Anthropometric characteristics as predictors of coronary heart disease in women

Authors


Ling Yang, Clinical Trial Service Unit & Epidemiological Studies Unit, University of Oxford, Richard Doll Building, Old Road Campus, Oxford OX3 7LF, UK.
(fax: +44 (0)1865 743985; e-mail: ling.yang@ctsu.ox.ac.uk).

Abstract.

Objectives.  Obesity and other anthropometric measures are clearly related to risk of coronary heart disease (CHD), although debate remains as to which measures are most important and how the impact of obesity varies over the life course.

Aim.  We aimed to investigate these issues in a large cohort of Swedish women. The Women’s Lifestyle and Health Cohort Study includes 49 259 women, aged 30–50 years at baseline (1991–1992) when an extensive questionnaire was completed.

Methods.  Women were given standard instructions for self-measurement of anthropometric characteristics. Women were followed through linkages to national registries until December 2003, during which time 256 cases of incident fatal CHD or nonfatal myocardial infarction occurred.

Results.  Waist circumference was associated with increased CHD risk after multivariate adjustment for confounders (HR = 1.9; 95% CI:1.1–3.3; highest versus lowest quartile), whereas height, weight and hip circumference were not. Measures of obesity were strongly related to CHD, and after mutual adjustment, waist-hip ratio (HR = 1.9, 95% CI: 1.2–3.2) was more closely related to CHD risk than BMI (HR = 1.5, 95% CI: 1.0–2.4). Risk of CHD was increased in women who remained heavy, those who were heavy at age 18, and those with low birth weight.

Conclusions.  In conclusion, there is strong evidence for supporting control of obesity, in particular avoidance of abdominal obesity, as a strategy to prevent CHD.

Background

Coronary heart disease (CHD) is the leading cause of death globally, with 7.2 million deaths occurring worldwide every year [1]. Abundant evidence from epidemiological studies, mostly in men, has firmly linked higher body weight with an increased risk of CHD [2, 3]. Unfortunately, an escalating pandemic of overweight and obesity has occurred globally, and there are currently more than 1 billion overweight adults, 300 million of whom are obese [1]. Obesity presents a major public health challenge, especially in many European countries and the United States. About 30 000 deaths per year in the UK and 10 times that in the USA are attributable to obesity [4, 5].

Body mass index (BMI) is the most easily measured and most commonly used surrogate measure of obesity. However, substantial differences in percentage of fat and lean-body mass could exist between individuals with the same BMI [4], and a high BMI due to high lean mass is related to physical activity and independently contributes to reduced risk of CHD [4, 6]. It is also unclear whether people who are overweight are at increased risk of CHD, or whether increased risk is restricted to those who are obese. Recent data suggest that central obesity [measured by waist-hip-ratio (WHR) or waist circumference] are more important than general obesity with respect to CHD risk [7–10], as abdominal adiposity is linked to metabolic abnormalities that can cause CHD [8, 11, 12]. The debate on which marker of central obesity is most closely related to CHD has moved forward since the INTERHEART case–control study [8] showed clearly that WHR outperforms BMI, although more epidemiological studies with a prospective design are required [13, 14].

Other anthropometric variables are also related to risk of CHD. A number of studies have shown an inverse association between birth weight and risk of CHD in adult life [15–18]. Taller people also seem to be at lower risk of CHD than those who are short in stature [19]. However, little evidence exists whether body size indices in childhood or young adulthood predict future CHD risk in women although there is a lot of concern about the future effect of the current epidemic of childhood obesity on adult CHD risk. Obesity in adulthood may be particularly deleterious for people born with a low birth weight, but there are few and inconsistent results from studies exploring this interaction [17].

The aim of this study was to investigate the association between different anthropometric characteristics amongst different life periods, and CHD risk using data from the Swedish Women’s Lifestyle and Health (WLH) study, a large prospective cohort.

Methods

Study population

As described previously [20], the WLH cohort was initiated during 1991–1992, when 49 259 women, aged 30–50 years residing in the Uppsala Health Care Region returned a completed mailed questionnaire. In this study, we excluded 868 women who lacked vital status information, or had emigrated or died before the start of follow-up, 225 women who experienced a myocardial infarction (MI) before the start of follow-up and 114 women who has no anthropometric information. The final study population included 48 052 women.

