The association of smoking with drinking pattern may provide opportunities to reduce smoking among students

Abstract

There is evidence that smoking and alcohol consumption are paired behaviours among university students, but we know little about how New Zealand students engage in these behaviours. We estimated prevalence of daily and occasional smoking among university students, and associations of smoking with drinking patterns, demographics and smokefree policies of the university. This research was conducted with 2822 university students, aged 17–25, from five New Zealand universities who participated in an online health survey in 2013. Fourteen percent reported they currently smoke occasionally, and 3% reported smoking on a daily basis. Increasing age, living situation, drinking more often, and drinking more alcohol during drinking occasions were all associated with greater levels of smoking. Understanding patterns of smoking in relation to alcohol consumption may help efforts to further reduce smoking prevalence. Specifically, policy makers should consider approaches that de-couple tobacco and alcohol consumption, such as expanding the smokefree perimeter of bars and disallowing the sale of tobacco at premises licensed to sell alcohol.

Keywords:

• Alcohol
• campus
• occasional smoking
• tobacco
• university

Introduction

In New Zealand, young adults aged 20 to 29 years have the highest prevalence of smoking (Ministry of Health 2010) and recent New Zealand research has found that substantial smoking initiation occurs among older youth and young adults (Edwards et al. 2013). Young adults at university may therefore be at higher risk of smoking initiation and subsequent addiction than other population groups.

Ten years ago, the rate of daily smoking among University of Otago students was 10%, with a further 10% reporting occasional smoking (‘occasional smoking’ is used in this research to describe smokers who smoke on some days but not every day) (Kypri & Baxter 2004). Nationally, current daily smoking has decreased among young adults aged 18 to 24 years from 24% in 2011/12 to 20% in 2012/2013 (Ministry of Health 2013), but international evidence shows that accompanying similar trends is an increase in occasional smoking among young adults (Schane et al. 2009).

While recent studies suggest occasional smoking is stable over a university year in the United States (Caldeira et al. 2012), longer-term patterns have not been documented. For example, occasional smoking may lead to regular daily smoking and addiction (DiFranza et al. 2002; Moran et al. 2004; Scragg et al. 2008); this outcome is consistent with the higher prevalence of smoking among young adults compared to other age groups and with evidence that the age of smoking initiation has increased (Edwards et al. 2013). Young adults who transition from school to university or workplace settings experience fundamental changes in their social contexts and identity that may promote smoking among occasional smokers, and foster their progression to daily smoking (Kenford et al. 2005). Nicotine dependence can occur during this transition (DiFranza et al. 2000), even among occasional smokers (Dierker et al. 2007). Irrespective of whether young adults become daily smokers, occasional smoking remains of concern as it is associated with poor health outcomes among university students (Caldeira et al. 2012).

Alongside smoking experimentation, many New Zealand university students drink in a heavy episodic pattern. In 2013, 70% of a New Zealand undergraduate sample reported drinking alcohol in the previous four weeks, and almost two-thirds reported one or more binge drinking episodes in the previous four weeks (Connor et al. 2014). Yet while several studies conclude that alcohol consumption and smoking among young adults (Hoek et al. 2013) and university students (Jackson et al. 2010) are paired behaviours, we know little about how New Zealand students engage in these behaviours. New Zealand's restrictive tobacco marketing environment, where all tobacco advertising and sponsorship is banned under the Smokefree Environments Act, lies in stark contrast to the liberal alcohol marketing and consumption environment (Casswell & Maxwell 2005). These quite different regulatory frameworks highlight a marketing context where (permitted) promotion of one product may foster uptake of another. To explore determinants of smoking among a potentially at-risk young adult group, we estimated current daily and occasional smoking among university students in New Zealand. Associations with drinking patterns, demographics and smokefree status of the university were also examined to identify opportunities to reduce smoking among students.

Methods

Survey

The 2013 Tertiary Student Health Survey used a confidential, computerised online questionnaire to collect cross-sectional health and behaviour data, with invitations sent to a stratified random sample of 5759 students aged 17–25 years at five New Zealand universities. The sampling and recruitment procedures replicated earlier surveys, which have been described previously (Kypri et al. 2009). Participants were invited by letter and email to complete the survey, and up to four reminders were sent to non-responders. An example of the survey can be viewed at http://ipru3.otago.ac.nz/hdpdemo/ns2013/.

Sample and survey measures

The current sample includes all respondents who provided responses to questions on their smoking behaviour.

