the Metropolit 1953 Danish male birth cohort - Semantic Scholar

1 downloads 0 Views 78KB Size Report
May 14, 2008 - attenuated the association of father's social class with adult behav- iour, while ... nation by a medical doctor.19 The results of these tests.
Published by Oxford University Press on behalf of the International Epidemiological Association ß The Author 2008; all rights reserved. Advance Access publication 11 June 2008

International Journal of Epidemiology 2008;37:1367–1374 doi:10.1093/ije/dyn115

SOCIAL EPIDEMIOLOGY

Childhood social circumstances and health behaviour in midlife: the Metropolit 1953 Danish male birth cohort Merete Osler,1* Nina S Godtfredsen2 and Eva Prescott3

Accepted

14 May 2008

Background It has been suggested that the association between social disadvantage in childhood and adult mortality could come about through processes related to the family environment in which the child is raised. This study examines the relationship of fathers’ social class with health behaviour in middle age and assesses the potential mediating role of cognitive function, educational status and social integration in young adulthood in these relationships. Methods

We used data from the Metropolit cohort which includes 11 532 Danish men born in 1953 with information on fathers’ social class at participants’ birth and assessments of cognitive performance, education and social integration in early adulthood. In 2004, 6292 of these men participated in a follow-up survey on health and behaviour. Logistic regression was used to investigate the association of father’s social class with smoking, alcohol drinking, leisure-time physical activity and the intake of fruit and vegetables in midlife.

Results

Middle-aged men with fathers from higher social classes were more often ex-smokers, wine drinkers and daily consumers of fruit or vegetables than men with working class fathers. Leisure activities and high alcohol consumption were not related to father’s social class. Cognitive function and educational achievement at age 18 attenuated the association of father’s social class with adult behaviour, while indicators of social integration had very little impact on the associations.

Conclusions Father’s social class influences adult smoking, alcohol preference and food intake, and a major part of the effect is mediated through cognitive function and education. Keywords

1

Social class, cohort, smoking, alcohol drinking, cognitive function

Department of Social Medicine, University of Copenhagen, Denmark. 2 Department of Lung medicine, Bispebjerg University Hospital, Copenhagen, Denmark. 3 Department of Cardiology, Bispebjerg University Hospital, Copenhagen, Denmark. * Corresponding author. Department of Social Medicine, Øster Farigmagsgade 5, 1014 Copenhagen K, Denmark. E-mail: [email protected]

A number of cohort studies has suggested that childhood socio-economic circumstances contribute to different causes of death.1 In particular, childhood conditions appear related to mortality from lifestyle related diseases such as respiratory and cardiovascular diseases,1–3 and it has been suggested that the associations could come about through processes related to the family environment in which the child is raised. Thus, children from socially disadvantaged

1367

1368

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

families are more often exposed to role models with an unhealthy behaviour such as smoking and heavy drinking. Children from disadvantaged families also demonstrate lower levels of education and cognitive function.4–6 Early educational and cognitive experiences may contribute to social functioning and emotional wellbeing, which in turn could influence health behaviour in adulthood.4,7–16 However, most of the longitudinal studies with socially related factors in childhood associated with adult health behaviour have been based on birth cohorts from the UK,7–9,11–13,15–17 and have considered only one behaviour4,7–9,11,12,14,17 or have focused on the relation between cognitive function and behaviour.8–11,13–16 The purpose of the present analysis is to: (i) add to the literature on socio-economic status early in life and health behaviour in middle age by examining the relationship of fathers’ social class at participants’ birth with smoking, alcohol drinking, leisure-time physical activity and intake of fruit and vegetables later in adult life; (ii) assess the role of the potential confounding or mediating variables of cognitive function, educational status and indicators of social integration in early adulthood in these relationships.

Methods The Metropolit cohort is defined as the 11 532 men, born in 1953 in the Copenhagen Metropolitan area, who were living in Denmark in 1968. The cohort is described in detail elsewhere.18 Data from birth certificates, including information on birth dimensions, mother’s marital status and father’s occupational status at the time of delivery were manually collected for all members of the original study population, in 1965. For decades, nearly all Danish men have had to register with the conscription board when they are about 18 years old. Here they will fulfil a cognitive test, the Boerge Prien’s test, and undergo a health examination by a medical doctor.19 The results of these tests have been collected manually from all Danish conscript district registers for 11 108 of the 11 494 cohort members who were alive and living in Denmark in 1971. Information on selected social characteristics were obtained for each year for all cohort members since 1968 by linkage to the Danish civil registration system (CRS) and health registers. In 2004, the 9507 members of this cohort who were still alive with an available address in Denmark were sent a questionnaire to which 6292 (66.2%) responded.

