Childhood Social Environment and Risk of Drug and Alcohol Abuse in a Cohort of Danish Men Born in 1953

Transcription

1 American Journal of Epidemiology Copyright ª 2006 by the Johns Hopkins Bloomberg School of Public Health All rights reserved; printed in U.S.A. Vol. 163, No. 7 DOI: /aje/kwj084 Advance Access publication January 27, 2006 Original Contribution Childhood Social Environment and Risk of Drug and Alcohol Abuse in a Cohort of Danish Men Born in 1953 Merete Osler 1,2, Merete Nordentoft 3, and Anne-Marie Nybo Andersen 4 1 Department of Social Medicine, Institute of Public Health, University of Copenhagen, Copenhagen, Denmark. 2 Epidemiology Unit, Institute of Public Health, University of Southern Denmark, Odense, Denmark. 3 Department of Psychiatry, Bispebjerg Hospital, Copenhagen, Denmark. 4 National Institute of Public Health, Copenhagen, Denmark. Received for publication May 30, 2005; accepted for publication October 28, In a 32-year follow-up study, the authors analyze how social circumstances during early life, childhood social participation, and school performance affect the risk of being admitted to a hospital or dying from a diagnosis closely related to drug or alcohol abuse in young adulthood. A total of 11,376 Danish males born in 1953, for whom data from birth certificates and conscription board examinations had been traced, were followed until 2002 through linkage to the Danish Psychiatric, National Patient, and Cause of Death registries. At age 12 years, 7,877 subjects completed a questionnaire on social participation and school performance. During follow-up, 12 percent of these were given a diagnosis indicating drug or alcohol abuse. Having a single mother and a working-class father were each associated with an increased risk of drug or alcohol abuse in adult life. At age 12 years, those who disliked school, scored low on a school test, or preferred to visit a youth club during leisure time showed a greater risk of adult substance abuse. These associations were slightly attenuated when adjusted for educational status at conscription. Deprived social circumstances during childhood, poor school performance in early adolescence, and attending a youth club seemed to be independent markers of substance abuse in adult life. alcoholism; cohort studies; social environment; substance-related disorders Abbreviation: ICD, International Classification of Diseases. There are considerable economic, health, and social costs associated with drug and alcohol abuse (1), and extensive research has been devoted to identifying factors associated with substance use and abuse (1 11). Prospective studies that explore early determinants of substance abuse later in adult life, however, are scarce. A few studies have shown a relation between socioeconomic disadvantages during childhood and increased risk of death due to substance abuse (12 15). Several nonexclusive explanations for this association exist. One such explanation could be the conditions of the family environment during early life. Poor material conditions and social support, which tend to cluster in families primarily for social reasons, would influence the behavior and health of the members of these families (11, 12). Several investigations, mostly cross-sectional studies, have shown that poor socioeconomic circumstances, parental substance abuse, parental divorce, negative attitudes towards school, family poverty, and peer relationships during childhood are associated with substance use in adolescence. The effects of these factors can either act independently of each other or be mediated through the child s own social position later in life. In the present study, we examine the effect of parental socioeconomic position, childhood social participation, satisfaction with school, and school performance on later continuous and problematic substance use, which is indicated Correspondence to Dr. Merete Osler, Epidemiology Unit, Institute of Public Health, University of Southern Denmark, J. B. Winsløws Vej 9B, 5000 Odense C, Denmark ( mosler@health.sdu.dk). 654

