Socio-economic position and coronary heart disease risk factors in youthFindings from the Young Hearts Project in Northern Ireland
FRANK J. VAN LENTHE 5
COLIN A. BOREHAM 5
JOS W.R. TW 0 5
JOHN J. STRAIN 5
J. MAURICE SAVAGE 5
GEORGE DAVEY SMITH 5
0 Department of Public Hearth, Erasmus University Rotterdam. The Netherlands 2 Department of Sport and Leisure, University of Ulster, Northern Ireland 3 Institute for Research In Extramural Mediane, Vrije Unlverslterl Amsterdam, The Netherlands 4 Northern Ireland Centre for Diet and Health, University of Ulster, Northern Ireland 5 Department of Child Health, Institute of Clinical Sciences, Royal Victoria Hospital Belfast. Northern Ireland 6 Department of Social Medidne, University of Bristol, United Kingdom University Rotterdam , P O. Box 1738, 3000 DR Rotterdam , The Netherlands, te\
1 , l.W.R. Twisfc
2 , JJ. Strain
3 , J.M. Savage
4 , G. Davey Smith
5 FJ. van Lenthe
Background: This study investigates the existence of socioeconomic differentials in behavioural and biological risk factors for coronary heart disease in young people from Northern Ireland, taking into account differences in biological maturation. Methods: A school-based prospective study, with measurements in 1989/1990 and 1992/1993. Socio-economic position was based on occupational level of the main family breadwinner. Behavioural risk factors included were physical inactivity, the intake of total energy, dietary fat and a number of mlcronutrients. Biological risk factors included were blood pressure, body fatness, llpoproteins and cardio-pulmonary fitness. Biological maturation was based on Tanner's stages. Participants: 251 boys and 258 girls who were measured at the age of 12 years and re-examined at the age of 15 years. Results: Cross-sectional analyses showed that socio-economic differences in cholesterol intake (In boys) and physical inactivity and total energy intake (in girls) were present at 12 and 15 years of age, while differences in fat and fruit intake and smoking behaviour (in boys and girls) became established at the age of 15 years, with unfavourable levels in subjects in the manual group. Longitudinal analyses confirmed that differences in behavioural risk factors exist or develop during adolescence. No clear pattern of differences in biological risk factors was found by socio-economic position. Adjustment for biological maturation did not materially alter the results. Conclusion: Differences in lifestyle by socio-economic position seem to become established in adolescence. These differences however, are not (yet) reflected in differences in biological risk factors by socio-economic position.
adolescence; behavioural risk factors; biological risk factors; CHD; socio-economic position
oronary heart disease (CHD) is still the leading cause
of mortality in many industrialised societies. It is now
established from prospective research that CHD
morbidity and mortality differ by socio-economic
position, with higher rates in subjects in the lower
socioeconomic strata.1 The socio-economic patterning of
CHD is reflected in socio-economic differences in the
prevalence of biological risk factors, such as hypertension,
hypercholesterolaemia, low cardio-respiratory fitness,
obesity and behavioural risk factors, such as physical
inactivity, smoking and micro-nutrient deficiency in
adulthood.2 There are, however, additional exposures
which may increase theriskof CHD among people in less
favourable circumstances, including low birth weight,
poor growth in childhood and psychosocial factors acting
It is known that the process underlying CHD,
atherosclerosis, starts in childhood and youth. Elevated
levels of biological risk factors are already apparent in
youth and, to a certain degree, predict the degree of risk
in adulthood.^ Further development of unfavourable
patterns of behavioural risk factors in youth is a matter of
concern. Hence, in order to improve population-based
preventive strategies there is a need to identify the
determinants of these biological and behavioural risk
factors during youth. Despite the relative consistency in
the socio-economic gradient in risk factors of CHD in
adulthood, it is less clear at which stage in life these
socio-economic differences are established.
Studies of the association between socio-economic
position and biological risk factors in childhood and
adolescence have thus far produced inconsistent results.
