Psychological, physical and social damage have been associated to patterns of high alcohol consumption among adolescents, like risky consumption (RC), characterized by high consumption of alcohol, or heavy episodic drinking (HED), described as the consumption of high quantities of alcohol in a short period of time.1–5

There have been many alerts about the high prevalence of these drinking habits among adolescents6–8 and many risk factors have been identified as familiar background,9,10 age of onset,11,12 peer influence,13,14 and expectancies.11,15

On the other hand, these own patterns of alcohol consumption may consist in risk factors for some aspects, like academic attainment. RC and HED may interfere directly or indirectly on educational achievement by reducing the number of hours dedicated for studying, by affecting the motivation and school attitudes or even limiting neurocognitive abilities.16,17

As the adolescent development is related to multiple dimensions and the alcohol consumption patterns and its associated factors can be modified at different ages,18 it is desirable to know the trajectory of these drinking habits and their associated risk factors throughout the years.

Up to now, there is increasing longitudinal data about prevalence and risk factors of alcohol drinking among subjects that are in process of development and independence.19–23 Nonetheless, in Spain, such longitudinal studies are still scarce. In one-year follow-up study with high school students (13–20 years old), Fuentes-Almendras et al24 identified increment of drinking. Higher drinking frequency was related to older ages and to being male. In another two-year follow-up study, 15 years-old male outscored female on week alcohol consumption.25

Therefore, in an attempt to extend the data provided by a previous work of our group,11 the aim of this study is to evaluate the incidence of RC and HED in the Cohort of Spanish university students at two-year follow-up and to identify predictors of these patterns of alcohol consumption and the association between these patterns and academic achievement.

The response rate at two-year follow-up was 64.1%. Table 1 shows the characteristics of total sample (n=1369) and the characteristics of two-year follow-up sample (n=877). The table does not show significant differences for any variable. The observed reduction in the follow-up sample might be explained by two motives of similar weight: dropping out of college; and absences from school or degree change.

The incidence of RC and of HED at two-year follow-up were 24.92% (95% CI: 21.96–27.86) and 4.01% (95% CI: 2.84–5.41) respectively. The condition of being male, having a non-abstainer mother, besides of the early age of onset of use, and high positive expectancies are associated with the higher incidence rates of RC and HED (table 2).

Table 3 shows the prevalences of alcohol consumption. The prevalence of RC shows significant changes, 37.1% vs. 54.6%, prevalence ratio=1.47 (95% CI: 1.34–1.61). On the contrary, the prevalence of HED drops from 12.2% to 8.7%, prevalence ratio=0.70 (95% CI: 0.55–0.92). In relation to abstemious variable, the prevalence at initial study and at two-year follow-up do not show significant differences.

The prevalence rates of RC at two-year follow-up are significantly higher than at initial study for all categories of subjects (table 3). The prevalence of RC present higher increase among the categories of subjects with lower prevalence at initial study (table 3). On the contrary, the prevalence of HED at two-year follow-up is significantly lower than at initial-study (table 3).

In relation to incidence of RC (table 4), the multivariate model shows that being male, high and medium positive expectancies, and age of onset of use at 15 years, all constitute risk factors for incidence of RC. On the contrary, living with parents shows a protective effect on RC. Table 4 also shows that being male and having high positive expectancies are risk factors for incidence of HED. Finally, table 5 shows the influence of RC and HED on academic achievement. Both patterns are risk factors of low academic achievement.

One out of four subjects that did not present RC at initial study does show it at two-year follow-up. Up to our knowledge, this proportion is among the higher ever cited in literature (Schulenberg et al18 have reported about 20%). In our study, the condition of being male, having a non-abstainer mother, and the early age of onset of use and high positive expectancies are associated with the higher rates, near 50%. Further study more, the higher incidences of onset of RC occur among categories of subjects that presented higher prevalence at initial study. This contributes for the expansion of prevalence of RC at two-year follow-up.

