Secondary Logo

Journal Logo

Original Article

Breastfeeding During Infancy and the Risk of Cardiovascular Disease in Adulthood

Rich-Edwards, Janet W.*†‡; Stampfer, Meir J.†‡§; Manson, JoAnn E.†‡∥; Rosner, Bernard‡¶; Hu, Frank B.†‡§; Michels, Karin B.*†‡; Willett, Walter C.†‡§

Author Information
doi: 10.1097/
  • Free


Human breast milk differs from infant formula in several ways that affect the growth and lipid profiles of human infants.1 Breast milk is low in protein and high in fat when compared with cow's milk and high in cholesterol when compared with commercial infant formula.2 Furthermore, human milk contains hormones and growth factors that may prove to be important to development of the cardiovascular system.3,4

Infant breastfeeding has been associated with lower levels of cardiovascular risk factors in childhood. In developed countries, breastfed infants are leaner throughout childhood and adolescence1,5–7 than babies fed cow's milk or commercial infant formulas. Some studies,8–10 including 1 randomized trial of breastfeeding among preterm infants,8 have demonstrated lower blood pressures among children who were breastfed, whereas others have detected no such associations.11 Breastfed infants have lower insulin levels than bottlefed infants.12 Although blood levels of cholesterol are higher among breastfed infants than formula-fed infants,13–15 this difference disappears with cessation of breastfeeding16–18 and may change to favorable lipid profiles in adolescence.19–21

It is unclear whether the generally advantageous cardiovascular profile in breastfed children persists into adulthood. Studies have been inconsistent with regard to breastfeeding history and levels of adult adiposity22–24 and serum cholesterol21,23–27 and have found no association with regard to blood pressure.23,26,28,29 Breastfeeding may confer better glucose tolerance23,26 and reduced risk of diabetes mellitus.22 On the other hand, a recent study indicated a troubling dose-response association between increased length of breastfeeding and reduced arterial distensibility in young adults.27

Only 2 studies have investigated whether method of infant feeding is associated with the risk of cardiovascular disease itself. Mortality from coronary heart disease among Englishmen born between 1911 and 1930 was higher among those exclusively breastfed or exclusively bottlefed for a full year, compared with those weaned from the breast in the first year of life.30 In the United States, follow up of a cohort of gifted children born between 1904 and 1915 revealed no associations between breastfeeding history and cardiovascular mortality.31 We investigated the association of having been breastfed with fatal and nonfatal coronary heart disease and stroke among 87,252 American women born between 1921 and 1946.


Study Design

The Nurses’ Health Study is a prospective cohort study established in 1976. More than 121,000 married female registered nurses, ages 30 to 55 years, initially responded to mailed questionnaires regarding their medical histories and lifestyles. Details have been published elsewhere.32,33 In 1992, participants were asked whether they were breastfed as infants (yes, no, or not sure) and, if so, the duration of breastfeeding (≤3 months, 4–8 months, ≥9 months, and duration unknown). Because there were relatively few women breastfed for ≤3 and 4–8 months, we combined these 2 groups for most of our analyses. In 1992, participants also reported their ethnicity, their birthweight, and whether they were full-term, 2 or more weeks premature, or a twin or triplet. Biennial questionnaires from 1992 until 2000 collected updated information on age, weight, cigarette smoking, alcohol consumption, use of oral contraceptives and postmenopausal hormones, menopausal status, hypertension, diabetes mellitus, and elevated cholesterol level. Periodically, we updated physical activity and parental history of coronary heart disease. For this analysis, we used energy-adjusted saturated fat consumption in 1990, which was measured by semiquantitative food frequency questionnaire. The 1976 baseline questionnaire collected information on adult height and parental occupation when the participant was age 16 years.