It has been suggested that this may be, at least partly, due
to differences in the ages of the populations under study.9
Given the fact that differences inriskby socio-economic
EUROPEAN JOURNAL OF PUBLIC HEALTH VOL 11 2001 NO. 1
position are apparent in adulthood, it could be
hypothesised that they develop during adolescence. It has
further been hypothesised that socio-economic
differences in behavioural risk factors precede the development
of socio-economic differences in some biological risk
factors during youth.10
Adolescence is characterised by the transition from
childhood into adulthood, through the process of biological
maturation. There is evidence that mean levels of
biological and behavioural risk factors differ according to the
timing of biological maturation.11'12 If diere is an
association between biological maturation and socio-economic
position, as for example found in a study in Finland,13
differences in mean values of biological and behavioural
risk factors by stage of biological maturation could mask
or explain the associations between socio-economic
position and these risk factors and further contribute to
the inconsistencies in findings presented thus far.
Northern Ireland faces one of the worlds' highest
mortality rates for CHD. To provide detailed
information for the need and direction of preventive strategies in
youth, a prospective cohort study, the Young Hearts
Project, was initiated in Northern Ireland in 1989/199O.15
The present study investigates die existence of social class differences in behavioural and biological risk factors at the age of 12 and 15 years, taking into account differences in biological maturation.
MATERIAL AND METHODS
The Young Hearts Project is a school-based prospective
study, aimed at investigating die development of biological
and behavioural risk factors for CHD. Details of die study
design and sampling procedure have been presented
elsewhere.15 In die original cross-sectional survey in
1989/1990, random samples of 12 and 15 year-old boys
and girls participated in die study (boys 12 year: N=251;
boys 15 year: N=252; girls 12 yean N=258; girls 15 year:
N=254; overall response rate: 78%). In 1992/1993, the
initially 12 year-old boys and girls were re-examined
under identical conditions (response rate: 87%). Reasons
for non-participation in die follow-up were refusal (N=9),
illness (N=25); moving from school (N=17) or odier
reasons (N=3). Only data for children aged 12 years at
baseline were included in diis study. Analyses for each
risk factor were performed in participants without missing
information on diat risk factor at bodi periods of
measurement, socio-economic position at die age of 12 years and
biological maturation at bodi periods of measurement.
As an indicator of die socio-economic position of die
children, information about die occupation of die main
breadwinner in die family was collected at die first period
of measurement and categorised using die Standard
Occupational Classification of die Office of Population
Censuses and Surveys Statistics (OPCS).16 The original
six categories (professionals; managerial and technical
occupations, skilled non-manual occupations; skilled
manual occupations; partly skilled occupations and
unskilled occupations) were dichotomised into a
nonmanual (upper diree classes) and manual (lower diree
classes) social class.
Behavioural risk factors
Information about physical activity was obtained from a
questionnaire, including aspects of everyday physical
activity (such as methods of transportation to and from
school, activities during breaks in die school days and
sports participation after school), which was previously
used in anodier large-scale study in Northern Ireland.17
In die computation of die activity score, frequency,
intensity and duration of die activities were taken into
account. This resulted in a weighted activity score,
varying between 0 and 100. Aldiough validation of die
questionnaire is complex in die absence of a gold
standard, associations widi physical fitness suggest die
questionnaire to be a good instrument for the purpose of
A dietary history mediod was used to collect information
about the nutritional intake.1" Questions were
openended and a photographic atlas, including more dian 170
photographs, was used to determine portion sizes. From
this information, total energy and nutrient intake were
calculated, using a computerised database. In die present
study, total energy intake (MJ/day) and fat intake
(expressed as a percentage of total energy intake) were
included. Given dieir protective effect against CHD, we
further included die intake of vegetables and fruit (borii
expressed as a percentage of total energy intake), fibre
(g/day), cholesterol intake (mg/MJ/day) and die
polyunsaturated to saturated fat ratio (P:S ratio).20'21
Information about smoking was obtained from a confidential
recall questionnaire, including questions about die
number of cigarettes smoked per week. Those who admitted
smoking at least one cigarette per week were characterised
Biological risk factors
Systolic (SBP) and diastolic blood pressure (DBP) were
measured twice from die right arm, using a Hawksley
random zero sphygmomanometer and widi die subjects
sitting quietly beforehand for at least five minutes. At die
age of 12 years, SBP was determined as die mean value of
die first Korotkoff phase. DBP was determined as die
mean value of die fourth Korodcoff phase (at die age of
12 years) or die fifth Korodcoff phase (at die age of 15
A non-fasting blood sample was drawn from die ante
cubital vein. From diis sample, total serum cholesterol
(TC) was estimated by an enzymatic technique
PAP, Boehringer, Mannheim). The concentration of high-density lipoproteins (HDL) was estimated by phosphotungstic magnesium reagents.
