Introduction

Heavy drinking among university students has been shown identified as a major public health burden, mainly based on North American and European samples (Dantzer et al. 2006; El Ansari et al. 2013; Perera and Torabi 2012; Sebena et al. 2011; Slutske 2005; Venegas et al. 2012; Wicki et al. 2010). Few studies of university students have included low, middle income and emerging economy countries, indicating that the prevalence of hazardous alcohol use in Australasia, Europe and South America appears similar to that in North America, but is lower in Africa and Asia (Isralowitz and Hong 1988; Karam et al. 2007). Heavy drinking was found to be among male and female university students in Colombia (46 and 24 %, respectively; past 2 weeks: ≥5 drinks among men and ≥4 among women) (Dantzer et al. 2006), China (16.7 and 5.4 %, ≥5 drinks in past 30 days, Kim et al. 2009; 37.4 and 11.6 %, ≥5 drinks in past 30 days, Ji et al. 2012), Malawi (48.8 and 5.0 %, past 1 week: ≥6 drinks among men and ≥5 among women) (Zverev 2008), Nigeria (31.1 %, ≥6 drinks in the past 30 or less than 30 days) (Abayomi et al. 2013); South Africa (27 and 3 %, past 2 weeks: ≥5 drinks among men and ≥4 among women) (Dantzer et al. 2006), Thailand (32 and 7 %; past 2 weeks: ≥5 drinks among men and ≥4 among women) (Dantzer et al. 2006), Uganda (34.1 and 23.4 %, ≥6 drinks among men and women in past 12 months drinkers) (Stafström and Agardh 2013), Venezuela (32 and 15 %; past 2 weeks: ≥5 drinks among men and ≥4 among women) (Dantzer et al. 2006).

Various factors have been identified to be associated with heavy drinking among university students, including (1) socio-demographic factors such as male gender (Dantzer et al. 2006; Ji et al. 2012; Kim et al. 2009; Stafström and Agardh 2013; Wicki et al. 2010; Zverev 2008), age, religious affiliation, national per capita alcohol consumption, income, living away from home (Dantzer et al. 2006; Ji et al. 2012; Kim et al. 2009; Miskulin et al. 2010; Perera and Torabi 2012; Vantamay 2009; Wicki et al. 2010); (2) knowledge about health-related consequences (Dantzer et al. 2006) and attitudes towards alcohol use (Dantzer et al. 2006; Vantamay 2009), (3) health related variables such as tobacco use (Deressa and Azazh 2011; Kim et al. 2009; Wicki et al. 2010), illicit drug use (Atwoli et al. 2011; Wicki et al. 2010), gambling (Cheung 2014), physical activity (Wicki et al. 2010), and low life satisfaction (Murphy et al. 2005; Paul et al. 2011). The protection/risk model based on both psychosocial and behavioural protective and risk factors, derived from Problem-Behavior Theory has been used to explain heavy episodic drinking among university students (Jessor et al. 2006).

The aim of this study was to investigate heavy drinking and social and health correlates in university students in low, middle income and emerging economy countries over the same time period, using a standard questionnaire allowing for direct comparison.