Validation of Self-Reported Breastfeeding History

The validity of self-reported breastfeeding history in infancy was evaluated in a sample of 3515 participants whose mothers were mailed surveys regarding the infancy and early childhood of their daughters. Compared with their mothers’ self-report, the sensitivity of the daughter participants’ self-report of ever having been breastfed was 71% and specificity was 90%. For duration of breastfeeding, the Spearman correlation coefficient between mother and daughter reports was 0.60. The accuracy of self-reported breastfeeding history was somewhat less than that of participants of the Nurses’ Health Study II, who were younger.34 Another study has found that, although errors in maternal recall tend to be correlated with errors in daughters’ recall, maternal recall of breastfeeding is reasonably accurate when compared with health records, even after 20 years.19

End Point Documentation

The end points for our analysis were incident coronary heart disease (nonfatal myocardial infarction or fatal coronary heart disease) and stroke, occurring between the 1992 questionnaire and June 1, 2000. We sought permission to review the medical records from the participants who reported a diagnosis of nonfatal coronary heart disease or nonfatal stroke on follow-up questionnaires. Physicians who were unaware of the risk-factor status of participants conducted systematic reviews of the records. Nonfatal myocardial infarctions were confirmed if they met the criteria of the World Health Organization: symptoms of myocardial infarction plus either diagnostic electrocardiographic changes or elevated cardiac-enzyme levels.35 We designated as probable events those infarctions reported by participants that required hospitalization and were corroborated by additional information from letter or telephone interview but for which we could not obtain hospital records. Nonfatal stroke was classified according to criteria established by the National Survey of Stroke36 and required evidence of a typical neurologic deficit of sudden or rapid onset that persisted for more than 24 hours. Strokes were further subclassified as hemorrhagic (subarachnoid or intraparenchymal), ischemic (thrombotic or embolic), or of unknown cause.

Deaths were identified from state vital records and the National Death Index, or were reported by the subject's family or postal authorities. Mortality follow up is more than 98% complete using these methods.11 Confirmed deaths from coronary heart disease were those classified in hospital records or autopsy as having been caused by myocardial infarction, or those for which coronary heart disease was listed as the cause of death and evidence of previous coronary disease was available. We designated as probable coronary heart disease those cases in which coronary heart disease was the underlying cause on the death certificate and for which no records were available. We also included as coronary heart disease any sudden deaths with no other plausible cause. Death resulting from stroke was documented by physician review of hospital records, autopsy reports, or the listing of stroke as the underlying cause on the death certificate. We combined confirmed (n = 864) and probable (n = 485) cases of coronary heart disease, and confirmed (n = 696) and probable (n = 492) strokes. The results did not differ appreciably when only confirmed cases were considered.

We examined the incidence of coronary heart disease and stroke separately, and then combined them as a total cardiovascular disease end point. We also examined ischemic and hemorrhagic strokes separately. In addition, we estimated hazard ratios for a combined ischemic cardiovascular disease end point that included coronary heart disease and ischemic stroke. When a participant had more than 1 cardiovascular event from 1992 to 2000 (eg, coronary heart disease and stroke), only the first event was included in our analyses. We examined the association between having been breastfed and the risk of first self-reported physician diagnosis of hypertension from 1992 to 2000. We also include analyses of the nonfatal incident cardiovascular disease that occurred between the cohort's inception in 1976 and the questionnaire regarding breastfeeding in 1992, as well as the combined retrospective and prospective data from 1976 to 2000.

Statistical Analysis

The main prospective analysis includes the 87,252 women who reported their history of having been breastfed on the 1992 questionnaire and who had no history of major coronary heart disease or stroke before 1992. Women who died before 1992 (n = 5691, including 1032 cardiovascular deaths) provided no information on breastfeeding and, therefore, could not be included in any analysis. Women who experienced a cardiovascular end point before 1992 were excluded from the main prospective analysis (n = 2128). We also excluded from all analyses women who did not respond to the 1992 questionnaire (n = 11,792), who responded to a shortened version of the questionnaire that lacked the breastfeeding questions (n = 14,271) or who skipped over the breastfeeding questions (n = 567).

Age-adjusted means and prevalence rates of baseline characteristics were obtained by direct standardization to the overall age distribution in years.37 We used a Cox proportional hazards model to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) adjusted for age in months and for changing cardiovascular risk factors in adulthood.