Assessment of the behavioural variables From the questionnaire based follow-up in 2004 we used the self-reported information from validated questions20 on current smoking to assign the participants to one of 3 categories: current smoker, ex-smoker and neversmoker. We also included information on whether father or mother had been

a smoker when the study participants were aged 12. Alcohol consumption was assessed as the number of drinks consumed per day, and heavy drinking defined as more than 21 drinks per week, while abstainers were those drinking 0 drinks per week, who had also reported that they did not drink alcohol. Preferred beverage type (beer, wine and spirits) was defined as one beverage type making up more than 50% of the total consumption of alcohol. Similar to the findings of a previous Danish study14 a preference for spirits was rare (3.7%) and unrelated to social class and other covariates. Thus, a preference for wine can be interpreted as a non-preference for beer. Further, we included data on the frequency of vegetable and fruit intake (dichotomized as yes, daily/not daily). Physical activity during leisure was assessed using a 4-item self-report of the frequency and intensity of walking, running and biking (a common form of transport in Denmark). Sedentary activity was defined as mainly reading, watching television or having other sedentary activities during leisure in contrast to walking, running, biking or other physical activities for at least 4 h a week.

Assessment of social class, cognitive performance and social integration From the birth records we used information on father’s occupational social class at participant’s birth. Validity of such reports are assumed to be high as compared to later recall.21,22 Subjects were assigned, on the basis of father’s occupation, to 23 strata: nonurban self-employed (4 strata); urban self-employed (6 strata); white collar workers (5 strata); blue collar workers (5 strata); pensioners; students and unknown. The preliminary data analyses showed that the estimates for self-employed and white collar workers were very close, and they were consequently combined. The unknown group (n ¼ 303) which also included the small group of pensioners (n ¼ 1) and students (n ¼ 67) and the blue collar workers were combined because their estimates were very similar. This left two categories: a high social class, which included self-employed and salaried employed, persons, and a working class, which included unskilled and skilled workers as well as the group of unknowns. Information on subjects’ own educational achievement was obtained from information collected at the conscript board examination. This ranged from basic school (7 years of schooling) to at least 14 years of schooling (equivalent of qualification for college/ university).6 The 45 min validated cognitive test, taken at the conscript board, which most cohort members (90%) attended in 1972–73, comprises four categories: letter matrices, verbal analogies, number series and geometric figures. The score is the total number of correct answers for 78 questions, and this tests has been shown to correlate well with the Wechsler Adult Intelligence Scale.23 As measures of social integration, we used information from the CRS on participants’

CHILDHOOD SOCIAL CIRCUMSTANCES AND HEALTH BEHAVIOUR

marital status at age 30 and labour market participation, i.e whether or not the participant was employed or was a student/trainee at age 22.

Statistical analyses Associations between father’s social class and each health behaviour outcome were analysed using logistic regression models. The analysis of ex-smoking was based on men who had ever smoked, while the analysis of drinking preference included only alcohol drinkers. For each behavioural outcome, we conducted a series of multivariable models to examine the effect of cognitive function, education and social integration on any associations. Four models adjust for the effect of each of the covariables and are used to examine support for a mediating effect of each of these variables. For example, an attenuation of any association with adjustment for cognitive function would lend support to the hypothesis that social disadvantage during childhood is associated with low cognitive function, which in turn results in an increased risk of the behaviour under study. Further, we mutually

1369

adjusted for all the covariables to determine whether these had independent effects or whether any one of these was more robustly associated with our outcomes. Finally, information on parental smoking was included in the models with smoking behaviour as the outcome. The initial data analyses supported that the cognitive test score and years of school education were entered into the logistic models as continuous variables. All analyses were conducted using Stata version 8.0 (Stata Corporation, Texas 2002).