2 Social Environment and Substance-related Disorders 655 by the risk of being admitted to a hospital or dying from a diagnosis related to long-term drug or alcohol abuse in adult life. We also explore whether any such effects are independent of the subject s own education later in life. MATERIALS AND METHODS Study population According to official statistics, 12,270 boys were born within the metropolitan area of Copenhagen in These persons formed the population of the Danish longitudinal study (Project Metropolit), which has been described in detail elsewhere (16). Briefly, 11,376 members of this population who were alive and living in Denmark in 1968 were registered with a unique, personal identification number (referred to as a CPR number ) when the Civil Registn System was established. In 1965, a total of 7,877 (69.2 percent) of these males participated in a school-based survey. Data sources Data from birth certificates, including information on the date and place of birth, mother s marital status, and father s occupational status at the time of delivery, were collected manually for all the members of the original study population. The school-based survey, in 1965, included a questionnaire administered by the subjects teachers. Each pupil provided his name, birth date, place of birth, and father s occupation. The survey involved tests of cognition and questions regarding social aspins and leisure-time activities. As part of the conscription procedure, all Danish men undergo physical and mental examinations when they are about 18 years of age (17). In 2004, we collected data from the conscript registers for all cohort members who were alive in 1971 (n ¼ 11,337). Variables TABLE 1. Diagnoses and ICD* codes used for outcome definitions in a study of 7,877 Danish men born in 1953 Parental characteristics. Mother s marital status at the time of delivery was treated in three categories: married; unmarried (i.e., single, divorced, widowed); and unknown. Father s occupation at the time of birth was recorded in 23 strata (nonurban self-employed (four strata); urban selfemployed (six strata); salaried employed (five strata); skilled workers and unskilled workers (five strata); pensioners; students; and unknown). In the preliminary data analyses, we computed risk estimates for various combinations of the 23 strata. This exercise suggested that they could be combined into three categories (high/middle, which included the self-employed and salaried employed; working class, which included skilled and unskilled workers; and unknown, which included the groups of pensioners (n ¼ 1), students (n ¼ 44), and fathers not known). School and leisure-time characteristics. Social participation at age 12 years was measured by four among 33 different items from the school survey identifying preferred leisure-time activity. These items were the following: 1) to be at home with friends; 2) to be at home with family; 3) to meet with friends; or 4) to visit a youth club. School satisfaction at age 12 years was measured by the question of whether or not the boy liked to go to school. Cognitive performance was measured by the Härnquist school test (14). This test consisted of spatial, arithmetic, and verbal subtests. Each subtest contained 40 problems, with one point awarded for each correct answer. Conscript characteristics. Educational level, primarily reflecting school education, was registered on a scale ranging from leaving school after the eighth grade to attaining the approximate equivalent of the British advanced-level general Certificate of Education. In the present analysis, this information was recoded into three categories. Follow-up Diagnosis ICD-8* code ICD-10* code Drug related Opioids , F11.0 F11.9 Cannabinoids F12.0 F12.9 Sedatives/hypnotics , F13.0 F13.9 Cocaine F14.0 F14.9 Other stimulants F Hallucinogens F16.0 F16.9 Other and multiple drugs , F18.0 F19.9 Alcohol related Alcohol psychosis and abuse syndrome F10.0 F K70.0 K70.9 Cirrhosis of the liver , I85.0 I85.9 Esophageal varices , * ICD, International Classification of Diseases; ICD-8, ICD, Eighth Revision; ICD-10, ICD, Tenth Revision. In August 2004, the Metropolit cohort was followed up for vital status through record linkage with the Civil Registn System registry. If a subject was not alive or living outside Denmark, the date of death or the date of emign/disappearance was obtained. Information on the time of admission to psychiatric wards from 1969 to January 2003, as well as diagnosis on discharge, was obtained from the Danish Psychiatric Central Register. This register has compiled computerized data on admissions to psychiatric hospitals and to psychiatric departments in general hospitals, in Denmark, since April 1969 (18). The National Patient Registry provided information on admission to somatic wards since 1978, when this registry was established (19). Causes of death from 1968 to December 2001 were determined through record linkage with the Cause of Death Registry (20). Diagnoses were classified according to the International Classification of Diseases (ICD), Eighth Revision, for the years and according to the ICD, Tenth Revision, from 1994 onward. Drug and alcohol abuse were defined according to the ICD codes in table 1.