Estimates of the association of breastfeeding with coronary heart disease and stroke, adjusted for age and smoking and additionally corrected for misclassification in self-reported breastfeeding history, are also presented. These were derived by the method of Zucker and Spiegelman.38


Prevalence of Breastfeeding

The nurses in this cohort were born between 1921 and 1946. Eighty-four percent of the nurses reported whether or not they had been breastfed; of these, the majority (64%) said they had been breastfed. The prevalence of breastfeeding declined markedly during this time period, when the use of evaporated cow's milk (usually modified by adding corn syrup and water) as a breastmilk substitute was increasing.39 Women in our cohort born before 1930 were almost twice as likely to have been breastfed compared with women born after 1940 (79% vs. 44%).

Of the breastfed women, 42% reported the duration of breastfeeding. Twenty-two percent of those women stated that they had been breastfed for up to 3 months, 34% for 4 to 8 months, and 44% for at least 9 months.

Cardiovascular Risk Factors by Breastfeeding History

Table 1 shows the age-adjusted prevalence of cardiovascular risk factors by breastfeeding history. Older nurses were more likely to have been breastfed longer. Preterm and low-birthweight infants were less likely to have been breastfed and were weaned earlier than the full-term, normal birthweight infants. Adolescent socioeconomic status, as reflected in the percentage of nurses whose fathers were professionals or managers, was higher among those who were not breastfed and was inversely associated with breastfeeding duration.

Age-Adjusted Characteristics of Study Population, by History of Having Been Breastfed

Few adult cardiovascular risk factors varied by breastfeeding history. Women who had been breastfed were slightly less likely to smoke and to drink alcohol than women who were not breastfed. There appeared to be no important differences between women who were or were not breastfed in the other risk factors examined.

Coronary Heart Disease

From 1992 until 2000, 1099 cases of coronary heart disease occurred among women who reported whether or not they had been breastfed, including 344 fatal and 755 nonfatal cases. The age-adjusted hazard ratio for coronary heart disease was 0.87 (95% CI = 0.77–1.00) for women who were breastfed compared with women who were not breastfed (Table 2). Adjustment for cigarette smoking and birthweight dampened the hazard ratio for breastfed women to 0.92 (0.80–1.05). The addition of adult BMI to the multivariate model made no meaningful change to the hazard ratio for breastfeeding (adjusted HR = 0.91; CI = 0.80–1.04).

Hazard Ratios and 95% Confidence Intervals for Coronary Heart Disease and Stroke Reported Prospectively (1992–2000), by History of Having Been Breastfed

We then examined duration of breastfeeding. We found a small decrease (-7%, -23–13%) in risk of coronary heart disease among women who were breastfed less than 9 months and a slightly larger decrease (-16%, -31–3%) among women breastfed at least 9 months (Table 2).


There were 940 strokes, of which 142 were fatal and 798 were nonfatal. Women who were breastfed had an age-adjusted hazard ratio for total stroke of 0.89 (0.77–1.02) compared with those who were not breastfed (Table 2). Adjustment for smoking and birthweight somewhat dampened these associations, whereas further adjustment for BMI made no material difference to the hazard ratios.

There was no dose-response association between duration of breastfeeding and risk of stroke (Table 2). Although there was an 18% decrease in risk among those breastfed less than 9 months, having been breastfed for at least 9 months was not associated with risk of stroke.

We also examined stroke subtypes (Table 2). Although there was a 14% reduction in risk of ischemic stroke among breastfed women, the reduction was evident only for those breastfed less than 9 months. There was no apparent association between breastfeeding history and risk of hemorrhagic stroke.

Total Cardiovascular Disease

Combining coronary heart disease and stroke, the adjusted hazard ratios were 0.91 (0.83–1.01) for all breastfed women, 0.88 (0.76–1.02) for those breastfed for less than 9 months, and 0.92 (0.79–1.06) for women breastfed for at least 9 months.