The prevalence of RC at two-year follow-up is also higher than at initial study, while prevalence of HED is lower than at initial study. These results suggest that HED is associated with adolescents and RC with young subjects. According to Jackson et al.30 and to Windle et al.31 adolescents' HED reaches its peak around 20 years-old and then tends to stabilize or decrease. Another longitudinal study has reported diminished relative risk for binge drinking over time for 12- to 20-year-old males.23

The prevalence of RC at two-year follow-up shows average increases of 50% on all variables, except for variables “age of onset of use” and “positive expectancies”, that show increases of 100% (table 3). These results are consistent with different studies.12,32

The prevalence of HED shows more marked decreases among categories of subjects with higher prevalence at initial study, except for the variable “positive expectancies”. Furthermore, among subjects with higher positive expectancies, the HED prevalence diminishes only 20%, opposite to the decrease of 30% showed by all subjects (table 3). Once more, the results of this work underline the role of positive expectancies in establishing and maintaining different patterns of alcohol consumption.

Both, socio-demographic characteristics and alcohol expectancies are associated with incidence of RC and with incidence of HED.

In relation to gender, at two-year follow up, opposite to initial study results, “being male” becomes a risk factor for incidence of RC, despite of the lower cut-off point used for RC in females. In line with these results, a work with Turkish college students has informed that alcohol use was predicted by being male33 and another Swedish study informed that being male has been related to frequent high alcohol consumption.34

Living with their parents is a protective factor for incidence of RC. However, this protective effect is lower at two-year follow-up (OR=0.48) than at initial study (OR=0.39).11 In contrast, this variable does not show an effect on incidence of HED.

The association found between the maternal educational level and the incidence of RC at two-year follow-up is similar as at initial study.11 It indicates that higher maternal educational level is still a risk factor for RC at early adulthood.

The observed association between early age of onset of alcohol use and incidence of RC or incidence of HED could be explained by tolerance generated by alcohol use. Other possible explanation for this association could be that the parents who “permit” alcohol use at an earlier age are also more likely to be more permissive with excessive consumption.9,10

The relationship between positive expectancies and alcohol consumption is similar to the one found at initial study. These results are consistent with those described in several studies, even though they have used different measurement scales for expectancies.

Finally, our results show that RC and HED are associated with low academic achievement. Inconsistent with these results, teacher-rated poor academic ability has been considered a predictor of adolescent HED among both women (OR=1.17, p<0.05) and men (OR=1.22, p<0.001).17 However, other study observed association between adolescent high HED and usual grades in school among men (OR=0.510, p<0.01) but not among women (OR=0.828, p>0.05).31

This lower achievement could be attributed to the confusion introduced by less academic effort of subjects who present these patterns of consumption. However, the fact that the models have been adjusted for grade of university access exam suggests absence of this confusion. Further, neuropsychological and psychophysiological evaluation of the subjects at two-year follow-up will facilitate the identification of the extension at which this lower academic achievement can be attributed to the neurocognitive consequences of alcohol consumption.

There are two possible limitations in our study: (1) Selection bias, because of the loss of subjects in the follow-up. However, the absence of significant differences between the initial sample and the follow-up sample (table 1) suggests absence of this bias. (2) Given that the question about expectancies is not specifically validated, expectancies may have not been correctly measured.

Our results show that HED in Spain is a pattern of alcohol consumption mainly associated with adolescence, while RC is associated with youth. Both patterns are predictors of academic achievement. Public Health strategies should focus on modifying expectancies and limit access to alcohol at young ages.

This research was supported by Consellería de Innovación e Industria, Xunta de Galicia, grant number PGIDIT05CS021103PR and INCITE08XIB211015PR, by Plan Nacional sobre Drogas, Ministerio de Salud y Consumo, Spain, grant number 2005/PN014, and by Ministerio de Ciencia e Innovación, Spain, grant number EDU2008-03440.

The authors N. Mota and A. Crego hold a predoctoral scholarship from Consellería de Innovación e Industria, Xunta de Galicia, and from Ministerio de Educación, Spain (grant number AP2006-03871), respectively.

F. Cadaveira, M. Corral, F. Caamaño-Isorna and S. Rodríguez Holguín have designed the study. A. Crego, M. Parada, N. Mota and R. Álvarez-Gil have collected data. F. Caamaño-Isorna and R. Álvarez-Gil have analysed and interpreted data. R. Álvarez-Gil and N. Mota have written the article. F. Caamaño-Isorna has approved the final version for publication.

No conflicts of interest.