3 656 Osler et al. TABLE 2. Number of cases of drug and alcohol abuse among Danish men born in the metropolitan area of Copenhagen in 1953 diagnosis Participated in school survey (n ¼ 7,877) Did not participate in school survey (n ¼ 3,399) diagnosis Drug and/or alcohol abuse diagnosis diagnosis diagnosis Drug and/or alcohol abuse diagnosis No. % No. % No. % No. % No. % No. % Psychiatric register Somatic register Death register Total* * Number of persons with appearance in at least one of the above-mentioned registers. Statistical methods Associations between social variables in early life and drug or alcohol abuse were analyzed using Cox s proportional hazards regression model with age as the underlying time scale. The events were defined as death from or firsttime discharge from a hospital with a substance abuserelated diagnosis, as described in table 1. Entry time was the subject s age on April 1, 1969, and follow-up ended at the time of event, time of emign from Denmark, death from nonevents, or January 1, 2003, whichever came first. Since follow-up in the National Patient Registry started in 1978 and most conscription board examinations took place in 1971 and 1972, we repeated all analyses with the age in 1978 as the entry. These two approaches, however, gave essentially the same results. In the present paper, therefore, we report only the first. The proportional hazards assumption was evaluated for all variables by comparing estimated -lognormal(-lognormal) survivor curves over the different categories of variables being investigated versus lognormal (analysis time) and by tests based on the generalization as described by Grambsch and Therneau (21). Statistical analyses were computed using STATA, version 7, software (Stata Corpon, College Station, Texas). RESULTS By follow-up, a total of 316 (4.0 percent) of the 7,877 men who participated in the school survey had died or been discharged from a hospital with a diagnosis related to drug abuse (table 2). The median age for first diagnosis was 25.6 years. For alcohol abuse, the respective values were 696 (8.8 percent) and 28.2 years, respectively. The risk of substance abuse was slightly higher among men who did not attend the school survey. The distributions of father s occupational class and mother s marital status at the subjects birth are shown in table 3, along with the unadjusted and mutually adjusted hazard s for drug and alcohol abuse according to these two characteristics. Both indicators of poor socioeconomic circumstances during early life were associated with an increased risk of discharge with a drug abuse diagnosis before and after mutual adjustments (table 3). A similar pattern of associations was seen for alcohol abuse. Crude analyses for those of the 3,499 nonparticipants for whom data were available on the mother s marital status and father s occupational class at the subjects birth produced risk estimates in the same direction as those based on the 7,877 subjects who participated in the school survey. Compared with those whose parents were of higher social positions, men who were born to single mothers or had working-class fathers were more likely to select youth-club visits as their favorite leisure-time activity, to dislike school, and to have lower scores on the school test at age 12 years. Preferring to be at home, with either friends or family, during leisure time was not associated with paternal occupational class or mother s marital status (data not shown). Those with youth-club visits as their leisure-time preference had a significantly increased risk of drug and alcohol abuse when adjustments were made for other school and leisure-time characteristics (table 4). Further, those who disliked school or scored low on the school test at age 12 years had an increased risk of substance abuse. The effects of indicators of social circumstances during early life were only slightly attenuated by the inclusion of indicators of social participation, satisfaction with school, and performance at age 12 years in the multivariate model (table 5). The effect of the latter characteristics did not vary in relation to social indicators (no significant interactions). The lowest educational level at the conscription board examination was most often found among men who were born to single mothers, had working-class fathers, preferred to visit youth clubs, did not like school, or scored low on the school test at age 12 years. Educational status at that examination was also inversely associated with the risk of both drug and alcohol abuse, and the risk estimates for social circumstances during early life, childhood social participation, and school performance changed slightly when adult educational level was included in the model (table 6). DISCUSSION In this cohort of Danish men born in 1953 who grew up during a youth rebellion that featured rock music, drugs, and sexual liben, we found that the risk of being admitted to a hospital or dying from a condition related to drug and alcohol abuse was surprisingly high. Based on rather stringent criteria of abuse, namely, death or hospital admission and diagnoses associated with long-term use, the risk was 12