We also examined a combined end point of ischemic cardiovascular events, which included coronary heart disease and ischemic stroke. Breastfed women had an adjusted hazard ratio of 0.91 (0.81–1.01) for ischemic events compared with women who had not been breastfed.

Further adjustment for prematurity, multiple birth, and socioeconomic status in childhood (represented by parental occupations) did not alter the associations between breastfeeding and total cardiovascular disease. Similarly, further adjustment for physical activity, alcohol intake, consumption of saturated fat, parental history of coronary heart disease before age 60, menopausal status and the use of postmenopausal hormones, and race/ethnicity did not materially alter these associations. Adjustment for the potential intermediate variables of hypertension, elevated cholesterol, BMI, or diabetes substantially changed the observed associations. Adjustment for year of birth made no difference to estimated hazard ratios, indicating that there was no cohort effect.

Because it had been suggested by animal experiments that atherosclerosis may be advanced in breastfed infants weaned to a high-fat diet,40 we stratified by quintile of saturated fat intake in adulthood. We found no evidence that the association of breastfeeding and cardiovascular disease varied by saturated fat intake (hazard ratios for breastfeeding by ascending quintile of fat intake were: 0.88, 0.85, 1.04, 0.99, and 0.88).

Retrospective Analysis of Nonfatal Events

Many nonfatal events occurred between the inception of the cohort in 1976 and the 1992 questionnaire that documented breastfeeding history. This enabled us to examine retrospectively 898 nonfatal coronary heart disease and 595 nonfatal strokes that were not included in the prospective analysis reported here. The adjusted hazard ratio from the retrospective analysis for coronary heart disease was similar to that from the prospective analysis: 0.89 (0.76–1.05) for breastfed women. There was no suggestion of a protective effect of breastfeeding in the retrospective stroke analysis, with an adjusted hazard ratio for nonfatal stroke of 1.08 (0.88–1.33) for breastfed women.

Figure 1 depicts the adjusted hazard ratios and 95% confidence intervals for the association of breastfeeding with total cardiovascular disease from 1976 to 2000, as reported retrospectively and prospectively. The greater numbers of end points permitted us to examine separately those breastfed less than 4 months and those breastfed 4–8 months. The adjusted risk of cardiovascular disease decreased 1% (0–2%) for every month breastfed.

Multivariate hazard ratios (solid boxes) and 95% confidence intervals (vertical bars) for cardiovascular disease, represented retrospectively or prospectively (1976–2000), by history of having been breastfed and duration of breastfeeding. Adjusted for birthweight and adult cigarette smoking.


A strong association has been reported between increasing duration of breastfeeding (beyond 4 months) and decreasing arterial distensibility among young adults.27 We therefore examined whether breastfeeding was associated with adult hypertension in our cohort. From 1992 to 2000, there were 13,841 cases of newly diagnosed hypertension reported among women without a history of hypertension. There was no association between breastfeeding history and incident hypertension. Compared with those who were never breastfed, the age-adjusted hazard ratios of hypertension were 1.00 (0.96–1.04) for all breastfed women, 0.98 (0.93–1.04) for those breastfed less than 9 months, and 1.02 (0.96–1.09) for those breastfed 9 months or more. Further adjustment for birthweight, smoking, and BMI did not change these hazard ratios.


In this large cohort of women born between 1921 and 1946, there was inconsistent evidence of a weak reduction in risk of cardiovascular disease among women who had been breastfed as infants. We observed hazard ratios for breastfed women that were consistently less than 1, suggesting a lower risk of ischemic events among women who had been breastfed. However, the associations were modest in magnitude (on the order of a 10% reduction in risk), generally lacked statistical significance, and provided no consistent dose-response pattern. Such weak associations could be explained by chance, by confounding factors we did not measure, or by residual confounding from factors that we measured imperfectly.