Smoking status: Respondents were asked ‘How often do you smoke now?’, with the following response options: ‘I have never smoked / I am not a smoker now'; ‘Less often than once a month'; ‘At least once a month'; ‘At least once a week'; and ‘At least once a day’. Smoking was classified into three levels: daily smokers, who currently smoked at least once a day, occasional smokers who currently smoked less than daily, and ‘not a current smoker’ who had never smoked or who did not smoke now.

Demographics: Respondents also provided demographic information including sex, age, ethnicity (where they were able to select multiple responses, which were prioritised in order from highest to lowest: Māori, Pacific, Asian, MELAA, Other, and finally European), whether they were an international or domestic student, and their type of residence during the semester.

Age of smoking initiation: Respondents were asked ‘At what age did you first smoke at least once a month’ with response options being: ‘9 years or younger'; ‘10–24’ (continuous variable); and ‘25 or older’.

Social smoking: Following Song and Ling (2011) social smokers were identified by asking self-identified non-smokers, occasional smokers (coded as above), and daily smokers to indicate whether they either did not see themselves as smokers or tended to smoke alone, with others, or equally alone and with others (Song & Ling 2011). Social smokers were defined as those who tended to smoke only with others.

Smokefree policy: Previous research undertaken by the lead author examined the smokefree policies at tertiary education institutions throughout New Zealand (Robertson & Marsh 2015). The status of these policies (100% smokefree, partially smokefree and no smokefree policy) was used to examine differences in smoking prevalence by tertiary institution.

Alcohol use: Respondents were asked about their alcohol use, including the number of days they consumed alcohol in a typical four-week period and the number of drinks consumed, on average, per drinking day. They were also asked about their frequency of heavy drinking (i.e. the number of days they consumed four or more drinks [for women] or six or more drinks [for men] in a typical four-week period).

Data analysis

Chi-squared tests were used to compare the distributions of smoking and drinking behaviours over the three levels of smoking. Ordinal logistic regression was then used to model associations between smoking status and sociodemographic and health behaviour variables. Models used all available data in each case. For each model, tests of proportionality were conducted and for each continuous variable, non-linearity was examined by adding a quadratic term to the model after centring the variable in question. This quadratic term was removed where it was not statistically significant. All analyses accommodated the complex survey design through clustering at the campus level and stratification by ethnicity (Māori [indigenous] vs. non- Māori). Variables were classified into blocks as follows: demographics; environment (smokefree campus status); and behaviour (alcohol consumption). Each variable was initially examined in an unadjusted model. Those variables with p < 0.25 from unadjusted models were then combined to create a ‘block’ model. Those variables with p < 0.25 in the demographic block model were added to those with p < 0.25 from the environment model; and those with p < 0.25 in that model were added to those with p < 0.25 from the behaviour model. All statistical analyses were performed using Stata 13.1 and two-sided p < 0.05 was considered statistically significant in all cases. No formal adjustment was made to the reported p-values following the variable selection process.

Results

A total of 2822 participants completed the minimum inclusion questions and were included in the analysis, giving a response rate of 49%. Two-thirds of those who took part were female, participants had a mean age of 20 years, and the majority were of European descent with 7% being Māori and 4% Pacific (Table 1). Over 80% of the sample were non-smokers, with 14% reporting they currently smoke occasionally, and 3% smoking on a daily basis.

Table 1. Demographics.

Variable	n (%)
Sex
Male	1025 (36.4)
Female	1797 (63.6)
Age—mean (SD)	20.0 (1.7)
Prioritised ethnicity
European	1534 (62.3)
Māori	686 (6.8)
Pacific	87 (4.0)
Asian	445 (23.8)
MELAA^a	32 (1.7)
Other	38 (1.5)
Student type
Domestic	2595 (91.3)
International	227 (8.7)
Residence
Flat	1355 (45.5)
Hall of residence	508 (15.9)
Parents	771 (33.4)
Boarding	82 (2.0)
Own home/renting	88 (2.8)
Other	18 (0.4)
Smokefree campus
No	501 (4.5)
Partial^b	1660 (61.1)
Yes	661 (34.3)
Smoking
Non-smoker	2264 (83.0)
Occasional smoker	464 (14.1)
Daily smoker	96 (2.9)
Drinking in last 12 months
No	334 (11.9)
Yes	2480 (88.1)
Missing	8
Drinking frequency
Days over 4 weeks—median (IQR)	2 (4)
Missing	17
Drinking amount
Drinks—median (IQR)	3 (5)
Missing	21
Heavy drinking frequency
Days over 4 weeks—median (IQR)	1 (3)
Missing	22

n is actual; %, mean, SD, median, and interquartile range (IQR) are survey-adjusted.