Results Table 1 shows the baseline characteristics of the cohort in relation to the prevalence of each health behaviour outcome at age 51. At follow-up, 42.5% of the men were current smokers and 32.4% were exsmokers. A total of 22.7% of the men were defined as heavy alcohol consumers, while 10.8% were abstainers. Among the alcohol drinkers 50.2% had wine as preferred beverage type. Further, 32.2% of the men consumed fruit or vegetables daily and 17.7% were

Table 1 The distribution (in %) of adult smoking, alcohol drinking, food intake and leisure acitivity at the age of 51 years in relation to social class, cognitive performance, and social integration earlier in life among 6292 Danish men born in 1953 Heavy Daily fruit or Sedentary Father leisure alcohol Prefers vegetable from high Current activity drinker winea intake social smoker Ex-smoker Abstainer (n ¼ 1999) (n ¼ 1102) class (%) (n ¼ 2287) (n ¼ 2038) (n ¼ 655) (n ¼ 1673) (n ¼ 2813) Father’s social class at birth High social class (n ¼ 2962)

37.6

35.7

9.2

28.3

57.3

36.2

17.0

Low social class (n ¼ 3330)

46.7

29.6

12.2

27.2

43.7

28.7

18.4

Cognitive function at age 18 Highest quartile (n ¼ 1558)

63.3

31.0

33.9

7.1

29.0

61.6

38.2

16.1

Lowest quartile (n ¼ 1291)

27.0

54.1

28.6

17.0

26.0

35.1

24.4

20.9

Highest or secondary (n ¼ 3110)

62.0

34.8

34.9

7.7

28.7

59.9

38.9

15.1

Basic (n ¼ 1596)

28.6

56.9

30.0

13.7

26.6

37.0

24.7

20.6

School education at age 18

Labour market participation at age 22 Yes (n ¼ 4748)

50.0

42.5

32.7

11.7

28.2

49.5

32.7

17.8

No (n ¼ 1544)

46.1

42.3

31.5

10.6

26.4

50.5

32.8

17.4

Marital status at age 30

a

Married (n ¼ 3103)

45.6

42.3

31.5

8.9

26.8

52.9

33.0

16.0

Single (n ¼ 2907)

43.8

41.0

33.9

12.2

26.8

48.4

32.1

18.7

Divorced (n ¼ 282)

38.9

58.5

25.9

16.1

31.5

40.4

23.6

26.4

All

47.1

42.5

32.4

10.8

22.7

50.2

32.2

17.7

As percentage of all drinkers.

1370

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 2 Odds ratios [OR (95% CI)] of smoking, alcohol drinking, food intake and leisure activity at the age of 51 years in relation to social and mental characteristics among Danish men born in 1953 Cognitive function at age 18 (per SD)

School education at age 18 (per year)

Labour market participation at age 22

Divorce at age 30

1.45 (1.31–1.60)

0.67 (0.63–0.70)

0.81 (0.80–0.84)

1.01 (0.90–1.13)

1.97 (1.54–2.51)

1.06 (0.95–1.19)

0.82 (0.71–0.90)

0.88 (0.84–0.97)

0.98 (0.86–1.11)

1.19 (0.86–1.11)

Crude

0.66 (0.59–0.74)

1.38 (1.28–1.48)

1.16 (1.13–1.20)

1.03 (0.90–1.18)

0.56 (0.42–0.75)

Adjusteda

0.83 (0.73–0.95)

1.17 (1.06–1.28)

1.09 (1.04–1.14)

1.06 (0.91–1.21)

0.55 (0.41–0.75)

Crude

1.37 (1.11–1.61)

0.65 (0.60–0.71)

1.29 (1.25–1.33)

1.04 (0.92–1.18)

0.81 (0.74–0.89)

Adjusteda

1.03 (0.85–1.34)

1.07 (0.98–1.16)

1.24 (1.19–1.29)

1.06 (0.93–1.22)

0.76 (0.69–0.85)

0.98 (0.87–1.11)

1.06 (0.97–1.17)

0.98 (0.93–1.02)

1.41 (0.99–1.31)

1.34 (1.02–1.78)

0.98 (0.88–1.11)

1.03 (0.97–1.11)

1.00 (0.99–1.03)

1.09 (0.95–1.25)

1.32 (1.00–1.73)

0.56 (0.50–0.62)

1.55 (1.45–1.65)

1.29(1.25–1.33)

1.04 (0.92–1.18)

0.81 (0.74–0.89)

0.80 (0.70–0.90)

1.07 (0.98–1.16)

1.23(1.18–1.28)

1.10 (0.95–1.25)

0.76 (0.67–1.00)

0.71 (0.63–0.79)

1.27 (1.19–1.35)

