of causality and future research should determine if there is a causal link between the correlates identified in this study. This will help to understand the impact of the specific correlations on harmful alcohol behaviours, and which can lead to a significant drop in alcohol consumption.

Social marketing interventions targeting prevention, treatment, and community support have shown promise in mitigating the detrimental effects of alcohol harm among military personnel and veterans (Funderburk et al., 2008; Martens et al., 2015). However, continued research and evaluation of these interventions are necessary to ensure their effectiveness and tailoring to the unique needs of this population. Additionally, a comprehensive and coordinated approach involving multiple stakeholders, including social marketers, healthcare providers, policymakers, and community organisations, is crucial to addressing this complex issue effectively.