Alcohol Consumption and Physical Activity in Austrian College Students—A Cross-Sectional Study

ABSTRACT

Background: The age of college students is considered as crucial for developing health-related behaviors, e.g., alcohol consumption or a physically active lifestyle. Previous research reported a positive relationship between alcohol consumption and physical activity (PA) in college students. However, the main body of research was done in students from the United States who might differ from European students. Objectives: Thus the aim of this study was to analyze the relationship between alcohol consumption and PA in a sample of Austrian college students. Methods: In a cross-sectional design, 861 Austrian students from various study fields responded to a web-based questionnaire. Self-reported alcohol consumption, PA, and relevant sociodemographic variables were assessed. Multiple regression analyses were used to study the relationship between alcohol consumption and PA. Results: In none of the regression models, a significant relationship between alcohol consumption and PA was found. There was a significant influence of sex, age, relationship status, education level, and study field on alcohol consumption. Male, older, and undergraduate students studying social sciences without a relationship reported higher alcohol consumption. Conclusions/Importance: The results do not support a general relationship between alcohol consumption and PA among urban Austrian college students of various study fields. Compared to other variables (e.g., sex, relationship status), PA seems to be less important in relation to the consumption of alcohol. This study challenges a global perspective on a positive relationship between alcohol consumption and PA and highlights the need for more cross-cultural investigations.

KEYWORDS:

• Alcohol consumption
• physical activity
• drinking
• health behaviors
• emerging adulthood
• binge drinking
• ordinal regression

Introduction

Behavioral influences, e.g., a physically active lifestyle or smoking, are among the most important determinants of health status (Institute of Medicine, 2001). Behaviors are considered to be modifiable; therefore people are able to positively affect morbidity and mortality by adopting healthy behaviors and avoiding risky behaviors (Mokdad, Marks, Stroup, & Gerberding, 2004). Previous research suggested that healthy behaviors cluster together, e.g., persons, who practice a physically active lifestyle, seem to be more likely to have a healthy nutrition (de Vries et al., 2008). Similarly, risky behaviors, e.g., smoking and heavy drinking (e.g. drinking alcohol on more than 4 days per week), tend to co-occur (Noble, Paul, Turon, & Oldmeadow, 2015; Poortinga, 2007b).

Both alcohol consumption and physical activity (PA) are behaviors, albeit only a physically active lifestyle is widely considered as a healthy behavior. Indeed, PA was shown to be a protective factor in the prevention of different types of diseases, including cardiovascular diseases (Gallanagh, Quinn, Alexander, & Walters, 2011; Thompson, 2003), depression (Mammen & Faulkner, 2013), anxiety disorders (Martinsen, 2008), and cancer (Moore et al., 2016).

The classification of alcohol consumption as a risky or a healthy behavior seems to be more problematic and highly dependent on the dose (Leasure, Neighbors, Henderson, & Young, 2015). Moderate levels of alcohol consumption (i.e., 25 g alcohol per day corresponding to approximately 2 drinks per day) were shown to be positively related with several health-enhancing effects including lower incidence of coronary heart disease and stress reduction (summarized in Piazza-Gardner and Barry (2012)). However, more recent publications indicate that the protective effects of moderate levels of alcohol consumption might be confounded by a selection bias (Britton & Bell, 2017; GBD 2016 Risk Factors Collaborators, 2017; Holmes et al., 2014). A greater degree of consensus seems to be about the effects of heavy drinking, which is widely considered as a risky and unhealthy behavior. It is estimated that heavy drinking is responsible for approximately 3.3 million deaths per year, which corresponds to 5.9% of all global deaths (World Health Organization, 2014). Furthermore, severe heavy drinking is associated with functional impairments, psychiatric problems and memory loss (Perreira & Sloan, 2002).

