Main

Barker et al. reported a negative correlation between birth weight and subsequent BP in six populations of different ages(1–4). They also suggested that the effect of birth weight on later BP amplifies from childhood to late adulthood(4). In their studies among children, they found the mean decrease in the subsequent BP to be 2.8 mm Hg for every 1-kg increase in birth weight(4) or 1 mm Hg for every SD increase in the ponderal index at birth(1), or to vary from 0.4 to 1.3 mm Hg when birth weight increased from the lowest to the highest tertile(2). In adult populations the BP decrease was somewhat greater, varying from 5.2 to 11 mm Hg for every 1-kg increase in birth weight(3, 4) or being 2.6 and 1.8 mm Hg among the female and male subjects, respectively, when birth weight increased from the lowest to the highest tertile(2). In addition to Barker's team, other groups have reported similar results(5–12). Two of these studies were made in the United Kingdom on the same study population(British Medical Research Council's national survey) that Barker et al. also used twice in their reports(5, 6), and four of them were made outside the United Kingdom(9–12). A lack of correlation between birth weight and later BP has been reported especially among children and young adults in six different populations(13–18). On the other hand, no positive correlation between birth weight and BP has been published.

Much of the discussion about the inverse relation between birth weight and BP has been based on Barker's results, but further studies are needed to prove whether there exists a true relationship between these two variables. We therefore wanted to analyze the relationship between birth weight and subsequent BP and to determine whether the relation amplifies throughout life among healthy children and adolescents. Apart from birth weight, several other pre- or postnatal factors may affect subsequent BP, and we therefore decided to evaluate the effects of the gestational age, the birth rank order, the duration of breast feeding, the mother's age, and the mother's high blood pressure during pregnancy on subsequent BP of the child.

METHODS

This study is part of a project titled “Cardiovascular Risk in Young Finns,” the protocol for which has been described in detail elsewhere(19, 20). The work was carried out in five Finnish university cities and their rural surroundings. The original population consisted of children and adolescents aged 3, 6, 9, 12, 15, and 18 y, who were followed for 6 y. A total of 3596 participants were examined in 1980 and 2991 subjects in 1983, and 2799 reported for the follow-up in 1986. The study plan was accepted by the ethical committees of all the participant universities, and all subjects gave informed consent for the study.

In each survey, the subjects filled out, together with their parents, a detailed questionnaire concerning their state of health and background. The questionnaires included information on the birth weight, the birth rank order, the duration of gestation, the duration of breast feeding, the mother's age at the time of the child's birth, and the mother's possible high BP during any pregnancy. The given information was checked by the study nurses during the day of physical examination, and the subjects were asked to bring with them their records from the well-baby clinics which include the birth weight data.

In the physical examination, BP was measured with an ordinary mercury sphygmomanometer in the first two surveys and with a random zero sphygmomanometer in the last survey(21). Korotkoff's first phase (K1) was used as a sign of systolic BP and Korotkoff's fifth phase(K5) as a sign of diastolic BP(22). The weight and height of the subjects were also measured, and pubertal development was evaluated according to the Tanner classification (pubic hair, P1-5) in each survey(23). In the last survey Tanner stage P5 was further divided into three subclasses (P5, P5 + 3 y, P5 + 6 y), depending on how many years the subject had been classified into this Tanner stage class.

The mean birth weight of the girls who participated in the first and in the last surveys was 3495 g (SD 554) compared with 3442 g (SD 513) of those lost from follow-up. The corresponding values for boys were 3512 g (SD 542) and 3590 g (SD 574), respectively. All of the statistical analyses were performed separately for girls and boys. In some analyses the subjects were divided into tertiles according to birth weight and means of BP, and weight-adjusted means of BP in each tertile of birth weight were calculated. Analyses of variance and covariance were used to determine the differences between the classified variables. The effect of birth weight on future BP was evaluated with correlation analysis and multiple linear regression analysis. The effects of pre- and postnatal factors on subsequent BP were described as mean differences from the baseline and in terms of multiple regression analysis.

RESULTS

The mean birth weight was 3450 g (SD 522) among the girls and 3575 g (SD 568) among the boys (Fig. 1 andTable 1). The mean birth weight ranged from 1300 to 3245 g in the lowest, from 3250 to 3645 g in the middle, and from 3650 to 5450 g in the highest tertile among girls and from 1040 to 3360 g in the lowest, from 3370 to 3800 g in the middle, and from 3820 to 5750 g in the highest tertile among boys.

Figure 1
figure 1

Distribution of birth weight and division of birth weight values into the lowest (I), middle (II), and highest (III) tertile categories among female and male subjects.