4 Social Environment and Substance-related Disorders 657 TABLE 3. in 1953 Risk of drug or alcohol abuse at age years according to parental characteristics among 7,877 Danish men born No. % No. Unadjusted Adjustedy No. Unadjusted Adjustedy Participated in school survey (n ¼ 7,877) Mother s marital status at birth Married 7, Single , 2.54* , 2.30* , 2.29* , 2.25* Unknown Father s occupational status at birth High/middle 3, Working 3, , , , , 1.75 Unknown , 3.04* , 3.22* , 2.43* , 2.16* Did not participate in school survey (n ¼ 3,399) Mother s marital status at birth Married 2, Single , 2.38* , , 2.71* , 2.47* Unknown Father s occupational status at birth High/middle 1, Working 1, , , , , 2.42 Unknown , 3.09* , 4.00* , 2.84* , 2.69* * p < (Wald s test). y The adjusted model includes marital status and occupational class. percent. Indicators of poor social circumstances during early life were associated with an increased risk of adult substance abuse. Youth-club visits as a preferred leisure-time interest, poor satisfaction with school, and a low performance test score at age 12 years also increased the risk of substance abuse. These effects were slightly attenuated after adjustment for educational status around the age of 18 years, which itself was inversely related to both drug and alcohol abuse. We found no effect associated with preferring to spend leisure time at home with family or friends. Strengths and limitations The present data relate to all males who were born in a well-defined area (covering one third of the Danish population) and who survived to the age of 15 years. We had prospectively collected information on life conditions and attitudes at age 12 years and at the conscription board examination. By using the population-covering registers, we managed to get complete follow-up information. Thus, hospital admission data were available for more than 95 percent of this nonselected population. Information on social participation, satisfaction with school, and school performance, however, was available for only 69.2 percent of the cohort members, and the risk of drug and alcohol abuse was slightly higher among nonparticipants (13.0 percent vs percent). On the other hand, the risk estimates for the parental variables that were available for all participants did not differ much for nonparticipants. The follow-up covered a period of more than 30 years, but it might take several years to develop severe complications of alcohol abuse. It seems likely, therefore, that future follow-up will capture more cases. We had information only on disorders diagnosed during admission to hospitals. Consequently, we assume that our outcome measure is more sensitive to dropout during follow-up than self-reports, since it does not depend on the subject s ability to answer a questionnaire. Our openal definition was rather conservative, because we used diagnoses closely linked to drug and alcohol abuse and not more broadly defined alcohol-related outcomes, such as gastric cancers, injuries, and acute alcohol intoxication. In Denmark, admission to hospitals is free, and it is likely that most of those with a diagnosis related to advanced abuse have been admitted to a hospital. It does remain possible, however, that some men affected by drug or alcohol use are

5 658 Osler et al. TABLE 4. Risk of drug or alcohol abuse at age years according to school and leisure-time characteristics at age 12 years among 7,877 Danish men born in 1953 Total No. % No. Unadjusted Adjustedy No. Unadjusted Adjustedy Social participation at age 12 years Visits youth club No 5, Yes 2, , 2.31** , 1.97** , 1.60** , 1.45* Meeting with friends No 3, Yes 4, , 1.77* , , 1.41* , 1.34 Being at home with family No 2, Yes 4, , , , , 1.32 Being at home with friends No 3, Yes 4, , 1.66* , , , 1.18 Likes to go to school at age 12 years Yes 3, Somewhat 3, , , , , 1.28 No , 3.88** , 3.43** , 2.20* , 2.10* Cognitive school test Highest third 2, Middle third 2, , , , , 1.65 Lowest third 2, , 2.76** , 2.67** , 2.57** , 2.47** * p < 0.05; **p < y The adjusted model includes all the variables in the table. untreated, treated solely in private specialist practice, or treated by a general practitioner. Such cases are not included in this study and might also be missed in studies based on self-reported substance abuse, because of the presumed low attendance rate in surveys among heavy abusers. One might also challenge the contemporary relevance of the social participation measures we used, since the Copenhagen youth club was a social pedagogic experiment designed, in the 1960s, to assist youths in deprived areas. In addition, it might be that some of the men in this cohort were just beginning their substance abuse at the time of their conscription board examinations. We used information on educational level and not on current occupation, since the latter is more likely to be a consequence of current interest. Further, half of the cases had