Given the clustering of health behaviors mentioned above, a negative association between the amount of PA and at least heavy alcohol consumption is expected. However, several studies with large sample sizes and in various age populations showed a positive relationship between PA and alcohol consumption (see Piazza-Gardner and Barry (2012) and Leasure, Neighbors, Henderson, and Young, (2015), for reviews). Piazza-Gardner and Barry (2012), based on the findings of 17 studies, concluded that drinkers are more physically active than their nondrinking peers. This was found consistently with different assessments of PA and alcohol consumption and was reported among adolescents, athletes, college students, and among the general population (Leasure, Neighbors, Henderson, & Young, 2015; Piazza-Gardner & Barry, 2012).

Out of the different age populations, college students are of specific interest. The age of college students falls in the transition from adolescence to adulthood and can be considered as crucial for the development of a healthy lifestyle and adopting health behaviors. Identity formation, including establishing behavioral patterns, “begins in adolescence but takes place mainly in emerging adulthood,” defined as the age of 18 to 25 years (Arnett, 2000), or, more recently, defined as the age of 18 to 29 years (Arnett, Zukauskiene, & Sugimura, 2014). Consequently, this age range has been studied intensively in relation to PA and alcohol consumption (Moore & Werch, 2008; Musselman & Rutledge, 2010; Seo, Nehl, Agley, & Ma, 2007; Vickers et al., 2004).

Musselman and Rutledge (2010) conducted a cross-sectional study with 296 college students in the United States using the three categories (low/moderate/high activity level) of the International Physical Activity Questionnaire (IPAQ) as an assessment tool for PA. Ordinal regression analyses provided strong evidence for the existence of a positive relationship between alcohol consumption and PA. Specifically, the odds ratio for being in a higher PA category was larger than 1.3 for alcohol frequency. This association persisted when the regression model was controlled for several covariates such as gender, age, and Greek Status. However, it has to be noted that this study was conducted “at a small, selective, liberal arts college located in a small rural community” (Musselman & Rutledge, 2010).

In summary, a large body of previous research suggests that someone who is physically more active tends to have higher alcohol consumption. In the light of clustering of health behaviors, this relationship must be considered as contradictory. However, it has to be mentioned that out of the 17 studies summarized in Piazza-Gardner and Barry (2012), only one study was conducted outside of the American region, in Great Britain (Wouter Poortinga, 2007a). All studies concerning the crucial age of college students were conducted in the United States.

It might be difficult to generalize these results globally when the vast majority of studies was conducted in the American region. Indeed, there is some evidence that intercultural differences exist both in alcohol consumption (Bloomfield, Stockwell, Gmel, & Rehn, 2003; Cheng & Anthony, 2017; Degenhardt et al., 2008; O'Brien, Hunter, Kypri, & Ali, 2008; World Health Organization, 2014) and in PA (Powell, Shima, Kazlauskaite, & Appelhans, 2010; Sjöström, Oja, Hagströmer, Smith, & Bauman, 2006). Consequently, the relationship between PA and alcohol consumption might be different in the European population compared to the U.S. population. Kopp et al. (2015) conducted a cross-sectional study based on a national health survey in Austria and could not confirm a positive alcohol consumption–PA relationship. Instead, higher alcohol consumption was found in men with low PA-levels in the general population. However, Kopp et al. (2015) included different age populations from 15 to over 85 years. To the best of our knowledge, no study focusing exclusively on European college students exists, although this age range is crucial for adopting health behaviors.

Thus the aim of this study was to analyze the relationship between alcohol consumption and PA in a sample of Austrian college students of various studies. In contrast to the findings of previous studies conducted in the United States or Canada (Piazza-Gardner & Barry, 2012), but according to the findings of Kopp et al. (2015) in the Austrian population, we hypothesized no (positive) relationship between alcohol consumption and PA.

Materials and methods

Sample and procedure

We collected the data via web-based questionnaire in a cross-sectional design. Information about PA level, alcohol consumption and socio-demographic characteristics was collected in 41 questions. The questionnaire was distributed by Email announcements of the University of Innsbruck and was received by a maximum of 27,400 students. Due to local rights, it is possible to unsubscribe from the Email announcements, what does not allow giving an exact number of recipients. The Email announcement solely consisted information about the study goal, namely the analysis of the relationship between alcohol consumption and PA in college students. No information about previous findings on the relationship between alcohol consumption and PA was given. Inclusion criterion was matriculation at the University of Innsbruck. No incentives were provided for participating in the study. The study procedure was approved by the Board for Ethical Questions in Science of the University of Innsbruck in accordance with the Declaration of Helsinki (No. 25/2015, date: 17.06.2015).

