Fetal growth and coronary heart disease in South India

THE LANCET

Fetal growth and coronary heart disease in South India

C E Stein, C H D Fall, K Kumaran, C Osmond, V Cox, D J P Barker

Summary Background Coronary heart disease is predicted to become the commonest cause of death in India within 15 years. People from India living overseas already have high rates of the disease that are not explained by known coronary risk factors. Small size at birth is a newly described risk factor for coronary heart disease, but associations between size at birth and the disease have not been examined in India. Methods We studied 517 men and women who were born between 1934 and 1954 in a mission hospital in Mysore, South India, and who still lived near to the hospital. We related the prevalence of coronary heart disease, defined by standard criteria, to their birth size. Findings 25 (9%) men and 27 (11%) women had coronary heart disease. Low birthweight, short birth length, and small head circumference at birth were associated with a raised prevalence of the disease. Prevalence fell from 11% in people whose birthweights were 5·5 lb (2·5 kg) or less to 3% in those whose birthweights were more than 7 lb (3·1 kg), p for trend=0·09. The trends were stronger and statistically significant among people aged 45 years and over (p=0·03 for birthweight, 0·04 for length, and 0·02 for head circumference). High rates of disease were also found in those whose mothers had a low body weight during pregnancy. The highest prevalence of the disease (20%) was in people who weighed 5·5 lb (2·5 kg) or less at birth and whose mothers weighed less than 100 lb (45 kg) in pregnancy. These associations were largely independent of known coronary risk factors. Interpretation In India, as in the UK, coronary heart disease is associated with small size at birth, suggesting that its pathogenesis is influenced by events in utero. The association w ith low maternal bodyw eight is further evidence that the disease originates through fetal undernutrition. Prevention of the rising epidemic of the disease in India may require improvements in the nutrition and health of young women.

Lancet 1996; 348: 1269–73 See Commentary page 1254

Medical Research Council Environmental Epidemiology Unit, Southampton General Hospital, Southampton SO16 6YD, UK (C E Stein MSE, C H D Fall MRCP , C Osmond PhD, Prof D J P Barker FRCP ); and Holdsworth Memorial Hospital, Mysore, South India (K Kumaran MB) Correspondence to: Dr C H D Fall

Vol 348 • November 9, 1996

Introduction Coronary heart disease is both common and on the increase in India. Rates of death from coronary heart disease are expected to overtake those from infectious disease by around the year 2010.1 High rates of coronary heart disease have also been recorded in Indian populations living in the UK and are largely unexplained by known coronary risk factors.2 Associations have been demonstrated between small size at birth and coronary heart disease and its risk factors, hypertension, non-insulin dependent diabetes, and abnormalities in lipid metabolism and blood coagulation.3–6 These associations are independent of adult lifestyle, including smoking, obesity, and social class, and have led to the hypothesis that the disease is programmed in utero. The fetal origins hypothesis proposes that adaptations made by the fetus in response to undernutrition lead to persisting changes in metabolism and organ structure. This hypothesis is supported by experimental evidence in which offspring of undernourished pregnant animals show permanent changes including raised blood pressure and abnormal glucose-insulin and lipid metabolism.7–9 Low rates of fetal growth and small size at birth are common in India. To explore associations between fetal growth and coronary heart disease in India we traced 517 men and women who were born in a hospital in the city of Mysore, South India, where detailed records have been kept on each birth since 1934.