Methods

Sample and Procedure

This cross-sectional study was carried out with a network of collaborators in participating countries (see “Acknowledgments”). The anonymous, self-administered questionnaire used for data collection was developed in English, then translated and back-translated into languages (Arabic, Bahasa, Chinese, French, Lao, Russian, Spanish, Thai, Turkish) of the participating countries. The study was initiated through personal academic contacts of the principal investigators. These collaborators arranged for data to be collected from intended 400 male and 400 female undergraduate university students aged 16–30 years by trained research assistants in 2013 in one or two universities in their respective countries. The universities involved were located in the capital cities or other major cities in the participating countries. Research assistants working in the participating universities asked classes of undergraduate students to complete the questionnaire at the end of a teaching class. Classes were selected through a stratified random sample procedure. The study was conducted in 2013. Participation rates were in most countries over 90 %. Ethics approvals were obtained from institutional review boards from all participating institutions. Countries included Barbados (n = 564), Grenada (n = 422), Jamaica (n = 737), Colombia (n = 816), Venezuela (n = 564), Cameroon (n = 627), Ivory Coast (n = 754), Madagascar (n = 776), Mauritius (n = 421), Namibia (n = 451), Nigeria (n = 723), South Africa (n = 740), Egypt (n = 801), Tunisia (n = 770), Turkey (n = 800), Russia (n = 789), Bangladesh (n = 781), India (n = 800), China (n = 1112), Indonesia (n = 750), Laos (n = 806), Philippines (n = 782), Singapore (n = 886), and Thailand (n = 858).

Measures

Alcohol use. Participants were asked about drinking alcohol, including beer, wine, spirits and any other alcoholic drink. Those who responded that they drink were asked to indicate “On how days over the past 2 weeks (14 days) did you have a drink?” and “On the days that you did drink, how many drinks did you have on average?” (Wardle and Steptoe 1991). Drinking volume in a typical day was then classified into three groups: non-drinkers, moderate drinkers (men ≤ 4 drinks and women ≤ 3 drinks in a sitting), and heavy drinkers (men ≥ 5 drinks and women ≥ 4 drinks in a sitting) (Dantzer et al. 2006; Wechsler and Nelson 2001).

Knowledge of alcohol use effects was asked with two questions, “do you believe heart disease is influenced by alcohol use”, and “do you believe high blood pressure is influenced by alcohol use?” Response option was “yes” or “no” (Wardle and Steptoe 1991).

Health beliefs in the importance of not drinking too much alcohol were assessed with one question. The response option ranged from 1 = of very low importance to 10 = of very high importance (Wardle and Steptoe 1991).

Tobacco use was assessed with the question: Do you currently use one or more of the following tobacco products (cigarettes, snuff, chewing tobacco, cigars, etc.)? (WHO 1998).

Illicit drug use was assessed with the question, “How often have you taken drugs in the past 12 months; other than prescribed by the health care provider.” Response options included 1 = 0 times, 2 = 1–2 times, 3 = 3–9 times and 4 = 10 or more times.

The South Oaks Gambling Screen (SOGS), a standardized measure of pathological gambling and gambling behaviours in their lifetime (Lesieur and Blume 1987) was used to assess nine different gambling behaviours, e.g., “Played cards for money.” Response options ranged from 1 = not at all to 3 = Once a week or more. Cronbach alpha for this nine item scale was 0.91 in this sample.

Physical activity was assessed using the self-administered International Physical Activity Questionnaire (IPAQ) short version, for the last 7 days (IPAQ-S7S). A sample item is, “During the last 7 days, on how many days did you do vigorous physical activities like heavy lifting, digging, aerobics, or fast bicycling?” We used the instructions given in the IPAQ manual for reliability and validity, which is detailed elsewhere (Craig et al. 2003). We categorized physical activity (short form) according to the official IPAQ scoring protocol (IPAQ 2006) as low, moderate and high.

Subjective general health status was assessed with the question, “In general, would you say your health is…?” (Rated from 1 = excellent to 5 = poor) and Life satisfaction, “All things considered, how satisfied are you with your life as a whole?” (Rated from 1 = very satisfied to 5 = very dissatisfied (Wardle and Steptoe 1991).

Socio-demographic questions included age, gender, residential status, religious affiliation and socioeconomic background. Response options for religious affiliation were: 1. Traditional religion, 2. Christian (Protestant, e.g., Lutheran, Anglican, etc.) 3. Christian (Catholic), 4. Hindu, 5. Moslem, 6. Buddhist, 7. No religion, 8. Other (specify). Socioeconomic status was assessed by rating their family background as wealthy (within the highest 25 % in “country”, in terms of wealth), quite well off (within the 50–75 % range for their country), not very well off (within the 25–50 % range from “country”), or quite poor (within the lowest 25 % in their country, in terms of wealth) (Wardle and Steptoe 1991).

