Identifying predictors and prevalence of alcohol consumption among university students: nine years of follow-up

RESEARCH ARTICLE Identifying Predictors and Prevalence of Alcohol Consumption among University Students: Nine Years of Follow-Up Lucı ´a Moure-Rodrı ´guez 1 *, Marı ´a Piñeiro 1 , Montserrat Corral Varela 2 , Socorro Rodrı ´guez- Holguı ´n 2 , Fernando Cadaveira 2 , Francisco Caamaño-Isorna 1 1CIBER de Epidemiologı ´a y Salud Pu ´blica (CIBERESP), Department of Preventive Medicine, Universidade de Santiago de Compostela, Santiago de Compostela, Spain, 2Department of Clinical Psychology and Psychobiology, Universidade de Santiago de Compostela, Santiago de Compostela, Spain *[email protected] Abstract Aim To evaluate the prevalence of alcohol consumption among university students during late adolescence and young adulthood and to identify the associated factors. Material and Methods Cohort study among university students in Spain (n = 1382). Heavy Episodic Drinking (HED) and Risky Consumption (RC) were measured with the Alcohol Use Disorders Identification Test (AUDIT) at ages 18, 20, 22, 24 and 27 years. Data on potential factors associated with alcohol use were obtained with an additional questionnaire. Multilevel logistic regression for repeated measures was used to obtain adjusted OR (Odds Ratios). Results The rates of prevalence of RC were lower, but not statistically significant, in women. The age-related changes in these rates were similar in both genders, and the prevalence of RC peaked at 20 years. By contrast, the prevalence of HED was significantly lower in women and peaked at 18 years in women and at 22 years in men. Multivariate models showed that early age of onset of alcohol use (OR = 10.6 and OR = 6.9 for women; OR = 8.3 and OR = 8.2 for men) and positive expectations about alcohol (OR = 7.8 and OR = 4.5 for women; OR = 3.6 and OR = 3.3 for men) were the most important risk factors for RC and HED. Living away from the family home was also a risk factor for both consumption patterns among women (OR = 3.16 and OR = 2.34), while a high maternal education level was a risk factor for RC among both genders (OR = 1.62 for women; OR = 2.49 for men). Conclusions Alcohol consumption decreases significantly at the end of youth, with higher rates of prevalence and a later peak among men. Prevention strategies should focus on beliefs and PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 1 / 15 a11111 OPEN ACCESS Citation: Moure-Rodrı ´guez L, Piñeiro M, Corral Varela M, Rodrı ´guez-Holguı ´n S, Cadaveira F, Caamaño-Isorna F (2016) Identifying Predictors and Prevalence of Alcohol Consumption among University Students: Nine Years of Follow-Up. PLoS ONE 11(11): e0165514. doi:10.1371/journal. pone.0165514 Editor: Hajo Zeeb, Leibniz Institute for Prvention Research and Epidemiology BIPS, GERMANY Received: May 30, 2016 Accepted: October 13, 2016 Published: November 3, 2016 Copyright: ©2016 Moure-Rodrı ´guez et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper. Our study was approved by the Bioethics Committee of the Universidade de Santiago de Compostela. Subjects were informed verbally and in written format, as part of the questionnaire, that participation was voluntary, anonymous, and the possibility to opt-out was available at any time. Subjects also were informed about the e-mail when they should write in order to opt-out. Finally, members of the research team explained to the