Methods Holdsworth Memorial Hospital is a mission hospital governed by the Church of South India. It was built as a maternity hospital in 1905 and is in a poor, crowded area of the city. Since 1934 birthweight, length, and head circumference have been recorded routinely for babies born in the hospital. These measurements were made by one of three midwives over the period of our study, under an agreed protocol. The birth records also contain the parents’ names, occupations, address, religion or caste, and the mother’s obstetric history. The records do not provide information on gestational age or placental weight. Some of the mothers first came to the hospital in labour whereas others attended the antenatal clinic, and their records include their weight during pregnancy. 8883 babies were born alive in Holdsworth Memorial Hospital between 1934 and 1954. In 1993 we identified people born in the hospital during this time by carrying out a house-to-house census in a 2 square mile area around the hospital. 7800 households were visited by four fieldworkers. 1311 people said that they had been born as singletons in the hospital during the 20-year period. We identified the birth records of 536 of them. We had difficulty in tracing birth records because most people were unaware of their date of birth and knew their age only vaguely, and the birth records did not contain the infant’s name. People were therefore matched to their birth records through their parents’ names, addresses, and occupations, through the number, sex, and order of older siblings, and through the mother’s medical history. 517 (96%) of the 536 people agreed to take part in the study. After a 12 h overnight fast they attended the hospital for examination and investigations. The presence of coronary heart

1269

THE LANCET

Risk factor

Men

Women

Coronary heart disease Present (n=25)

Absent (n=241)

164·9 61·1 22·5 91·1 19·7

165·9 63·1 22·8 91·3 20·5

Percentage with diabetes

24

12

Serum lipids Total cholesterol (mmol/L) LDL cholesterol (mmol/L) HDL cholesterol (mmol/L) HDL/LDL cholesterol ratio (%) Triglyceride (mmol/L)

5·3 3·7 0·88 24 1·8

4·9 3·1 0·92 30 1·7

Anthropometric factors Height (cm) Weight (kg) Body-mass index (kg/m2) Waist/hip ratio (%) Subscapular skinfold thickness (mm)

All (SD) (n=266)

Coronary heart disease

All (SD)(n=251)

Present (n=27)

Absent (n=224)

151·1 57·5 25·1 84·9 27·4

151·6 57·0 24·8 83·1 26·2

13 (34)

26

15

16 (37)

5·0 (1·0) 3·2 (0·8) 0·92 (0·22) 29 (1·4)* 1·7 (1·7)*

4·7 3·0 0·93 31 1·7

4·9 3·2 1·01 32 1·5

4·9 (0·9) 3·2 (0·8) 1·01 (0·24) 32 (1·4)* 1·5 (1·7)*

165·8 (6·2) 62·9 (13·2) 22·8 (4·1) 91·3 (6·3) 20·4 (8·8)

151·5 (6·4) 57·0 (12·2) 24·8 (4·9) 83·3 (6·0) 26·3 (9·2)

Blood pressure Systolic (mm Hg) Diastolic (mm Hg) Percentage with hypertension

136 82 48

131 80 17

131 (17) 80 (11) 20 (40)

133 76 41

132 77 24

133 (18) 77 (11) 26 (44)

Plasma fibrinogen (g/L)

298

311

310 (79)

341

339

339 (60)

72 4

61 12

62 (49) 11 (32)

0 0

0 0

0 (1) 0 (0)

72

45

48 (50)

48

48

48 (50)

Lifestyle Percentage of smokers Percentage with moderate or heavy alcohol consumption Percentage in lower social class

Values are means except where stated. *Geometric standard deviation.

