Article Text

Download PDFPDF

The effect of childhood socioeconomic position on alcohol-related disorders later in life: a Swedish national cohort study
  1. Karl Gauffin1,2,
  2. Tomas Hemmingsson3,4,
  3. Anders Hjern1,5
  1. 1Centre for Health Equity Studies, Stockholm University/Karolinska Institutet, Stockholm, Sweden
  2. 2Department of Public Health Sceinces, Karolinska Institutet, Stockholm, Sweden
  3. 3Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden
  4. 4Centre for Social Research on Alcohol and Drugs, Stockholm University, Stockholm, Sweden
  5. 5Clinical Epidemiology/Department of Medicine, Karolinska Institutet, Stockholm, Sweden
  1. Correspondence to Karl Gauffin, Centre for Health Equity Studies, Stockholm University/Karolinska Institutet, Stockholm SE-10691, Sweden, karl.gauffin{at}


Background Alcohol use is the third most important global-health risk factor and a main contributor to health inequalities. Previous research on social determinants of alcohol-related disorders has delivered inconsistent results. We aimed to investigate whether socioeconomic position (SEP) in childhood predicts alcohol-related disorders in young adulthood in a Swedish national cohort.

Methods We studied a register-based national cohort of Swedish citizens born during 1973–1984 (N=948 518) and followed them up to 2009 from age 15. Childhood SEP was defined by a six-category socioeconomic index from the Censuses of 1985 and 1990. HRs of alcohol-related disorders, as indicated by register entries on alcohol-related death and alcohol-related medical care, were analysed in Cox regression models with adjustment for sociodemographic variables and indicators of parental morbidity and criminality.

Results Low childhood SEP was associated with alcohol-related disorders later in life among both men and women in a stepwise manner. Growing up in a household with the lowest SEP was associated with risk for alcohol-related disorders of HR: 2.24 (95% CI 2.08 to 2.42) after adjustment for sociodemographic variables, compared with the highest SEP group. Adjusting the analysis for parental psychosocial problems attenuated the association to HR 1.87 (95% CI 1.73 to 2.01).

Conclusions The study demonstrates that low SEP in childhood predicts alcohol-related disorders in young adulthood. Alcohol abuse needs to be addressed in policies to bridge the gap of health inequalities.

  • Alcohol
  • Social Inequalities
  • Longitudinal Studies
  • Social Class
  • Social Epidemiology

Statistics from


The importance of childhood factors for alcohol-related disorders has been recognised by multiple studies. Several childhood predictors for alcohol-related disorders have been discussed, including genetic factors,1 developmental disorders2 and child abuse and neglect.3 ,4 Some studies have also investigated the effect of childhood socioeconomic factors on alcohol-related disorders later in life.5–10

When examining the social determinants of health, childhood socioeconomic position (SEP) may be worth extra attention not only due to children's increased vulnerability but also as it is a strong predictor of social and health-related development over the entire life course.11 In the study of alcohol-related disorders, the focus on childhood SEP entails a methodological advantage, as the possibility of reversed causality or social drift as a consequence of alcohol abuse does not need to be considered as in studies focusing on adult health determinants.

Previous research on the association between childhood SEP and alcohol-related disorders has delivered paradoxical results. While one British study found that social advantage in childhood increased the weekly consumption quantity in adolescence (both sexes) and the risk for harmful drinking in adulthood (women),12 another study using a sample from the same cohort found that low childhood SEP was related to problem drinking later in life.13 Another recent birth cohort study from England showed that household income was positively associated with problematic adolescent drinking among girls, whereas higher maternal education reduced the risk for alcohol-related problems in boys.14 A systematic review by Wiles et al9 shows that studies rarely address the relation between childhood SEP and alcohol abuse later in life explicitly and the findings are partly inconsistent. Moreover, the majority of these studies are based on small-sized or medium-sized cohorts and include self-reported data on alcohol consumption, making them vulnerable to recall and disclosure bias. On a global level, there seems to be a need for evidence regarding the social determinants of alcohol-related disorders, especially given the fact that alcohol use has been ranked as the third most important global health risk factor.15 The lack of research in this field is also apparent in the final report by the WHO Commission on Social Determinants of Health, which barely addresses the determinants of alcohol-related disorders.11

The Swedish national health survey conducted by the National Institute of Public Health did not find any socioeconomic differences in actual alcohol consumption levels.16 Regarding alcohol consumption of young people, a recent representative survey on alcohol and drug consumption in adolescents in Stockholm showed that self-reported alcohol consumption and binge drinking are higher among those living in the more affluent areas of Stockholm.17 It is unclear, however, whether this association is also reflected in the register-based indications of alcohol-related disorders later in life. These surveys provide an interesting starting point for this study, in which we aim to investigate whether SEP in childhood predicts alcohol-related disorders in young adulthood in a Swedish national cohort.