Anthropometrical measurements included body height and weight. From diese data BMI was calculated as weight (kg) divided by die height squared (m ). Further, four skinfold diicknesses (biceps, triceps, subscapular and
Soao^conomic position and CHD in youth
suprailiac) were estimated, according to Durnin and
Rahaman, and their sum (SSF) was also used as an
indicator of body fatness.23 Participants carried out a 20
metre endurance shuttle run test and the number of
completed laps was used an indicator of cardio-respiratory
fitness. A detailed description about the measurement
procedures is presented elsewhere.
Biological maturation was based on Tanner's pubic hair
stages.25 At the age of 12 years, boys and girls in the first
Tanner stage were characterised as slowly maturing, while
those in higher stages were characterised as rapidly
maturing. At the age of 15 years, rapidly maturing boys
and girls were those in the final stage, while slowly
maturing boys and girls had not yet reached that stage.
Independent sample t-tests were used to describe the
mean values ofriskfactors by socio-economic position at
the age of 12 and 15 years separately. In addition, a
longitudinal data-analysis technique was used to test the
overall significance of differences in the risk factors by
socio-economic position at both periods of time
simultaneously. Therefore, the technique of generalised
estimating equations (GEE) was used.26 In this technique,
data from several periods of measurements are included
in one analysis, in which adjustment is made for the
dependence of the data in the same subjects. As a result,
the number of data points in the analyses - and hence die
power of the tests - increases substantially. Its use in
epidemiological studies has been described in detail by
Twisk et al. ' In short, it may be regarded as a (linear)
regression technique widi time-dependent variables (for
example age) and time-independent variables (for
example gender). We used GEE to evaluate the effects of
socio-economic position at the age of 12 years
(time-independent variable) on the behavioural and biological
risk factors measured at both periods of time
(timedependent variables), adjusted for the period of
measurement (time-dependent variable). The resulting regression
coefficients for socio-economic position can be
interpreted similarly as in cross-sectional linear regression
analysis, although estimated using both periods of
measurement. In die model a (socio-economic position *
period of measurement) interaction term was included to
test whether the association between socio-economic
position and behavioural and biological risk factors
changed between die periods of measurement.
In order to investigate die effects of biological maturation
on die association between socio-economic position and
die biological risk factors, we first investigated die
association between socio-economic position and biological
maturation, using chi-square tests. Subsequently, analysis
of variance was used to investigate die effect of biological
maturation at die age of 12 and 15 year separately, while
GEE was used to test die effect of biological maturation (a time-dependent variable) over die entire period of research.
Tables I and 2 present die mean values of die behavioural
risk factors in boys and girls, respectively. In general, die
patterns are radier similar for boys and girls. Total energy
intake in subjects in the manual group was significantly
higher compared to diat in subjects in die non-manual
group at die age of 12 years, and also at die age of 15 years
in girls (but not in boys). Dietary cholesterol intake was
higher in die manual group in boys and girls at die age of
12 years, and also at die age of 15 years in boys (but not
in girls). At die age of 15 years, dietary fat intake (as a
percentage of die total energy intake) was significantly
higher in die manual group compared to die non-manual
group. At that age, boys and girls in die manual group
consumed significantly less fruit compared to subjects in
die non-manual group. In boys, the P:S ratio was
significantly lower in die manual group at die age of 12 years.