Initially, data from 1,174 students were collected (minimum response rate: 4.3%). Due to missing data and data processing rules, the final sample consisted of 861 students with complete data in all relevant variables. The final sample consisted of 56.3% female students, with a mean age of 24.0 (SD = 4.8) years, and a mean study duration = 3.5 (SD = 2.3) years. 230 (26.7%) students studied natural sciences, 420 (48.8%) social sciences, 162 (18.8%) humanities, and 49 (5.7%) applied sciences.

Measurements

Self-rated alcohol consumption

Self-rated alcohol consumption was assessed using three items concerning frequency, quantity, and heavy drinking in the past 30 days (Musselman & Rutledge, 2010). Alcohol frequency was quantified by the number of times of alcohol consumption in the past 30 days (“How many times did you drink alcohol in the past 30 days?”). The response options (coding in brackets) were: “Did NOT drink alcohol in the past 30 days” (0), “Once in the past 30 days” (1) “2-3 times in the past 30 days” (2), “Once or twice a week” (3) “3-4 times a week” (4), “5-6 times a week” (5), “Nearly every day” (6), and “Every day” (7). Alcohol quantity was quantified by the usual number of drinks on one occasion (“In the past 30 days, when you drank, how many drinks did you usually have on any one occasion?”). The response options (coding in brackets) were “Did NOT drink alcohol in the past 30 days” (0), “1 drink” (1), “2 drinks” (2), “3 drinks” (3), “4 drinks” (4), “5 drinks” (5), “6 drinks” (6), “7 drinks” (7), “8 drinks” (8), “9 drinks” (9), “10 drinks” (10), “11 drinks” (11), “12 drinks” (12), and “13 or more drinks” (13). Heavy drinking was quantified by the number of times of five or more drinks in a single sitting (“In the past 30 days how many times have you had five or more drinks in a single sitting?”) with identical response options and coding to those of alcohol frequency. A fourth measure of alcohol consumption, alcohol amount in the past 30 days, was calculated as a product of the number of times of alcohol consumption in the past 30 days and alcohol quantity (drinks per past 30 days). Since the units of the variable alcohol frequency differ in the response options (30 days, week, day), transforming alcohol frequency before multiplying with the variable alcohol quantity was necessary. The mean value was taken for response options with ranges (i.e., “2–3 times in the past 30 days”), resulting in the following transformations: 0 was transformed to 0, 1 was transformed to 1, 2 was transformed to 2.5, 3 was transformed to 6, 4 was transformed to 14, 5 was transformed to 22, 6 was transformed to 26, 7 was transformed to 30. After multiplying, a value range from 0 to 403 drinks per past 30 days was possible for the variable alcohol amount in the past 30 days. A drink was defined as 330 ml beer, 150 ml wine, or a shot of liquor. More detailed information can be found in Musselman and Rutledge (2010).

Self-rated PA

The amount of PA was assessed with the long form of the International Physical Activity Questionnaire containing 27 items (IPAQ, http://www.ipaq.ki.se/ (Craig et al., 2003)). Participants were asked to rate the frequency and usual duration of vigorous, moderate, and walking activity in the domains of work, housekeeping, transport, and leisure time during the last week. Based on the participants’ responses, four continuous scores (total, vigorous, moderate, and walking activity) and one categorical score (level of PA) for the volume of self-rated PA were calculated according to the Scoring protocol of the IPAQ (http://www.ipaq.ki.se/).

The continuous scores were expressed as estimated energy expenditure in multiples of the resting metabolic rate in minutes (MET-min) for vigorous, moderate, and walking activity separately. To account for the different intensities, different weighs were used for the intensities. Estimated total activity was calculated by summarizing all scores of the different intensities. All recommended data processing and data truncation rules were applied.