Data Analysis

The data were analysed using STATA 11.00 (StatCorp LP, College Station, TX, USA). Descriptive statistics were used for reporting the proportion of moderate and heavy drinking. The product–moment correlation was used to compare the country per capita alcohol consumption with heavy drinking. Logistic regression was used to assess the association between sociodemographic variables, health knowledge and beliefs, health variables and heavy drinking. Since the study used a clustered design, country was included as a clustering variable in the regression models.

Results

Sample Characteristics

The sample included 17,590 university students (41.8 % men and 58.2 % women), with a mean age of 20.8 years (SD 2.9). Overall, 71.6 % were non-drinkers, 17.1 % moderate and 11.3 % heavy alcohol drinkers in the past 2 weeks. The overall prevalence of heavy drinkers among university students differed by country, ranging from below 3 % in Cameroon (3.4 % for men and 1.9 % for women, respectively), Nigeria (0.5 and 0.3 %), Egypt (0.3 and 0.2 %), Bangladesh (3.6 and 0.6 %), India (0.9 and 1.5 %), China (2.9 and 2.6 %) and Indonesia (0.0 and 0.2 %) to above 20 % in Colombia (62.2 % for men and 43.4 % for women, respectively), Venezuela (26.5 and 22.8 %), Namibia (25.9 and 21.5 %), South Africa (31.5 and 12.6 %), Russia (37.1 and 39.6 %) and Laos (46.5 and 26.3 %). Men were more frequently than women moderate (P < 0.001) and heavy drinkers (P < 0.001). Although the preponderance of heavy drinking among men were true for students from nine study countries, there were no significant gender differences in heavy drinking in 15 countries. The proportion of heavy drinkers was positively correlated across study countries with the per head country pure alcohol consumption (recorded and unrecorded), ranging from 0.2 in Bangladesh to 15.1 in Russia (WHO 2014) (r = 0.22, P < 0.001).

Associations with Heavy Alcohol Use

In multivariate logistic regression analysis, among men, older age [22–30 years: odds ratio (OR) 1.91, confidence interval (CI) 1.50–2.43], coming from a not well off or poor family background (OR 1.35, CI 1.15–1.86), living in an upper middle income or high income country (OR 1.84, CI 1.36–2.42), having a Christian and other or no religious affiliation (OR 1.35, CI 1.01–1.78), weak beliefs in the importance of limiting alcohol use (OR 2.12, CI 1.76–2.56), higher country per capita alcohol consumption (OR 1.12, CI 1.09–1.15), other substance use (tobacco and illicit drug use) (OR 2.29, CI 1.86–2.83, and OR 1.69, CI 1.36–2.06), poor subjective health status (OR 0.74, CI 0.68–0.81), poor life satisfaction (OR 0.52, CI 0.41–0.66), and high physical activity consumption (OR 1.31, CI 1.07–1.61), were associated with heavy drinking. In multivariate logistic regression analysis, among women, older age (22–30 years: OR 1.76, CI 1.25–2.48), coming from a not well off or poor family background (OR 1.63, CI 1.20–2.23), living in an upper middle income or high income country (OR 2.70, CI 1.95–3.75), having a Christian and other or no religious affiliation (OR 1.32, CI 1.04–1.73), weak beliefs in the importance of limiting alcohol use (OR 2.61, CI 2.01–3.39), higher country per capita alcohol consumption (OR 1.29, CI 1.21–1.38), other substance use (tobacco and illicit drug use) (OR 2.15, CI 1.48–3.11, and OR 1.50, CI 1.13–2.01), frequent gambling (OR 2.34, CI 1.33–4.09), and poor life satisfaction (OR 0.49, CI 0.35–0.71) were associated with heavy drinking. Residential status and knowledge variables (alcohol-heart disease and alcohol-high blood pressure link) were not associated with heavy drinking.