1.20 (1.16–1.24)

1.09 (0.96–1.26)

0.64 (0.48–0.85)

0.94 (0.83–1.07)

0.90 (0.82–0.99)

1.23 (1.18–1.29)

1.16 (1.02–1.33)

0.67 (0.49–0.92)

Father’s social class low vs high Current smoking Crude Adjusted

a

Ex-smoking

Abstainer

Heavy alcohol drinker Crude Adjusted

a

Preferred wine drinker Crude Adjusted

a

Daily intake of fruit and vegetables Crude Adjusted

a

Sedentary leisure activity Crude

1.10 (0.97–1.26)

0.85 (0.79–0.91)

0.89 (0.86–0.92)

1.02 (0.88–1.19)

1.71 (1.30–2.24)

Adjusteda

0.90 (0.78–1.05)

1.01 (0.91–1.12)

0.88 (0.84–0.93)

1.04 (0.88–1.22)

1.55 (1.17–2.08)

a

Father’s social class, cognitive function, school education, labour market participation and divorce at age 30 in the logistic regression model.

sedentary during leisure (Table 1). Table 2 shows the unadjusted associations of father’s social class and the early life characteristics with each behavioural outcome. Men with fathers from the working class, were at a higher ‘risk’ of being a current smoker [OR ¼ 1.45 (1.31–1.60)] and abstaining from alcohol [OR ¼ 1.37 (1.11–1.61)], but at a lower ‘risk’ of being an ex-smoker [OR ¼ 0.66 (0.59–0.74)], a preferred wine drinker [OR ¼ 0.56 (0.50–0.62)], or a daily consumer of fruit and vegetables [OR ¼ 0.71 (0.63–0.79)] compared with men with fathers from the high social class. Father’s social class was not related to heavy drinking [OR ¼ 0.98 (0.87–1.11)] or sedentary leisuretime activity [OR ¼ 1.10 (0.97–1.26)]. Cognitive performance, education and being divorced at age 30 were associated with father’s social class (Table 1) and with most behaviour outcomes (Tables 1 and 2), and Table 3 shows how father’s social class relates to the behavioural outcomes after adjustment

for each of these potentially confounding or mediating variables. The associations between father’s social class and health behaviour at age 51 were weakened somewhat when cognitive performance or education were added separately to the multivariable model, but were retained except for abstention. Adjustment for indicators of social integration had very little impact on the relationships. In multivariable models which adjusted for all variables the effect estimates for ex-smoking [ORlow vs high ¼ 0.83 (0.73–0.95)] and preferred wine drinking [ORlow vs high ¼ 0.80 (0.700.90)] did not cross unity, while the associations were lost for current smoking [ORlow vs high ¼ 1.06 (0.95–1.19)] and daily intake of fruit and vegetables [ORlow vs high ¼ 0.94 (0.83–1.07)] (Tables 2 and 3). In the fully adjusted models cognitive performance, own education and divorce at age 30 remained as predictors of current smoking, ex-smoking, abstention, preferred wine drinking and intake of fruit and vegetables.

CHILDHOOD SOCIAL CIRCUMSTANCES AND HEALTH BEHAVIOUR

1371

Table 3 Multivariable association (OR) of father’s social class and smoking, alcohol drinking, food intake and leisure activity at the age of 51 years among Danish men born in 1953

Current smoking

Labour Cognitive School market Divorce at function, education, participation, age 30, Multiply, Unadjusted adjusted adjusted adjusted adjusted adjusteda 1.45 (1.31–1.60) 1.17 (1.04–1.31) 1.12 (1.01–1.26) 1.45 (1.31–1.61) 1.44 (1.30–1.61) 1.06 (0.95–1.19)

Ex-smoker

0.66 (0.59–0.74) 0.78 (0.69–0.89) 0.79 (0.71–0.90) 0.66 (0.59–0.74) 0.66 (0.59–0.74) 0.83 (0.73–0.95)

Abstainer

1.37 (1.11–1.61) 1.11 (0.92–1.34) 1.06 (0.92–1.34) 1.37 (1.11–1.67) 1.36 (1.15–1.61) 1.03 (0.85–1.34)

Heavy alcohol drinker

0.98 (0.87–1.11) 1.00 (0.88–1.14) 0.99 (0.87–1.11) 0.98 (0.87–1.09) 0.98 (0.87–1.10) 0.98 (0.88–1.11)