Exposure classification

At baseline the women completed a detailed self-administered questionnaire. Under standard instructions about how to make measurements, women were asked about their birth weight, perceived body shape at age 7 years, current waist and hip circumferences, adult height and weight at age 18 as well as at cohort enrolment (1991–1992). Participants self-reported on established coronary risk factors, including: cigarette smoking (duration and intensity), exercise (scale 1–5), years of education, alcohol consumption (glasses per week of different types of alcohol), oral contraceptive (OC) use (type and duration) and doctor diagnosis of diabetes or high blood pressure.

Follow-up and CHD endpoints

Follow-up of the cohort was achieved through linkages with nationwide registers, using the unique national registration number, to ensure virtually complete follow-up. Information on death and emigration was collected through linkage to Statistics Sweden. CHD information was collected through linkage to the National Hospital Discharge Register [coded as MI (ICD-9 = 410)] and the National Causes of Death Register (coded as CHD deaths). Person-years were calculated from the start of follow-up, i.e. the date of receipt of the returned questionnaire to the primary diagnosis of fatal CHD or nonfatal MI, date of emigration or death or the end of follow-up (31 December, 2003), whichever came first.

Statistical analysis

Weight at enrolment (kg), height (cm), waist and hip circumferences (cm), BMI (kg/m2) for both age 18 and at enrolment, WHR were categorized into quartiles. An indicator variable for the change in body shape between childhood and adulthood was created and categorized into remained thin; decreased weight, remained average weight, increased weight and remained fat [21]. Changes in BMI units (U) between age 18 and study enrolment were defined as decreased, increased by up to 1.4, 1.5–4.0 or >4.0 U.

Hazard ratios (HR) and corresponding 95% confidence intervals (CI) were calculated as measures of relative risks, using Cox proportional hazards models. Multivariate models were conducted for each anthropometric characteristic. All analyses were adjusted for age at baseline, which was categorized by 5-year intervals. The second models were also adjusted for established coronary risk factors, which may act as confounders, including: educational attainment (≤9, 10–12, 13–15, ≥16 years), alcohol consumption (0, <1.7, 1.7–4.4, ≥4.4 g day−1), cigarette smoking (never smoker, <5, 5 to <10, ≥10 pack years), ever used OC (ever versus never) at enrolment and overall physical activity level (very low, low, normal, high, very high). The third models were adjusted for the previous indicators, as well as self-reported hypertension and diabetes, to assess potential mediators of the association between anthropometric variables and CHD. The final models included all the variables and obesity measures (BMI or WHR), to assess the effects of anthropometric variables independent of other anthropometric markers. Only single-born subjects were included in the study of the association between CHD risk and birth weight (1130 women who had twin sibling were excluded). The interaction between birth weight and adult body size was investigated by examining the ‘tertile crossing’ effect from birth to adulthood.

Ethics

The study was approved by the Data Inspection Board in Sweden and by the regional Ethical Committee. Consent was assumed by the return of the postal questionnaire.

Results

During the average 12.0 years follow-up, a total of 256 events (229 nonfatal MI and 27 fatal CHD) occurred amongst 48 052 women (575 000 person years). Compared with the total population, CHD cases were more likely to experience fewer years of education, heavier smoking, higher prevalence of diabetes and hypertension, lower levels of physical activity and alcohol consumption. A high WHR was also associated with lower education, higher exposure to smoking, lower levels of physical activity and a higher prevalence of diabetes and hypertension (Table 1).