^aMiddle Eastern/Latin American/African.

^bPartial smokefree campus means the policy either permits smoking on campus, such as in designated smoking areas, or contains at least one element in the policy inconsistent with de-normalising smoking.

Associations between smoking and drinking

Among smokers, there were associations between smoking in social situations and occasional versus daily smoking (p < 0.001). Occasional smokers were more likely to smoke with others than alone or in a mixed social environment (Table 2). There was also an association between age of initiation and occasional versus daily smoking (p < 0.001), where occasional smokers were more likely to initiate in the 17–20 years age range and less likely to smoke before 17 years or after 20 years (Table 2). Occasional and daily smokers were significantly more likely than non-smokers to be current drinkers, to drink more than twice a week, to drink more on each drinking occasion and to meet the threshold for heavy drinking occasions. Post hoc tests found no evidence of differences between occasional smokers and daily smokers for three of the four drinking behaviours, although occasional smokers were more likely than daily smokers to drink more than 10 drinks on average per occasion (p = 0.050).

Table 2. Associations with social smoking, age of initiation and drinking behaviour.

	Non-smokers n (%)	Occasional smokers n (%)	Daily smokers n (%)	p-value for association
Social smoking*
I am not a smoker		208 (46.1)	0 (0.0)	<0.001
I tend to smoke alone		10 (1.6)	12 (12.8)
I smoke mainly with others		217 (46.3)	27 (30.1)
I smoke equally alone and with others		25 (6.0)	56 (57.1)
Age of smoking initiation
16 years and under		36 (22.1)	53 (56.3)	<0.001
17–20 years		117 (73.6)	39 (38.7)
21–25 years		11 (4.3)	4 (5)
Current drinkers
No	327 (16.6)	2 (1.0)	5 (3.0)	<0.001
Yes	1931 (83.4)	458 (99.0)	91 (97.0)
Drinking more than twice per week (>8 times in typical 4-week period)
No	2139 (96.1)	389 (89.0)	77 (86.8)	0.002
Yes	116 (3.9)	69 (11.0)	15 (13.2)
Drinking more than 10 drinks on average per occasion (typical 4 weeks)
No	2127 (95.6)	356 (83.3)	72 (87.9)	<0.001
Yes	125 (4.4)	102 (16.7)	19 (12.1)
Having more than 4 heavy drinking occasions (4+/6+ drinks for women/men) in a typical 4-week period
No	2070 (93.5)	307 (69.8)	72 (80.0)	<0.001
Yes	183 (6.5)	149 (30.2)	19 (20.0)

Bold indicates significant results p < 0.05.

Ordinal logistic regression models

From unadjusted models, increasing age, any drinking in the past 12 months, drinking more often, drinking more on drinking days, and more frequent heavy drinking were all associated with higher odds of more frequent smoking (Table 3). Living with parents or boarding was protective for smoking relative to both living in a hall of residence and living in a shared house or apartment.

Table 3. Associations with smoking.

Block	Variable		Unadjusted models		Block models		Demographics and environment		Fully adjusted
			OR (95% CI)	p-value	OR (95% CI)	p-value	OR (95% CI)	p-value	OR (95% CI)	p-value
Demographics	Sex			0.244		0.095		0.089		0.219
		Female	1.00		1.00		1.00		1.00
		Male	1.22 (0.82–1.81)		1.36 (0.92–2.00)		1.36 (0.93–1.99)		0.84 (0.60–1.17)
	Age	Per year	1.15 (1.05–1.26)	0.012	1.16 (1.03–1.32)	0.028	1.16 (1.04–1.30)	0.022	1.23 (1.07–1.40)	0.013
	Ethnicity			0.002		0.013		0.010		0.157
		European	1.00		1.00		1.00		1.00
		Māori/Pacific	1.08 (0.66–1.78)		1.20 (0.77–1.87)		1.20 (0.77–1.88)		1.29 (0.75–2.20)
		Other	0.40 (0.34–0.47)		0.41 (0.27–0.61)		0.40 (0.30–0.54)		0.81 (0.65–1.00)
	Student type			0.243		0.874
		Domestic	1.00		1.00
		International	0.83 (0.57–1.21)		0.96 (0.50–1.86)
	Residence			0.011		0.015		0.004		0.002
		Parents/boarding	1.00		1.00		1.00		1.00
		Hall of residence	2.26 (1.50–3.41)		2.43 (1.57–3.76)		2.34 (1.89–2.89)		1.62 (1.44–1.81)
		Flat	2.52 (1.70–3.73)		2.09 (1.33–3.27)		2.01 (1.14–3.58)		1.26 (0.80–1.97)
Environment	Smokefree			<0.001				0.604		0.395
		No or partial	1.00				1.00		1.00
		Yes	0.57 (0.50–0.65)				0.93 (0.66–1.32)		1.16 (0.75–1.81)
Behaviour	Drinking in last 12 months			0.001
		No	1.00
		Yes	14.77 (6.29–34.72)
	Drinking frequency		1.21 (1.12–1.30)	0.002	1.11 (1.07–1.15)	0.001			1.10 (1.06–1.14)	0.002
	Drinking amount		1.19 (1.16–1.23)	<0.001	1.12 (1.09–1.15)	<0.001			1.11 (1.10–1.13)	<0.001
	Heavy drinking		1.36 (1.18–1.57)	0.004	1.11 (0.97–1.27)	0.094			1.13 (0.98–1.30)	0.072