The categorical score categorizes the participants in one of the three levels of PA based on standard scoring criteria available at http://www.ipaq.ki.se.

a)	Low activity level: not meeting any the criteria for moderate or high activity level.
b)	Moderate activity level: not meeting any of the criteria of high activity level and achieving ≥ three days of vigorous-intensity activity of ≥20 minutes per day or ≥ five days of moderate-intensity activity and/or walking of ≥30 minutes per day or ≥5 days of any combination of walking, moderate-intensity or vigorous intensity activities with a minimum of ≥ 600 MET-min per week
c)	High activity level: achieving vigorous-intensity activity on ≥ three days and accumulating ≥1500 MET-min per week or ≥7 days of walking, moderate-intensity or vigorous intensity activities accumulating ≥3000 MET-min per week.

Even though there was some controversy about the validity of the IPAQ (Hallal & Victora, 2004; Lee, Macfarlane, Lam, & Stewart, 2011), the IPAQ is widely used in national and international surveys (Bauman et al., 2011; Guthold, Ono, Strong, Chatterji, & Morabia, 2008; Sjöström et al., 2006) and Craig et al. (2003) concluded that the IPAQ showed acceptable validity values (r = .8).

Covariates

The following covariates were collected because of their potential association with alcohol consumption: sex (male, female), age (<21 years, 21 years and above), relationship status (no relationship, married/de facto relationship), education level (undergraduate, postgraduate), and study field (natural sciences, social sciences, humanities, applied sciences).

Statistical analyses

All statistical analyses were performed using SPSS v. 23 (IBM, New York, United States). A series of multiple linear regression analyses was conducted to test the relation between PA and alcohol consumption. The four variables for alcohol consumption were used as dependent variables in four separate multiple linear regression analyses. PA in amount of METs per week and the covariates were included as independent variables.

Since the direction of the effects between PA and alcohol consumption is not clear and might be bidirectional, we also followed the approach of Musselman and Rutledge (2010) and included a series of ordinal regression analyses (also termed as cumulative logistic regression) with PA level as dependent variable. Four separate ordinal regression analyses were conducted for each variable for alcohol consumption as independent variables. The ordinal regression analyses allowed a comparison with the results of Musselman and Rutledge (2010).

The level of significance was set at p < .05 (two-tailed). Data were presented as mean (SD) and frequencies (percentage) unless otherwise stated.

Results

Descriptive statistics

After data processing, 861 participants were categorized to low, moderate or high activity level and 32 (3.7%) showed low, 248 (28.8%) moderate, and 581 (67.5%) high PA level. Mean age was 24.0 (4.8) years. Descriptive values are displayed in Table 1 for the total sample and by PA level.

Table 1. Mean physical activity variables, alcohol consumption, and covariates for the total group and by activity level.

			Physical activity level
	Total sample (n = 861)		Low (n = 32)		Moderate (n = 248)		High (n = 581)
	Mean or %	SD	Mean or %	SD	Mean or %	SD	Mean or %	SD
Self-rated physical activity
Total activity (MET min per week)	5236	4100	494	540	1784	702	6970	3919
Vigorous activity (MET min per week)	1967	2625	119	419	321	480	2772	2848
Moderate activity (MET min per week)	2089	2248	204	380	834	595	2728	2461
Walking activity (MET min per week)	1180	1350	171	208	629	544	1470	1517
Self-rated alcohol consumption
Alcohol frequency (times/30 days)	2.8	1.2	3.1	1.4	2.7	1.2	2.9	1.2
Alcohol quantity (usual number of drinks/alcohol consumption)	3.8	2.9	3.9	3.1	3.7	2.8	3.8	2.9
Alcohol amount (drinks/30 days)	31.8	40.0	33.3	37.2	28.6	33.2	31.8	38.4
Heavy drinking (times/30 days)	1.5	1.2	1.7	1.4	1.4	1.1	1.5	1.2
Covariates
Sex (% male)	43.7		50.0		39.0		45.3
Age (% 21 years and above)	70.3		59.4		60.9		75.0
Relationship status (% single)	49.2		53.1		49.0		49.0
Education (% postgraduate)	22.6		21.9		19.8		23.9