Preferred wine drinker

0.56 (0.50–0.62) 0.66 (0.59–0.75) 0.76 (0.68–0.85) 0.56 (0.50–0.62) 0.55 (0.49–0.61) 0.80 (0.70–0.90)

Daily intake of fruit 0.71 (0.63–0.79) 0.80 (0.71–0.91) 0.89 (0.79–1.00) 0.71 (0.63–0.79) 0.71 (0.63–0.79) 0.94 (0.83–1.07) and vegetables Sedentary leisure activity

1.10 (0.97–1.26) 1.00 (0.87–1.16) 0.93 (0.81–1.07) 1.10 (0.96–1.25) 1.08 (0.95–1.23) 0.91 (0.78–1.05)

a Father’s social class, cognitive function, school education, labour market participation and divorce at age 30 in the logistic regression model.

Eighty-seven per cent of the men recalled that at least one of their parents had been a smoker when they were young. Parental smoking was most common among current smokers (92%) and in the lowest social class (90%). The inclusion of parental smoking in the model did not influence the effect of father’s social class (data not shown).

Discussion This population based cohort study showed that men with fathers from the higher social classes had a healthier lifestyle: they were more often ex-smokers, wine drinkers and daily consumers of fruit or vegetables than men with fathers from the working class. Adjustment for cognitive function and educational status at age 18 years explained the major part of the relations, while indicators of social integration had very little impact. Father’s social class was not associated with heavy alcohol consumption and leisure activity in midlife. A number of investigators have examined the relation between childhood social circumstances and adult smoking. In a cohort study from New Zealand, father’s social class was associated with smoking behaviour at age 25 even after adjustment for the subject’s own educational level, childhood cognitive function and parental smoking.4 However, in the British 1958 cohort a relation between childhood social circumstances and current smoking at age 41 was eliminated by adjustment for educational qualifications obtained at age 23.7 A similar pattern was seen among men in the Whitehall II study,24 where smoking became weakly associated with father’s

social class when own employment grade was adjusted for. Further, a Finnish cohort study showed no relation between childhood conditions and smoking in middle-aged men.5 A few cohort studies have examined the link between social circumstances early in life and later alcohol consumption, food intake and physical activity. In Finnish men and women adverse childhood conditions were associated with low intake of fruit and vegetables and low leisure activity, but not with alcohol consumption. However, the effect of potential confounding/mediating variables was not examined in this study. In the 1946 British cohort, parental education was positively associated with frequency of physical activity at age 36.17 Two recent studies from UK have related childhood socioeconomic status to diet later in life. In the Boyd Orr cohort, childhood social class was not associated with a healthy diet score,12 while a high parental social class was significantly related to self-reported vegetarianism in the 1970 birth cohort study.11 In the latter high education and high childhood cognitive function were also associated with vegetarianism, but the effect on childhood social class of adjustment for these variables was not reported. Most other cohort studies have focused on the influence of childhood cognitive function on health behaviour8–10,13–16 and have not reported separately on the relation to childhood social class.

Study strengths and limitations The present study includes all males born in a well-defined area (covering one-third of the Danish population), and who survived to the age of 51 years. We had nearly complete and prospectively collected

1372

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

information on early life conditions and measures of cognitive function at age 18. There may be some inaccuracy in self-reported behaviour items. This is a problem that affects all research using such information. If errors in reporting are non-differential (independent of father’s social class) this might lead to an underestimation of the relations. At the followup survey in 2004, there were 384 (3.3%) emigrations, 623 (6.9%) with unknown (47) or protected (576) address and 1018 (8.9%) deaths. Further, 3215 (33.8% of those eligible) did not respond in the questionnaire survey. Non-participation was associated with having a single mother at birth and low childhood cognitive function.25,26 The incidence of hospitalizations for alcohol abuse, tobacco-related lung disease and depression were also higher among non-responders. When we analysed the associations between information on early life factors and these health outcomes, which were available for all cohort members, we found that the risk estimates did not differ significantly between responders and the entire cohort. Thus, we concluded that the non-response did not bias the exposure–risk associations. In order to explore the mechanisms underlying the association between social class and cognitive function and the relatively high rates of unhealthy behaviour in these middle-aged men, we would have liked more information on the circumstances underlying this behaviour such as information on peer behaviour, and psychological attributes such as personality factors and coping styles. These characteristics might have provided other important mediating pathways for the associations that we have observed, but our study is based on exposure information collected in the past, and unfortunately we had no data on this in the files. We did have information on admission to a psychiatric hospital during young adulthood, but this is considered to be too extreme a marker of minor mental distress which might explain a health behaviour such as smoking. On the other hand, the inclusion of parental smoking and indicators of social integration in the statistical models did not change the effect of father’s social class. However, one might question the relevance of labour market participation at age 22 and marital status at age 30 as indicators of social integration. To our surprise the estimates for the unmarried men were close to those for the married. However, there might be some misclassification of this variable, since in this population of men it was quite common to cohabitate with a partner without being formally married. This may have diluted the effect of this variable. Our cohort consists of men only, and therefore the results are not necessarily applicable to women. However, in previous studies on the effect of childhood socio-economic status or cognitive function on different health behaviour the associations have been rather similar in men and women.4,7,12,13