Contrary to expectation, fibre intake was significantly
higher in girls in die manual group at die age of 15 years.
Total physical activity was borderline lower in girls in die
manual group at die age of 12 years (p=0.06) and
significandy lower at die age of 15 years. At die age of 12 years,
only 5 boys and 3 girls smoked at least one cigarette per
week, and therefore, no meaningful comparison could be
made between die percentage of smokers in die manual
and non-manual groups. In die manual group, a
significantly higher percentage of boys and girls smoked dian
in die non-manual group at die age of 15 years.
The cross-sectional results were confirmed in die
longitudinal analysis (not shown), including die measurements
at 12 and 15 years of age simultaneously. A significant
effect of socio-economic position on total energy intake
was found (p=4.04, p<0.01) in boys. The fact diat diis
difference was only apparent at die age of 12 years was
reflected in a significant (socio-economic position *
period of measurement) interaction effect (P? -1.88,
p<0.01). A main effect of socio-economic position was
also found for die intake of dietary fat (P=1.05, p?0.02),
dietary cholesterol 0=52.5, p=0.00), fruit (P= -0.42,
p=0.03) and the P:S ratio (P= -0.04, p=0.01). Although
t-tests showed only effects of socio-economic position on
fat intake, fruit intake and die P:S ratio at one period of
measurement, no significant interaction effects were
found in die longitudinal analyses. For girls, main effects
were found for physical activity (P= -3.04, p<0.01), total
energy intake (P=1.02, p=0.01) and the intake of dietary
fat 0=1.10, p=0.01), cholesterol (P=93.0, p=0.01) and
fruit (P= -0.87, p=0.01). The finding that diere is only
an effect of socio-economic position on dietary
cholesterol intake at the age of 12 years was reflected in
a significant interaction (P= -42.7, p=0.02). No
significant interaction effect was found for the intake of
dietary fat, fruit and fibre intake. Due to the limited
number of smokers at the age of 12 years, we did not
investigate die effects of smoking on socio-economic
Tables 3 and 4 present the mean values of the biological risk factors by socio-economic position in boys and girls, respectively. In boys, there appeared to be no statistically
a P S ratio - poly-unsaturated to saturated fat ratio
b: Number of smoking subjects too small
c p-value of t-test, assuming unequal variances, based on Levine's test for equality of variances
d. p-value of the chi-square test for the association between smoking behaviour and socio-economic position
a: PS raoo - poly-unsaturated to saturated fat ratio
b: Number of smoking subjects too small
a p-value of t-test, assuming unequal variances, based on Levine's test for equality of variances
d: p-value of the chi-square test for the association between smoking behaviour and socio-economic position
SBP: systolic blood pressure; DBP: diastolic blood pressure;
TC: total serum cholesterol; HDL: high density lipoproteins;
BMI: body mass index; SSF. sum of four skinfolds;
20 MST- 20 metre shuttle run test
a: p-value of t-test, assuming unequal variances, based on Levine's test for equality of vanances
SBP: systoik blood pressure; DBP: diastolic blood pressure;
T C total senun cholesterol; HDL high density lipoproteins;
BMI: body mass index; SSF sum of four skinfolds; 20 MST - 20 metre shutde run test
EUROPEAN JOURNAL OF PUBLIC HEALTH VOL. 11 2001 NO. 1
significant differences in risk factors levels by
socio-economic position at the age of 12 years. For SBP, DBP, BMI,
SSF and the 20-MST, no significant differences were
found at the age of 15 years either, with still very small
absolute differences in mean values, except for TC. The
non-significant absolute difference in TC between die
non-manual and manual groups at the age of 12 years was
reduced at die age of 15 years. At that age, and contrary
to expectation, HDL was significantly higher and die
TC/HDL ratio was significantly lower in the manual
group compared to the non-manual group. At die age of
15 years, boys and girls in die manual group were
significantly shorter. No odier significant differences were
found in mean values of the risk factors by socio-economic
position at die age of 12 and 15 years in girls and die
absolute differences remained very small. In die
longitudinal analysis, a main effect of socio-economic
position was found for TC (P=> -0.32, p=0.05) in boys after
inclusion of a significant (socio-economic position * period
of measurement) interaction term. The increasing
difference in HDL by socio-economic position over time was
reflected in a significant interaction ([3=0.07, p=0.02). In
die absence of a main effect of socio-economic position,
a significant interaction effect was found for die TC/HDL
ratio in boys (p=0.07, p=0.02). A significant effect of
socio-economic position on body height was found in
bodi boys (P= -2,29, p=0.02) and girls (P=20.0, p=0.02).