Full size image

The mean BPs were calculated for each tertile of birth weight(Table 2). According to an analysis of variance, the mean systolic and diastolic BPs did not differ between the birth weight categories(p values varying from 0.06 to 0.88). After adjustment for current weight, however, the mean systolic BPs did differ between the birth weight categories among both sexes, and the mean diastolic BPs differed among the boys (p values ≤0.0 1) in such a way that the highest BPs were found in the lowest birth weight categories (Table 2).

Table 2 Mean systolic and diastolic BP (mm Hg) in 1986 divided into tertiles according to birth weight
Full size table

When the correlation between birth weight and systolic BP was analyzed in each of the three surveys, a weak but statistically significant negative correlation was found after standardization for current weight in the last survey among girls (regression coefficient -2.05) and in each three surveys among boys (regression coefficients varying from -1.84 to -2.50)(Table 3). Birth weight and diastolic BP correlated significantly only in the last survey among boys (regression coefficient,-1.11) (Table 3).

Table 3 Correlation (r) between birth weight(kg) and BP (mm Hg) and regression coefficients (b) from a multiple regression model in 1980, 1983, and 1986
Full size table

The age range in our population in 1986 varied from 9 to 24 y. We were therefore able to analyze the effect of puberty on the relation between birth weight and BP. A significant negative correlation between birth weight and systolic BP was found after adjustment for current weight at the Tanner stage P5 + 3 y among both sexes (regression coefficients -2.36 and -2.83). No significant correlations were found between birth weight and diastolic BP at the different stages of puberty.

The mean adjusted systolic BPs were calculated in the three birth weight categories, taking into account gestational age, the duration of breast feeding, and the history of mother's high BP during pregnancy(Table 4). In the full-term birth group, the mean systolic BP decreased from the lowest to the highest birth weight category(Table 4). Systolic BP also tended to decrease in relation to the duration of breast feeding among both sexes, the mother's age among the female subjects and the history of mother's high BP during pregnancy(Table 4). Furthermore, those with a history of mother's high BP during pregnancy had a higher systolic BP in every birth weight category compared with those whose mother's did not report high BP during pregnancy (Table 4). Systolic BP did not change uniformly from the lowest to the highest birth weight category in relation to the birth rank order or the mother's age among the male subjects or in the premature birth group.

Table 4 Mean systolic BP (mm Hg) adjusted for current weight in three classes of birth weight according to time of birth, duration of breast feeding, and maternal history of high BP during pregnancy in year 1986
Full size table

The mean changes in systolic BP in 1986 in the different classes of current weight and age, birth weight, duration of breast feeding, gestational age, birth rank order, mother's age, and history of mother's high BP during pregnancy were evaluated (Table 5). The strongest determinants of the change in systolic BP were current weight and age(Table 5). Furthermore, future systolic BP changed significantly from the baseline when the subject had been born more than 6 wk preterm among the girls (6.0 mm Hg; 95% CI, 0.4-11.6) and when the birth rank order was over four (2.7 mm Hg; 95% CI, 0.6-4.7 among the girls and 5.0 mm Hg; 95% CI, 2.2-7.8 among the boys). The mean systolic BP decreased with an increasing duration of breast feeding, being -4.5 mm Hg among the girls and-6.5 mm Hg among the boys (Table 5). Systolic BP also decreased with increasing birth weight, but not statistically significantly(-0.5 mm Hg; 95% CI, -2.1 to 1.1 among the girls and -1.3 mm Hg; 95% CI, -3.4 to 0.7 among the boys) (Table 5). Blood pressure was higher among the subjects whose mothers were over 40 y of age at the time of the child's birth (3.7 mm Hg; 95% CI, -0.5 to 7.9 among the girls and 2.9 mm Hg; 95% CI, -3.2 to 9.1 among the boys) and who had a history of mother's high BP during pregnancy, but these changes did not reach statistical significance(Table 5).

Table 5 Mean change of systolic BP (mm Hg) from the baseline according to current weight and age, birth weight, time of birth, duration of breast feeding, birth rank order, mother's age, and maternal history of high BP during pregnancy in 1986
Full size table

According to the multiple regression analysis, systolic BP correlated significantly with current weight, age, birth weight, and mother's high BP during pregnancy among both sexes and with birth rank order over four among the male subjects (Table 6).

Table 6 Multiple regression analysis of current weight and age, birth weight, time of birth, duration of breast feeding, birth rank order, mother's age, history of maternal high blood pressure during pregnancy, and systolic BP in year 1986
Full size table

The effect of weight, age, birth weight, time of birth, duration of breast feeding, birth rank order, mother's age, and history of mother's high BP during pregnancy on BP change during the 3- or 6-y follow-up was evaluated with the multiple regression analysis. The only factor that had a significant influence on BP change was mother's age.

