Size at birth and blood pressure: cross sectional study in 8-11 year old childrenBMJ 1997; 314 doi: http://dx.doi.org/10.1136/bmj.314.7079.475 (Published 15 February 1997) Cite this as: BMJ 1997;314:475
- Stephanie J C Taylor (), honorary research fellow in public health medicinea,
- Peter H Whincup, senior lecturer in clinical epidemiologya,
- Derek G Cook, reader in epidemiologyb,
- Olia Papacosta, research statisticiana,
- Mary Walker, honorary research fellowa
- a Cardiovascular Research Group Department of Primary Care and Population Sciences Royal Free Hospital School of Medicine London NW3 2PF
- b Department of Public Health Sciences St George's Hospital Medical School London SW17 0RE
- Correspondence to: Dr Taylor
- Accepted 29 November 1996
Objective: To identify which patterns of fetal growth, represented by different measurements of size at birth, are associated with increased blood pressure in children aged 8-11 years.
Design and setting: School based, cross sectional survey conducted in 10 towns in England and Wales in 1994.
Subjects: 3010 singleton children (response rate 75%) with physical measurements and information on birth weight from parental questionnaires. Hospital birth records were examined for 1573.
Main outcome measures: Systolic and diastolic blood pressure at age 8-11 years.
Results: In the whole group birth weight was inversely related to systolic pressure (regression coefficient -1.48 mm Hg/kg; 95% confidence interval -2.20 to -0.76) after adjustment for current body size. There was no significant association between birth weight and diastolic pressure. The association with systolic pressure was much stronger in girls (-2.54 mm Hg/kg; -3.60 to -1.48) than in boys (-0.64 mm Hg/kg; -1.58 to 0.30), with a significant difference between the sexes (P=0.006). Among the other neonatal measures, head circumference and placental weight were inversely associated with subsequent blood pressure in girls, and placental ratio (placental weight:birth weight) was positively associated with blood pressure in boys. Neither ponderal index at birth nor length:head circumference ratio was related to blood pressure in either sex.
Conclusions: In these contemporary children the association between birth weight and blood pressure was apparent only in girls. There was no evidence that measures of size at birth, which may be related to nutrition at critical periods of pregnancy (thinness at birth or shortness in relation to head circumference), are related to blood pressure in the offspring.
This study confirms the well established inverse association between birth weight and childhood blood pressure, but there seems to be a strong difference between the sexes towards the end of the first decade of life—the association being strong in girls and weak or absent in boys
Only highly correlated, simple measures of size at birth seem to be associated with blood pressure in girls; more complex measures of size at birth, including ponderal index and shortness in association to head circumference, are not associated with childhood blood pressure in either sex at 8-11 years
The association between placental weight and placental ratio and blood pressure in children of 8-11 years is inconsistent
These findings do not support the hypothesis that fetal undernutrition “programmes” raise blood pressure in the second and third trimesters
Current body size is a much more important determinant of blood pressure in childhood than size at birth
It has been proposed that events occurring before birth may influence the risk of cardiovascular disease in later life.1 Associations between small size at birth and increased risk of adult cardiovascular mortality have been described,2 3 and there is strong evidence of an inverse association between birth weight and later blood pressure in both adults and children.4 The mechanism underlying this association is unknown, but fetal undernutrition at critical periods of intrauterine development may produce a permanent increase in blood pressure by a process known as “programming.”5 6 The “fetal origins” hypothesis suggests that undernutrition in the first, second, or third trimester may result in a symmetrically small baby; a thin, light baby; or a baby with a normal birth weight who is short in relation to its head circumference; and that these particular patterns of size at birth may be related to raised blood pressure.7 It has also been proposed that babies with large placentas, especially when associated with low birth weight, are at particular risk of hypertension in later life.8 Few studies, however, have actually examined the associations between different measures of size at birth or placental size and blood pressure in contemporary children. Here we present a population based study of these associations in 8-11 year old children.