Table 1.   Baseline characteristics of the WLH cohort, 1991–1993
 Total cohort (%)CHD cases (%)WHR quartile (%)
<74 cm (%)74–77 cm (%)77–81 cm (%)>81 cm (%)
Education (years)
 <99280 (19)106 (41)1211 (14)1453 (16)1630 (19)2147 (25)
 10–1218479 (39)89 (35)2995 (35)3325 (37)3487 (40)3536 (40)
 13–1512757 (27)39 (15)2555 (30)2607 (29)2380 (27)2034 (23)
 ≥166556 (14)12 (5)1599 (19)1395 (16)1134 (13)905 (10)
Smoking status
 Never smoker19545 (41)50 (20)3912 (46)3938 (44)3502 (40)3412 (39)
 <5 pack years 7232 (15)19 (7)1334 (16)1368 (15)1323 (15)1089 (12)
 5 < 10 pack years6966 (15)25 (10)1171 (14)1236 (14)1344 (15)1201 (14)
 ≥10 pack years14151 (30)160 (63)2056 (24)2360 (27)2574 (29)3052 (35)
Ever use of oral contraceptive40032 (83)201 (79)7082 (83)7448 (84)7320 (85)7172 (82)
Diabetes622 (1)33 (13)71 (1)84 (1)91 (1)17 (2)
High blood pressure4462 (9)72 (28)635 (8)634 (7)794 (9)115 (13)
Physical activity
 Very low 2001 (4)20 (8)211 (3)223 (3)294 (3)520 (6)
 Low4885 (10)36 (14)719 (9)756 (9)891 (10)1202 (14)
 Normal27384 (57)160 (63)4611 (54)5055 (57)5103 (58)5200 (59)
 High7747 (16)21 (8)1736 (20)1697 (19)1498 (17)1113 (13)
 Very high3850 (8)15 (6)908 (11)854 (10)663 (8)450 (5)
Alcohol drinking (g day−1, Mean ± SD)3.5 ± 4.52.9 ± 4.03.6 ± 4.13.7 ± 4.33.6 ± 4.53.5 ± 5.0

A modest positive association was found between adult weight and CHD risk, which reduced to null after adjustment for other CHD risk factors (Table 2). No association was found between adult height and CHD risk. Larger waist circumference was strongly associated with an increased age-adjusted risk of CHD (HR = 3.0, 95% CI: 1.8–4.9), which remained significant after multivariate adjustment for health behaviour, hypertension and diabetes (HR = 1.9, 95% CI: 1.1–3.3). Age-adjusted risk of high hip circumference was also positively associated with CHD, but this association disappeared after further adjustment.

Table 2.   Adult anthropometric characteristics and CHD risk in the WLH cohort, 1991–2003
  Total cohort (%) CHD cases (%)Age adjusted Adjusted for age and health behavioursaAdjusted for age, health behaviours and diabetes and hypertensionbAdjusted for age, health behaviours, diabetes and hypertension and measures of obesityc
  1. aAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise. bAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise, plus diabetes and hypertension. cAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise, diabetes and hypertension, plus BMI.

Weight at enrolment (kg)
 <5811697 (24)42 (16)ReferenceReferenceReferenceReference
 58–6310793 (22)39 (15)0.9 (0.6–1.4)0.9 (0.6–1.5)1.0 (0.6–1.5)0.9 (0.5–1.5)
 63–7012179 (25)64 (25)1.3 (0.9–1.9)1.2 (0.8–1.7)1.0 (0.7–1.6)0.7 (0.4–1.3)
 ≥7012303 (26)101 (39)1.9 (1.3–2.7)1.6 (1.1–2.3)1.2 (0.8–1.9)0.6 (0.3–1.2)
 P for trend  <0.0001<0.0001NSNS
Height (cm)
 <16210444 (22)62 (24)ReferenceReferenceReferenceReference
 162–16612230 (25)76 (30)1.0 (0.7–1.4)1.1 (0.8–1.5)1.1 (0.8–1.6)1.2 (0.8–1.8)
 166–17011597 (24)52 (20)0.8 (0.5–1.1)0.9 (0.6–1.2)0.9 (0.6–1.3)0.9 (0.6–1.4)
 ≥17012858 (27)58 (23)0.8 (0.6–1.2)0.9 (0.6–1.3)0.9 (0.6–1.4)1.0 (0.6–1.5)
 P for trend  <0.05NSNSNS
Waist (cm)
 <706116 (13)18 (7)ReferenceReferenceReferenceReference
 70–759767 (20)40 (16)1.3 (0.8–2.3)1.4 (0.8–2.5)1.2 (0.7–2.2)1.3 (0.7–2.6)
 75–8110670 (22)31 (12)0.9 (0.5–1.6)0.9 (0.5–1.6)0.9 (0.5–1.6)0.9 (0.5–1.9)
 ≥819196 (19)95 (37)3.0 (1.8–4.9)2.6 (1.5–4.3)1.9 (1.1–3.3)1.9 (0.9–4.0)
 P for trend  <0.0001<0.0001<0.01<0.05
Hip (cm)
 <947966 (17)32 (13)ReferenceReferenceReferenceReference
 94–988120 (17)22 (9)0.6 (0.4–1.0)0.6 (0.3–1.0)0.5 (0.3–1.0)0.5 (0.3–1.0)
 98–1039881 (21)50 (20)1.0 (0.6–1.6)1.1 (0.7–1.6)1.0 (0.6–1.6)0.9 (0.5–1.5)
 ≥1039106 (19)77 (30)1.5 (1.0–2.3)1.5 (1.0–2.2)1.1 (0.7–1.7)0.8 (0.4–1.5)
 P for trend  <0.0001<0.001NSNS