On the other hand, random misclassification of self-reported breastfeeding status, especially breastfeeding duration, could have obscured stronger protective associations of breastfeeding history with cardiovascular disease. We had validation data from the comparison of nurse self-reports to their mothers’ reports of whether or not the nurse had been breastfed. Using the method of Zucker and Spiegelman,38 we were able to correct the multivariate hazard ratios and confidence intervals for misclassification of breastfeeding exposure. Correcting for exposure misclassification, the multivariate hazard ratio for coronary heart disease was 0.77 (95% CI = 0.59–0.98) for breastfed women compared with bottlefed women. The corrected hazard ratio for stroke was 0.71 (0.55–0.92). These corrected estimates suggest a much stronger protective association of breastfeeding for cardiovascular disease than the uncorrected estimates, reflecting the modest sensitivity (71%) of self-reported breastfeeding status. Future studies that are able to examine recorded, rather than self-reported, breastfeeding history may demonstrate important associations between breastfeeding in infancy and reduced cardiovascular disease in adulthood.

Only 2 previous studies have examined the association of breastfeeding history and adult cardiovascular disease outcomes. In a retrospective follow-up study of 5471 men born from 1911 to 1930 in Hertfordshire, England, an association between infant feeding and mortality from ischemic heart disease was observed.30 Men who had been exclusively breastfed for more than 1 year and men who had been exclusively bottlefed had higher standardized mortality ratios for ischemic heart disease when compared with breastfed men weaned before 1 year or men who received mixed breast- and bottlefeeding for the first year of life. Because these data were drawn from infant health records and death certificates, it was not possible to control for such factors as adult cigarette smoking. On the other hand, the collection of breastfeeding history from infant records minimized misclassification of breastfeeding exposure in this study. The pattern of results obtained in the Hertfordshire study is not easily interpreted: it is not clear what biologic mechanism would lead to higher ischemic heart disease among those exclusively breastfed and those exclusively bottlefed for the first year, in comparison with those who were fed both by breast and by bottle before age 1. In our data, particularly after control for confounding factors, there was no evidence of a protective effect of breastfeeding for less than 9 months.

The Terman Life Cycle Study enrolled more than 1500 gifted boys and girls between 1904 and 1915 in California; 647 men and 523 women from this cohort whose breastfeeding history was known were followed for mortality through 1986.31 The 127 deaths from cardiovascular disease in the Terman study were not associated with breastfeeding history or with duration of breastfeeding for either men or women. Although breastfeeding history was recalled by parents when the children were enrolled at age 11 years, it was undoubtedly more precise and complete than our study. Like the Hertfordshire study, the Terman study was structured to examine only fatal cardiovascular disease and was unable to control for adult risk factors.

Our study was considerably larger than either of the previous studies, enabling us to examine separately coronary heart disease, stroke, and stroke subtypes. We were also able to examine both nonfatal and fatal events. Because the Nurses’ Health Study was primarily designed to investigate the relation of adult behaviors with the risk of cancer and cardiovascular disease, we had good information on adult risk factors that might have acted as confounders (such as smoking) or intermediate variables (such as BMI, diabetes, hypertension, and hypercholesterolemia).

We lacked information on infant feeding history for a large percentage of the cohort, including all participants who died of cardiovascular disease before the 1992 questionnaire. This would have biased the hazard ratios only if the association between breastfeeding and cardiovascular disease was different among those who did not know their breastfeeding status or who died of cardiovascular disease at an early age. Several findings argue against this possibility: we observed similar results for fatal and nonfatal disease and for the retrospective (1976–1992) and prospective (1992–2000) analyses; and the hazard ratios for breastfeeding history (as a yes/no variable) were consistent for women who did not know the duration of their breastfeeding. Another limitation is that most of the participants of the Nurses’ Health Study are white, reflecting the composition of the nursing population in 1976. However, adjustment for race/ethnicity did not alter findings, and we know of no reason why these results would not be generalizable across ethnic groups.

In our study, as in the 2 previous studies, we do not know what was in the bottles of those who were not breastfed. Changing fashions in breastmilk substitutes over the course of the century may account for some differences in findings between studies. Furthermore, it is not clear whether these findings would pertain to infants fed modern commercial infant formulas.