students in the classroom that the expectations about alcohol and on delaying the age of onset. Women are at particular risk for these consumption patterns if they live away from their parents. Belonging to a highincome family is a strong risk factor for RC. Introduction Alcohol consumption constitutes a public health problem worldwide, and Europe has the highest rate of consumption per capita [1]. In Spain, alcohol consumption is deeply rooted in social customs and until recently was traditionallyassociated with family and socialevents, with mainly men partaking[2]. However, in the last decade,heavy episodicdrinking(HED), characterizedby the intake of large amounts of alcohol ina short periodof time(producingblood alcohol levels of at least 0.8 g/l), has replaced the traditional pattern of consumption among young people [2]. Although in the general population alcohol consumption is more prevalent in men than in women, this distinctionis becomingblurred in young adults, particularlyregarding HED, the prevalence of which is similar in both sexes [3]. To date, the age-related distributionof prevalence rates has followed a bell-shaped curve,peakingat around 21 years in the US [4] and at between19 and 24 years dependingon thecountryand genderin European studies, usually with earlier peaks in females [3,5]. One of the latest and most widely cited reviews on the topic [6] has mainly cited the following variables for HED in university students: gender, age, ethnicity, religion,expectationsabout alcohol, age of onset of alcoholconsumption, consumption of other drugs,health and stress, personality, physical activity, socioeconomiclevel, living circumstances,and any family history of alcoholism.Prevalencerates of HEDfound in thesestudies can be explained by the importanceof different risk factors in different European countries. The brains of adolescents and young adults are particularlysusceptibleto the neurotoxic effectsof HED [7–10]. Cohortstudies have shown an association betweenthis pattern of alcohol consumption and cognitive[7], structural[8] and neurofunctionaleffects[9]. Moreover, in the university environment, HED has been associatedwith poorer academicperformance[11], greater consumption of medicines[12], higher incidenceof injuries [13] and a higherincidence of risky sexual behavior [10], relative to control subjects.Finally, it is important to take into account that codesof behavior acquired duringadolescencetend to be maintained in adulthood[14]. The objectiveof this study was to evaluate the prevalenceof alcohol consumption among university students duringadolescenceand young adulthoodand to identifythe associated factors. Materials and Methods Design, Population and Sample We carriedout a cohort study among university students (Compostela Cohort,Spain), between November 2005 and February2015. We usedcluster sampling to selectthe participants.Thus, at least one of the first-year classes was randomly selectedfrom each of the 33 university faculties/schools(a total of 53 classes). The number of classes selectedin each university faculty or schoolwas proportional to the number of students. All students present in the class on the day of the surveywere invited to participate in the study (n = 1382). Abstinent students were excluded from the association analysis, although the numbers are included in the sample Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 2 / 15 data would be guarded