In girls, no odier significant main or interaction effects
were found in die longitudinal analysis.
Table 5 shows die number and percentages of rapidly and
slowly maturing boys and girls by socio-economic
position. There was no association between socio-economic
position and maturation among 12-year-old boys, while a
higher proportion of non-manual girls was classified as
slow maturers at that age. At die age of 15 years, a higher
proportion of non-manual boys was classified as rapid
maturers dian manual boys. None of diese differences,
however, appeared to be statistically significant. Due to
these weak findings, adjustment for biological maturation
was not expected to affect die association between
socioeconomic position and biological risk factors. This was
confirmed in die analyses of variance. Such an
adjustment did not affect die reported significant differences by
socio-economic position group in boys and girls. There
appeared to be no association between smoking behaviour
and biological maturation in boys at die age of 15 years
and hence, biological maturation could not affect die
reported associations between smoking behaviour and
socio-economic position. No significant interactions
between biological maturation and socio-economic
position were found for die behavioural variables.
Inclusion of biological maturation in die longitudinal analyses did not materially alter die results.
The present study investigated die effects of socio-eco
nomic position on behavioural and biological risk factors
at the age of 12 and 15 years in boys and girls from
Northern Ireland, both separately and simultaneously.
Furthermore, we were able to investigate if adjustment for
biological maturation affected die association between
socio-economic position and behavioural and biological
risk factors. In general, we found diat behavioural risk
factors were more unfavourable in subjects in die manual
group compared to subjects in die non-manual group,
particularly at die age of 15 years. This was, except for an
increased body height in die higher compared to the
lower socioeconomic group at die age of 15 years, not yet
reflected in differences in biological risk factors by
socioeconomic position in youth. HDL was found to be higher
and die TC/HDL ratio lower in boys in die manual group.
There appeared to be only a weak association between biological maturation and socio-economic position and consequently, adjustment for biological maturation minimally affected diese findings.
Some mediodological problems may have influenced our
results. First, non-response may reduce die external
validity of die results. The sampling procedure aimed at
selecting a sample of approximately 250 children in four
age-sex groups (12 and 15-year-old boys and girls), taking
into account geographical spread and different categories
of schools in Northern Ireland. These numbers amounted
to a 2% random sample of each group in the province.
The overall response rate was 78%; reasons for non
a: Rapidly mature (12 yeais of age). Tanner stage 2?5; slowly mature: Tanner stage 1,
rapidly mature (15 years of age): Tanner stage 5; slowly mature; Tanner stage 1-4
participation were obtained from 196 non-responders:
objection to blood sampling, reluctance to do any part of
the study, recent illness and parental opposition were the
main reasons for refusal. There was therefore no
indication of selective non-participation, based on health.
Further, a prospective study always faces the problem of
selective dropout. The external validity could therefore
also be threatened by selective dropout of subjects
according to risk factor levels or socio-economic position.
Reasons for dropout, as mentioned earlier, did not suggest
the occurrence of any selection in compliance. Further,
due to the limited number of subjects in the study we
could only create two socio-economic groups. Hence,
differences in risk factor levels could not be compared by
the original six social class categories of the OPCS.