Subjects and methods
Sampling procedures–Details of the study design and methods have been presented elsewhere.9 The study was based on 10 towns, five with exceptionally high adult cardiovascular mortality (Wigan, Burnley, Port Talbot, Rochdale, and Rhondda) and five with exceptionally low adult cardiovascular mortality (Esher, Chelmsford, Leatherhead, Bath, and Tunbridge Wells). In each town we selected a stratified random sample of 10 junior schools. Fifty children from the upper two years of each school were invited to attend for measurement.
Survey procedures–All the relevant local research ethics committees approved the study and informed, written parental consent was obtained. Low and high mortality towns were visited alternately between April and November 1994. All measurements were made by one of two pairs of trained field nurses.
Blood pressure and other physical measurements–Two seated blood pressure measurements were carried out on the right arm by using a Dinamap 1846SX automated oscillometric monitor (Critikon, United States),10 11 which was calibrated daily and compared with a mercury column sphygmomanometer weekly. There was no evidence of measurement drift during the study. A single cuff size was used to ensure the minimum cuff bladder width to arm circumference ratio recommended by the American Heart Association12 was met for 89% of the study population. The oldest 40% of the children were also asked to provide a blood sample after the application of local anaesthetic cream. Heights were measured to the last complete millimetre by using a portable stadiometer (CMS). Weights were measured to the last complete 0.1 kg by using an electronic weighing scale (Soehnle) with the children in light clothing without shoes. Blood samples were taken after the physical measurements were completed.
Parental questionnaire and birth records–Parents were sent a detailed questionnaire that included questions on the child's birth weight, place of birth, gestation, siblings, parents' occupations, and mother's height. Social class was coded from parental occupation by using the Office of Population Censuses and Surveys classification of occupations (1980). Consent to examine birth records was also requested, and these records were sought at every hospital in which 20 or more study children (5% of the study children in each town) had been born. Details extracted from birth records included weight, head circumference, and length at birth, placental weight, and the mother's antenatal blood pressures. Birth records were not sought for children born at home.
Statistical methods–All data were analysed with the SAS system statistical software package (SAS Institute, North Carolina). Ponderal index (weight/height3) was used as a measure of weight for height at 8-11 years because it was independent of both height and age in these data; it was also used as a measure of thinness at birth. Placental ratio (placental weight:birth weight) and shortness for head size (birth length:head circumference) were used when appropriate. Systolic and diastolic blood pressures were based on the average of two readings. Multiple linear regression was used to adjust associations between blood pressure and birth weight (as well as other indices) for structural and confounding variables. Body measures at birth and in childhood were included as continuous variables or fitted in fifths or thirds, when appropriate, by using linear regression techniques. All models included adjustment for sex, town, and who measured the blood pressure as categorical variables and age as a continuous variable. An adjustment for the slightly higher mean blood pressures seen in children anticipating blood tests (even after adjustment for age) was fitted as a categorical variable.
After invitation 3728 children participated in the study, and 3719 had a complete set of measurements (response rate 75%). Questionnaires including data on birth weight were returned for 3181 children (85% of participants); 3010 of these children were singleton births and only 39 (1.2%) were home births. Permission to examine birth records was granted for 3107 children (98% of those returning questionnaires), and, of these, 2350 children (76%) were born in local hospitals. Birth records for 1573 of the 2350 children described above were located. Subjects were excluded from analyses of placental weight and placental ratio if birth records stated that the placenta was incomplete (64 children). Table 1) shows the characteristics of the 3010 study children and of the subset with birth records retrieved. There were no significant differences between the mean ages, birth weights, anthropometric measurements, and blood pressures in these two groups. There were significant differences between the sexes: the boys were thinner and lighter at 8-11 years and were heavier, longer, and had greater head circumferences at birth. In the 1550 children with birth weights from both questionnaire and birth records the agreement between the two measures was close (mean (SD) difference 1.7 g (117.9 g); correlation coefficient 0.98). In 1508 (97.3%) of these children the difference between the two weights was less than 200 g.