carefully by the research team. Therefore, for these last two reasons we consider that we do not upload the dataset to a stable, public repository. However, the authors agree to make freely available any materials and data described in the publication upon reasonable request to the Corresponding Author. Funding: Funding for this study was provided by the Spanish National Plan on Drugs (PND, SPI/ 3462/2010). PND played no further role in this study. Competing Interests: The authors have declared that no competing interests exist. description.This study was approved by the BioethicsCommitteeof the Universidade de Santiago de Compostela.Subjects were informedboth verbally and in written format, as part of the questionnaire, that participationwas voluntary, anonymous, and the possibility to opt-out was available at any time.Subjects were informedthat they were free to fillor refuseto fillthe questionnaire. This procedure was approved by the BioethicsCommittee. Data Collection Procedure Two team researchers visited each first-year classroom in November 2005 and invited all students present in the class to participate in the study. Participants were evaluated via a selfadministeredquestionnaire in the same classroom (1st questionnaire).In November 2007, the same team of researchers visited the third-yearclassroom in order to follow-up with the students. Participants were re-evaluated via a self-administeredquestionnaire (2nd questionnaire). The questionnaires were linked using birth date, sex, school,and class. Students who provided a phone number in the first or secondquestionnaire were furtherevaluated by phone at 4.5-, 6.5-, and 9.0- year follow-ups (3rd, 4rd and 5rd questionnaire).On all five occasions, alcoholuse was measuredwith the Galicianvalidated version of the AUDIT [15,16]. In addition to the AUDIT, we used another questionnaire that asked about the potential factors associated with alcohol use (educationallevel and alcohol use of parents, alcohol-related problems and age of onset of use). One of the items in thesecondquestionnairespecificallyreferredto expectationsabout alcohol use. In this question the students were required to rank 7 positive and 7 negative expectationsabout the effectsof alcohol. This question was generated using items from a questionnaire previously used with young Spanish adults [17]. More details about data collectionare available in the following references [11,12,18]. Definition of variables Independentvariables. Severalsocio-demographicvariables were considered: gender, place of residence (parental home/away from the parental home), and maternal educational level (primary school/highschool/university). Alcohol use in the family was included as mother’s alcohol use (doesn’t consume/consumes).Four categorieswere definedfor age of onset of use (after 16 years old, at 16, at 15, before the age of 15). Cannabis consumption at the beginningof the study was measuredwith the question “Do you consume cannabis when you go out? Never; Sometimes;Most of the times; Always”. The categorieswere recategorizedinto No ("never")and Yes ("sometimes"or “most of the time” or “always”). Tobacco consumption at the beginningof the study was also measured as a dichotomous variable, without any temporal reference: No/Yes. Finally, takingthe number of positive and negative expectationsinto account, a score ranging from 0 to 14 was generated (0 beingthe maximum of negative expectanciesand 14 the maximum of positive expectancies).The scores were dividedinto tertiles. Dependentvariables. 