DISCUSSION

The decrease in systolic BP when birth weight increased from the lowest to the highest tertile was significant only after standardization for current weight in accordance with the results of Whincup et al.(7, 8). In our study the magnitude of the decrease was about 2 mm Hg. In Barker's studies, the rise in BP upon decreasing birth weight was somewhat higher than in our study, but not always statistically significant(1–4). Four studies that support Barker's theory have been reported outside of the United Kingdom(9–12). Gennser et al.(9) studied 104 hypertensive and normotensive 28-y-old men in Sweden and found that the risk of high diastolic BP was higher among the men who had a history of low birth weight for gestational age. Kolaceket al.(10) studied the relation between birth weight and BP in 465 18-y-olds in Croatia and found an inverse correlation similar in strength to that shown by our regression model. Launer et al.(11) studied 476 4-y-old children in Holland and found a U-shaped relation instead of a linear one between birth weight and future BP. Hashimoto et al.(12) studied 195 Japanese children and found that children whose weight at birth were over 3520 g had 3 mm Hg lower mean systolic BP at 3 y of age than those whose weight at birth was 2990 g or less.

We suggest that the relation found between birth weight and systolic BP may have clinical significance among healthy children and adolescents, because our analyses showed statistically significant correlations after adjustment for current weight among female subjects in the last survey and among male subjects in each survey. The suggestion that birth weight is a weak regulator of future BP is supported by the studies in which no significant relation between birth weight and future BP has been found(13–18). Most of these studies have been made on children or young adults, and some have been criticized for not having taken into account the pubertal development of the subjects(24). We analyzed the relation between birth weight and BP at different pubertal stages and found significant correlations adjusted for current weight among the postpubertal (P5 + 3 y) female and male subjects. It seems that during puberty the inverse relation between birth weight and future systolic BP is weaker than after puberty.

In our study, an ordinary mercury sphygmomanometer was used to measure BP in 1980 and 1983 and a random zero sphygmomanometer in 1986. According to our results, diastolic BP and birth weight did correlate in the last survey among the male subjects only. It therefore seems that there is a relation between birth weight and diastolic BP among the boys but not girls, even when BP is measured with random zero sphygmomanometer, which has been reported to be more accurate in measuring BP(21).

Law et al.(4) hypothesized on the basis of their four populations of different ages that the effect of birth weight on future BP increases throughout life. A combination of correlation coefficients from different populations may, however, lead to false conclusions because of a possible cohort effect. According to the theory of Law et al. birth weight should also correlate with the subsequent change of BP in a follow-up study. Such a relation was found by Whincup et al.(25) in a study of 549 initially 5-7-y-old children who were examined again when they were 9-11 y old. Our results on BP changes over 3-6 y did not support this hypothesis.

Although there was a negative correlation between birth weight and subsequent BP, BP was also determined by several other factors such as current age and weight, the duration of breast feeding, and birth rank order over four presented as a mean difference from the baseline.

The relationship between birth rank order and BP has been reported to be of the same strength as the relation between the total sibship size and BP(8), and the effect of birth rank on subsequent BP has thus been suggested to be rather a postnatal effect. Our results show that the effect of birth rank order on subsequent systolic BP plays a role after the birth rank order of four.

The importance of breast feeding was first brought up by Barker, who found mortality from ischemic heart disease to be lower among adults who had been breast-fed compared with those who had been bottle-fed in infancy(26). Efforts have been made in the later studies to analyze the correlation between breast feeding and future BP, but no significant relation has been found(7, 10, 27). Nevertheless, we suggest on the basis of our results that the duration of breast feeding might decrease later BP or protect from an increase in BP.

It has been reported that offspring of women with hypertension during pregnancy have elevated BP in childhood compared with controls from normotensive pregnancies(7, 17), and that the difference in the children's systolic BP was 2 mm Hg between the groups of mothers with the lowest and highest pressures(2). Whincup et al.(8) found BP to be significantly higher in the subjects with a maternal history of high BP than in those without, and pointed out that whether or not maternal high BP was observed during the pregnancy made little difference for the outcome. They also found that the association between the maternal history of high BP and childhood BP was of similar strength compared with that of the paternal history. On the other hand, children's BP has earlier been found to relate more closely to the mother's than the father's BP and has been ascribed to X-linked genes(28). Our results show that the mean systolic BP did not differ significantly from the baseline. On the other hand, those with a history of mother's high BP during pregnancy had higher BPs in the different birth weight categories than those from normotensive pregnancies. According to multiple regression analysis, there was also a statistically significant positive correlation between the child's subsequent systolic BP and the mother's high BP during pregnancy. This is in accordance with the results of Hashimoto et al.(12) who reported an increase of 0.12 mm Hg in the children's systolic BP with each increment of 1 mm Hg in the systolic BP of their mothers.

In conclusion, our results based on healthy children and adolescents lend support to Barker's theory of low birth weight as a predicting factor for future BP. However, also other pre- and postnatal factors seem to be important determinants of subsequent BP.

Table 1 Baseline characteristics of the subjects in 1980
Full size table