Data on birth weight from parental recall and blood pressure
Figure 1) summarises the association between birth weight and blood pressure with the corresponding multiple linear regression coefficients in table 2). When we adjusted for age and sex alone there was no apparent association between birth weight and systolic blood pressure. Current body size, however, was a potential confounder of the association between birth weight and blood pressure (table 3)), both height and ponderal index being related not only to blood pressure but also to birth weight. When we made adjustments to birth weight:blood pressure for height and ponderal index an inverse association between birth weight and systolic blood pressure became apparent; there was a fall in systolic blood pressure at 8-11 years of about 1.5 mm Hg for every kilogram increase in birth weight (fig 1, table 2)). Even after we adjusted for current body size there was only a weak inverse association between birth weight and diastolic blood pressure, and this was not significant. Further adjustments for room temperature, number of siblings, maternal height, gestational age, maternal blood pressure during pregnancy, and social class did not alter these associations substantially (not shown).
Because the inverse association between birth weight and systolic blood pressure was apparent only after adjustment for height and ponderal index, all further analyses were adjusted for these two variables. There was no suggestion that the association between birth weight and blood pressure differed at different childhood heights (not shown). This remained true when the sexes were examined separately. The association did decrease slightly with increasing ponderal index, but this effect was seen only in the girls (not shown).
The association between birth weight and blood pressure seemed to be similar in children born at term and preterm (before 37 weeks' gestation), although in the smaller preterm group the association was not significant (table 2)). Analysis by sex revealed that the association seemed to be concentrated in the girls and very weak or absent in the boys. Test results for a sex difference in the association were significant for both systolic and diastolic pressure.
Data from birth records
Birth weight–Because of the clear sex difference in the association found in the whole dataset, data on the subset of children whose birth records were retrieved were analysed separately by sex. Table 4) presents the associations between the different birth measurements and blood pressure at 8-11 years, adjusted for height and ponderal index, along with the results of formal tests for differences in the regression coefficients (b) between the sexes. Again birth weight was inversely associated with blood pressure in the girls rather than the boys. The strength of the association was weaker in this subset, and this remained the case when these associations were re-examined in the same children by using their parentally recalled birth weights (b -1.20 mm Hg kg; 95% confidence interval -2.64 to 0.24 for systolic blood pressure in girls; -0.16 mm Hg kg; -1.44 to 1.12 for boys).
Head circumference and length at birth–Among the girls, head circumference at birth showed a significant inverse association with systolic blood pressure at 8-11 years and a weak, non-significant inverse association with diastolic blood pressure (table 4)). Head circumference and birth weight, however, were closely correlated (r=0.64; P<0.0001), and we could not differentiate their relative importance in relation to subsequent blood pressure in the girls. An increase of 1SD in either of these two measures (the standard regression effect) resulted in a fall in systolic blood pressure of about 1 mm Hg. Length at birth, which was also strongly correlated with birth weight (r=0.56; P<0.0001), also tended to be inversely associated with both systolic and diastolic blood pressure in the girls, although these associations were not significant. Among the boys, none of these simple birth measures was related to blood pressure, although formal evidence of sex interaction was weak (table 4)).
Ponderal index at birth and birth length:head circumference ratio–Neither of the two ratio birth measures examined (ponderal index and length:head circumference ratio) showed any consistent significant association with blood pressure in either sex.
Placental weight and placental ratio–Among the girls, placental weight showed a significant inverse association with systolic blood pressure, but it was also closely correlated with birth weight (r=0.58; P<0.0001), and the association between placental weight and blood pressure became non-significant when birth weight was included in the model (not shown). Placental ratio did not seem to be related to blood pressure in the girls. In the boys placental weight and placental ratio tended to be positively related to blood pressure; associations between placental weight and diastolic blood pressure and placental ratio and systolic blood pressure were significant despite the absence of an association between birth weight and blood pressure in the boys. The association between placental weight and blood pressure was unchanged by the addition of birth weight into the model (not shown). Tests for interaction suggested that the associations between placental weight and blood pressure were significantly different in the two sexes (table 4)).