1. Risky consumption (RC). Dichotomous variable generated from AUDIT score. A different cut-off value was established according to gender:= >5 for women; and = >6 for men. These cut-offs are recommendedin the Galicianvalidated version of the AUDIT. 2. Heavy episodicdrinking(HED). This is a dichotomous variable generated from the third AUDIT question “How often do you have 6 or more alcoholic drinks per occasion?”, which was codedas follows: never= 0, less than once a month = 0, once a month = 1, once a week= 1, daily or almost daily = 1. The sensitivity and specificityof this question with this Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 3 / 15 cut-off value are respectively0.72 and 0.73, and the area under the curveis 0.767 (95% CI: 0.718–0.816)[19]. Statistical analysis We used multilevellogisticregression for repeated measures to obtain adjusted Odds Ratios (OR) for independent variables from the final RC and HED models. Confidenceintervals of 95% (95% CI) were calculated. These models are more flexiblethan traditional models and therefore allow us to work with correlated data. As this is the case,there is a dependencystructure. The same subject was measuredseveral times, and the responses were strongly correlated. The university faculty/schooland classroom were considered to be random variables. The follow-up time was included as an offset term. Maximal models were generated, including all theoreticalindependentvariables. From these maximal models, final models were generated. Final models included all significantvariables or non-significant variables when their exclusion changed the OR of other variables by more than 10%. Data were analyzed using GeneralizedLinearMixed Models in R statistical software. Results The characteristicsof the sample of women and men are summarisedin Tables 1and 2respec- tively. There were no significantdifferencesbetweenthe sexes in relation to any of these variables. The rates of prevalence of RC were always lower in women, although the differenceswere not statisticallysignificant.The age-related changes in these rates were similar in men and women, and the prevalenceof RC peakedat 20 years. By contrast, the prevalence of HED was significantlylower in women of all ages and peakedat 18 years in women and at 22 years in men. The rates of prevalenceof RC and HED in each age group of women and men are shown in Tables 3and 4respectively. Figs 1and 2present the trends of prevalencesof RC and HED, for both sexes. In relation to the factors associated with RC and HED, the multivariate modelrevealedthe followingas risk factors in both sexes:early onset of alcohol consumption (OR = 10.7 and OR = 6.90 in women compared with OR = 8.30 and OR = 2.23 in men) and positive expectationsabout alcohol (OR = 7.80 and OR = 4.54 in women compared with OR = 3.59 and OR = 3.26 in men). These data are presented in Tables 5and 6. Older age of participants constituted a protective factor (OR = 0.06 and OR = 0.13 in women compared with OR = 0.11 and OR = 0.34 in men), and the effectwas greater in women than in men for both patterns of consumption. Finally, in relation to the sociodemographic characteristics of the family, high maternal educational level acted as a risk factor only for RC (OR = 1.62 in women compared with OR = 2.49 in men), while the place of residence was significant only in women