Possible mechanisms and implications of our findings The associations of childhood social circumstances with adult current smoking, ex-smoking, drinking preference and intake of fruit and vegetables were attenuated somewhat when educational status or cognitive function at age 18 were adjusted for, but the effect estimates for ex-smoking and drinking preference were retained. The independent effect of childhood social circumstances on ex-smoking and drinking preference could reflect that psychological attributes and social norms such as peer role models—not captured by education and cognitive function—also influence the relation. Education and low cognitive function are closely correlated. Both factors may reflect one’s ability to process information and assess risk, one’s type of occupation and the society and culture in which one lives. A culture of unhealthy coping strategies and lack of resources in lower social classes and among those with lower intellectual abilities may influence behaviour later in life. However, the relation between father’s social class and behaviour was independent of indicators of social integration despite the association of divorce at age 30 with both smoking, drinking and dietary behaviour. Thus, our study does not support the assumption that these measures of social functioning are important mediators. It has been suggested that cognitive function precede educational achievement, and both might be improved through early learning. However, it is also worth considering the environmental determinants that generate the behaviour. Further, there are genetic sources of variation in social class, intelligence and behaviour,6,27,28 and the covariation between these domains could be mediated by genes. In conclusion, in the present study men with fathers from the higher social class were more often ex-smokers, wine drinkers and daily consumers of fruit and vegetables than men with fathers from the working class. Adjustments for cognitive function and educational status in young adulthood explained a major part of the relations, and these associations offer insight into the mechanisms underlying the link between father’s social class and mortality.

Acknowledgements We 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. This work was supported by the Danish Heart Association, the Lundbeck Foundation, the Danish Health Insurance Funds, the Danish Pharmaceutical Funds and Else and Aage Wedell-Wedellsborgs Fund. Conflict of interest: None declared.

CHILDHOOD SOCIAL CIRCUMSTANCES AND HEALTH BEHAVIOUR

1373

KEY MESSAGES  Middle-aged men with fathers from higher social classes were more often ex-smokers, wine drinkers and daily consumers of fruit or vegetables than men with working class fathers.  Cognitive function and educational status at age 18 explained the major part of the relations, while indicators of social integration had very little impact.  These associations offer insight into the mechanisms underlying the link between father’s social class and mortality.