Table 1: Coronary risk factors in men and women in Mysore disease was ascertained through the Rose/WHO chest pain questionnaire10 and a standard 12-lead electrocardiogram (ECG). The ECGs were coded according to the Minnesota protocol11 by two trained coders in the UK, who were unaware of birth or current measurements of the participants. Coronary heart disease was defined as the presence of one or more of the following: typical angina according to the Rose/WHO questionnaire; ECG Minnesota codes 1-1 or 1-2 (Q and QS codes); or a history of coronary artery angioplasty or bypass graft surgery. Blood samples were taken after the participants had been fasting for 12 h for measurements of plasma glucose and fibrinogen and serum total cholesterol, high-density-lipoprotein (HDL) cholesterol, and triglyceride concentrations. Except for 39 people already known to be diabetic, whose fasting blood glucose concentrations were all higher than 7·8 mmol/L, 478 participants were given 75 g oral glucose load and further blood samples were taken for plasma glucose measurement 30 and 120 min later. 467 of the participants completed the oral glucose tolerance test. Plasma glucose concentrations were analysed in Mysore by a standard glucose oxidase method. Serum total cholesterol, HDL cholesterol, and triglyceride concentrations were measured at Addenbrooke’s Hospital, Cambridge, UK by standard enzymatic methods. Serum low-density-lipoprotein (LDL) cholesterol fibrinogen concentrations were calculated. 12 Plasma concentrations were analysed at the Wolfson Institute, St Bartholomew’s Hospital, London, UK, by the Clauss method, with an electrical impedance endpoint.13 Blood pressure was measured with an automated recorder (Dinamap) after the participants had been lying down for at least 5 min. Height, weight, hip and waist circumferences, and biceps, triceps, subscapular, and suprailiac skinfold thicknesses were measured by one of two observers. Information on smoking habits, alcohol consumption, and socioeconomic status was obtained by questionnaire. Alcohol consumption was converted to the total number of units per week (1 unit=7·5 g). Socioeconomic status was assessed by the Kuppuswamy score, a standard questionnaire method for urban Indian populations, which uses information on family size, housing, education, occupation, and income.14 Before the study we revised the classification of incomes so that they were in line with contemporary earnings in Mysore. The observers and interviewers at the hospital were unaware of the birth measurements of the participants. Before we started the study the

1270

procedures for the measurements were standardised and the fieldworkers trained.

Statistical analysis The birthweights of all 517 participants had been recorded. Birth length was recorded for 512 and head circumference at birth for 511. Birthweight and mother’s weight were measured in pounds and ounces, and length and head circumference were measured in inches; we maintained these units for our analyses. The measurements were often rounded, producing clumping of data around whole numbers. For tabulations we have preserved divisions of birthweight used for analyses of European data but divided length and head circumference as closely as possible by tertiles. Data were analysed with multiple linear and logistic regression. Serum triglyceride concentrations and the ratio of HDL cholesterol to LDL cholesterol had a skewed distribution and so were log-transformed to symmetry.

Results The age range of the 517 men and women was 38–60 years (mean 47 years). 184 (36%) were Hindu, 279 (54%) were Muslim, and 54 (10%) were Christian. The men and women were short, light, and thin (table 1) compared with British people of the same age.15 They were smaller at birth in all measurements compared with babies born in the UK,16,17 the difference being most pronounced for birthweight. Their mean (SD) birth measurements were birthweight 6·1 (0·9) lb, birth length 18·9 (1·2) inches, and head circumference 13·2 (0·7) inches. The corresponding figures for all babies born in the hospital during the same period were 5·9 lb, 18·8 inches, and 13·1 inches. 25 (9%) men and 27 (11%) women had coronary heart disease. 14 (2·7%) men and women had Q waves on ECG, 41 (7·9%) had symptoms of angina, and one had a history of coronary artery bypass surgery. The prevalence of coronary heart disease rose with age, from 8% in men younger than 45 years to 10% in men aged 45 and over, and from 6% to 13% in women of the same ages. The prevalence was similar in Muslim (10%), Hindu (10%), and Christian (13%) participants.

Vol 348 • November 9, 1996

THE LANCET

All participants n

Angina Birthweight (lb) 聿5·0 5·1–5·5 5·6–6·0 6·1–6·5 6·6–7·0 >7·0 All weights p for trend*

62 8 87 10 111 9 121 8 77 8 59 3 517 8

Birth length (inches) 聿18 18·1–19 >19 All lengths p for trend*

193 11 169 7 150 6 512 8

Head circumference at birth (inches) <13 80 11 13 227 9 肁13·5 204 5 All circumferences 511 8 p for trend*

n

% with CHD

10 11 13 11 9 3 10 0·09

43 56 79 81 55 48 362

14 16 13 12 7 4 11 0·03

5 1 2 3

15 7 8 10 0·03

155 110 94 359

15 9 9 11 0·04

1 3 3 3

13 11 7 10 0·08

66 165 128 359

15 12 8 11 0·02

Q waves

CHD

2 2 5 4 1 0 3

For conversion to SI units: 1 lb=0·454 kg, 1 inch=2·54 cm. *With allowance for age and sex. CHD=coronary heart disease defined by presence of angina or Q waves.