Linear regression on alcohol consumption

Total PA (in MET min per week) was no significant predictor in any of the alcohol consumption variables in the present sample of college students, compare Table 2. Sex was significant in all alcohol variables indicating that males reported to consume alcohol more often, to consume a higher quantity per drinking occasion and per the last 30 days, and to be more commonly heavy drinkers. Age below 21 years was significantly associated with a lower alcohol frequency compared to 21 years and above. Postgraduates reported a significantly lower alcohol quantity compared to undergraduates. Students being single reported a significantly higher alcohol amount and more episodes of heavy drinking in the last 30 days compared to students in relationship.

Table 2. Multiple linear regression models with alcohol consumption as dependent variable.

	Alcohol frequency		Alcohol quantity		Alcohol amount		Heavy drinking
	β	p	β	p	β	p	β	p
Total physical activity (MET min per week)	−0.03	.336	0.02	.544	0.06	.069	−0.01	.762
Sex	0.17	<.001	0.28	<.001	0.27	<.001	0.30	<.001
0 = female, 1 = male
Age	0.09	.012	−0.05	.131	< −0.01	.939	0.02	.656
0 = below 21 years, 1 = 21 years and above
Education level	0.02	.680	−0.08	.021	−0.02	.551	−0.02	.512
0 = undergraduate, 1 = postgraduate
Relationship status	−0.01	.727	−0.07	.021	−0.07	.032	−0.10	.002
0 = single, 1 = relationship
Study field
0 = natural sciences (reference)
1 = social sciences	0.04	.319	0.09	.026	0.09	.030	0.09	.023
1 = humanities	−0.04	.364	0.03	.427	0.02	.530	0.02	.596
1 = applied sciences	0.04	.239	0.03	.343	0.02	.474	0.05	.125

Note: β: standardized regression coefficient, bold values represent significant coefficients.

Figure 1 shows the alcohol amount in drinks per 30 days by sex and by level of PA. The other alcohol consumption variables showed similar sex differences. Male students reported a mean of 44.2 (45.9) alcoholic drinks per month compared to female students, who reported a mean of 22.3 (31.9) alcoholic drinks per month.

Figure 1. Mean alcohol amount in the past 30 days by sex and activity level. Error bars represent standard deviations.

Ordinal regression level on PA

No significant results were found when the level of PA (low, moderate, high) was modelled as a function of alcohol consumption, p > .186, indicating no important influence of alcohol consumption on the level of PA. Table 3 shows all ordinal regression models with PA (in MET min per week) as the dependent variable. Since PA was not significant in any of the models no other covariates were added to the model.

Table 3. Ordinal regression analysis with dependent variable level of physical activity.

	Alcohol frequency				Alcohol quantity				Alcohol amount				Heavy drinking
	OR	CI lb	CI ub	p	OR	CI lb	CI ub	p	OR	CI lb	CI ub	p	OR	CI lb	CI ub	p
Alcohol variable	1.08	0.96	1.21	.187	1.01	0.96	1.07	.621	1.00	1.00	1.01	.343	1.02	0.90	1.15	.772

Note: OR: odds ratio, CI lb: 95% confidence interval lower bound, CI ub: 95% confidence interval upper bound.

Discussion

Main results

This study does not provide evidence for a positive relationship between PA and alcohol consumption among urban Austrian college students of various study fields. In the present sample, more than 65% of the students reported a high PA level, i.e., being active with at least moderate intensity for at least 1.5 to 2 hours of total activity per day. Alcohol consumption was related to sex, age, education, and relationship status. Higher alcohol consumption was found in male, undergraduate students below 21 years without relationship. PA seems to play a minor role in relation to alcohol consumption in Austrian college students.

