Maternal and paternal alcohol abuse or at-risk drinking was associated with moderately higher levels of attention and conduct problems, both at the crude level and when demography and the other parent's consumption was controlled for. There seems to be a dose-response trend, as the at-risk groups consistently scored between abusers and light drinkers on these outcomes. Heavy drinking in the parents did not predict dissatisfaction with school in general or with academic results in any of the analyses, even though this study has high power. Parental alcohol use predicted poor adjustment only on the impulse control-related dimensions attention and conduct. It is not surprising that we find associations with these dimensions. Children of alcoholics have previously been found to have elevated risks for attention and conduct problems [7, 18] and the related diagnoses ADHD  and conduct disorder . Previous studies show strong genetic components in the link between externalizing behaviour in parents and the children, such as parental drinking and behavioural control in the offspring [15, 16]. Also, social strains such as negative life events, family conflict or dysfunction, disruption of routines, or neglect can also foster maladjustment [19, 20]. From the social strain perspective, a lack of association between parental alcohol abuse and satisfaction at school could simply be explained by the school representing an escape from a troublesome home environment for some adolescents.
Maternal drinking was particularly predictive of high attention and conduct problem scores in our data. Our results are consistent with previous findings that maternal drinking has a greater impact on children than paternal drinking [8, 37]. If not simply due to statistical fluctuations, the apparent heightened risk associated with mothers compared to fathers may be explained by impairment of the primary caregiver role, commonly undertaken by the mother, or by drinking during pregnancy. The present study does not have data on drinking during pregnancy, but previous studies have found that moderate prenatal exposure to alcohol increases the risk of conduct problems [10, 18] and learning difficulties . Also, since women drink less than men on average, pathological drinking in mothers may indicate a more severe stressor or higher heritable vulnerability to impulse control problems than drinking in fathers.
Children of abstainers had fewer attention, conduct, and academic problems than children of light drinkers. This finding stands in contrast to the results from a British study . One may speculate that different factors lead to abstention in Norway and in the United Kingdom. As light drinking among parents is unlikely to constitute a social strain, we believe it is more likely that the difference between abstainers and light drinkers stems from lifestyle or personality factors rather than alcohol use per se.
Previous studies disagree on whether other parental psychopathology confounds the association between parental drinking and psychosocial functioning among their children [25–28]. The associations seen in the present study cannot be ascribed parental mental distress as it did not act as a confounder: adding this variable to the analysis did not substantially alter associations between parental drinking and school adjustment. Parental psychopathology may be more severe in studies finding such confounding. In addition, a high number of untreated cases in the general population, likely to be included in the present study, occur without severe comorbidity .
Adolescent mental distress was a strong predictor of attention problems, and a moderate predictor of conduct problems. Adolescent mental distress was moreover associated with maternal drinking, and the association between maternal abuse and attention problems was reduced when this variable was added to the analyses. It may therefore be considered a partial mediator for maternal alcohol use on attention problems. Before adjusting for adolescent mental distress, maternal drinking was more strongly associated with attention problems than was paternal drinking. Hence, it may be that maternal drinking has some additional effect on attention problems that is mediated by the adolescents' mental distress. However, most associations were independent of this distress: similar mediation was not seen between paternal drinking and attention problems, and the mediation on conduct seems to be small or non-existent.
The adolescent self-report of having witnessed parental drunkenness was a stronger predictor of maladjustment than was parental alcohol report. Adolescent report of seeing their parents drunk was associated with both parental report of drinking and outcome, and all effect sizes were reduced when this variable was added to the analyses. Hence, it is likely that seeing one's parents drunk mediates a non-trivial part of the association between parental alcohol use and school adjustment. One interpretation of this would be that being with intoxicated parents is harmful in itself, and that this question measures the subjective burden of having an alcoholic parent. Alternatively, this question may tap into variation in alcohol problems that is not captured by our parental alcohol measure. However, unlike with parent-reported measures, associations between predictors and outcome both reported by the adolescents may also partially reflect mood-congruent response consistency.
Children who had alcohol abusing fathers with a dissolved relationship were particularly at risk for attention problems. This may be an example of the principle that an accumulation of risk factors is especially harmful [25, 55]. The risk seems to be equal across age and gender, as no interaction effects were found on these variables.
As Young-HUNT data were collected during school hours, the adolescent sample is fairly representative of adolescents in the county, with most non-response resulting from sick leave. Parental response rates were lower. Although people who are struggling with many problems at once, or with very severe problems, seem to be underrepresented in population surveys , alcohol use only moderately predicts non-participation in the HUNT study . In addition, simulations have shown that associations between variables are only moderately weakened by high rates of selective non-response . We therefore believe that all consumption groups are adequately represented in the sample, and that it is suited for studying alcohol use within the general population.
Alcohol consumption is usually underreported in population studies . If this underreporting changes the ranking of individuals, misclassification occurs. However, the alcohol consumption measure showed good reliability, with consistent scores over a long period (11 year test-retest correlation was 0.63). The prevalence of abuse in this study was also lower than usually reported , and due to the representativeness of the sample  and the strict inclusion criteria for the abuse groups, the large majority of people classified as abusers are likely to be true cases. False negatives, however, can lead to an underestimation of the number of exposed adolescents.
A strength in our study was that mothers and fathers reported their alcohol use and mental distress independently, thereby avoiding inflated effect sizes due to single responders reporting on all measures. There may, however, be correlated errors between measures reported by the same person. Since this was a general health study covering a large number of topics, respondents were not aware of the purpose of the alcohol questions, which has probably also reduced the risk of response bias.
We did not detect any confounding by comorbid parental disorders, perhaps because we only measured internalizing symptoms in the parents. It may be that parental externalizing behaviour or antisocial personality characteristics in reality confound or mediate the effects of parental alcohol abuse . However, there were no data available on parental psychopathology besides of internalizing symptoms. In addition, as this study is cross-sectional, we cannot conclude on causal mechanisms or persistence of the problems. The inclusion of a missing category was necessary to avoid excluding many problem drinkers whose spouses did not participate. This implies a not fully complete control of mothers' and fathers' unique contributions to school adjustment.