Overweight and obesity, measured using BMI, were independently associated with an increased age-adjusted risk of CHD, with a dose–response relationship (Table 3). This association was reduced after adjustment for CHD risk factors and biological mediators, and was no longer significant after additional adjustment for WHR. There was a strong and highly significant dose–response increasing risk for CHD with increasing WHR. The associations were weakened (though remained significant) in multivariate models adjusted for BMI (HR = 1.8, 95% CI: 1.1–3.1, highest versus lowest quartile).

Table 3.   Obesity measures and CHD risk in the WLH cohort, 1991–2003
 Total (%) CHD cases (%)Age adjusted Adjusted for age and health behavioursaAdjusted for age, health behaviours and diabetes and hypertensionbAdjusted for age, health behaviours, diabetes and hypertension and measures of obesity
  1. aAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise. bAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise, plus diabetes and hypertension. cAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise and diabetes and hypertension, plus WHR. dAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise and diabetes and hypertension, plus BMI.

BMI at enrolment (kg/m2)
 <219585 (20)30 (12)ReferenceReferenceReferenceReferencec
 21–2313298 (28)38 (15)0.8 (0.5–1.3)0.8 (0.5–1.3)0.7 (0.4–1.2)0.7 (0.4–1.4)
 23–2510533 (22)54 (21)1.3 (0.8–2.4)1.2 (0.7–1.9)1.2 (0.7–1.9)1.1 (0.6–2.0)
 ≥2512767 (27)117 (46)2.2 (1.5–3.3)1.9 (1.3–2.8)1.5 (1.0–2.4)1.2 (0.7–2.1)
 P for trend  <0.0001<0.0001<0.05NS
Waist-hip-ratio
 <0.7378497 (18)29 (11)ReferenceReferenceReferenceReferenced
 0.737–0.7728923 (19)34 (13)1.2 (0.7–2.0)1.3 (0.8–2.2)1.3 (0.7–2.2)1.3 (0.8–2.3)
 0.772–0.8118759 (18)37 (14)1.3 (0.8–2.1)1.3 (0.8–2.1)1.3 (0.7–2.2)1.3 (0.7–2.2)
 ≥0.8118771 (18)80 (31)2.8 (1.8–4.3)2.4 (1.5–3.9)1.9 (1.2–3.2)1.8 (1.1–3.1)
 P for trend  <0.0001<0.0001<0.01<0.05

Figure 1 showed the BMI stratified association between WHR and CHD risk. Higher WHR consistently increased CHD risk, regardless of BMI category. Women with high BMI and high WHR were at substantially increased risk of CHD, indicating an interaction between BMI and WHR in the development of CHD (P for interaction <0.0001). High BMI was also associated with CHD, except amongst women in the lowest WHR who were not at increased risk of CHD.

Figure 1.

 Age-adjusted incidence rates for CHD according to BMI and WHR tertiles in WLH cohort, 1991–2003 (P for interaction <0.0001).