their first admission at age 25 years or later, and risk estimates were essentially the same when we used 1978 as the entry time in statistical analyses. Although substance abuse in cohorts born in the 1950s is more common in men than in women, it is an obvious limitation to our study that women were not represented in the data set. In addition, no information was available on some important determinants such as parental abuse. On the other hand, previous studies have shown that relations between adverse childhood experiences and alcohol abuse exist independently of parental abuse (4, 10). Comparison with other studies Few of the increasing number of studies concerning childhood socioeconomic circumstances and adult mortality have included outcomes of substance abuse (12). A large, male cohort in Finland showed more alcohol-related deaths among men with fathers from a lower social class (13), and, in the Oslo mortality study, housing conditions during childhood were associated with psychiatric deaths due largely to alcohol or drug dependence (14). A Danish register study, which included 84,765 children born in 1966, showed that parental abuse of alcohol was associated with an increased

6 Social Environment and Substance-related Disorders 659 TABLE 5. Risk of drug or alcohol abuse at age years according to parental, school, and leisure-time characteristics among 7,877 Danish men born in 1953y Mother single at birth , , 2.17* Father s occupational class Working vs. high/middle , , 1.58** Unknown vs. high/middle , , 2.02 Visits youth club at age 12 years , 2.01** , 1.44* Meeting with friends , , 1.43 Being at home with family , , 1.31 Being at home with friends , , 1.18 Likes to go to school at age 12 years Somewhat vs. yes , , 1.20 No vs. yes , 3.61** , 2.03** Cognitive school test at age 12 years Middle vs. highest third , , 1.53 Lowest vs. highest third , 2.37** , 2.22** * p < 0.05; **p < y The adjusted model includes all the variables in the table. TABLE 6. Risk of drug or alcohol abuse at age years according to parental, school, and leisure-time characteristics among 7,877 Danish men born in 1953y Mother single at birth , , 2.03* Father s occupational class Working vs. high/middle , , 1.32 Unknown vs. high/middle , , 1.76 Visits youth club at age 12 years , 1.91** , 1.37 Meeting with friends , , 1.45* Being at home with family , , 1.34 Being home with friends , , 1.19 Likes to go to school at age 12 years Somewhat vs. yes , , 1.13 No vs. yes , 2.90** , 1.74 Cognitive school test at age 12 years Middle vs. highest third , , 1.23 Lowest vs. highest third , , 1.48 Educational level at conscript High Middle , , 2.75 Low , , 6.44 Did not attend/unknown , 19.54** , 4.24** * p < 0.05; **p < y The adjusted model includes all the variables in the table.

7 660 Osler et al. risk of hospital admission for a diagnosis related to drug addiction (crude odds ¼ 3.3, 95 percent : 2.4, 4.5). However, the effect was explained mainly by factors that were closely linked to parental alcohol abuse (e.g., long-term parental unemployment, low education, violence in the family, and mental illness) (4). In the Adverse Childhood Experience Study (10, 11), a retrospective cohort study, 9,346 adults who visited a primary care clinic completed a survey about nine adverse childhood experiences, including childhood mental and physical abuse, domestic violence, parental divorce, and growing up with a drugabusing or mentally ill household member. Each of the adverse childhood experiences was associated with selfreported adult alcoholism, and the number of adverse experiences had a graded relation to alcoholism, independent of parental alcohol abuse (10). Some cross-sectional and short-term follow-up studies have analyzed social determinants of adolescent substance use, and a wide range of factors (e.g., the personality of the young, factors concerning their family situation, peer-related factors, school and leisure-time activities) have been shown to contribute to the development of adolescent addictive behavior (2, 9). A few studies have included factors from more than one category in order to distinguish relative effects. Those studies that are available, however, have associated high alcohol consumption and drug use in adolescence with low socioeconomic position, parental substance abuse, poor family relations, time spent with peers, peer support and risk behavior, lack of school adoption, low self esteem, and high alcohol use in young adulthood (1 9). The present study suggests that the family situation, peer factors, and school-related factors also predict to continuous and problematic substance use, as indicated by a higher risk of substance abuse requiring medical treatment in adulthood. Interpretation The present study indicates that social disadvantages during childhood, dissatisfaction with school, and poor school performance are related to the risk of hospital admission or death from drug or alcohol abuse in young adult men. To our surprise, preference for youth-club participation was also associated with higher