Swedish population registers include a large number of social and health-related indicators and provide researchers in social epidemiology with excellent sources for longitudinal population-based studies. Every Swedish resident is given a unique personal identification number (PIN) at the time of birth or immigration. PINs enable record linkage, which makes it possible to follow a Swedish citizen from birth to death in the population registers. This study has been approved by the regional ethics committee in the Stockholm region.

Study population

The study cohort includes all men and women born in Sweden between 1973 and 1984 who were alive and registered as Swedish residents at 15 years of age (n=948 518). Owing to the limited information about migrants’ socioeconomic circumstances in childhood, residents not born in Sweden were excluded from the cohort. The individuals in the study population are linked to their parents via the multigeneration register.


In this study, we used the socioeconomic index (SEI) developed by Statistics Sweden. This is a well-recognised and frequently used measure reflecting both the level of education and position in the working place.18 Childhood SEP was specified by the parental SEI classification from the Swedish Population and Housing Censuses of 1985 and 1990. The earlier census was used to indicate the childhood SEP of the cohort born during 1973–1978 and the latter census for the cohort born during 1979–1984. In other words, the age of the study population was between 6 and 12 years at the time of measurement. This study includes six socioeconomic groups: three levels of non-manual professionals, two levels of manual workers and an unclassified level. The latter group is very heterogeneous including farmers, students, homemakers as well as unemployed and self-employed individuals and people in early retirement. Childhood SEP was determined by the highest SEP of any adult in the household.


In this study, alcohol-related disorders were defined as a binary outcome (yes/no) and indicated by at least one entry in registers on alcohol-related medical care and alcohol-related mortality. Alcohol-related medical care includes data from the national patient register on alcohol-related hospitalisation (1973–2009) and outpatient care (2002–2009). Alcohol-related medical diagnoses that do not necessarily imply long-term alcohol abuse, such as accidental alcohol intoxication, are excluded from the measure. Records on alcohol-related mortality were collected from the national cause of death register (1973–2008).


We collected a number of potential intervening variables from the national censuses and from the register of the total population held by Statistics Sweden including: sex, year of birth, area of residency, maternal country of birth and parent civil status. Indicators for parental psychosocial problems such as parental alcohol and/or drug abuse, parental psychiatric disorder and parental criminality were retrieved from the cause of death register, the national patient register and the national register of criminal convictions. All parental factors were analysed separately for mothers and fathers and treated as binary variables (yes/no). Parental alcohol and/or drug abuse was indicated by any entry of alcohol and/or drug-related death, medical care or criminality. Alcohol/drug-related criminality implies driving under the influence of alcohol/drugs or any crime related to illicit substances. Parental psychiatric disorder was indicated by medical care or cause of death records with a diagnosis related to psychiatric illness and/or self-inflicted injuries. Parental criminality was defined as any criminal conviction leading to a sentence to prison, probation or forensic psychiatric care.

Statistical analysis

The analysis in this study was conducted using the Cox proportional hazard models. Person time was calculated from age 15 until first indication of alcohol-related disorders, death, record of emigration or end of follow-up in 2008/2009.

The analysis was adjusted for potential intervening variables in three separate models. Model 1 shows the association between SEP in childhood and alcohol-related disorders later in life, only adjusting for sex and year of birth. Model 2 additionally adjusts for area of residency, maternal country of birth and single parent household. Model 3 includes indications of parental alcohol and/or drug abuse, parental psychiatric disorder and parental criminality. All regression analyses show the combined outcome measure of alcohol-related medical care and death. Outcome specific analyses are available on request.


Table 1 presents the study population divided by SEP of the parental household. The population with a lower socioeconomic background was over-represented in the groups from rural areas, in groups with a non-Swedish mother, in groups growing up in a single parent household and also among those who had a parent with indications of alcohol and/or drug abuse, psychiatric disorder and criminality.

Table 1

Study population stratified by socioeconomic background (N=948 518)

In total, 11 056 persons (1.17%) had a register indication of alcohol-related medical care or death. Tables 2 and 3 show the incidence of alcohol-related disorders in men and women, respectively. Alcohol-related disorders were generally more common in the male population, but for both men and women they were more common in the lower socioeconomic groups.