Most (64%) of the women in our cohort reported a birth weight over 3 kg and only 5% women reported a low birth weight (≤2.5 kg). A significant increased risk for CHD was found amongst women with low birth weight, which was not explained by further multivariate adjustment (HR = 2.0; 95% CI: 1.1–3.6) (Table 4). Above average body size in childhood (at age 7) or overweight in young adulthood (at age 18) had a significant increased age-adjusted risk for CHD, which became nonsignificant after further adjustment for health behaviours, mediators and adult BMI (Table 4).

Table 4.   Early life body sizes and CHD risk in the WLH cohort, 1991–2003
  Total cohort (%) CHD cases (%)Age adjusted Adjusted for age and health behavioursaAdjusted for age, health behaviours and diabetes and hypertensionbAdjusted for age, health behaviours, diabetes and hypertension and measures of obesityc
  1. aAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise. bAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise, plus diabetes and hypertension. cAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise and diabetes and hypertension, plus BMI. dRestricted to single-born subjects.

Birth weight (kg)d
 <2.52088 (4)18 (7)1.6 (0.9–2.9)1.7 (1.0–3.0)2.0 (1.1–3.7)2.0 (1.1–3.6)
 2.5–3.08285 (18)43 (17)ReferenceReferenceReferenceReference
 >3.030073 (64)133 (53)0.9 (0.6–1.3)0.9 (0.6–1.3)1.0 (0.7–1.5)1.0 (0.7–1.5)
Body size at age 7
 Thin19076 (40)95 (37)1.0 (0.7–1.3)0.9 (0.7–1.2)1.0 (0.8–1.4)1.1 (0.8–1.5)
 Normal24091 (50)124 (48)ReferenceReferenceReferenceReference
 Above average4613 (10)35 (14)1.6 (1.1–2.3)1.4 (0.9–2.0)1.3 (0.9–2.0)1.2 (0.8–1.8)
 P for trend  NSNSNSNS
BMI at age 18 (kg/m2)
 <1910795 (22)47 (18)ReferenceReferenceReferenceReference
 19–208149 (17)33 (13)1.0 (0.6–1.5)1.1 (0.7–1.7)1.1 (0.7–1.8)1.1 (0.7–1.8)
 20–2213861 (29)74 (29)1.3 (0.9–1.8)1.4 (0.9–2.0)1.2 (0.8–1.8)1.1 (0.7–1.7)
 ≥2211493 (24)72 (28)1.5 (1.1–2.2)1.4 (1.0–2.1)1.3 (0.9–2.0)1.1 (0.7–1.6)
 P for trend  <0.001<0.001<0.05NS

Women who were consistently overweight over their lifetime had a significantly increased risk of CHD with an age-adjusted HR of 3.5 (95% CI: 1.8–7.2) that, however, became nonsignificant after further adjustment for health behaviours, diabetes and hypertension (Table 5). There was no further association between CHD risk and changes in BMI between age 18 and at study enrolment. There was no evidence of an interaction between BMI and birth weight in the development of CHD (P for interaction =0.98) (Table 6). Nor was there any evidence for interaction between birth weight and young adulthood BMI or adult WHR and CHD risk (data not shown).

Table 5.   Body shape change over time and CHD risk in the WLH cohort, 1991–2003 (restricted to singleton births)
  Total cohort (%) CHD cases (%)Age adjustedAdjusted for age and health behavioursaAdjusted for age, health behaviours and diabetes and hypertensionb
  1. aAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise. bAdjusted for age, smoking, alcohol, years of education, ever use of oral contraceptives, and exercise, plus diabetes and hypertension.