risks. This might reflect the fact that 1960s Copenhagen youth clubs were attended primarily by low-income youths. Multiple factors, such as social relations and psychological stress, have been suggested as mediators in the relation between early social circumstances and substance abuse later in life (11, 12). In our study, the occupational class variable might reflect material wealth, parental attitudes, and behavior, while the risk associated with single mothers also reflects limited resources in childhood. These factors, in turn, might influence social participation, school performance, and the behavior of offspring. Our findings indicate that prevention of drug and alcohol abuse in families and schools should start early. Medical and public health practitioners and teachers should also have an increased awareness of adverse childhood conditions and signs of poor performance. Since dissatisfaction with school and poor school performance in early adolescence seem to be rather strong predictors of adult substance abuse, parents and schools should be aware of these early markers and their social origins. Further, both parents and school teachers should appreciate the importance of a pleasant and inspiring school environment. In addition, youth clubs should consider students access to alcohol and drugs in the institution. The high prevalence of drug and alcohol abuse in these Danish men also indicates that the liberal Danish drug policy, especially in relation to alcohol (22), is problematic and needs to be revised. ACKNOWLEDGMENTS Supported by the Danish Heart Association, the Lundbeck Foundation, and the Danish Health Insurance funds. The authors thank all those who initiated and/or continued the Metropolit Study: K. Svalastoga, E. Høgh, P. Wolf, T. Rishøj, G. Strande-Sørensen, E. Manniche, B. Holten, I. A. Weibull, and A. Ortman. Conflict of interest: none declared. REFERENCES 1. Galea S, Ahern J, Vlahov D. Contextual determinants of drug use risk behaviour: a theoretical framework. J Urban Health 2003;80(suppl 3):iii Hoffmann JP, Su SS. Paternal substance use disorder, mediating variables and adolescent drug use: a non-recursive model. Addiction 1998;93: Friedman AS, Glassman BA. Family risk factors versus peer risk factors for drug abuse. A longitudinal study of an African American urban community. J Subst Abuse Treat 2000;18: Christoffersen MN, Soothill K. The long-term consequences of parental alcohol abuse: a cohort study of children in Denmark. J Subst Abuse Treat 2003;25: Poulton R, Caspi A, Mine BJ, et al. Association between children s experience of socioeconomic disadvantage and adult health. Lancet 2002;360: Ary DV, Tildesley E, Hops H, et al. The influence of parent, sibling and peer modeling and attitudes on adolescent use of alcohol. Int J Addict 1993;28: Wills TA, Vaughan R. Social support and substance use in early adolescence. J Behav Med 1989;12: Poikolainen K, Tuulio-Henriksson A, Aalto-Setälä T, et al. Predictors of alcohol intake and heavy drinking in early adulthood: a 5-year follow-up of year-old Finnish adolescents. Alcohol Alcohol 2001;36: Andersen A. Epidemiological studies on alcohol use in adolescence. (PhD thesis). Copenhagen, Denmark: University of Copenhagen, Anda RF, Whitfield CL, Felitti VJ, et al. Adverse childhood experiences, alcoholic parents, and later risk of alcoholism and depression. Psychiatr Serv 2002;53: Felitti VJ, Anda RF, Nordenberg D, et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. Am J Prev Med 1998;14: Galobardes B, Lynch JW, Davey Smith G. Childhood socioeconomic circumstances and cause-specific mortality in

8 Social Environment and Substance-related Disorders 661 adulthood: systematic review and interpretation. Epidemiol Rev 2004;26: Pensola TH, Martikainen P. Cumulative social class and mortality from various causes of adult men. J Epidemiol Community Health 2003;57: Claussen B, Davey Smith G, Thelle D. Impact of childhood and adulthood socioeconomic position on cause specific mortality: the Oslo Mortality Study. J Epidemiol Community Health 2003;57: Nordentoft M, Wandall-Holm N. 10 year follow up study of mortality among users of hostels for homeless people in Copenhagen. BMJ 2003;327: Osler M, Andersen AM, Lund R, et al. Revitalising the Metropolit 1953 Danish male birth cohort: background, aims and design. Paediatr Perinat Epidemiol 2004;18: Green A. The Danish Conscription Registry: a resource for epidemiological research. Dan Med Bull 1996;43: Munk-Jørgensen P, Mortensen PB. The Danish Psychiatric Register. Dan Med Bull 1997;44: Andersen TF, Madsen M, Jørgensen J, et al. The Danish National Patient Register. A valuable source of data for modern health science. Dan Med Bull 1999;46: Juel K, Helweg-Larsen K. The Danish register of cause of death. Dan Med Bull 1999;46: Grambsch PM, Therneau TM. Proportional hazard tests and diagnostics based on weighted residuals. Biometrika 1994; 81: Chenet L, McKee M, Osler M, et al. Alcohol policy in the Nordic countries. Why competition must have a public health dimension. BMJ 1997;314: