This study was carried out as part of the uniHcos project (a dynamic cohort of university students to study drug use and other addictions), a multicenter project studying habits and lifestyles and their development during university years by creating a dynamic cohort of freshmen in their first term.23 One of uniHcos’ main objectives is to study drug use habits.

We carried out a cross-sectional observational study based on data from students enrolled in the first year of various degrees offered by Spanish universities participating in the uniHcos project (León, Cantabria, Jaen, Vigo, Granada, Huelva, Salamanca, Valladolid, Alicante, and Valencia). These students received an email with a link to information about the project and a self-administered questionnaire created by the SphinxOnline® platform. All students gave informed consent to participate and the project has the approval of the Ethics Committee of the University of León regarding the use of personal data.

The self-administered uniHcos questionnaire includes 373 items on various areas: sociodemographic characteristics (sex, age, marital status, etc.), habits and lifestyles (exercise, food, problematic Internet use, etc.), and consumption of legal or illegal substances (alcohol, tobacco and others); this questionnaire was based on that used by the EDADES survey.2

Since the purpose of uniHcos is to create a cohort and track it, we did not determine a minimum sample size for this study. We included in this study all students who completed the self-administered form and signed the informed consent between October 2011 and March 2015 (overall response rate 4.2%; N = 4166). We excluded students over 25 years of age (n = 314, 7.5%) due to characteristic differences with respect to the target population. Also we excluded, due to the differences found in previous analyses realized in the same sample, students residing in their own flat (n = 29, 0.8%) and those who could not determine their type of residence (57, 1.5%) both due to differences with the study population and the small number that these students represent regarding to the total sample; thus, the final sample included 3767 students. Furthermore, in order to avoid possible misinterpretations individual (psychological) factors were excluded of our analyses. These factors were excluded due its possible relationship with the family support and the difficulty to take this relationship into account adequately related to the indicators used in our survey to measure psychological factors.23

Data on substance use was collected by the uniHcos questionnaire, which includes items regarding consumption of the following illegal drugs on any previous occasion, during the previous year, and during the previous month: cannabis, non-prescription sedatives, central nervous system (CNS) stimulants (collectively considered as cocaine, ecstasy, speed, and hallucinogens), and other CNS depressants (collectively considered as heroin, GHB, and inhalants). Cannabis and non-prescription sedatives were evaluated separately, both because of the large number of people who reported having used them and because of the differences in consumption patterns compared to other drugs analyzed. We considered stimulants and other depressants together because few people reported having used them, and we also recorded the number of substances tried by each participant, evaluating each individual substance with a specific question. We considered poly-drug users to be those who reported having used two or more substances or groups of substances when one of them was a stimulant or a depressant, and non-users to be people who had not tried any of the drugs mentioned; the latter was used as the reference category for analysis.

We also evaluated the average age of onset of use or average age of first use for each drug and each group of drugs. In substance groups (stimulants and depressants), we considered the averages of the ages of onset for the different drugs in each group.

• •

  Gender: men; women.

• •

  Age: in years, analyzed as a continuous variable.

• •

  Employment status: studying and looking for work; studying and working; studying but not looking for work.

• •

  Residence: family home; university residence; rented apartment.

• •

  Family support: We used the valid and reliable FAMILY APGAR questionnaire to assess each participant's perception of the state of his family's functioning at a given time.24 The questionnaire consists of 5 Likert items with a 3-point scale from 0 to 2. We categorized the scores as follows: 7-10 points, normal support; 3-6 points, slightly dysfunctional; and 0-2 points, extremely dysfunctional. In addition to using this variable as a predictor, we also used it as a stratifying variable to evaluate consumption rates, with the aim of exploring its association with substance use.

A descriptive analysis of substance use and polyconsumption stratified by gender and family support were performed. Statistically significant differences were tested at the bivariate level using the chi-squared test for categorical variables and Mann-Whitney U test for mean age taken as a continuous variable with non-normal distribution. To analyze the relationship between any previous consumption and each of the independent variables, multivariate logistic regression models were fitted and adjusted odds ratios (aOR) and their respective 95% confidence intervals (95%CI) were calculated. All models were stratified by gender, and adjusted for all covariates. Statistical analyses were carried out using STATA v. 13 software.25

Regarding the prevalence of people who had ever consumed the substances evaluated, the results regarding the use of cannabis and non-prescription sedatives are notable. The prevalence of stimulant and depressant consumption was similar to that observed in previous studies, and even lower for non-prescription sedatives, with no differences in consumption between men and women; however, the prevalence of cannabis use (men: 49.2%; women: 42.0%) was higher than that observed in data from a representative sample of the Spanish youth population (about 40%).2,26,27 Similarly, we observed that everyone reporting polyconsumption had tried cannabis, which could be important for designing prevention strategies.