In summary, breastfeeding, especially prolonged breastfeeding, was associated with a small reduction in the risk of total cardiovascular events. If this reflects a causal effect of breastfeeding, even a 10% to 20% reduction in risk of coronary heart disease and ischemic stroke among breastfed infants would have important implications for cardiovascular disease prevention in both the developed and developing world. However, further studies with documented infant feeding history and long-term follow up of adults are needed before we can add cardiovascular disease prevention to the list of the health benefits provided by breastfeeding.


We are grateful to the statistical advice and programming of Drs. David Zucker and Donna Spiegelman, who generated the multivariate risk estimates corrected for exposure misclassification. We are also thankful for the expert programming assistance of Ms. Eileen Hibert.


1. Hamosh M. Does infant nutrition affect adiposity and cholesterol levels in the adult? J Pediatr Gastroenterol Nutr. 1988;7:10–16.
2. Hodgson PA, Ellefson RD, Elveback LR, et al. Comparison of serum cholesterol in children fed high, moderate, or low cholesterol milk diets during neonatal period. Metabolism. 1976;25:739–745.
3. Koldvosky O, Thornburg W. Hormones in milk. J Pediatr Gastroenterol Nutr. 1987;6:172–196.
4. Golding J, Emmett PM, Rogers IS. Does breast feeding have any impact on non-infectious, non-allergic disorder? Early Hum Dev. 1997;49(suppl):S131–S142.
5. Kramer MS. Do breast-feeding and delayed introduction of solid foods protect against subsequent obesity? J Pediatr. 1981;98:883–887.
6. von Kries R, Koletzo B, Sauerwald T, et al. Breast feeding and obesity: cross sectional study. BMJ. 1999;319:147–150.
7. Gillman MW, Rifas-Shiman SL, Camargo CA, et al. Risk of overweight among adolescents who were breast fed as infants. JAMA. 2001;285:2461–2467.
8. Singhal A, Cole TJ, Lucas A. Early nutrition in preterm infants and later blood pressure: two cohorts after randomized trials. Lancet. 2001;357:413–419.
9. Taittonen L, Nuutinen M, Turtinen J, et al. Parental and postnatal factors in predicting later blood pressure among children: cardiovascular risk in young Finns. Pediatr Res. 1996;40:627–632.
10. Wilson AC, Forsyth JS, Greene SA, et al. Relation of infant diet to childhood health: seven year follow up of cohort of children in Dundee infant feeding study. BMJ. 1998;316:21–25.
11. Rich-Edwards JW, Corsano KA, Stampfer MJ. Test of the National Death Index and equifax Nationwide Death Search. Am J Epidemiol. 1994;140:1016–1019.
12. Lucas A, Adrian TE, Blackburn AN, et al. Breast vs bottle: endocrine responses are different with formula feeding. Lancet. 1980;i:1267–1269.
13. Viikari J, Akerblom HK, Rasanen L, et al. Cardiovascular risk in young Finns. Acta Paediatr. 1990;365:13–19.
14. Jooste PL, Rossouw LJ, Steenkamp HJ, et al. Effect of breast feeding on the plasma cholesterol and growth of infants. J Pediatr Gastroenterol Nutr. 1991;13:139–142.
15. Kallio MJT, Salmenpera L, Simes MA, et al. Exclusive breast-feeding and weaning: effect on serum cholesterol and lipoprotein concentrations in infants during the first year of life. Pediatrics. 1992;89:663–666.
16. Freedman DS, Srinivasan SR, Cresanta JL, et al. Serum lipids and lipoprotein. Pediatrics. 1987;80:789–796.
17. Famon S, Roger RR, Ziegler EE, et al. Indices of fatness serum cholesterol at age eight years in relation to feeding and growth during early infancy. Pediatr Res. 1984;18:1233–1238.
