Short stature and coronary heart disease: a 35-year follow-up of the Finnish cohorts of The Seven Countries Study
Prof. Jaakko Tuomilehto, Diabetes and Genetic Epidemiology Unit, Department of Epidemiology and Health Promotion, National Public Health Institute, Mannerheimintie 166, FIN-00300 Helsinki, Finland (fax: + 358 947448 338; e-mail: email@example.com).
Abstract. Forsén T, Eriksson J, Qiao Q, Tervahauta M, Nissinen A, Tuomilehto J (National Public Health Institute, Helsinki, and University of Kuopio, Kuopio, Finland). Short stature and coronary heart disease: a 35-year follow-up of the Finnish cohorts of The Seven Countries Study. J Intern Med 2000; 248: 326–332.
Objectives. To examine whether short stature is associated with an increased risk of coronary heart disease.
Design. Follow-up study.
Setting. Two geographically defined areas in eastern and western Finland.
Subjects. A total of 1441 men who were free of coronary heart disease at the start of the follow-up.
Main outcome measures. Hazard ratios for fatal and non-fatal coronary heart disease
Results. Height was inversely related to fatal coronary heart disease and incident non-fatal coronary heart disease during the follow-up. These relationships persisted after adjusting for other major cardiovascular risk factors. Comparing the high-risk area in eastern Finland with the low-risk area in south-western Finland, no difference in fatal coronary heart disease and cumulative incidence of non-fatal coronary heart disease was seen in tall men. The increase in risk of coronary heart disease death was 19% for a 10 cm decrease in height (OR = 0.81, 95% CI = 0.68–0.95).
Conclusions. Our results show that short stature is an independent risk factor for coronary heart disease. Differences in stature partly explain the Finnish east–west difference in the incidence of coronary heart disease.
An association between adult height and mortality from coronary heart disease (CHD) has been detected in several studies [1–8]. The relationship between short stature and CHD in adult life is, however, controversial. In the Framingham study, an association between stature and risk of myocardial infarction was observed in women only . Furthermore, the NHANES I study produced negative findings when adjusting for age and education . Therefore, these reports indicate that the socioeconomic status and risk factors associated with social class are confounders of the relationship between short stature and CHD. The Physicians Health Study and the Helsinki Businessmen Study did, however, show an association between short stature and CHD [2, 6]. A recent study related short stature to clinical measures in 1046 men, showing that the shorter men had a higher prevalence and greater severity of angiographically verified CHD .
Previously, differences in CHD incidence between eastern and western Finland have been described. Annual mortality from CHD in eastern Finland (North Karelia) was 496 in 100 000, which is 1.7 times higher than the that in south-western Finland (Loimaa), where the incidence was only 300 in 100 000 [12, 13]. The Finnish cohorts of the Seven Countries Study (SCS) comprise 1711 men who have been followed up since 1959, primarily to identify risk factors for CHD [13, 14]. In these cohorts, we attempted to evaluate the effect of stature on CHD mortality and morbidity and also whether the difference in stature in men between eastern and south-western Finland could explain the difference in CHD risk between these two areas.
Subjects and methods
The original Finnish cohorts of the SCS consist of all 1711 men born during 1900–19, and living in two different areas in Finland in 1959: Ilomantsi in eastern Finland (n = 816), and Pöytyä and Mellilä in south-western Finland (n = 856). Of these, 1673 men took part in the first examination in 1959. A total of 1441 men were free of CHD at baseline in 1959. The cohorts were re-examined in repeat field surveys in 1964, 1969, 1974, 1984 and 1989. The participation rate was high, and between 90 and 98% of all surviving men of the original cohort were examined in the field surveys [15, 16].
The measurement techniques of the study have been described in detail elsewhere . Briefly, blood pressure at baseline was measured twice to the nearest 2 mmHg with a mercury sphygmomanometer, with the men in the supine position. The mean of the two readings of systolic and diastolic blood pressure was used in the analyses. According to the smoking status at baseline, the men were grouped into three categories: never-smokers, ex-smokers and current smokers of cigarettes or pipes. Men who had quit smoking less than 1 year before the examination were regarded as current smokers. The socioeconomic status was determined according to occupation and the men were grouped into four categories: (i) white collar; (ii) blue collar; (iii) farmer; and (iv) other farming and forestry workers.
Assessment of CHD
Prior history of CHD in 1959 was assessed using a questionnaire. CHD status in later years was determined in the repeat field surveys. The assessment of non-fatal CHD cases during the follow-up was based on medical history, physical examination and ECG using a slightly modified SCS classification , which has been described in detail elsewhere . According to this classification, individuals having one of the following were classified as having CHD: (i) typical ECG changes for myocardial infarction, (ii) typical myocardial infarction symptoms, (iii) typical angina pectoris, or (iv) typical ECG changes for myocardial ischaemia. Information about deaths was obtained from death certificates and medical records. These were coded according to the SCS protocol [13, 14].