Body shape change between childhood and adulthood
 Remained thin2956 (6)11 (4)1.1 (0.5–2.2)0.9 (0.5–2.0)1.0 (0.5–2.2)
 Decreased weight12999 (28)71 (28)1.3 (0.9–2.1)1.1 (0.7–1.8)1.0 (0.6–1.7)
 Remained average6509 (14)26 (10)ReferenceReferenceReference
 Increased weight21622 (46)121 (48)1.3 (0.8–2.0)1.1 (0.7–1.7)0.9 (0.6–1.5)
 Remained overweight804 (2)11 (4)3.5 (1.8–7.2)2.4 (1.2–4.9)1.6 (0.7–3.5)
BMI change between age 18 and study enrolment
 Decreased4932 (11)22 (9)1.2 (0.7–2.1)1.1 (0.6–1.9)1.2 (0.7–2.3)
 Increased 0–1.4 U6959 (15)25 (10)ReferenceReferenceReference
 Increased 1.5–4.0 U16346 (35)61 (24)0.9 (0.6–1.4)0.9 (0.6–1.5)1.0 (0.6–1.7)
 Increased ≥4.0 U14489 (31)107 (43)1.5 (1.0–2.3)1.3 (0.8–2.0)1.2 (0.7–2.0)
Table 6.   Age-adjusted HR (95%CI) for CHD risk from the interaction between birth weight and adult BMI in the WLH cohort, 1991–2003
Adult BMI (kg m2)Birth weight (kg)
No of population (CHD cases)Age-adjusted HR (95% CI)
<2.52.5–3.0>3.0<2.52.5–2.0>3.0
  1. Restricted to single-born subjects, P for interaction = 0.98.

<21.5621 (3)2457 (7)7973 (23)1.5 (0.4–5.4)0.9 (0.3–2.4)1.0 (0.5–2.0)
21.5–24.0625 (5)2691 (10)10336 (27)2.2 (0.7–6.4)Ref.0.8 (0.4–1.6)
≥24.0762 (10)2798 (24)10816 (76)3.2 (1.3–7.7)2.1 (1.0–4.4)1.8 (0.9–3.5)

Discussion

In this large Swedish prospective study, we found strong evidence for a positive association between overweight or obesity on CHD risk, whether measured by adult weight, waist circumference, WHR or BMI. The results of this study indicate that markers of central obesity, measured by waist circumference or WHR were more strongly related to CHD risk than BMI. The interaction between BMI and measures of abdominal obesity, shown in this study indicates that women with high levels of abdominal adiposity who are overweight are at particularly high risk of developing CHD, whilst women with a high BMI but little central obesity showed little increased risk of CHD. In contrast to other studies, adult height was not related to CHD. Low birth weight and being overweight as a young adult increased risk of CHD, as did having an above average body size in childhood or remaining overweight throughout life before further adjustment for diabetes and hypertension.

The strong association shown in this study between obesity and CHD is consistent with previous studies from both developed and developing countries [22–29]. There has been debate as to whether the effect of BMI on CHD is limited to those who are obese. The American Nurses’ Health Study reported that even women with BMI within the ‘normal’ weight range had a significantly increased risk of CHD compared with the leanest women [30, 31] and higher WHR was associated with more than three-fold increase in CHD risk (WHR ≥0.88 vs. <0.72) [12]. In contrast, the significantly increased risk of CHD was only found for the most obese women (BMI > 27.6 kg m2) in the Framingham study [32] and in the Iowa Women’s Health study (BMI > 30 kg m2) [33] whilst overweight women were not at increased risk of CHD. Our study confirmed the increased risk of CHD amongst women who were overweight, as well as for those who were obese.

Several pathophysiological pathways have been hypothesized to explain the role of obesity, which is a chronic metabolic condition, in increasing CHD risk. These include elevating levels of serum total cholesterol, LDL-cholesterol, triglycerides, fibrinogen, C-reactive protein, reducing levels of the ‘protective’ factor, HDL-cholesterol and the biological mediators for developing cardiovascular risk factors such as diabetes and hypertension [12, 22, 34]. There has been debate as to which measure of obesity is most pertinent in relation to CHD risk. Since BMI may not be a good measure of visceral fat, the key determinant of metabolic abnormalities contributing to CHD risk, waist circumference or WHR provide better information about visceral adipose tissue accumulation and may be more closely related to risk of CHD. The result found from our study support the conclusion in the American Nurses’ Health Study [12], the Iowa Women’s Health Study [7], the INTERHEART case-control study across 52 countries [8] and an Japanese study [11] that central obesity measured in waist circumference and WHR are stronger anthropometric predictors of CHD risk than BMI alone. Amongst the measures of central adiposity, WHR appears to be the most sensitive anthropometric marker for predicting the CHD risk in women, compared with waist circumference. This is also consistent with the Dallas Heart Disease in which WHR provided better discrimination than either BMI or waist circumference for the association between obesity and prevalent atherosclerosis [9]. In our study, although obese women were at a threefold age-adjusted increasing risk for CHD, this effect soon faded to null after adjustment for biological mediators, suggesting that the association between BMI and CHD was mediated by diabetes and hypertension. In contrast, waist circumference and WHR remained closely associated with CHD risk even after adjustment for other CHD risk factors, mediators and BMI.