Bold indicates significant results p < 0.05. Models including multiple independent variables are indicated by a box drawn around these variables, including the two separate block models.

Age, ethnicity and living situation each remained significantly associated with smoking in the adjusted block demographic model (Table 3). In the adjusted block behaviour model, drinking more often and drinking more during drinking occasions were each associated with higher odds of more frequent smoking, but more frequent heavy drinking was not.

In the fully adjusted model, age, living situation, drinking more often and drinking more per drinking occasion both remained significantly associated with smoking, but ethnicity was no longer associated with smoking (Table 3). While statistically significant in the unadjusted model, there was no evidence from the demographic and environment block model or the fully adjusted model of an association between smoking restrictions on campus and level of smoking. The following discussion draws implications from the associations and models.

Discussion

We found a strong link between smoking and alcohol consumption among New Zealand university students. Drinking more frequently and in larger volumes were associated with smoking both before and after adjusting for demographic and environment variables. This finding is consistent with earlier studies (Jackson et al. 2010) and highlights the need to address settings that allow combined use of tobacco and alcohol. Evidence that alcohol facilitates smoking experimentation and subsequent addiction among young adults (Saules et al. 2004; Jiang & Ling 2011) may explain tobacco companies' strong opposition to smokefree bars legislation and more recent calls to establish areas outside bars as smokefree.

Previous New Zealand research has shown that daily smoking among university students has declined and occasional smoking has increased (Kypri & Baxter 2004), consistent with an increase in occasional smoking among young adults in the United States (Schane et al. 2009). We cannot discern the cause of this trend from our cross-sectional data. It may result from tobacco control policies introduced over this period; for example, the introduction of smokefree indoor (and subsequently many outdoor) areas, excise tax increases, a retail display ban, mass media campaigns aimed at young people such as ‘Smoking not our future’, or from shifts in young adults' social contexts. Evidence that tobacco consumption is declining internationally among young adults suggests that denormalisation strategies, whether brought about by policy, social marketing or health promotion measures, have reduced smoking's acceptability.

Occasional smokers were more likely to smoke with others than were daily smokers, which is consistent with international evidence that smoking begins as a shared social experience, often facilitated by alcohol (Moran et al. 2004; Hoek et al. 2013). This finding follows a similar trend among young people in the general population (Pierce et al. 2009), and university students in the USA (Moran et al. 2004). Occasional smokers were also found to be heavier drinkers than daily smokers. This could possibly be explained if occasional smokers only smoke when they are intoxicated, while daily smokers may not necessarily be heavy drinkers. It may also be that occasional smokers have more discretionary income to spend on alcohol as they are not spending as much on increasingly expensive tobacco.

Despite the downward trend in daily smoking, the smoking practices we identified have important implications for New Zealand's smokefree 2025 goal (New Zealand Government 2011). Policies that de-couple the sale of tobacco and alcohol, increase the pervasiveness of smokefree settings, raise the age at which tobacco can be purchased and offer cessation support appropriate for young adults, could reduce prevalence among this key population group.

Removing tobacco sales from settings where alcohol may be purchased would reduce smokers' ability to purchase tobacco while drinking. Given that one-third of New Zealand tobacco retail outlets have a licence to sell alcohol, and 13% of all retail outlets are on-licence premises, this simple measure could dramatically change where tobacco is consumed (Marsh et al. 2013). In addition, prohibiting tobacco sales in on-licenced premises would alter how tobacco is consumed and recognise smokers' tendency to smoke more when drinking alcohol (Paul et al. 2010).

