In this large longitudinal cohort we found that shorter stature in children from age 7 to 13 years was associated with an increased risk of CHD in adulthood, and these associations were not modified by birth weight. Height z-scores predicted the later risk of CHD with approximately the same effect size as did BMI z-scores in these same data 
, but the direction of the effect and changes in strength of the associations across the age groups were opposite. The associations between height and CHD showed a clear age pattern: they were strongest at the age of 7 and decreased until the age of 13 in both boys and girls. This pattern was partly because of the effect of BMI; when the results were adjusted for BMI the association between height and further CHD incidence become stronger because of correlation between height and BMI (r from 0.18 to 0.30), and this change was more visible at older ages. These findings strongly suggest that further investigations of the underlying biological processes that the growth-CHD association is driven by are warranted. The associations were slightly stronger for CHD cases before 60 years of age and the fatal events than for all fatal and non-fatal events together. The associations are measured on the multiplicative hazard ratio scale, which is sensitive to the baseline hazards - the lower it is the stronger the association for a given excess incidence rate. Whether the growth patterns are truly differently associated with the various manifestations of CHD, ranging from first diagnosis of angina pectoris through immediate death, is an open question beyond the scope of our present study.
The association between short adult stature and increased CHD risk has been known since the 1980s 
and has been replicated in numerous studies since then 
. However studies on the association between height in childhood and later CHD risk are sparse. In the Hertfordshire study it was reported for the first time that low birth weight in men and women as well as low weight at one year of age in males was associated with a higher risk of cardiovascular diseases 
, and thereafter it has been found in several studies that low birth weight is associated with increased risk of cardiovascular disease 
as well as its risk factors 
. In another UK cohort of children 2 to 14 years of age at baseline, leg-length and height were inversely associated with CHD mortality in adulthood 
Our results show that it is necessary to disentangle the effects of a generally slow growth trajectory leading to short stature in adults and a fast growth trajectory during late childhood; both are associated with increased risk of later CHD, and are therefore masking the effects of each other in the final stature. Thus the decreasing strength of the association between height and the risk of CHD across childhood probably reflects a mixture of opposing effects. The inverse association between childhood height and risk of CHD in adulthood may be diluted by the opposite effect of rapid growth that increases the CHD risk. The tallest boys and girls at 13 years of age thus appear to be a heterogeneous group with regard to CHD risk. Those who were taller-than-average at 7 years of age and remained tall had the lowest CHD risk. In contrast, those who were shorter-than-average at 7 years of age and then grew and became taller-than-average at 13 years of age had an increased CHD risk, which even exceeded the risk of the children who remained shorter-than-average between 7 and 13 years of age. Rapid growth was associated with higher BMI, but the association between rapid growth and CHD risk remained even when adjusting the results for baseline BMI.
Interestingly, rapid growth between 11 and 13 years of age in boys and between 9 and 11 years of age in girls showed the strongest association with increased CHD risk. These ages closely coincide with the timing of puberty in the Danish population both measured as sexual maturation 
and as the onset of pubertal growth spurt 
. In this same cohort, the average age at the onset of pubertal growth spurt was 12.2 years in boys born from 1930 to 1940 and decreased to 11.8 years in those born from 1965 to 1970 whereas in girls the ages were 10.4 and 10.2 years, respectively 
. A Finnish study has also shown that rapid growth during these periods correlates closely with early sexual maturation 
. Thus it is likely that children in our cohort who experienced rapid growth at these ages also had an early onset of puberty. This is consistent with the finding of high risk of CHD in the boys and girls who were in the shortest height quartile at 7 years of age but in the tallest quartile at 13 years of age. Previous studies have shown that early onset of puberty is associated with the emergence of several risk factors for CHD including hypertension 
, obesity 
, adverse blood lipid profile 
and high levels of insulin 
in adulthood. Our study thus suggests that the growth rate during the age of pubertal development is associated with increased risk of CHD in adulthood, and this association is very similar in men and women despite their different timing of puberty.