Misclassification of subjects according to occupation or social
class of parents may have further resulted in a bias towards
Our finding of a higher percentage of smoking adolescents
in the manual group is in line with other studies. In a
representative sample of US adolescents, smoking was
more prevalent in the lower socio-economic groups,
either based on the educational level of the responsible
adult or the family income.29 De Vries, who also reported
significant differences in the percentage of smoking
adolescents between lOand 15 years of age by socio-economic
position, found that the perception of the association
between smoking and adverse health outcomes was more
clear in subjects in the high socio-economic position
group. Subjects in the low socio-economic position
group felt more social pressure to smoke. It is suggested
that peer pressure in adolescents is a more important
predictor of smoking than low socio-economic
position.31-32 Contrary to these findings, a recently
published study found no (linear) association between
smoking behaviour and socio-economic position.10 In the
latter study, sports participation was the only behavioural
factor linearly related to socio-economic position, which
was also reported in a study in the US. Sallis et al.
investigated this association in more detail and found
that total time per week of vigorous exercise out of school
did not differ between subjects in a high and low
socio-economic position group (defined by school
district), as was found for girls in our study.33 However,
subjects in the high socio-economic position group were
more frequently involved in a sport team and they had
more activity lessons (perhaps because they could access
such lessons more easily) and physical education classes
Our finding of a higher fat intake in adolescents in the
manual group seems to be relatively consistent with
findings from other studies. Lowry et al. found a higher
consumption of fat in food among girls with a lower
compared to a higher socio-economic position, but not in
boys.29 Further, significantly more 15-year-old subjects
from the non-manual compared to the manual class were
characterised as healthy eaters (eating relatively less fat
than carbohydrates) in Scotland.34 Other studies reported
higher fruit and vegetable consumption in adolescents
with a higher compared to a lower socio-economic
Despite these differences in diet composition by
socioeconomic position in adolescence, socio-economic
position differences in mean BMI or SSF were not found
in our study. In a recently published study, differences
were found in the prevalence of obesity by
socio-economic position in Belgian girls, mean aged 12.5 years, but
not in boys.36 BergstrSm et al. reported a higher mean
BMI in adolescents in families widi a low socio-economic
position in 14 to 17-year-old adolescent girls.32
An interesting finding of our study is that boys and girls
in the manual class are shorter compared to boys and girls
in the non-manual class at the age 15 years. Such
differences were not yet found at the age of 12 years. Although
boys and girls at the age of 15 years have probably not
reached their adult height, and hence manual class
children could theoretically catch up again in body
height, our finding is compatible with earlier research
showing similar differences in adulthood. In a study by
Kuh et al. (1989) father's occupation was still negatively
associated with adult body height of their children, after
adjustment for parental body height.37 Body height is
inversely associated with coronary heart disease
mortality.38 If the development of socioeconomic
inequalities in health is considered in a life course
perspective, the period of adolescence may contribute
through constrained growth in die lower compared to the
higher socioeconomic groups.
Except for HDL and the TC/HDL ratio in boys, we did
not find other statistically significant differences in mean
biological risk factor values by socio-economic position.
However, such differences are often reported in adults.
The question therefore remains at what stage in life
these differences become clear. It suggests diat the end
of puberty and early adulthood could be of major
importance for the development of socio-economic position
differences. Socio-economic position differences at the
end of puberty however, have been little investigated.
Future research needs to investigate if differences in bio
logicalriskfactors by socio-economic position develop at
that period of life. Further, it should also address the
mechanisms responsible for the development of
socioeconomic differences, and therefore use a prospective
In a previous report, we expressed serious concern about
die prevalence of hypertension, hypercholesterolaemia
and obesity in youth from Northern Ireland.15 The
implication of the present findings is that prevention of
CHD dirough biological risk factor modification in
adolescence should not be specifically targeted at a particular
socio-economic grouping, though it cannot be excluded
that such an approach would be beneficial at a later stage
of adolescence. Because manual class children show
unfavourable values for behaviouralriskfactors compared to
those in the non-manual class, programmes aimed at
adopting a healdiy lifestyle should already start in primary
school, continue into secondary school and pay particular
attention to children in the lower socio-economic groups.