References 1

2

3

4

5

6

7

8

9

10

11

12

Garlobardes B, Lynch JW, Davey Smith G. Childhood socioeconomic circumstances and cause specific mortality in adulthood: systematic review and interpretation. Epidemiol Rev 2004;26:7–21. Pensola TH, Martikainen P. The effect of parental social class, own education and social class on mortality among young men. Eur J Public Health 2002;12:29–36. Davey Smith G, Hart C, Blane D, Hole D. Adverse socioeconomic conditions and cause specific adult mortality: prospective observational study. Br Med J 1998;316: 1631–5. Fergusson DM, Horwood LJ, Boden JM, Jenkin G. Childhood social disadvantage and smoking in adulthood: result of a 25 year longitudinal study. Addiction 2007;102:475–82. Lynch JW, Kaplan GA, Salonen JT. Why do poor people behave poorly? Variation in adult health behaviours and psychosocial characteristics by stages of the socioeconomic lifecourse. Soc Sci Med 1997;44:809–19. Teasdale T, Owen DR. The influence of parental social class on intelligence and educational level in male adoptees and non-adoptees. Br J Educ Psychol 1986;56:3–12. Jeffens BJMH, Power C, Graham H, Manor O. Effects of childhood socioeconomic circumstances on persistent smoking. Am J Public Health 2004;94:279–85. Taylor MD, Hart CL, Smith GD et al. Childhood mental ability and smoking cessation in adulthood: prospective observational study linking the Scottish Mental Survey 1932 and midspan studies. J Epidemiol Comm Health 2003;57:464–65. Taylor MD, Hart CL, Smith GD et al. Childhood IQ and social factors on smoking behaviour, lung function and smoking-related outcomes in adulthood: linking the Scottish Mental Survey 1932 and the Midspan studies. Br J Health Psychol 2005;10:399–410. Kubicka L, Matejcek Z, Dytrych Z, Roth Z. IQ and personality traits assessed in childhood as predictors of drinking and smoking behaviour in middle-aged adults: a 24-year follow-up study. Addiction 2001;96:1615–28. Gale CR, Deary IJ, Schoon I, Batty GD. IQ in childhood and vegetarianism in adulthood: 1970 British cohort study. Br Med J 2007;334:245–47. Maynard M, Gunnell D, Ness AR, Abraham L, Bates CJ, Blane D. What influences diet in the early age? Prospective and cross-sectional analyses of the Boyd Orr cohort. Eur J Public Health 2006;16:316–24.

13

14

15

16

17

18

19

20

21

22

23

24

25

26

Batty GD, Deary IJ, Schoon I, Gales C. Childhood mental ability in relation to food intake and physical activity in adulthood: the 1970 British cohort study. Pediatrics 2007;119:e38–45. Mortensen LH, Sørensen TIA, Grønbæk M. Intelligence in relation to later beverage preference and alcohol intake. Addiction 2005;100:1445–52. Batty DG, Deary IJ, Schoon I, Gale CR. Mental ability across childhood in relation to risk factors for premature mortality in adult life: the 1970 British cohort study. J Epidemiol Comm Health 2007;61:997–1003. Batty GD, Deary IJ, MacIntyre S. Childhood IQ in relation to risk factors for premature mortality in middle-aged persons: the Aberdeen children of the 1950’s study. J Epidemiol Comm Health 2007;61:241–47. Kuh D, Cooper C. Physical activity at age 36 years: patterns and childhood predictors in a longitudinal study. J Epidemiol Comm Health 1992;46:114–19. Osler M, Lund R, Kriegbaum M, Christensen U, Nybo Andersen A-M. Cohort profile: The metropolit 1953 Danish male birth cohort. Int J Epidemiol 2006;35:541–46. Green A. The Danish Conscription Registry: a resource for epidemiological research. Dan Med Bulletin 1996;43:464–67. Christensen U, Kriegbaum M, Holstein B, Osler M. Documentation report. Volume 3: the 2004 questionnaire survey. Department of Public Health: University of Copenhagen, Copenhagen, 2007. Berney LR, Blane DB. Collecting retrospective data: accuracy of recall after 50 years judged against historical records. Soc Sci Med 1997;45:1519–25. Batty GD, Lawlor DA, Macintyre S, Clark H, Leon DA. Accuracy of adults recall of childhood social class: findings from the Aberdeen children of the 1950s study. J Epidemiol Comm Health 2005;59:898–903. Mortensen EL, Reinisch JM, Teasdale TW. Intelligence as measured by the WAIS and military draft board group test. Scand J Psychol 1989;30:315–18. Brunner E, Shipley MJ, Blane D, Davey Smith G, Marmot M. When does cardiovascular risk starts? Past and present socioeconomic circumstances and risk factors in adulthood. J Epidemiol Comm Health 1999;53:757–64. Osler M, Kriegnaum M, Christensen U, Lund R, Nybo Andersen A-M. Loss to follow-up did not bias associations between early life factors and adult depression. J Clin Epidemiol [Epub ahead of print May 19, 2008]. Osler M, Kriegbaum M, Holstein B, Christensen U, Nybo Andersen AM. Non-response in follow-up of a cohort of Danish men: consequences for associations of

1374

27

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

early life factors with lifestyle-related health outcomes. Ann Epidemiol 2008;18:422–24. Deary IJ, Spinath FM, Bates TC. Genetics of intelligence. Eur J Genetics 2006;14:690–700.

28

Osler M, Holst C, Prescott E, Sørensen TIA. Influence of genes and family environment on adult smoking behaviour assessed in an adoption study. J Genetic Epidemiol 2001;21:193–200.