Table 2: Percentages of men and women with coronary heart disease according to birth measurements

18 (3%) participants were in social class I, 251 (49%) in social class II, 123 (24%) in social class III, and 125 (24%) in social class IV. Social class V was not represented in this study sample. We grouped social classes I and II as upper class and classes III and IV as lower class. Lower-class men had a higher prevalence of coronary heart disease than upper-class men (table 1) but there was no difference between classes in women. Table 2 shows that the prevalence of coronary heart disease was higher in the men and women who had low birthweights, short birth lengths, or small head circumferences. Among participants, aged 45 years and over, each of these trends was statistically significant. The trends in older participants are shown as odds ratios in table 3. Short birth length was associated with increased prevalence of coronary heart disease at all birthweights and prevalence was therefore unrelated to ponderal index (birthweight/birth length3), a measure of thinness at birth. The trends in coronary heart disease with size at birth were similar in men and women.

Odds ratio (95% CI) Birthweight (lb) 聿5·0 5·1–5·5 5·6–6·0 6·1–6·5 6·6–7·0 >7·0

3·7 (0·7–19·5) 4·4 (0·9–21·4) 3·3 (0·7–15·8) 3·2 (0·7–15·4) 1·8 (0·3–10·3) 1·0 ..

Birth length (inches) 聿18 18·1–19 >19

1·9 (0·8–4·4) 1·1 (0·4–2·9) 1·0 ..

Head circumference at birth (inches) <13 13 肁13·5

2·1 (0·8–5·4) 1·6 (0·7–3·6) 1·0 ..

For conversion to SI units: 1 lb=0·454 kg, 1 inch=2·54 cm.

Table 3: Odds ratios for coronary heart disease in men and women aged 45 years and over according to weight, length, and head circumference at birth

Vol 348 • November 9, 1996

n

Participants 肁45 years

% with

Mother’s weight (lb) <88 88–96 97–104 105–114 >114 All p for trend

Birthweight (kg)

41 42 47 39 40 209

Length (cm)

Head circumference (cm)

Coronary heart disease prevalence (%)

2·548 2·686 2·796 2·868 3·035

47·7 46·6 47·9 49·1 48·1

33·1 33·2 33·3 33·8 33·9

12 13 11 0 8

2·785

47·9

33·4

<0·0001

0·06

9

0·004

0·07*

*Adjusted for age and sex. For conversion to SI units 1 lb=0·454 kg.

Table 4: Mean weight, length, and head circumference at birth and prevalence of coronary heart disease according to mother’s weight in pregnancy

