The Data were obtained from the 2010 Swedish Level-of Living survey (LNU). LNU is based on a nationally representative sample of 0.1% of the population aged 18–75 and is performed approximately every 10 years, with the 2010 dataset being the most recently available . Personal interviews performed in the respondent’s household are used to gather the data. The survey contains questions on a broad range of living conditions, including, e.g., sociodemographic characteristics, educational and occupational careers, and health. Starting with LNU 2000, two complementary surveys with other family members are also performed, namely Partner-LNU and Child-LNU. The Partner-LNU is offered to the spouses/partners of the main respondent and consists of a shortened version of the main survey. The Child-LNU is offered to adolescents aged 10–18 years who live in the main respondent’s household. Ethical approval has been obtained from the Regional Ethical Review Board of Stockholm (ref. no. 2009/1802-31/5).
For the current study, we combined data from the LNU, Partner-LNU, and Child-LNU of 2010. As data were gathered from the co-habiting partner of the main responding parent, variables based on parental information can include answers from both custodial and/or step-parents. The LNU 2010 survey included 7,253 individuals in the sampling frame and interviews were conducted with 4,415 of these, corresponding to a response rate of 60.9%. Furthermore, 3,347 spouses or partners to the main respondent were identified and 2,522 of these completed the Partner-LNU 2010 (75.4%). In total, 1,238 adolescents were offered to participate in the Child-LNU 2010 and 923 agreed to take part (71.9%). After further exclusion due to internal non-response on any of the variables used in the current study, the analytical sample consisted of 909 adolescents in 629 households. Among these, 640 adolescents (70.4%) had linked information from two parents. Among those living with two custodial parents (n = 634), 83.4% had linked information from both parents.
The dependent variables in this study, psychological complaints, somatic complaints, and perceived stress, were based on adolescents’ self-reports from the Child-LNU.
Psychological complaints were constructed from three items: ‘I often feel sad or down’; ‘I am often tense or nervous’; ‘I am often grouchy or irritated’. Response alternatives were: (1) ‘Matches exactly’, (2) ‘Matches roughly’, (3) ‘Matches poorly’, and (4) ‘Does not match at all’. Responses were then reversed and added to an index ranging from 3 to 12, with higher values indicating more psychological complaints (Cronbach’s alpha 0.65). Missing answers on at most one of the items were replaced with the individual mean of the remaining items.
Somatic complaints were constructed from three items: ‘In the last 6 months, how often have you had the following complaints?’: ‘Headache’, ‘Stomach-ache’, ‘Difficulties in falling asleep’. Respondents were given five response alternatives: (1) ‘Every day’, (2) ‘Several times a week’, (3) ‘Once a week’, (4) ‘About once a month’, and (5) ‘More seldom or never’. Responses were then reversed and added to an index ranging from 3 to 15, with higher values indicating more frequent somatic complaints (Cronbach’s alpha 0.55). Missing answers on at most one of the items were replaced with the individual mean of the remaining items. These measures of psychological and somatic complaints have been used previously .
Perceived stress was captured by the question: ‘During the last 6 months, how often have you felt stressed?’. Response alternatives were the same as for somatic complaints. Adolescents who felt stressed more than once a week were categorised as reporting perceived stress .
The independent variable, parental alcohol use, was based on parent-reported information and constructed from three items in the LNU and Partner-LNU. Parents were categorised as: abstainers; low consumers; moderate drinkers; or heavy drinkers. The first question was: ‘Do you at any time drink wine, strong beer/cider, or liquor?’ and those who answered ‘Yes’ responded to two additional questions: ‘During the last 12 months, about how often have you consumed some amount of alcoholic beverage, that is: wine, strong beer, strong cider or liquor?’ (with response alternatives: ‘Daily or almost daily’, ‘2–4 times a week’, ‘Once a week’, ‘2–3 times a month’, ‘Once a month’, ‘6–11 times a year’, ‘Less often’, and ‘never’) and ‘On such occasions, how many glasses do you usually drink? One glass can be 1 glass of wine, 1 bottle or can of beer, or 1 schnapps or drink’. Non-drinking parents were categorised as abstainers and those who drank less than once a month as low consumers. We wanted to keep these types of drinking patterns separate since current non-drinking or low consumption may be a result of prior heavy consumption and/or indicative of characteristics that may influence offspring health, such as severe health problems among parents or small social networks [25, 26]. Parents who drank daily, regardless of quantity, or 2–4 times a week and at least 3 glasses per occasion, were categorised as heavy drinkers. Using this cut-off, parents in the heavy-drinking category consume at least 6 glasses per week. While this figure is well below the Swedish guidelines for what counts as heavy drinking it is important to acknowledge that the response categories given contain variation that is impossible to avoid in the construction of the categories. One individual who answers ‘2–4 times a week’ could drink twice as often as another individual who ticks the same box, for example. The answers that respondents provide are also likely affected by recall bias due to difficulties in remembering how often and how much one drinks, the propensity to underestimate one’s consumption of alcohol because of social desirability, and the hard task of calculating an average number of drinking occasions and drinks for the last 12 months. The difficulty of accurately measuring individuals’ alcohol consumption in epidemiology is known . Consequently, the measure of parental alcohol use will, regardless of how the different categories are defined, always contain a non-trivial amount of uncertainty. To validate this study’s results, we will perform a series of sensitivity analyses, using different cut-offs for heavy drinking (presented in the Supplementary Material). The rest were categorised as moderate drinkers (i.e., those who drank at least monthly but at most weekly or 2–4 times a week but less than 3 glasses per occasion). These questions about alcohol use have been employed in prior studies, separately and combined, to assess drinking patterns among parents, young adults, and older adults [28,29,30]. For adolescents with answers from two parents, parental alcohol use was based on the respondent that reported the most frequent alcohol consumption. Thus, children described as having heavy-drinking parents in the text should be interpreted as having at least one parent categorised as a heavy drinker. Those with moderate-drinking parents were used as the reference category in the analyses since we were interested in the difference in health between children with heavy-drinking parents and children whose parents have more typical drinking habits. It is also the largest category and the one that is easiest to interpret. However, we also performed additional analyses changing the reference category to children with heavy-drinking parents.
Adolescents’ gender and age, as well as family structure, household social class, household cash margin, parental level of education, and parental unemployment were included as control variables, based on information in LNU and Partner-LNU. Family structure distinguished between adolescents living in households with two custodial parents, one custodial parent, or in reconstituted families (i.e. one custodial parent and one step-parent or with two foster parents). Household social class was based on parents’ occupations and classified into four categories: manual workers, farmers and entrepreneurs, intermediate/lower non-manual workers, and higher non-manual workers. For those with information from two parents, the classification was based on the principle of dominance order presented by Erikson . Lack of cash margin in the household was defined as at least one parent reporting not being able to come up with 14,000 SEK (∼ 1350 US$ or 1300 €) in a week if needed to. Parental level of education was categorised as ‘compulsory or vocational’, ‘secondary or lower tertiary’, and ‘university degree’, and was based on the parent with the highest level of educational attainment in the household. Parental unemployment was defined as at least one parent in the household being currently unemployed.
First, the associations between parental alcohol use and all control variables were examined through bivariate cross-tabulations with chi2-tests. Next, linear and binary logistic regression analyses were performed with psychological and somatic complaints and perceived stress as the dependent variables. For each outcome, crude analyses (including one independent/control variable at a time) and an adjusted model (mutually adjusting for all independent/control variables) were conducted. To account for the fact that some adolescents lived in the same households (being siblings or step-siblings), robust standard errors were estimated, clustering at the household level. Regression coefficients (b) and odds ratios (OR) with 95% confidence intervals (CI) are presented. All statistical analyses were carried out in Stata version 16.1.