Table 2

Indications of alcohol abuse in men (N=492 536)

Table 3

Indications of alcohol abuse in women (N=492 536)

Effect of childhood SEP

The results of the regression analysis are shown in table 4. Model 1 illustrates a socioeconomic gradient in HRs. The lower the SEP, the greater appeared the risk for alcohol-related disorders. Those growing up in a mid non-manual household had a slightly increased risk (HR 1.19, 95% CI 1.10 to 1.29), whereas the population born into an unskilled manual household had a significantly greater risk for alcohol-related disorders later in life compared with the reference group (HR 2.37, 95% CI 2.21 to 2.55). Additional analyses using paternal and maternal SEP separately as indicators for childhood SEP did not result in any significantly different effects for the outcome of interest compared with the analysis using the highest SEP of any adult in the household (additional analyses available on request).

Table 4

HRs (95% CI) for at least one indication of alcohol abuse in men and women

Effect of covariables

Adjusting for the area of residency, maternal country of birth and single-parent household only marginally attenuated the effect of childhood SEP on alcohol-related disorders. Model 3 shows that exposure to parental substance abuse, psychiatric disorder and/or criminality attenuated a part of the social gradient, although a large part of the effect related to SEP in childhood remained. In model 3, the population from the unskilled manual households still had higher risk for alcohol-related disorders later in life compared with the reference group (HR 1.87, 95% CI 1.73 to 2.01). Some interaction effects between childhood SEP and the covariables could be identified in the fully adjusted model. Sex mattered more in lower SEP groups, whereas maternal psychosocial problems and paternal and/or drug abuse had significantly stronger effects on those growing up in non-manual households (p>0.05). Regression analyses stratified on sex or any other covariable are available on request.


This national cohort study of almost 950 000 Swedish residents demonstrates a stepwise gradient in the relationship between childhood SEP and alcohol-related disorders in youth and young adulthood. A large and representative study population with minimal attrition and the use of childhood SEP as our indicator of social position, which rules out the possibility of reversed causality caused by social drift, makes our results more robust compared with the ambiguous results of previous studies that have addressed this research question (9).

The patterns of alcohol-related disorders found in our study were quite different from the equal alcohol consumption levels found across SEP groups in the national health survey,16 and the Stockholm school survey, which showed that adolescent drinking was more common in affluent areas of Stockholm.17 Rather, our results confirm those of a Finnish study which found that, after controlling for drinking pattern, severe alcohol-related consequences were also more common in lower socioeconomic groups.19

Thus, it seems that childhood socioeconomic factors do not determine alcohol consumption levels (exposure) to the same extent as they do the health consequences of high alcohol consumption. Like Schmidt et al,20 we hypothesise that a high SEP background may decrease the vulnerability to alcohol-related harm. In other words, these groups are more likely to have a social and material buffer protecting them from the negative consequences of their behaviour. The population from a low socioeconomic background, however, may suffer from alcohol-related disorders triggered by the combination of high alcohol consumption and conditions related to their SEP, such as nutritional deficiencies and psychosocial stress. Also, the stigma related to alcohol abuse is likely to disproportionately affect disadvantaged groups, thereby adding to their social marginalisation and exclusion.21

A part of the social gradient found could be explained by the covariables. Parental alcohol and/or drug abuse, psychiatric disorders and criminality were more common in lower socioeconomic groups and also increased the risk for alcohol-related disorders later in life. Both the mental-health condition of the parent and the social stigma related to substance abuse, psychiatric disorders and criminality could arguably create stressors for a child, which might lead to increased risk for alcohol-related disorders later in life. These conditions are also strongly interrelated, which further increases the likelihood of a negative outcome, and in the case of substance abuse, potential heritable factors should be considered. The identified interaction effects suggest that a low-socioeconomic background may have a slightly stronger impact on men compared with women in terms of their risk for alcohol-related disorders. This is consistent with studies finding an attenuated social gradient among women in terms of problematic alcohol consumption.22 The stronger effect of parental psychosocial problems on children from non-manual backgrounds could be interpreted as a higher proportional importance of factors related to the interplay between genetic and environmental factors. In low-socioeconomic groups, parental psychosocial problems may be one of many other risk factors for alcohol-related disorders.