The average age of the mothers was 24 years, and 108 (21%) were primiparous. The weights in pregnancy of 209 of the mothers were recorded, of whom 46 (22%) were primiparous. For mothers with more than one weight recorded, we analysed the first. The mean weight was 103 lb (47 kg, SD=8), recorded on average 13 (SD 9) weeks before delivery. Participants whose mothers’ weight was recorded were similar to the others except that 57% were of high social class compared with 48% in the remainder. Mothers of lower weight had babies who were smaller in each birth measurement and, as adults, had a raised prevalence of coronary heart disease (table 4). The highest prevalence of coronary heart disease (20%) was in people who weighed 5·5 lb or less at birth and whose mothers weighed less than 100 lb (table 5). By contrast there were no cases of coronary heart disease in people who weighed more than 6·5 lb and whose mothers weighed 100 lb or more. Table 1 shows the mean values of known coronary risk factors of the participants. 39 people were known to be diabetic. A further 34 men and 40 women had late-onset diabetes (2 h plasma glucose 肁11·1 mmol/L). 65 people had systolic blood pressure above 150 mm Hg. 76 people, whose mean systolic pressure was 146 mm Hg compared with 129 mm Hg in the other participants, were already being treated for hypertension. There were 54 men and 64 women who could be defined as hypertensive by either criterion. 166 (62%) men had smoked regularly at some time in their lives and 128 (48%) were current smokers: only one woman had ever smoked. 56 (21%) men reported that they drank alcohol; 9 (3%) had a moderate intake (22–35 units per week) and 21 (8%) had a heavy intake (more than 35 units per week). Only one woman drank alcohol. In summary, table 1 shows that in both sexes coronary heart disease was associated with diabetes and hypertension, low serum concentrations of HDL cholesterol, a low ratio of HDL cholesterol to LDL

Birthweight (lb)

聿5·5 5·6–6·0 6·1–6·5 >6·5 All

Mother’s weight (lb) <100

肁100

All

20% (40) 9% (22) 9% (23) 6% (16) 13% (101)

8% (13) 14% (22) 8% (26) 0 6% (108)

17% (53) 11% (44) 8% (49) 2% (63) 9% (209)

For conversion to SI units 1lb=0·454 kg.

Table 5: Frequency of coronary heart disease according to birthweight and mother’s weight in pregnancy

1271

THE LANCET

cholesterol, and with raised serum concentrations of triglycerides and raised systolic pressures. In men coronary heart disease was also associated with raised LDL cholesterol, short stature, smoking, and low social class. The association with low social class was not explained by a higher prevalence of other risk factors. In women the disease was associated with increased waist/hip ratio and subscapular and suprailiac skinfold thickness, but not with triceps or biceps skinfold thickness. Coronary heart disease was not associated with raised plasma fibrinogen concentrations or alcohol consumption. We examined the associations between size at birth and coronary heart disease taking into account other risk factors. In a simultaneous regression with age, sex, the prevalence of diabetes, hypertension, and smoking, serum HDL cholesterol and LDL cholesterol and triglyceride concentrations, height and social class, the association with short length remained statistically significant (p=0·04). The other significant associations in this analysis were with hypertension (p<0·001) and serum HDL cholesterol (p=0·03). Among men the relation with LDL cholesterol was also significant (p<0·001). Neither birthweight nor head circumference, when substituted for length, was significant (p=0·2 for both).

Discussion Mysore City has changed little in the past 50 years and still has no major industries. By European standards the city is overcrowded and has poor sanitation and housing, but there is no poverty of the kind seen in the slums of large Indian cities. The people of Mysore are relatively well nourished, although men and women are considerably shorter than in British populations, and the men are thinner. Tobacco smoking is common among men, but only one woman in our study smoked. The prevalence of coronary heart disease defined by standard criteria was found to be 10%, which is similar to rates in England and Wales18 and elsewhere in India.19 The prevalence of major Q-wave disease was 2·7%, similar to rates reported in other urban Indian populations.1 Coronary heart disease in Mysore was associated with some of the conventional risk factors including age, diabetes, hypertension, altered concentrations of serum lipid, short stature, and smoking but not with raised plasma fibrinogen concentrations or obesity. Prevalence rates were similar in men and women, as in other studies in India19–21 and studies of Indians living in the UK. Rates were similar in the different religious groups. There was a higher prevalence of the disease among men of lower social class, which was not explained by known risk factors. This finding is consistent with some previous studies in India,20 although others have shown an association with high income.21 All the Mysore birth measurements were low by western standards, although similar to Indian community averages.22 Birthweights were especially low, 29% of the babies weighing under 2·5 kg. Our findings that coronary heart disease was associated with low birthweight, small head circumference, and short body length at birth are consistent with associations between the disease and small size at birth reported in the UK and the USA.23–25 These associations were seen in babies born at term and therefore reflect low rates of fetal growth. Thinness at birth, shown by a low ponderal index, was also related to coronary heart disease in one of the UK studies.24 By 1272