for bothRC and HED (OR = 3.16 and OR = 2.34). These data are presented in Tables 5and 6. Discussion The findingsindicate high rates of prevalenceof both RC and HED in the population of university students under study. Moreover, multivariate models showed that early age of onset of drinkingand positive expectationsabout alcohol were the most important risk factors for RC and HED respectively. Finally, living away from the family home also constituted a risk factor Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 4 / 15 for bothpatterns consumption among women, whilea high maternal educational levelwas found to be a risk factor for RC in both women and men. The distributionof changes in prevalence of HED followed a bell-shaped curve,similar to that describedin previous studies [4], although with an earlier peakin women than in men. Although the HED pattern is generallyconsidered to peak during the vital life stage between adolescenceand young adulthood[20,21], which has beenexplained as a consequenceof the acquisition of new social roles and responsibilities at this stage [22], this was only confirmedin Table 1. Characteristics of female initial sample and follow-up samples. Percentage or mean (95%CI) Initial (18–19 years old) n = 992 2-year follow-up (20– 21 years old) n = 669 4-year follow-up (22– 23 years old) n = 461 6-year follow-up (24– 25 years old) n = 266 9-year follow-up (27– 28 years old) n = 325 p-value Maternal educational level Primary school 41.8 (38.4–45.3) 44.2 (40.1–48.4) 43.1 (38.3–48.3) 47.3 (41.3–54.1) 45.7 (40.1–51.8) High school 33.6 (30.2–37.1) 30.5 (26.4–34.7) 30.6 (25.8–35.8) 26.5 (20.4–33.3) 28.1 (22.5–34.2) University 24.6 (21.2–28.1) 25.3 (21.3–29.6) 26.3 (21.4–31.4) 26.1 (20.1–32.9) 26.2 (20.7–32.4) 0.642 Residence In parental home 24.7 (22.1–27.5) 22.9 (19.7–26.1) 22.2 (18.5–26.0) 22.1 (18.1–26.1) 20.9 (16.5–25.1) Away from the parental home 75.3 (72.6–78.0) 77.1 (74.0–80.3) 77.8 (74.1–81.6) 77.9 (73.9–81.9) 79.1 (74.9–83.5) 0.720 University entrance grade a 9–10 points 45.6 (42.3–49.1) 43.2 (39.3–47.5) 43.2 (38.4–48.3) 43.3 (38–3–48.8) 47.1 (41.3–52.9) 7 - <9 points 48.8 (45.5–52.3) 50.3 (46.4–54.6) 50.6 (45.8–55.6) 50.1 (45.1–55.6) 47.1 (41.3–52.9) 5 - <7 points 5.6 (2.2–9.0) 6.5 (2.5–10.7) 6.2 (1.4–11.2) 6.6 (1.6–12.1) 5.8 (0–11.6) 0.977 Positive expectations about alcohol Low 37.1 (33.4–40.9) 37.5 (33.2–42.1) 36.5 (31.4–42.0) 36.5 (30.9–42.3) 37.9 (31.7–44.3) Medium 34.0 (30.3–37.8) 32.6 (28.3–37.3) 34.6 (29.4–40.1) 35.4 (29.8–41.1) 34.8 (28.6–41.2) High 28.9 (25.2–32.7) 29.9 (25.5–34.5) 28.9(23.7–34.4) 28.1 (22.5–33.8) 27.2 (21.0–33.6) 0.999 Age of onset of use of alcohol After 16 years old 19.0 (16.5–21.8) 17.9 (14.9–21.3) 16.5 (13.0–20.5) 16.7 (12.1–22.5) 14.5 (10.5–19.2) At 16 years old 38.9 (35.6–42.2) 38.1 (34.1–42.2) 36.8 (32.0–41.7) 40.1(33.6–46.8) 36.6 (30.9–42.6) At 15 years old 25.6 (22.7–28.7) 25.9 (22.3–29.6) 26.5 (22.2–31.1) 26.4 (20.8–32.7) 28.3 (23.0–34.0) Before age of 15 years 16.5 (14.0–19.7) 18.1 (15.0–21.5) 20.3 (16.4–24.5) 16.7 (12.1–22.5) 20.7 (16.0–25.9) 0.438 Heavy episodic drinking b Never 61.2 (58.2–64.3) 61.3 (57.7–65.1) 59.0 (54.7–63.7) 59.4 (53.8–65.5) 60.0 (54.8–65.4) Less than once a month 20.9 (17.8–23.9) 20.9 (17.3–24.7) 23.4 (19.1–28.1) 22.2 (16.5–28.3) 22.5 (17.2–27.9) Monthly 9.8 (6.7–12.8) 9.1 (5.5–12.9) 9.1 (4.8–13.8) 9.8 (4.1–15.9) 9.8 (4.6–15.3) More frequently 8.2 (5.1–11.2) 8.7 (5.1–12.5) 8.5 (4.1–13.2) 8.6 (3.0–14.8) 