Possible explanations for discrepancy to existing studies

The present findings can be considered as contrary to most of the previous studies regarding the relationship between alcohol consumption and PA in different populations (summarized in Piazza-Gardner and Barry (2012) and Leasure, Neighbors, Henderson, and Young, (2015)). Out of the alcohol consumption variables, heavy drinking (also termed ‘binge drinking’ by some authors (Dinger, Brittain, & Hutchinson, 2014; Moore & Werch, 2008; Nelson, Lust, Story, & Ehlinger, 2009; Seo, Nehl, Agley, & Ma, 2007)) is of specific interest. Alcohol consumption per se, especially moderate levels of alcohol consumption, might not be considered as risky, but heavy drinking has to be classified as a health-risk behavior (Piazza-Gardner & Barry, 2012).

Musselman and Rutledge (2010) reported in an ordinal regression model that students with a higher PA level were 22% more likely to report higher heavy drinking. In this study, the ordinal regression model did not show differences in the alcohol consumption between PA groups. Since the study design (cross-sectional) and the methods of assessing alcohol consumption (self-reported, four questions) and PA (self-reported, IPAQ) were identical in the two studies and the alcohol consumption was comparable, explanations for the discrepancy have to be found somewhere else. Some differences in the sample have to be mentioned: In the sample in Musselman and Rutledge (2010) younger (mean age 19.8 years), only undergraduate students were asked, whereas in the present sample the mean age was 24.0 years with both undergraduate and postgraduate students. However, age and education level were found to be significant predictors for alcohol frequency, alcohol quantity and alcohol amount, but not for heavy drinking. We therefore conclude that age and education level may play a minor role in heavy drinking. What might be of higher interest concerning heavy drinking is the relationship status of the students. Being single was associated with significantly more episodes of heavy drinking compared to being in a relationship, thereby confirming findings in the general population (Power, Rodgers, & Hope, 1999). Relationship status was not reported by Musselman and Rutledge (2010) or other studies including samples in the emerging adulthood (see Piazza-Gardner and Barry (2012)), but might be of high importance also in students. Another important predictor in heavy drinking was study field. Musselman and Rutledge (2010) conducted their study in a rural area in a liberal arts college. The present sample consisted of students of various studies with a higher rate of heavy drinking in social science students compared to natural science students. Although no difference between humanities (including liberal arts) and natural sciences were found, our findings indicate that alcohol consumption might be different in various studies and the sample of college students should be considered as an inhomogeneous group itself.

Probably the most important discrepancy between existing studies and this study was the level of PA. Musselman and Rutledge (2010) reported 38% as highly active, whereas in this study 67% were classified as highly active. In the existing literature, there is a large range regarding the percentage of highly active people categorized by the IPAQ both in Austria and in the United States. In Austria, percentages between 26% (Sjöström et al., 2006) and 56% (Kopp et al., 2015) have been reported. Similarly, in the United States, percentages between 25% (Hallal et al., 2012) and 62% (Bauman et al., 2009) of highly active people have been reported. These wide variations are already known and can be partly explained by recall biases due to the self-reported values (Haase, Steptoe, Sallis, & Wardle, 2004). However, cross-cultural differences might also contribute to differences between the United States and Austria not only for PA but also for the relationship between PA and alcohol consumption. This assumption is supported by Cheng and Anthony (2017), who reported cross-cultural differences in alcohol consumption in adolescents, and by findings of a large population-based study in an adult Austrian sample: Contrary to studies in the U.S. population, Kopp et al. (2015) could not find a relationship between PA and alcohol consumption. One might speculate that differences in the minimal age to buy alcohol legally or the presence of an alcohol prohibition phase in United States might partly explain cross-cultural differences. The present findings do not support a globally uniform perspective on a positive PA-alcohol consumption relationship.

Practical recommendations for Austrian college students

In all variables assessing alcohol consumption, male sex was associated with higher alcohol consumption. Thereby, findings from previous studies stating higher alcohol consumption in males in different populations were confirmed (Erol & Karpyak, 2015; Wilsnack et al., 2000). The results provide further evidence to tailor alcohol prevention programs to men compared to women (Kopp et al., 2015).

Concerning heavy drinking, relationship status might be a variable of concern in Austrian students. Heavy drinking was reported more often in students being single compared to students being in a relationship. Consequently, relationship might be considered as a protective factor against heavy drinking. This was already reported in the general population (Power, Rodgers, & Hope, 1999), but was less noticed in college students at present.