contrast, in Mysore short length was related to the disease at all birthweights. Small head circumference, short length at birth, and low birthweight are thought to result from fetal adaptations to undernutrition with reduction in growth of the head, body length, or soft tissues.6 These adaptations occur at different stages of gestation, since growth of the head precedes growth in length. This is the first study in which the association between mother’s weight and coronary heart disease in the offspring could be studied. The mean weight of mothers in Mysore was 47 kg, which is strikingly below weights of pregnant women in the UK. We found that mothers with the lowest weights had the smallest babies who as adults had the highest rates of coronary heart disease. This study therefore provides further evidence that the fetal growth failure that leads to coronary heart disease is a consequence of fetal undernutrition.3 Our study is based on a group of adults who were born in one hospital in Mysore, had survived childhood, were identified during our census, and who gave sufficient information to enable their birth records to be found. The study group is therefore unrepresentative of all people in the city, although their birth measurements were similar to those of all people born in the hospital. Although our survey was carried out in a poor area of the city, 52% of the participants were upper rather than lower social class, perhaps because tracing information obtained from upperclass people is more accurate. Our analysis, however, is based on internal comparisons; bias would be introduced only if the associations between fetal growth and coronary heart disease differed in those born in and outside the hospital, and in those traced and not traced. We have no reason to suspect such differences. Our study sample comprised people who survived to middle age. We know, however, from studies in the UK that low birthweight is associated with increased coronary heart disease mortality at all ages,23 and the findings in Mysore are therefore unlikely to reflect differential survival. The associations between size at birth and coronary heart disease were apparent despite the inevitable inaccuracies of birth measurements made in a busy mission hospital. These inaccuracies would tend to introduce random error and weaken the associations. We defined coronary heart disease by standard criteria for epidemiological studies; misclassification of some individuals is possible, but would tend to weaken the associations found. Simultaneous regression analyses suggested that the associations between reduced fetal growth and conventional coronary risk factors do not explain the association between reduced fetal growth and coronary heart disease. We therefore conclude that this association must be partly mediated by processes other than the known risk factors, which is consistent with the fact that known risk factors cannot explain the high rates of coronary heart disease in Indian people living overseas.2 Our results suggest that any strategy to prevent further increases in rates of coronary heart disease in India must include measures to improve the exceptionally low rates of fetal growth that prevail across the country. Fetal growth depends on the supply of nutrients,6 and better growth may therefore require improvement in the nutrition and health of young women. We thank the men and women who participated in the project; B D R Paul, Director of Holdsworth Memorial Hospital, Mysore, for his support; Mr Venkatachalam of the Medical Records Department who preserved the records and allowed us to use them: and Mr Surendrakumar

Vol 348 • November 9, 1996

THE LANCET for his expert laboratory help; Miss Shylaja, Miss Annamma, Mrs Geetha, Mr Jayakumar, T Lawrence, J Saldhana, Miss Kalpana, Mrs Chechyamma, Miss Harini, Mrs Poornima, and S Ahmed for fieldwork; N Keen and C Rose who coded the ECGs, and laboratory staff of the Department of Clinical Biochemistry, Addenbrooke’s Hospital, Cambridge, and the Coagulation Laboratory at the Wolfson Institute, St Bartholomew’s Hospital, London. The study was funded by the Wellcome Trust.