7.7 (2.5–13.1) 0.999 AUDIT: Total (mean) 5.4 (5.2–5.7) 5.6 (5.1–5.8) 5.6 (5.2–6.0) 5.6 (5.0–6.1) 5.3 (4.9–5.8) 0.884 Cannabis consumption 18.6 (16.2–21.1) 19.0 (15.9–22.0) 20.6 (16.8–24.4) 18.0 (13.2–22.9) 18.8 (14.4–23.2) 0.942 Tobacco consumption 31.0 (28.1–34.0) 31.5 (27.9–35.1) 34.3 (29.8–38.7) 31.2 (25.4–37.0) 32.9 (27.7–38.2) 0.786 a Variable with a scale of 1 to 10 b Question 3 of the AUDIT. doi:10.1371/journal.pone.0165514.t001 Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 5 / 15 women in the present study. We found that one in fivemen of age 27 years continued to follow this pattern of alcohol consumption. These findingsshould be considered in the current social and economicsituation in Spain, which is undergoingan economiccrisisthat is particularly affectingyoung adults and greatly limiting their access to the job market [23]. This may cause delay in the acquisition of new roles and responsibilitiesthat are understoodto be part of this new life stage of passage into adulthood[20]. Finally, in relation to differencesbetweenthe sexes, HED was not as prevalent in females as in males. However, it is possible that the Table 2. Characteristics of male initial sample and follow-up samples. Percentage or mean (95%CI) Initial (18–19 years old) n = 371 2-year follow-up (20– 21 years old) n = 206 4-year follow-up (22– 23 years old) n = 139 6-year follow-up (24– 25 years old) n = 81 9-year follow-up (27– 28 years old) n = 90 p-value Maternal educational level Primary school 32.0 (26.5–37.8) 35.8 (28.4–43.3) 41.6 (32.8–50.8) 43.0 (31.6–54.8) 41.6 (31.5–53.5) High school 27.6 (22.1–33.3) 27.4 (19.9–34.9) 25.5 (16.8–34.7) 24.1 (12.7–35.8) 27.0 (16.8–38.9) University 40.3 (34.8–46.0) 36.8 (29.3–44.3) 32.8 (24.1–42.0) 32.9 (21.5–44.7) 31.5 (21.3–43.4) 0.449 Residence In the parental home 29.7(25.1–34.5) 27.8 (21.9–34.1) 28.8 (21.6–36.4) 31.6 (23.9–40.6) 28.9 (20.0–38.3) Away from the parental home 70.3 (65.7–75.1) 72.2 (66.3–78.5) 71.2 (64.0–78.9) 68.4 (60.7–77.4) 71.7 (62.2–80.5) 0.949 University entrance grade a 9–10 points 50.7 (45.3–56.2) 47.5 (40.6–55.0) 50.0 (42.0–59.2) 50.0 (41.4–60.1) 51.7 (41.6–62.7) 7 - <9 points 42.7 (37.3–48.2) 43.6 (36.6–51.0) 43.5 (35.5–52.7) 42.2 (33.6–52.3) 42.7 (32.6–53.7) 5 - <7 points 6.6 (1.1–12.1) 8.9 (2.0–16.4) 6.5 (0–15.7) 7.8 (0–17.8) 5.6 (0–16.6) 0.996 Positive expectations about alcohol Low 29.7 (23.7–36.0) 33.0 (25.1–41.0) 34.2 (25.0–44.3) 35.4 (25.3–46.4) 31.6 (20.3–43.7) Medium 38.0 (32.0–44.4) 30.7 (22.9–38.8) 31.7 (22.5–41.8) 32.3 (22.2–43.4) 30.4 (19.0–42.5) High 32.3 (26.3–38.7) 36.3 (28.5–44.4) 34.2 (25.0–44.3) 32.3 (22.2–43.4) 38.0 (26.6–50.0) 0.705 Age of onset of alcohol use After 16 years old 18.1 (12.5–24.1) 16.8 (9.2–24.7) 15.5 (6.9–25.5) 16.4 (6.0–29.7) 18.2 (7.8–30.3) At 16 years old 36.9 (31.2–42.8) 41.0 (33.5–49.0) 44.0 (35.3–54.0) 50.7 (40.3–64.0) 48.1 (37.7–60.1) At 15 years old 21.6 (15.9–27.5) 20.2 (12.7–28.2) 21.6 (12.9–1.6) 23.9 (13.4–37.2) 20.8 (10.4–32.8) Before age of 15 years 23.4 (17.8–29.4) 22.0 (14.4–30.0) 19.0 (10.3–9.0) 9.0 (0.0–22.3) 13.0 (2.6–25.1) 0.381 Heavy episodic drinking b Never 39.1 (34.0–44.7) 43.2 (36.4–50.6) 42.4 (34.5–51.7) 46.9 (37.0–58.9) 45.6 (35.6–56.5) Less than once a month 25.3 (20.2–31.0) 20.4 (13.6–27.8) 21.6 (13.7–30.8) 21.0 (11.1–33.0) 21.1 (11.1–32.1) Monthly 12.7 (7.5–18.3) 14.6 (7.8–22.0) 13.7 (5.7–22.9) 17.3 (7.4–29.3) 15.6 (5.6–26.5) More frequently 22.9 (17.8–28.6) 21.8 (15.0–29.2) 22.3 (14.4–31.6) 14.8 (4.9–26.8) 17.8 (7.8–28.8) 0.905 AUDIT: Total (mean) 7.8 (7.2–8.4) 7.4 (6.6–8.2) 7.3 (6.4–8.2) 6.5 (5.4–7.6) 7.1 (6.0–8.2) 0.784 Cannabis consumption 27.0 (22.3–31.6) 27.7 (21.3–34.0) 25.9 (18.3–33.5) 23.5 (13.6–33.3) 24.4 (15.0–33.9) 0.885 Tobacco consumption 27.5 (22.8–32.2) 21.8 (16.0–27.7) 23.0 (15.7–30.4) 23.5 (13.6–33.3) 24.4 (15.0–33.9) 0.636 a Variable with a scale of 1 to 10 b Question 3 of the AUDIT. doi:10.1371/journal.pone.0165514.t002 Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 6 / 15 prevalence of HED in women may have beenunderestimated, as the 3rd question of the AUDIT considers consumption of 6 or more drinks. This same hypothesis of new roles and responsibilitiescan be one explanation for the important decreaseobservedat age of 24 years for both consumption patterns, irrespectiveof group or gender. After collegeyears, when high prevalences of consumption are observed, young people becomeconcernedabout their futures, and tryto enter the job market and acquire new responsibilities. In fact, during this ages, students who entered college with better grades,declared no HED or RC practices at all. Moreover, the slight increase at age 27 may be due to new adult alcohol consumption. Regarding the factors associated with RC and HED, early onset of alcohol consumption was found to be the main risk factor for bothpatterns of drinking.This is consistent with the findings of previous studies,in whichearly onset of drinkinghas beenassociatedwith higherconsumption and negative consequences during adolescenceand with problems related to alcohol Table 3. Main characteristics of the female subjects and alcohol consumption at age 18, 20, 22, 24 and 27 years. Percentage of Risky Consumption Percentage of Heavy Episodic Drinking Age in years Age in years 18–19 20–21 22–23 24–25 27–28 18–19 20–21 22–23 24–25 27–28 Maternal educational level Primary school 47.2 46.1 39.1 12.0 18.9^ 17.1 12.6 11.2 5.6 4.7^ High school 53.5 55.9 42.9 11.4 16.5^ 18.5 19.3 20.0 4.3 4.4^ University 57.3*58.9*51.7 13.0 29.4^ 18.7 21.4*18.3 1.4 5.9^ Residence In parental home 42.2*40.1*36.3 9.5 14.7^ 13.9 11.2*12.7 3.2 0*^ Away from the parental home 54.9 56.1 45.4 12.9 22.6^ 19.4 18.5 16.5 4.5 6.2^ University entrance grade a 9–10 points 40.4 41.5 29.6 0 22.2^ 5.8 12.2 22.2 0 5.6 7 - <9 points 50.8 49.5 37.6 11.7 18.5^ 16.2 16.9 12.2 3.1 2.1^ 5 - <7 points 53.6 57.3 51.3*12.1 22.6^ 20.8*17.9 19.6 3.7 6.8^ Positive expectations about alcohol Low 27.5 33.3 25.0 5.4 10.9^ 7.1 6.9 7.4 2.2 3.6 Medium 64.3 60.6 49.3 18.8 18.8^ 19.4 17.6 18.6 6.2 3.0^ High 71.0*64.5*59.8*12.1*32.9*^ 27.4*26.7*19.7*3.0 6.3^ Age of onset of alcohol use After 16 years old 30.5 42.3 28.8 7.9 2.5^ 7.9 8.7 10.6 7.9 0 At 16 years old 57.3 57.0 44.2 14.3 22.8^ 13.4 18.1 17.7 5.5 3.0^ At 15 years old 69.2 70.0 61.3 11.7 28.2^ 28.5 22.7 20.8 0 7.7^ Before age 15 years 78.9*63.8*56.8*23.7 33.3*^ 38.7*26.7*21.0 7.9 10.5^ Cannabis consumption at 18 years old No 42.9 45.8 37.2 9.6 14.8^ 13.3 12.0 12.6 3.7 2.7^ Yes 89.2*79.5*66.3*22.9*47.5*^ 38.4*37.0*27.4*6.2 14.8*^ Tobacco consumption at 18 years old No 37.7 42.4 34.0 9.8 16.1^ 10.5 12.2 12.9 3.8 3.7^ Yes 82.1*73.5*60.8*16.9 30.8*^ 34.4*26.5*20.9*4.8 7.5^ Total 51.5 52.2 43.2 12.2 20.9^ 17.9 16.7 15.7 4.1 4.9^ a Variable with a scale of 1 to 10. *Significant differences in relation to category, p<0.05. ^ Significant differences in relation to age group, p<0.05. doi:10.1371/journal.pone.0165514.t003 Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 7 / 15 abuse and dependencyin adults [24]. For example, Cranford [25] showed that the risk of HED in university students was up to four times higher in students whohad begundrinkingat earlier stages (OR = 4.15), establishing this variable as an important predictor of HED independently of race or gender. Greater parental permissivenesshas been indicated as an explanatory factor for this association.Previous studies have shown that alcohol consumption in adolescents is affectedby the perceptionthat the young people have of their parents’ attitude to drinking[26,27]. Childrenof more permissive parents beginto drink at earlier ages and undertake risky drinkingbehavior during adolescenceand early adulthood. In the present study, positive expectationsabout alcohol acted as a risk factor for RC and HED in both sexes. This is consistent with previously reported findings, in relation to both the RC pattern [28,29] and the HED pattern [30,31]. We also found differences betweenthe sexes in relation to expectationsabout alcohol, as previously reported[32,33], with this factor having a greater influencein females. Table 4. Main characteristics of the male subjects and alcohol consumption at ages 18, 20, 22, 24 and 27 years. Percentage of Risky Consumption Percentage of Heavy Episodic Drinking Age in years Age in years 18–19 20–21 22–23 24–25 27–28 18–19 20–21 22–23 24–25 27–28 Maternal educational level Primary school 47.4 59.7 45.6 14,7 24.3^ 29.3 36.1 31.6 8.8 18.9^ High school 60.0 58.2 65.7 10.5 29.2^ 38.0 41.8 51.4 10.5 25.0^ University 65.8*70.3 57.8 38.5*42.9^ 39.0 40.5 48.9 30.8 17.9 Residence In parental home 51.4 56.1 47.5 18.5 19.2^ 34.9 41.9 30.0*14.8 7.7^ Away from the parental home 60.5 64.9 58.6 24.1 35.9^ 35.7 31.6 48.5 18.5 25.0^ University entrance grade a 9–10 points 47.8 61.1 44.4 20.0 0 21.7 22.2 33.3 0 0 7 - <9 points 51.3 55.7 55.0 26.5 36.8^ 31.3 34.1 36.7 20.6 28.9 5 - <7 points 64.6*69.8 56.5 19.0 28.3^ 41.0 45.8 50.7 16.7 13.0^ Positive expectations about alcohol Low 33.7 42.4 39.0 10.3 32.0^ 20.2 25.4 19.5 3.4 20.0 Medium 64.0 70.9 55.3 36.8 41.7^ 37.7 41.8 42.1 31.6 29.2 High 73.2*75.4*68.3*26.1 23.3^ 46.4*46.2*65.9*17.4*16.7^ Age of onset of alcohol use After 16 years old 32.8 62.1 61.1 18.2 28.6^ 17.2 37.9 38.9 9.1 21.4 At 16 years old 60.2 71.8 66.7 38.2 32.4^ 31.4 40.8 47.1 32.4 21.6 At 15 years old 73.9 68.6 56.0 12.5 31.2^ 46.4 40.0 52.0 12.5 18.8^ Before age of 15 years 89.3*86.8 68.2 0 50.0^ 66.7*65.8*59.1 0 20.0^ Cannabis consumption at 18 years old No 45.4 57.3 47.6 17.7 26.5^ 22.1 27.5 35.0 11.3 14.7^ Yes 92.0*86.0*77.8*36.8 45.5^ 72.0*68.4*66.7*36.8*36.4*^ Tobacco consumption at 18 years old No 47.6 56.5 51.4 17.7 30.9^ 25.7 34.2 36.4 11.3 20.6^ Yes 85.3*84.4*68.8 36.8 31.8^ 61.8*55.6*65.6*36.8*18.2^ Total 58.0 62.6 55.4 22.2 31.1^ 35.6 38.8 43.2 17.3 20.0^ a Variable with a scale of 1 to 10. *Significant differences in relation to category, p<0.05. ^ Significant differences in relation to age group, p <0.05. doi:10.1371/journal.pone.0165514.t004 Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 8 / 15 Fig 1. Prevalences of risky consumption among female and male Spanish university students from 18–19 years old to 27–28 years old. doi:10.1371/journal.pone.0165514.g001 Predictors and Prevalence of Alcohol Use among University Students PLOS ONE | DOI:10.1371/journal.pone.0165514 November 3, 2016 9 / 15