The statistical analyses were performed with SPSS statistical software for Windows version 7.5.1 The chi-square test was used to compare differences between groups with categorical data. The differences between the group means of the continuous variables were tested for significance using Student’s t-test. Relative risk ratios were obtained using the Cox proportional hazards model.
The total number of person-years at risk was calculated amongst men who were free of CHD at baseline in 1959. Men who had developed CHD during the follow-up contributed only half of the total number of person-years. The cumulative incidence of CHD was calculated according to the tertiles of height and presented as rate per 1000 person-years of follow-up.
The Ethical Committee at the National Public Health Institute and the University of Kuopio approved the study.
The baseline characteristics of the study population are shown in Table 1. The levels of cardiovascular risk factors were higher in the men from eastern Finland. Men in the west were, on average, 3.2 cm taller than men in the east (P < 0.0001) and they also had significantly higher BMI (P < 0.0001, Table 1). There were 1396 (n = 1711) deaths during the follow-up from 1959 to 1994, of which 532 were due to CHD and 135 from other cardiovascular disease (CVD).
Table 1. Baseline measurements of the study group at the beginning of the follow-up in 1959 (n = 1673)
|Age (years)||50.3||5.5||49.2||5.6||P < 0.001 |
|Standing height (cm)||171.2||6.0||167.9||6.1||P < 0.001 |
|Body weight (kg)||70.7||11.1||65.9||9.9||P < 0.001 |
|Body mass index (kg m–2) ||24.1||3.3||23.3||3.0||P < 0.001 |
|Serum cholesterol (mmol L–1) ||6.6||1.3||6.9||1.4||P < 0.001 |
|Systolic blood pressure (mmHg)||140||20||148||20||P < 0.001 |
|Diastolic blood pressure (mmHg)||82||11||89||11||P < 0.001 |
|Number of men (%)||Number of men (%)|
| Never||208 (24.3)|| ||94 (11.5)|
| Ex-smoker||141 (16.5)|| ||124 (15.2)|| ||P < 0.0001 |
| Current smoker||507 (59.2)|| ||598 (73.3)|
| White collar||47 (5.2)|| ||104 (12.6)|
| Blue collar||209 (23.6)|| ||130 (15.8)|| ||P < 0.0001 |
| Farmer||522 (58.9)|| ||159 (19.3)|
| Other farming and forestry worker||100 (11.3)|| ||421 (51.2)|
| Retired||8 (0.9)|| ||9 (1.1)|
Height was inversely related to CHD mortality ( Tables 2 and 4) and the cumulative incidence of non-fatal CHD ( Table 3). The increase in risk of CHD death was 19% for a 10 cm decrease in height (OR = 0.81, 95% CI = 0.68–0.95). Adjustment for baseline age, weight, smoking, systolic blood pressure, cholesterol, occupation or area of residence did not alter the detected associations between CHD and height. Height was also significantly and inversely related to the cumulative incidence of non-fatal CHD ( Table 4).
Table 2. Relative risk (RR) of death from coronary heart disease (CHD), any cardiovascular disease (CVD) and all-cause mortality in relation to tertiles of height in men during the 35-year follow-up (n = 1441)
|Death from CHD|
| Unadjusted||1.59 (1.27–1.98)||1.45 (1.17–1.80)||1||1649||514||0.0001|
| Age-adjusted||1.48 (1.18–1.85)||1.39 (1.17–1.71)||1||1649||514||0.001|
| Multivariate a||1.36 (1.05–1.77)||1.33 (1.06–1.67)||1||1617||507||0.03|
|Death from any CVD|
| Unadjusted||1.60 (1.31–1.96)||1.53 (1.26–1.85)||1||1649||647||<0.0001|
| Age-adjusted||1.47 (1.20–1.80)||1.45 (1.19–1.76)||1||1649||647||0.0001|
| Multivariate||1.33 (1.05–1.69)||1.38 (1.13–1.69)||1||1617||638||0.006|
| Unadjusted||1.32 (1.16–1.52)||1.22 (1.07–1.39)||1||1651||1347||0.0001|
| Age-adjusted||1.21 (1.06–1.39)||1.15 (1.01–1.31)||1||1651||1347||0.01|
| Multivariate||1.00 (0.85–1.17)||1.05 (0.92–1.21)||1||1619||1319||0.6|
Table 4. Cumulative risk (%) of non-fatal coronary heart disease (CHD), fatal CHD, fatal cardiovascular disease (CVD) and all-cause mortality (rate per 1000 person-years) during the 35-year follow-up from 1959 to 1994 according to tertile of height. Numbers of cases are in parentheses
|< 167 cm||58 (116)||51 (65)||15 (78)|| 9 (28)||17 (89)||11 (32)||71 (120)||58 (69)|
|167–172 cm||59 (104)||38 (91)||15 (74)||10 (50)||16 (79)||12 (60)||61 (96)||46 (97)|