In conclusion, in 12 and 15-year-old boys and girls, no
clear differences seem to be established in biological risk
factors of CHD, except for a greater body height in
non-manual compared to manual boys and girls at the age
of 15 years. Youth in manual groups however, have
already, or develop, a less healdiy lifestyle between 12 and
15 years of age compared to subjects in the non-manual
group. It is suggested that socio-economic position
differences in behavioural risk factors develop during
adolescence, and will result in differences in biological
risk factors at the end ofpuberty or during early adulthood.
We are currently following up this cohort to test this hypothesis.
This study is financially supported by the British Heart Foundation,
Wellcome Trust and Northern Ireland Chest, Heart and Stroke
1 Marmot MG , Rose G , Shipley M , et al. Employment grade and coronary heart disease in British civil servants . J Epidemiol Comm Hearth 1978 ; 32 : 244 - 9 .
2 Marmot MG , Davey Smith G , Stansfeld S , et al. Health inequalities among British civil servants: the Whitehall II study . Lancet 1991 ; 337 : 1387 - 93 .
3 Davey Smith G . Down at Heart: the meaning and implications of social inequalities in cardiovascular disease . J Royal Coll Phys 1997 , 31 : 414 - 24 .
4 Strong JP , McGill HC . The pediatric aspects of atherosclerosis . J Atheroscler Res 1969 ; 9 : 251 - 6 .
5 Newman WP , Freedman DS , Voors AW , et al. Relation of serum lipo-protein levels and systolic blood pressure to early atherosclerosis . N Engl J Med 1986 ; 314 : 138 - 44 .
6 Lenthe FJ van, Kemper HCG , Twisk JWR . Tracking of blood pressure in children: a literature review . Am J Human Biol 1994 , 6 : 389 - 99 .
7 PorVka KVK , Vikari JSA , Akerblom HK . Tracking of serum-HDL cholesterol and other lipids in children and adolescents: the Cardiovascular Risk in Young Finns Study . Prev Med 1991 , 20 : 713 - 24 .
8 Gunnell DJ , Frankel SJ , Nanchahal K , et al. Childhood obesity and adurt cardiovascular mortality: a 57-y follow-up study based on the Boyd Orr cohort . Am J Clin Nutr 1998 , 67 : 1111 - 8 .
9 Sobal J , Stunkard AJ . Socio-economic status and obesity: a review of the literature . Psychol Bull 1989 ; 105 : 26O - 75 .
10 Tuinstra J , Groothoff JW , Heuvel WJA , et al. Socio-economic differences in hearth risk behaviour in adolescence: do they exist? Soc Sci Med 1998 , 47 : 67 - 74 .
11 Post GB , Kemper HCG . Nutrient intake and biological maturation during adolescence: The Amsterdam Growth and Health Longitudinal Study . Eur J Clin Nutr 1993 ; 47 : 400 - 8 .
12 Lenthe FJ van, Kemper HCG , Mechelen W van. Rapid maturation in adolescence results in higher levels of indicators of obesity in adulthood: The Amsterdam Growth and Health Study . Am J Clin Nutr 1996 ; 64 : 18 - 24 .
13 Rimpela AH , Rimpela MK . Towards an equal distribution of health? Socioeconomic and regional differences of the secular trend of the age of menarche In Finland from 1979 to 1989 . Acta Pediatr 1993 , 82 : 87 - 90 .
14 Uemara K , Pisa Z. Trends in cardiovascular disease mortality in industrial countries since 1950 . World Health Stat Quart 1988 ; 41 : 1S5 - 78 .
15 Boreham C , Savage JM , Primrose D , et al. Coronary risk factors in schoolchildren . Arch Dis Child 1993 ; 68 : 182 - 6 .
16 Office of Population Censuses and Surveys . Standard Occupational Classification Vol 3 , 1991 .
17 Boreham CA, Twisk JWR , Savage JM , Cran GW , Strain JJ . Physical activity, sports participation, and risk factors in adolescents . Med Sci Exerc Sports 1997 ; 29 : 788 - 93 .