An interesting observation was found in the influence of the study field on heavy drinking. The present findings indicate that alcohol consumption is higher in social science students compared to natural science students. To the best of our knowledge, the study field does not receive attention in alcohol prevention programs for college students.

Students with an age of 21 years and above reported higher alcohol frequency and undergraduate students reported to drink more alcohol on a usual drinking occasion. However, neither age nor education level reached a significant level in the alcohol variable heavy drinking. Since it is especially heavy drinking which can be considered as a risky healthy behavior and lower levels of alcohol consumption might also have positive health effects (Corrao, Rubbiati, Bagnardi, Zambon, & Poikolainen, 2000; Fuller, 2011), these variables seem to be less important in alcohol prevention programs for college students. This speculation is supported by heterogeneous findings about the type of relationship between PA and alcohol consumption. Some studies presented a linear relationship (Moore & Werch, 2008), others showed a curvilinear relationship (Buscemi, Martens, Murphy, Yurasek, & Smith, 2011; Lisha, Sussman, FAPA, & Leventhal, 2013). Lisha, Sussman, FAPA, and Leventhal, (2013) pointed out that the relationship between PA and alcohol consumption might be dependent on the alcohol use pathology continuum with a positive association for moderate alcohol consumption and a negative association for alcohol dependence.

In none of the regression models, PA was significantly associated with alcohol consumption. The present findings in students confirm previous findings, where no relationship between alcohol consumption and PA was found (Kopp et al., 2015). Consequently, for Austrian college students, PA might play a minor role in alcohol prevention programs.

Limitations

The following limitations have to be considered when interpreting the findings: Firstly, probably the most critical point of the study is the low number of inactive students (n = 32), which might reflect a possible selection bias. Every student of the university received the link to the questionnaire, but there was no obligation to answer it. Highly active students who were interested in the topic might have answered more likely. Another possible explanation is that the urban area of Innsbruck with its surrounding mountains has the reputation of being a “sport city”. A former study on health status in women conducted in Innsbruck reported 71% to exercise regularly (Ulmer, Deibl, Jakel, & Pfeiffer, 2001). Also data from an Austrian survey presented 58% of the population to be highly active (Kopp et al., 2015). Thus, the sample might actually represent the Innsbruck college sample but might be questionable for other areas. Secondly, comparable to previous studies, we assessed PA and alcohol consumption by self-report of the students, which might introduce a recall bias. Alcohol consumption was operationalized by a questionnaire with uncertain psychometric values to be able to compare our findings with the findings of Musselman and Rutledge (2010). Future studies might consider using a validated measure for quantifying alcohol consumption (e.g., Alcohol Use Disorder Identification Test, Babor et al., 2001). Thirdly, we used a cross-sectional design und thus cannot assess causal effects between alcohol consumption and PA. Future studies might consider a prospective study design.

However, there are several strengths in the study to be mentioned: The present sample size was fairly large compared to some previous studies (Moore & Werch, 2008; Musselman & Rutledge, 2010) and we included various study fields in our sample. Therefore, we were able to examine possible effects of the study field. To the best of our knowledge, this is the first study including specifically college students in Austria and thus might help to gain a broader understanding of cross-cultural differences regarding alcohol consumption and PA.

Conclusions

The current results provide no evidence for a positive relationship between PA and alcohol consumption. Compared to other variables (e.g., sex, relationship status), PA might be less important in relation to the consumption of alcohol in Austrian college students. Programs to reduce alcohol consumption in Austrian college students might focus less on PA, but more on male students without relationship. The influence of different study fields on alcohol consumption needs further investigation. This study challenges a global perspective on a positive PA-alcohol consumption relationship which was mainly shown in studies conducted in the United States. It highlights the need for more cross-cultural investigations. Future studies in European samples might consider including more reliable measures both for alcohol consumption, e.g., daily diary (Conroy et al., 2015), and PA, e.g., accelerometer.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the article.

Competing interests

All authors declare that no competing interests exist.

Acknowledgments

We want to thank Prof. Rutledge for her kind reply to questions related to the alcohol consumption variables.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.