Weight gain is associated with unfavourable metabolic changes such as raising blood glucose, blood pressure, HDL-cholesterol and triglycerides [35]. However, inconsistent results have been reported for the effect of weight changes over time on CHD risk, as some studies show increased risk for CHD with weight changes [36–39], whilst others show no effect [40, 41]. The American Nurses’ Health Study reported that even a modest weight gain (4–10 kg) during adulthood increased risk of CHD by 27%, compared with women with a stable weight [42]. In our study, an age-adjusted significant increasing risk only appeared in women who remained overweight over time whilst other changes in body size had no importance. The variation across studies may be caused by differences in the methodology used to define the weight change or the time period over which weight change has been measured [41]. The null association between height and CHD risk found in our study is inconsistent with evidence that taller people are at lower risk of heart disease [19], though other studies have also shown null associations [43, 44].

The association between low birth weight and CHD in this cohort is consistent with other studies [15, 16, 18, 45, 46]. Barker proposed the ‘fetal origins’ hypothesis to suggest that the vulnerability of people with low birth weight to CHD is because of adaptations that the foetus makes when undernourished [47]. It is hypothesized that birth weight is inversely associated with CHD risk factors, such as raised blood pressure, dyslipidemia and glucose intolerance [18], and may also reflect adverse parental socio-economic status.

The lack of association between childhood body shape (age 7) and the risk of CHD in later life is consistent with results from a large Scottish birth cohort study in which the childhood BMI was measured when subjects were at mean age of 4.9 years [48]. A number of interpretations of the interaction between birth weight and adult body size have been suggested, such as rapid weight gain, rapid ‘centile crossing’ after birth as a marker of growth restriction in utero or a true biological interaction of early growth and later growth [17]. In contrast to other studies [15, 17, 49, 50], our study did not find a strong evidence of this interaction, although heavier adult women with low birth weight had an increased CHD risk. In our study, birth weight grouping was based on broad categorization and there were rather small numbers of subjects in each interaction subgroup between birth weight and adult body size, which means that the effects found in our study may due to the chance.

The major limitations of our study is the potential misclassification of exposure measurement as it was based on self-report, and this could have lead to weakening of the reported associations [51]. In addition, anthropometric measures in adulthood were only measured at baseline and not during the follow-up. Adjustment for mediators of the association between anthropometric variables and CHD was incomplete as we were not able to collect biochemical blood indices, such as blood lipid and blood cholesterol. Interpretation of the independent effects of BMI and WHR on CHD is complicated because of the high intercorrelation between these variables. Only categorical birth weight data, based on generally accepted cut-points was available in the questionnaire. This categorization may not have been ideal for our cohort, as only a small proportion (4.5%) were in the lowest birth weight category (<2.5 kg). This may have resulted in underestimation or distortion of the effect from low birth weight and the interaction between birth weight and adult body size. Furthermore, we do not know whether weight changes over time were intentional or unintentional, and so were not able to assess whether changes in weight were an antecedents or consequences of illness. Almost all the women (96%) were resident in Nordic countries for the first 7 years of life, but there was no detailed information on ethnicity.

The main strength of the present study is the large size of the study population, the prospective study design, and virtually complete follow-up information through linkages to national registers. Furthermore, we collected and compared detailed information on different anthropometric measurements for adiposity during adulthood, and self-reported changes from childhood and early adulthood and potential confounders which allowed comparison of the effects of different anthropometric characteristics on CHD risk.

In conclusion, results from this rich cohort study lends support to a growing body of evidence supporting control of obesity starting at a young age as one of the population primary prevention strategies for CHD, along with control of other major risk factors. Avoidance of abdominal obesity is a particular priority.

Conflict of interest statement

No conflict of interest was declared.

Ancillary