18 Riddoch C , Savage JM , Murphy N , Cran BW , Boreham CA. Long term hearth implications of fitness and physical activity patterns . Arch Dis Child 1991 ; 66 : 1426 - 33 .
19 Strain JJ , Robson PJ , Livingstone MBDE , et al. Estimates of food and macronutrient intake in a random sample of Northern Irelands adolescents . Br J Nutr 1994 , 72 : 343 - 52 .
20 Rimm EB , Ascherio A , Giovannucci E , et al. Vegetable, fruit, and cereal fiber intake and risk of coronary heart disease among men . JAMA 1996 , 275 : 447 - 51 .
21 Ness AR , Powles JW . Fruit and vegetables, and cardiovascular disease: a review . Int J Epidemiol 1997 ; 26 : 1 - 13 .
22 Lopez-Virella MF , Stone P , Ellis S , et al. Cholesterol determination in high density lipoproteins separated by three different methods . Clin Chem 1977 , 23 : 882 - 4 .
23 DurninJVGA, Rahaman MM. The assessment of the amount of fat in the human body from measurements of skinfold thickness . Br J Nutr 1967 ; 21 : 681 - 9 .
24 Boreham CAG , Paliczka VJ , Nichols AK . A comparison of the PWC 170 and 20-MST tests of aerobic fitness in adolescent schoolchildren . J Sports Med Phys Fitness 1990 ; 30 : 19 - 23 .
25 Tanner JM . Growth at adolescence. Oxford: Blackwell, 1962 .
26 Zeger SL , Liang KY , Albert PS . Models for longitudinal data: a generalized estimating equations approach . Biometrics 1988 ; 44 : 1049 - 60 .
27 Twisk JWR , Kemper HCG , Mellenbergh GJ , et al. Relation between the longitudinal development of lipoprotein levels and lifestyle parameters during adolescence and young adulthood . Ann Epidemiol 1996 , 6 : 246 - 56 .
28 Rothman KJ . Modern epidemiology . Boston: Little, Brown & Co, 1986 .
29 Lowry, R , Kann L , Collins JL , etal. The effects of socio-economic status on chronic disease risk behaviors among US adolescents . JAMA 1996 ; 276 : 792 - 7 .
30 Vries H de. Socio-economic differences in smoking: Dutch adolescents' beliefs and behaviour . Soc Sci Med 1995 ; 41 : 419 - 24 .
31 Pedersen W , Lavik NJ . Role modelling and cigarette smoking: vulnerable working class girls? Scand J Soc Med 1991 ; 19 : 110 - 5 .
32 Bergstrom E , Hernell O , Persson LA . Cardiovascular risk indicators cluster in girls from families of low socio-economic status . Acta Pediatr 1996 ; 85 : 1083 - 90 .
33 Sallis JF , Zakarian JM , Hovell MF , et al. Ethnic, socio-economic and sex differences in physical activity among adolescents . J Clin Epidemiol 1996 ; 49 : 125 - 34 .
34 Anderson AS , Macintyre S , West P . Dietary patterns among adolescents in the West of Scotland . Br J Nutrition 1994 ; 71 : 111 - 22 .
35 Neumark-Sztainer D , Story M , Resnick MD , et al. Correlates of inadequate fruit and vegetable consumption among adolescents . Prev Med 1996 ; 25 : 497 - 505 .
36 Spiegelaere M de , Dramaix M , Hennart P . Social class and obesity in 12 year old children in Brussels: influence of gender and ethnic origin . Eur J Pediatr 1998 ; 157 : 432 - 5 .
37 Kuh D , Wadsworth M. Parental height childhood environment and subsequent adurt height in a national birth cohort Int J Epidemiol 1989 ; 18 : 663 - 8 .
38 Davey-Smrth G , Hart C Upton M , Hole D , Gillis C. Watt G , Hawthorne V . Height and risk of death among men and women: aetiological implications of associations with cardiorespiratory disease and cancer mortality . J Epidemiol Comm Hearth 2000 : 54 : 97 - 103 .
Received 18 February ( 999 , accepted 15 August 2000