Childhood SEP may affect alcohol-related disorders later in life, either directly (eg, through exposure effects during a sensitive period in early life) or indirectly (through adult SEP). Other studies confirm the association between low SEP in adulthood and alcohol-related disorders;7 ,23 ,24 however, this correlation appears contextual and dependent on the study population.25–27 The present study has focused on the relation between childhood SEP and alcohol-related disorders later in life. It is possible, however, that adult SEP may actually be the most important predictor, and that childhood socioeconomic factors only have an indirect effect on alcohol-related disorders.28 Educational performance is one plausible mediator between childhood SEP, on the one hand, and adult SEP and risk for alcohol-related disorders, on the other hand, and should be investigated more closely. Alternatively, cumulative disadvantage from low SEP in childhood, adolescence and adulthood that increases exponentially over the life course may be the main cause of social inequality with regard to alcohol-related disorders.20 Future research is needed to separate the effect of childhood and adult socioeconomic factors.

Strengths and limitations

The registry design of this study has some significant advantages in relation to many previous studies concerned with similar research questions. Perhaps the most obvious benefit is related to the considerable size of the study cohort enabling detailed analyses without losing statistical power. The focus on the effect of early-life factors, which abolishes the risk for reverse causality between exposure and outcome, is another advantage of the study. This design serves as a solution of a major methodological challenge in the study of SEP and alcohol-related disorders: the risk for social drift following alcohol abuse.9 The register follow-up minimises attrition, which is a problem in most studies on substance-related disorders, and allows us to study a sufficiently large population with a graded indicator of SEP, in the presence of a number of potential confounders. However, the validity of our outcome and to what extent it relates to alcohol abuse in the population needs to be considered in the interpretation of our results. Our outcome measure is based on proxy indicators for problematic alcohol consumption. The proportion of hidden alcohol abuse is likely to be high, potentially even more so in specific population groups. For example, those from a high socioeconomic background may have other resources enabling them to cope with alcohol-related problems without assistance from hospital-based care, which could potentially create an overestimation in our social gradient. The fact that the gradient was quite similar in mortality and hospital data (see tables 2 and 3), however, seems to indicate that the impact of this bias was minor. Unlike studies based on survey data, we were not able to control for reported consumption patterns. However, the survey on alcohol and drug consumption in Stockholm youth gives no reason to believe that destructive consumption patterns in adolescence lie behind the social gradient of alcohol-related disorders in young adulthood. Rather, the survey indicated the opposite, as binge drinking was reported to be more common in the affluent areas of the city.17

Implications for research, policy and practice

The results illustrate that alcohol-related disorders are not only a public-health problem but are also closely linked to health inequality. The association between alcohol-related disorders and low socioeconomic background leads to increased morbidity and premature mortality for this population, which directly contributes to widening health inequalities. This is important to take into consideration when developing policies addressing social and health-related inequality, as well as in specific interventions targeting harmful alcohol use. This study's focus on childhood factors underlines the importance of initiating interventions in early life.

Furthermore, studies such as this highlight the importance of taking alcohol abuse into account in strategic documents on health inequality following the report by the WHO Commission on Social Determinants of Health.11 Exploring the relation between SEP and alcohol-related disorders requires the consideration of multiple factors along the spectrum of social determinants of health—reaching from individual and behavioural approaches to more environmental and structural explanations. An adequate policy response taking the social determinants of alcohol-related disorders into account would thus not only promote public health, but would also contribute to reducing the social gap in health and promoting health equality within and across generations.

What is already known on this subject

  • Previous research has found an association between low-socioeconomic position (SEP) and alcohol-related disorders, but it remains unclear whether this association is caused by downward social mobility following alcohol abuse, the so-called social drift. Also, the relation between childhood SEP and alcohol-related disorders later in life has not been clearly elucidated.

What this study adds

  • The study was able to find a clear stepwise gradient in the association between childhood socioeconomic position and alcohol-related disorders later in life—the lower the socioeconomic background, the higher the risk for alcohol-related disorders in young adulthood. Using childhood socioeconomic position eliminates the risk for a bias due to reversed causality and social drift. Our study underlines the importance of considering alcohol in efforts to close health inequalities within and across generations.


We are indebted to Andrea Dunlavy for her helpful comments on the language in the manuscript.



  • Contributors The study was conceived by AH, who also designed the study in collaboration with KG. KG conducted all statistical analyses under supervision of AH. KG wrote the first draft of the manuscript, while all authors interpreted the data, contributed to the writing of the subsequent versions of the manuscript and approved the final version.

  • Funding This work was supported by the Swedish Council for Working Life Research (grant number: 2009-1669).

  • Competing interests None.

  • Ethics approval Regional ethics committee in the Stockholm region.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.