It is noteworthy that rapid growth and increased CHD risk was seen already between 7 and 9 years of age in both boys and girls thus well preceding the onset of puberty. A Finnish study found that participants diagnosed with hypertension at 63 years of age had a lower birth weight and had showed catch-up growth in height and weight until 11 years of age when compared to normotensive men and women. The individuals, however, who were newly diagnosed as hypertensive in the clinical examination showed a different growth pattern since they had low birth weight, but did not show catch-up growth 
. In this same Finnish cohort it was also found that in females short length at birth was associated with higher risk of CHD in adulthood, and this association was strongest in those women who had experiences catch-up growth until 7 years of age 
We found no evidence supporting that only children with a low birth weight combined with later rapid growth in height have a higher CHD risk; the association between height in childhood and later CHD risk was largely independent of birth weight. This result is consistent with a previous Finnish study, which found no evidence of catch-up growth from 2 to 10 years of age in the persons who had had coronary event in adulthood 
. The results rather suggest that the height growth velocity in childhood carries a risk for CHD independent of birth weight. These associations likely reflect the combined effects of nutritional, environmental, genetic and hormonal factors, such as the secretion of insulin-like growth factor, on growth 
. For example, it has been found that leg length in childhood is more strongly associated with CHD risk than trunk length, which may be because leg length is a more sensitive indicator of environmental factors than trunk length 
. There is also evidence based mainly on Finnish studies that growth in utero and infancy may be associated with the proportion of lean body mass and lipid metabolism in adulthood 
The association between short stature-to-age in childhood and later CHD risk may at least partly reflect the role of socio-economic factors, since it is well known that low socioeconomic position in childhood is associated with shorter adult stature 
and that clear socioeconomic differences in CHD risk are present in Northern-Europe and the USA 
. We did not have access to information on parental socio-economic position, so we could not directly test this hypothesis. However, a previous Nordic study found an association between adult stature and CHD risk within discordant twin pairs suggesting that socio-economic background cannot explain the association 
. Further, we found that the association between stature and CHD was similar across birth cohorts despite a clear secular trend in height and improvements in the standard of living in Danish society across the birth cohorts. Taken together, these results suggest that the association between height and CHD risk is not strongly modified by socioeconomic position, and we neither find reasons to suspect that rapid growth in childhood is associated with low socio-economic position. Moreover, the study design and completeness of follow-up make it very unlikely that any socioeconomic selection bias exists.
The biological pathways underlying the association of CHD incidence with growth are still largely unknown. In addition to the above mentioned hypotheses on the role of nutritional, metabolic, hormonal and socioeconomic factors as mediators of this association, there are also other possible explanations. It has been suggested that short stature might be a CHD risk factor itself since it correlates with narrow coronary arteries, which may further predispose to CHD 
. However our results did not give support for this hypothesis since we found that the association between height at 13 years of age and further CHD incidence depended on the growth velocity from 7 to 13 years of age. Further, there may be other mechanisms that are currently unknown. To answer these types of questions data sets containing information on growth in childhood and detailed metabolic as well as other biological measurements in adulthood are required. Future research can be undertaken to investigate these potentially mediating pathways in depth by using information from detailed examinations in much smaller subgroups of individuals selected on the basis of various growth patterns in childhood.
Our study has several strengths, but also limitations. We used a very large population-based cohort with annual height measures from 7 through 13 years of age. The health examinations were performed at all public and private schools and thus cover virtually every schoolchild in the Copenhagen municipality born from 1930 to 1976. Our follow-up data cover also all study participants because of universal health care system in Denmark. Thus, it is very unlikely that any selection bias exists at baseline or during follow-up. In the present cohort, we lack information on conventional risk factors for CHD, and it will be a task for future studies to elaborate which metabolic or other biological factors may underlie the observed associations between childhood growth and further CHD risk. Further since we do not have any measurements of sexual maturation, such as age at the onset of menarche, and we cannot determinate peak growth velocity for late mature children our results on the associations between timing of puberty and further CHD incidence can be regarded only tentative. It is also noteworthy that our study population represents an ethnically homogeneous Caucasian population with tall average stature 
. There is little research on the associations between childhood height and CHD risk in other ethnic groups. However, for adult stature the association was very similar in a study of Korean men 
to that found in studies conducted in adult Caucasian populations 
suggesting that the association between growth and CHD risk would be present also in other ethnic groups as well.
To summarize, we found that shorter stature in childhood was associated with increased CHD risk in adulthood, that it grew weaker with age, and that the association was independent of birth weight. Rapid pre-pubertal growth, but especially rapid growth close to puberty, was associated with increased CHD risk, demonstrating the importance of physical development of the child for the risk. Physical development in childhood thus captures important information on the childhood environment, such as socio-economic factors and nutrition, which are important for further cardio-vascular health. Thus our results support the importance of optimal living conditions in childhood for a future healthy life. Unraveling the biological pathways of this association and the exploration of the utility of the association in prevention of CHD are tasks for future research.