18. Huttunen JK, Saarinen UM, Kostianen E, et al. Fat composition of the infant diet does not influence subsequent serum lipid levels in man. Atherosclerosis. 1983;46:87–94.
19. Kark JD, Troya G, Friedlander Y, et al. Validity of maternal reporting of breast feeding history and the association with blood lipids in 17 years olds in Jerusalem. J Epidemiol Community Health. 1984;38:218–225.
20. Bergstrom E, Hernell O, Persson LA, et al. Serum lipid values in adolescents are related to family history, infant feeding, and physical growth. Atherosclerosis. 1994;117:1–13.
21. Owen CG, Whincup PH, Odoki K, et al. Infant feeding and blood cholesterol: a study in adolescents and a systematic review. Pediatrics. 2002;110:597–608.
22. Pettitt DJ, Forman MR, Hanson RL, et al. Breastfeeding and incidence of non-insulin-dependent diabetes mellitus in Pima Indians. Lancet. 1997;350:166–168.
23. Ravelli ACJ, Van der Meulen JHP, Osmond C, et al. Infant feeding and adult glucose tolerance, lipid profile, blood pressure, and obesity. Arch Dis Child. 2000;82:248–252.
24. Marmot MG, Page CM, Atkins E, et al. Effect of breast-feeding on plasma cholesterol and weight in young adults. J Epidemiol Community Health. 1980;34:164–167.
25. Kolacek S, Kapetanovic T, Zimolo A, et al. Early determinants of cardiovascular risk factors in adults. A. Plasma lipids. Acta Paediatr. 1993;82:699–704.
26. Fall CH, Osmond C, Barker DJP, et al. Fetal and infant growth and cardiovascular risk factors in women. BMJ. 1995;310:428–432.
27. Leeson CPM, Katternborn M, Feanfield JE, et al. Duration of breast feeding and arterial distensibility in early adult life: population based study. BMJ. 2001;322:643–647.
28. Kolacek S, Kapetanovic T, Luzar V. Early determinants of cardiovascular risk factors in adults. B. Blood pressure. Acta Paediatr. 1993;82:377–382.
29. Whincup PH, Cook DG, Shaper AG. Early influences on blood pressure: a study of children aged 5–7 years. BMJ. 1989;299:587–591.
30. Fall CHD, Barker DJP, Osmond C, et al. Relationship of infant feeding to adult serum cholesterol concentration and death from ischaemic heart disease. BMJ. 1992;304:801–805.
31. Wingard DL, Criqui MH, Edelstein SL, et al. Is breast feeding in infancy associated with adult longevity? Am J Public Health. 1994;84:1458–1462.
32. Hennekens CH, Speizer FE, Rosner B. Use of permanent hair dyes and cancer among registered nurses. Lancet. 1979;1:1301–1303.
33. Stampfer MJ, Willet WC, Colditz GA. A prospective study of postmenopausal estrogen therapy and coronary heart disease. N Engl J Med. 1985;313:1044–1049.
34. Troy LM, Michels KB, Hunter DJ, et al. Self-reported birthweight and history of having been breast-fed among younger women: an assessment of validity. Int J Epidemiol. 1996;25:122–127.
35. Rose GA, Blackburn H. Cardiovascular Survey Methods, 2nd ed. Geneva: World Health Organization, 1982. (WHO monograph series No 58.)
36. Walker AE, Robins M, Weinfeld FD. The National Survey of Stroke: clinical findings. Stroke. 1981;12(suppl I):I-13–I-44.
37. Kleinbaum D, Kupper L, Morgenstern H. Epidemiologic Research: Principles and Quantitative Methods. Belmont, CA: Lifelong Learning Publications; 1982;336:
38. Zucker DM, Spiegelman D. A simple relative risk estimator for survival data with an imperfectly measured binary covariate. Biometrics. In press.
39. Fomon SJ. Infant Nutrition, 2nd ed. Philadelphia: WB Saunders; 1974.
40. Mott GE, Lewis DS, McGill HC. Programming of cholesterol metabolism by breast or formula feeding. The childhood environment and adult disease. Ciba Foundation Symposium. 1991;156:56–76.
© 2004 Lippincott Williams & Wilkins, Inc.