|> 172 cm||41 (61)||41 (119)|| 7 (20)|| 8 (47)|| 7 (22)|| 8 (51)||46 (63)||47 (129)|
| ||(1959–89)|| ||(1959–94)|| ||(1959–94)|| ||(1959–94)|| |
|< 167 cm||74 (111)||56 (61)||17 (118)||14 (56)||22 (146)||18 (69)||42 (286)||37 (145)|
|167–172 cm||59 (90)||46 (92)||18 (114)||13 (87)||22 (140)||17 (118)||40 (253)||36 (247)|
|> 172 cm||47 (59)||45 (118)||10 (42)||11 (97)||13 (53)||14 (120)||32 (134)||33 (281)|
Table 3. Relative risk (RR) of non-fatal coronary heart disease (CHD) in relation to height during the 25-year follow-up
|Unadjusted||1.37 (1.15–1.64)||1.16 (0.98–1.39)||1||1340||755||0.002|
|Adjusted for area of residency||1.30 (1.09–1.56)||1.14 (0.95–1.36)||1||1340||755||0.02|
|Age-adjusted||1.31 (1.10–1.57)||1.11 (0.93–1.32)||1||1340||755||0.01|
|Multivariate-adjusted a||1.35 (1.10–1.66)||1.14 (0.95–1.37)||1||1318||744||0.01|
Height was inversely related to all-cause mortality in the univariate analysis when adjusting only for age, but this relationship disappeared when adjustment for other risk factors was made ( Table 2).There was also an inverse relationship between height and CVD mortality, although this diminished to some extent in the multivariate model ( Table 2).
Table 4 shows the differences in non-fatal and fatal CHD and all-cause mortality between the two areas in relation to height, expressed as cases per 1000 person-years. The follow-up started in 1959 and data are presented for the intermediate years (1974, 1980, 1984 and 1989) as well as for the last year (1994) for which data are available. The incidence of both non-fatal and fatal CHD was higher in eastern than in south-western Finland at any time of the follow-up. Although the effect of stature on CHD was similar in the east and the west, there was an interaction between height and area. The east–west difference in cumulative incidence of non-fatal CHD was only seen amongst the short men, since the incidence of CHD in the tallest tertile of men (> 172 cm) was similarly low in both the east and the west ( Table 4). The absolute risk difference for non-fatal CHD between the lowest and highest tertiles of height was larger in the east (32 per 1000 person-years) than in the west (15 per 1000 person-years).
We have shown an inverse relationship between adult height and fatal and non-fatal CHD in middle-aged men. Our study sample was the Finnish part of the SCS, where the frequency of CHD was found to be the highest in the world [13, 14]. The SCS was one of the first studies which showed that much of this high risk was related to the three main CHD risk factors: serum cholesterol, smoking and blood pressure. It was therefore interesting to see that the relationship between height and CHD persisted even after adjusting for occupation and other major risk factors for CHD, resulting in a 19% increased risk of CHD death for a 10 cm decrease in height. Height was also related to CVD and all-cause mortality after adjusting for age. Even when adjusting for other risk factors, the association with CVD remained, although the relationship with all-cause mortality disappeared. These findings are in agreement with results from several previous studies [1–7], although some studies have shown negative findings [9, 10]. Our study included only men, but a recent study showed that the risk of a recurrent adverse CHD event in women is 2.1-fold higher amongst the 25% shortest women .
The random errors of measuring height are likely to have been smaller than for other CHD risk factors. Therefore, the effect of height in relation to other CHD risk factors may be slightly overestimated in a multiple regression analysis. On the other hand, the exactness of height measurements makes a falsely observed association unlikely (type 1 error). Therefore, the relationship found between the risk of CHD and height is valid. It is not height itself which is responsible for this association, but the underlying mechanisms that affect height. These factors and the mechanisms through which they act remain unknown, although plausible theories on this issue exist.