Discussion

The study found, among a large sample of university students from 24 low, middle income and emerging economy countries across Asia, Africa and the Americas, a large group of past 2 weeks non-drinkers, a third drinkers and 11.3 % heavy alcohol drinkers, which compares as lower than students in North America and Europe (El Ansari et al. 2013; Dantzer et al. 2006; Perera and Torabi 2012; Sebena et al. 2011; Slutske 2005; Venegas et al. 2012; Wicki et al. 2010). However, the study found a large country variation in the overall prevalence of heavy drinkers among university students, which in part confirms findings from previous studies of above 20 % in Colombia (Dantzer et al. 2006), Venezuela (Dantzer et al. 2006), Nambia (Peltzer 2009), South Africa (Dantzer et al. 2006), and Russia (Stickley et al. 2013). In contrast to previous studies among university students in China (Kim et al. 2009; Ji et al. 2012), Nigeria (Abayomi et al. 2013) and Thailand (Dantzer et al. 2006), this study found lower proportions of heavy drinking in these countries. Contrary to the relatively high alcohol consumption in the general population in Cameroon (8.4 L of pure alcohol per capita consumption) and Nigeria (10.1 L), this study found a low prevalence of heavy drinking among university students in Cameroon and Nigeria. A low proportion of heavy drinking was expected among students from countries who are predominantly Hindu or Muslim, including Egypt, Bangladesh, India, Indonesia, Mauritius, Turkey, and Tunisia. In all these countries a low prevalence of heavy drinking among university students was found except for Tunisia, where 21.9 % of the men were heavy drinkers. The finding that in Tunisia despite a low proportion of drinkers in the study student population, a high proportion of heavy drinkers were found seems in concordance with a study on the drinking pattern of adults in African countries, namely the likelihood that current drinkers were heavy drinkers (an ‘all-or-none’ drinking pattern) increased with increasing prevalence of lifetime abstinence (Clausen et al. 2009). Reasons for this may be that informal social control of drinking pertains to any drinking rather than heavy drinking occasions (Clausen et al. 2009). In Egypt the low prevalence of heavy drinking in students correlates with the low alcohol consumption in the general population (WHO 2014).

In concordance with a number of other studies (e.g., Miskulin et al. 2010; Vantamay 2009; Wicki et al. 2010), the study found across the countries that men engaged more frequently in heavy drinking than women. However, the preponderance of heavy drinking among men were only true for students from nine study countries, while there were no significant gender differences in 15 countries. No gender differences in heavy drinking were found in students from low prevalence heavy drinking Hindu or Muslim cultures, medium prevalence in Caribbean countries and also a high prevalence of heavy drinking in students from Namibia and Russia. Gender differences in heavy drinking may be influenced by biological and cultural factors such as self-restraint of drinking by women (Wilsnack et al. 2009) and women’s position in society (Rahav et al. 2006). It is possible that in the Caribbean study countries and Namibia and Russia, with a medium and high prevalence of heavy drinking among both men and women, a higher women’s position in the society reduces the difference between men and women drinking rates. In relation to the age or year of study, we found that older students and later year of study (not shown in the results) were more likely to report heavy drinking. This is in agreement with a study among Nigerian university students, where older age was associated with problem drinking (Abayomi et al. 2013), but in contrast to some previous studies in European university students (Sebena et al. 2011) where the higher study year was associated with lower levels of heavy episodic drinking. Unlike in some other studies (e.g., Sebena et al. 2011), this study did not find that living with parents or guardians was associated with less frequent heavy drinking.