References 1

Chadha SL, Radakrishnan S, Ramachandran K, Kaul U, Gopinath N. Epidemiological study of coronary heart disease in urban population of Delhi. Indian J Med Res 1990; 92: 424–30. 2 McKeigue PM, Marmot MG. Mortality from coronary heart disease in Asian communities in London. BMJ 1988; 297: 903. 3 Barker DJP. Fetal origins of coronary heart disease. BMJ 1995; 311: 171–74. 4 Law CM, de Swiet M, Osmond C, et al. Initiation of hypertension in utero and its amplification throughout life. BMJ 1993; 306: 24–27. 5 Hales CN, Barker DJP, Clark PMS, et al. Fetal and infant growth and impaired glucose tolerance at age 64. BMJ 1991; 303: 1019–22. 6 Barker DJP, Gluckman PD, Godfrey KM, Harding JE, Owens JA, Robinson JS. Fetal nutrition and cardiovascular disease in adult life. Lancet 1993; 341: 938–41. 7 Langley SC, Jackson AA. Increased systolic blood pressure in adult rats induced by fetal exposure to maternal low protein diets. Clin Sci 1994; 86: 217–22. 8 Snoeck A, Remacle C, Reusens B, Hoet JJ. Effects of a low protein diet during pregnancy on the fetal rat endocrine pancreas. Biol Neonate 1990; 57: 107–18. 9 Hales CN, Desai M, Ozanne SE, Crowther NJ. Fishing in the stream of diabetes: from measuring insulin to the control of organogenesis. Biochem Soc Trans 1996; 24: 341–50. 10 Rose GA, Blackburn H. Cardiovascular Survey Methods. Geneva: World Health Organisation, 1968. 11 Prineas RJ, Crow RS, Blackburn H. The Minnesota code manual of electrocardiographic findings: standards and procedures for measurement and classification. Boston: John Wright, 1982.

Vol 348 • November 9, 1996

12 Friedwald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma without the use of the preparative ultracentrifuge. Clin Chem 1972; 18: 499–502. 13 Clauss A. Gerinnungsphysiologische Schnellmethode zur Bestimmung des Fibrinogens. Acta Haematol 1957; 17: 237–46. 14 Kuppuswamy B. Manual of socio-economic status scale. Delhi: Manasayan Publication, 1962. 15 Office of Population Censuses and Surveys. The dietary and nutritional survey of British adults. London: HM Stationery Office, 1990. 16 Law CM, Barker DJP, Richardson WW, et al. Thinness at birth in a northern industrial town. J Epidemiol Community Health 1993; 47: 255–59. 17 Gairdner D, Pearson J. A growth chart for premature and other infants. Arch Dis Child 1971; 46: 783–87. 18 Reid DD, Hamilton PJS, Keen H, Brett GZ, Jarrett RJ, Rose G. Cardiorespiratory disease and diabetes among middle aged male civil servants. Lancet 1974; i: 469–73. 19 Kutty VR, Balakrishnan KG, Jayasree AK, Thomas J. Prevalence of coronary heart disease in the rural population of Thiruvananthapuram district, Kerala, India. Int J Cardiol 1993; 39: 59–70. 20 Gupta R, Gupta VP, Ahluwalia NS. Educational status, coronary heart disease, and coronary risk factor prevalence in a rural population of India. BMJ 1994; 309: 1332–36. 21 Sarvotham SG, Berry JN. Prevalence of coronary heart disease in an urban population in Northern India. Circulation 1968; 37: 939–53. 22 Mohan Man, Shiv Prasad SR, Chellani HK, Kapani V. Intra-uterine growth curves in North Indian babies: weight, length, head circumference and ponderal index. Indian Pediatr 1990; 27: 43–51. 23 Osmond C, Barker DJP, Winter PD, Fall CHD, Simmonds SJ. Early growth and death from cardiovascular disease in women. BMJ 1993; 307: 1519–24. 24 Barker DJP, Osmond C, Simmonds SJ, Wield GA. The relation of small head circumference and thinness at birth to death from cardiovascular disease in adult life. BMJ 1993; 306: 422–26. 25 Rich Edwards J, Stampfer M, Manson J, et al. Birthweight, breastfeeding and the risk of coronary heart disease in the Nurses Health Study. Am J Epidemiol 1995; 141: S78.

1273