Given that much of students' drinking and smoking occurs off campus, extending the smokefree perimeter outside bars and restaurants could reduce casual smoking by making it less convenient and appealing. At present, bars provide attractive social spaces for smoking; new interventions could reverse this approach and recognise that, as spaces where smokers congregate, bars may represent settings for cessation interventions (Jiang & Ling 2013). Ling et al.'s (2014) innovative social branding campaigns, including branded events marketed to young adults through multiple communication channels, directly counter tobacco marketing activities and have successfully reduced smoking prevalence among high-risk young adult groups (Ling et al. 2014). These novel interventions may also be effective in university settings where students spend much of their social time.

Our findings highlight the important role university environments may play in encouraging smokefree lives and supporting smoking cessation. Internationally, smokefree campus policies have reduced tobacco use among tertiary students (Seo et al. 2011), though changes among less frequent smokers may take longer to become evident. To date, nine of the 29 New Zealand Tertiary Education Institutions have adopted a completely smokefree campus policy; and a further three institutions have ‘predominantly smokefree’ policies that allow some exceptions (Robertson & Marsh 2015). Despite the uptake of smokefree policies, few evaluations of these are available and little is known about their impact on smoking behaviours, particularly among occasional smokers. We found no relationship between presence of a smokefree policy and smoking behaviour. However, given that many of these policies were only recently implemented, it is possible that evidence of any behaviour change is not yet apparent. Further analyses, once these policies have become established, could clarify the effect smokefree settings have on smoking and its pairing with risky drinking.

In some cases occasional smoking will represent a stage of smoking cessation (Berg et al. 2012), and thus may be a time when young people would benefit from cessation interventions. However, many occasional smokers in this study did not view themselves as smokers, consistent with other findings (Hoek et al. 2013), and current smoking cessation support and treatment may not reach them or meet their needs effectively. Our findings suggest a need for smoking cessation support that is developed for and communicated to this substantial sub-group of young adults.

A strength of this study was the large randomly selected sample of students from five of New Zealand's eight universities. Data on demographics of the total university population from these five universities are not available; however, data from universities nationally show that a higher proportion of participants in this study were women (63.6%) than are in the student population (56.8%) (Ministry of Education 2014). However, the sample was representative of the ethnicity of university students enrolled nationally with 62.1% of European ethnicity, while 9.0% were Maori, and 6.3% were Pacific. These demographic variables were also consistent with a recent study of University of Otago students (Marsh et al. 2014). This research uses cross-sectional data, which does not allow us to discern whether heavier drinkers are more likely to be smokers, or smokers more likely to be heavier drinkers.

The response rate of 49% may have introduced some selection bias. Literature on non-response error suggests that those respondents involved in risky behaviours are less likely to participate in health surveys, possibly leading to an underestimation of the true prevalence of these behaviours (Kypri et al. 2011). An online survey was used to reduce social desirability bias (Crutzen & Göritz 2010) and increase the response rate. While the prevalence estimates presented here may have been affected, previous research has found little effect of biases in estimated associations due to response rates (Van Loon et al. 2003).

Future research should extend work that has already begun to explore how young smokers view the relationship between their smoking and drinking practices (Hoek et al. 2013). Qualitative research could examine how smoking and drinking behaviours evolve, particularly the contexts that support (or deter) these behaviours. Subsequent research could evaluate interventions that reduce the potential for these contexts to facilitate smoking initiation. Our findings now need to be tested using longitudinal data that allow the direction of the smoking–drinking relationship to be determined.

Conclusion

While New Zealand has reduced direct promotion of smoking, alcohol settings allow smoking to be modelled as an accepted and normal behaviour. Further, drinking alcohol may facilitate smoking uptake (Saules et al. 2004; Jiang & Ling 2011; Hoek et al. 2013) and increase tobacco consumption among university students. As a precautionary measure, policy makers could address both relationships by disallowing tobacco sales wherever alcohol is consumed and extending the smokefree areas outside bars so dual consumption becomes more difficult in social settings.

Acknowledgements

The researchers wish to acknowledge the university student participants who took part in this research. We also wish to extend our thanks to the university staff members who supported and enabled this research to be undertaken.

Disclosure statement

No potential conflict of interest was reported by the authors.