Further, the study found that coming from a not well off or poor family background and living in an upper middle income or high income country was associated with heavy drinking. Previous studies have shown that poorer economic background or perceived economic insufficiency was associated with heavy drinking (El Ansari et al. 2013), while Dantzer et al. (2006) found among university students in 16 developed and 5 developing countries that wealthier family background was associated with heavy drinking. On the other hand university students in this study are more likely to drink heavily if they come from a higher income level country compared to a lower income level country, which conforms to previous studies (Dantzer et al. 2006). Higher national per capita alcohol consumption was found, as in a previous study (Dantzer et al. 2006), related to greater frequency of heavy drinking in this study population. As expected, being a Hindu or Muslim was less likely than being a Christian to be associated with heavy drinking (Clausen et al. 2009; Wicki et al. 2010).

After adjusting for relevant socio-demographic variables, the current study findings indicate associations between heavy drinking and a number of negative or risky health-related variables including substance use (tobacco and illicit drug use), gambling, poor life satisfaction and poor subjective health, which is largely consistent with findings of previous research (Atwoli et al. 2011; Cheung 2014; Deressa and Azazh 2011; Kim et al. 2009; Murphy et al. 2005; Paul et al. 2011; Wicki et al. 2010). It is possible that these behaviours are all likely to be interrelated, and interventions may need to target these problems comprehensively (Paul et al. 2011). Contrary to what might be expected, male students who reported heavy drinking also reported some positive health behaviour, i.e., high physical activity, as found in some previous research (Paul et al. 2011). It is possible that as found in a previous study (Akmatov et al. 2011), where students from sport faculties had a higher risk of heavy drinking, that greater physical activity, including sports activities involves greater promotion and increase of heavy drinking.

In agreement with previous research (Dantzer et al. 2006; Paul et al. 2011), this study found heavy drinkers perceived less health risk related to drinking than individuals who drank less than them. It is also possible, however, that the perceived health risks of heavy drinkers in this young student sample is relatively low, considering the short time since the initiation of drinking (Paul et al. 2011).

Using the protection/risk model based psychosocial and behavioural protective and risk factors (Jessor et al. 2006), several elements, which were assessed in this study, seem to confirm the protection/risk model for heavy drinking in this university student population across different countries in three continents. For example, the problem behaviour of heavy drinking was associated with other behavioural risk factors including tobacco use, illicit drug use and gambling.

Study Limitations

This study had several limitations. The study was cross-sectional, so causal conclusions cannot be drawn. The investigation was carried out with students from one or two universities in each country, and inclusion of other centres could have resulted in different results. University students are not representative of young adults in general, and the alcohol use levels, sociodemographic and health variables may be different in other sectors of the population. A further limitation of the study was that all information collected in the study was based on self-reporting. It is possible that certain behaviours were under or over reported. Participants were asked about the number of days that they drank (i.e., frequency) and number of drinks per drinking day (i.e., quantity). These questions have shown in previous research an underestimation of the prevalence of heavy drinking, since often binge drinking is not factored in into such estimates (Stahre et al. 2006). Moreover, one other concern may be that students across different cultures are aware of what constitute a single serving or standard unit of alcohol. Kerr and Stockwell (2012) found that drinkers have difficulty in defining standard drinks and typically underestimate the intake volume. However, this possible underestimation of drinking quantity further emphasizes the relevance of the present study findings. Further, White and Hingson (2013) emphasize that “the discrepancies between self-reported and actual drinking levels are not large enough to question the general findings of college drinking surveys.” Also, theoretical concepts such as constructs of the theory of planned behaviour of binge drinking (Elliott and Ainsworth 2012) or the protection/risk model (Jessor et al. 2006) were not assessed and should be included in future surveys.

Conclusions

This study confirms low to high levels of heavy drinking in different cultures across Africa, Asia and the Americas. Various factors identified, such as socioeconomic status, weak beliefs in the importance of limiting alcohol and co-occurrence of substance use and gambling can be used to guide interventions to reduce heavy drinking among university students. The public health implication of the findings may be that alcohol use health promotion should also include other problem behaviours such as tobacco use, illicit drug use and gambling.