For all drugs evaluated, we observed much higher prevalence in consumption during the previous year and month than in the general youth population.2 This again highlights the start of university as a particularly important moment, both in terms of drug use among young people as well as for prevention, since many who reported having tried drugs at some point in their life had also consumed in the last year or month.

Notably, the high prevalence of cannabis consumption could be related to differences between the university population and the general youth population, an also to a decrease in perception of the risk associated with consumption observed in recent years. Various studies have found that low perception of risk is associated with higher rates of drug use.28–31The ages of first use were slightly lower than those found in the general population, around 18 years for first cannabis use and slightly higher for other drugs.2 In addition, while the age of first cannabis use was similar in men and women, that for stimulants was lower among women. These results differ from those of previous studies, where the age of onset was generally lower among men, and provide new evidence to support previous results that point to a possible decrease in the age of onset of consumption, which could be more pronounced among women.19 This should also be taken into account for subsequent studies because people who use drugs at an earlier age have greater risk of developing addiction later, such that consumption may have a greater impact on health and cognitive development at older ages.1,5

We observed that more than 70% of college students perceived that their family support was normal, which is consistent with previous reports from studies in non-university populations.24,32 Additionally, we also observed that when perceived support is worse, the proportion of people who reported having consumed illegal drugs is higher. These findings provide new evidence to support the hypothesis that family functioning influences drug use,3,11,14,16 and highlights the importance of considering this factor when evaluating consumption. This factor is also involved in prevention because improving support at the family level or providing families with strategies that culminate in increased perceived support may help prevent substance use.3

The fact that family functioning is more strongly linked to illegal drug use among women than men supports gender differences in associated factors and consumption patterns, as proposed in previous studies.16,19 Among other factors, this difference could be related to the social role of women in countries with traditional family structures like Spain, which non-explicitly assigns women the chief role within the family.33,34 Thus, policies that promote equality and reduce social inequalities between men and women may help prevent unhealthy habits, like drug use, as well as physical and mental health problems.3

In terms of factors associated with having tried drugs previously, we found lower rates of cannabis use among women who lived in the family home, as previously noted by Molina et al.21 This may be because women who live in the family home have an authority figure who has greater control over them, compared to those living in university residences or shared flats.

Regarding employment status, we found higher a prevalence of cannabis use among in both men and women who were studying and looking for work, and of stimulant use and polyconsumption among women. This has not been observed in previous studies, as far as we know, and represents a new perspective on the relationship between drug use and employment in university students, although it may be related to increased stress and economic vulnerability, among other things. Moreover, the results compared to other employment situations evaluated in this study are different to those obtained in previous studies where employment was found to be protective against substance use.17,18 Therefore, in future studies assessing factors associated with drug use among university students, especially among women, it may be useful to consider home address and employment status, the influence of parental or institutional control, and economic stress.

Finally, the differences we found in factors related to consumption could have important implications for prevention. As suggested previously,19,35 it might be appropriate to consider introducing the gender perspective in prevention programs, both in school and at home. Also, we consider it important in future studies to take into account other variables not considered in our analyses as the country of origin of the students and their families, or the social class that they belong. These variables could be related to drug use in the university student population (as protective factors or as risk factors) due to the differences previously found between countries and social classes,36–39 and might allow having a wider view in order to develop more specific drug-use prevention programs.

Due to the population characteristics, type of questionnaire, and the objectives of the uniHcos project, the main limitation of our study is that we were unable to use a probabilistic sampling approach, which may affect our inferences about the student population in this sample and the corresponding results. Despite this limitation, the study population could be considered valid for our objective, both in this paper and in the uniHcos project, because our aims are to describe the illegal drug use in this population and to analyze some related factors, and the aim of the uniHcos project is to create a cohort of students to study long-term habits and lifestyles. Furthermore, we are currently working to improve recruiting programs and the response rate so that in the near future we expect a substantial increase in the size of the sample and its representativeness.

Another possible limitation is that the questionnaire used has not yet been validated. However, this questionnaire consists of validated questions and scales from previously validated national questionnaires, such as the National Health Survey of Spain or the EDADES survey among others.2

Other limitation of our study is its design. While cross-sectional studies can determine prevalence, they cannot establish causality between drug use and the other variables considered, especially family support. Nevertheless, based on biological plausibility and the results of previous longitudinal studies, the observed direction of effect, at least in the case of family support, might be correct.40 It may be appropriate in the future to verify these results with longitudinal data.

Finally, we emphasize a possible limitation related to drawing conclusions about consumption during the previous year and month, since models were only fit for consumption at any previous time in life. Despite this limitation, we fit models using this variable because one of the objectives of this study was to determine whether family support and other factors were related to drug use on any previous occasion. Moreover, our approach allows us to include more participants in the sample, and thereby obtain more robust conclusions.