Impaired growth in utero and in early childhood is related to an increased risk of CHD later in life [18, 19]. This may be linked to the socioeconomic situation, at least to a certain degree. The higher level of CHD risk factors and mortality from CHD in individuals in lower social classes compared with higher social classes may be due to events that affect growth in utero and subsequently influence childhood and adult height. The genotype determines the maximum attainable adult, but this has not yet been reached at a population level. Therefore, the attained height is also affected by environmental factors and the increase in mean height during the last decades is due to a change in environment and not to changes in genotype. The main modifiable factors associated with social class that can influence growth in childhood are deficits in nutrition, infections and hormonal disorders. The habitual nutritional intake in individuals from the lower social classes is known to be less healthy than that of people in higher socioeconomic groups [20, 21]. Furthermore, people in lower socioeconomic classes are likely to have more infections than those from higher social classes because of crowded housing. Therefore infections may play a role in the pathogenesis of CHD at a very early stage. The few studies linking childhood socioeconomic factors to CHD have, however, produced both positive [22–24] and inconclusive  findings. Our ecological data indirectly support the association between socioeconomic factors and adult height, since the men in eastern Finland were living in an area which, during the entire 20th century, was poorer and had a lower personal income compared with the more affluent south-western part of the country . These regional socioeconomic differences within Finland have, however, decreased with time, although they have not disappeared .
Growth hormone secretion in childhood strongly affects the growth, but the role of growth hormone once final height has been achieved is not known. Individuals with hypopituitarism have increased cardiovascular mortality but the administration of growth hormone later in life only seems to increase the level of cardiovascular risk factors [28–32]. A primary deficiency of growth hormone would therefore seem to be an unlikely cause of the increased risk of CHD in short individuals.
Although mortality from CHD has been decreasing in Western countries during the past decades, social class differences in mortality [33–35] as well as the relationship between height and social class still persist [36–39]. This might be due to intergenerational effects of intrauterine or early childhood growth retardation. We have previously shown that maternal obesity is strongly linked to an increased risk of CHD in the offspring of short mothers but not of tall mothers . The mothers with short stature are likely to have suffered from growth retardation in utero or in childhood, caused by deficient nutrition and/or infections. Short mothers constrain the growth of their fetuses in utero, which leads to low birthweight [41, 42], and thus the childhood nutritional status of the mother can have an effect on the offspring. Also, fetal growth retardation has been shown to predispose to a short stature in adulthood .
Liao et al.  suggested that making adjustments for socioeconomic factors such as income and education would make the relationship between short stature and CHD disappear. This is plausible, since height correlates positively with socioeconomic status , and thus adjusting for both these variables simultaneously in a multivariate model can abolish their effects. Detecting a relationship between height and CHD in a socioeconomically more homogenous group than that in the study of Liao et al. does, however, reduce the number of possible confounders that exist in more heterogeneous groups. Although we do not have access to the data on income and education in our cohorts, they do consist of men from two small geographically well-defined homogenous rural areas and therefore the likelihood of socioeconomic bias is reduced.
The finding that the Finnish men in the west were significantly taller and had a lower risk of CHD is also in agreement with the hypothesis that adult height is an independent determinant of CHD within the population [1–5]. The increase in height in Finnish men has been reported to be ≈ 1.3 cm per decade since 1899. Despite this secular increase in height, the 2 cm difference in height between men in south-western Finland compared with those in the eastern part of the country has persisted . Moreover, in this study, height had a similar effect on the risk of CHD for both eastern and western Finnish men. Whilst a large proportion, but not all, of the difference in the CHD risk between eastern and western Finland has been explained by the major CHD risk factors – cholesterol, blood pressure and smoking – these results now show that short stature may account for much of the previously remaining unexplained difference. The detected economic differences  between the eastern and the south-western regions are also likely to have led to differences in nutrition. Nutritional factors affecting growth in utero and subsequent stature could therefore explain part of the difference in the incidence of CHD. However, a genetic origin for the growth retardation and short stature is also possible.
One of the most striking results from this study was that there was no east–west difference in fatal and non-fatal CHD in men whose height was 172 cm or more. The reason for the similar risk of CHD in the tallest eastern and western men is unknown, but the difference in the short men could mirror the inferior nutritional intake of earlier generations in eastern compared with western Finland. Possibly, the association between height and CHD may disappear in the future as the population grows in stature. However, the more recent analysis of younger birth cohorts of men from the same regions also found that height independently predicted CHD .
We conclude that short stature is independently associated with CHD mortality and incidence and CVD mortality in men. Short stature is therefore a risk marker for CHD. Differences in stature partly explain the Finnish east–west difference in CHD incidence. Growth retardation, either during early childhood or in utero, is likely to be responsible for the short stature and for the subsequent increased risk of CHD.
Received 23 December 1999; revision received 19 June 2000; accepted 18 July 2000.