Relative importance of current body size and birth weight on childhood blood pressure–We compared the relative strengths of the associations between systolic blood pressure and current size or birth weight by examining their standard regression effects. At 8-11 years an increase of 1SD in birth weight, childhood height, or childhood ponderal index was associated with a change in systolic blood pressure of -0.8 (-1.2 to -0.4), 3.4 (2.9 to 3.8), and 3.4 (3.1 to 3.8) mm Hg, respectively. Similar results were found when the unit of comparison was the interquartile range rather than the SD and also when these analyses were confined to the girls.
In this study we observed an inverse association between birth weight and systolic blood pressure, but this was concentrated in the girls and hardly apparent in the boys. The studies of data from birth records suggested that, among the girls only, three highly correlated measures—birth weight, head circumference, and placental weight—were inversely associated with blood pressure. Neither ponderal index at birth nor length:head circumference ratio showed any association with blood pressure in either sex. Placental weight, which was strongly correlated with birth weight in both sexes, was inversely associated with blood pressure in girls but tended to be positively associated in the boys, and placental ratio was also positively associated with blood pressure in boys.
This population based study included children with a wide range of birth weights and social circumstances. In agreement with the results of previous research13 the validity of parental recall in this study was good. Although the association between birth weight and blood pressure seemed slightly weaker in the subgroup with data from birth records when it was compared with that in the full group of study children, mean birth weights and physical measurements between the two groups were similar, suggesting that the selection of children with birth record data was not biased with respect to these variables. As we relied on birth measures taken in various hospitals, differences in measuring techniques will inevitably have introduced some error,14 which should be predominantly random. The consistent overestimation of systolic blood pressure by the Dinamap,11 although influencing comparisons with other studies, should not have affected estimates of differences within the study population.
Relations to other studies
Birth weight and blood pressure–As in other studies of children,4 adjustment for the effect of current size was necessary before the association between birth weight and blood pressure became apparent. This adjustment is justified in children because while childhood body size seems to confound the association, in adulthood this confounding is less apparent.8 The strength of the association between birth weight and blood pressure seen in the present study was weaker than that in our earlier study of children of a similar age15 and of comparable strength to those in 5-7 year old children.16 17 This study, therefore, does not support our earlier suggestion of amplification of the association during the first decade.15 The greater strength of the association in girls at this age has been suggested by two previous studies18 19 and remains unexplained. The inverse association is not seen in adolescents,4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 and this may be because blood pressure tracking is perturbed as a result of rapid growth during this period.22 Our study, which found stronger associations among girls, who were closer to puberty than the boys,23 does not support this theory. Furthermore, there is no evidence that body build is acting as a stronger confounder of the association in girls as correlation coefficients relating childhood height and ponderal index to birth weight and blood pressure were similar in both sexes.
Other measurements of size at birth and blood pressure–Several previous studies have looked at measures of neonatal size other than birth weight and later blood pressure24 25 26 27 28 29 30; five of these looked at the associations in children.24 25 26 28 30 Head circumference tended to be inversely related to systolic blood pressure in two studies,25 29 but the associations between other measures of size at birth (including length,25 28 29 30 ponderal index,25 26 27 28 29 and head circumference:length ratio25 27) and subsequent blood pressure have all proved more inconsistent. Only one31 of five subsequent studies15 25 29 32 has confirmed the original finding of the highest subsequent blood pressures arising in light babies with big placentas.8 Our findings on placental weight and placental ratio are in agreement with the results of Barker's study on placental weight8 but only among the boys in whom no association between birth weight and blood pressure was observed. In the present study, apart from birth weight and univariate measures of size (particularly head circumference), associations between other measures of size at birth and blood pressure were weak and inconsistent.
These data on the associations between size at birth and blood pressure in contemporary children raise important questions about the fetal origins hypothesis. Firstly, they do not support the possibility of amplification during the second half of the first decade. Secondly, at 8-11 years the association is almost entirely restricted to the girls; a finding that is unlikely to be an artefact of adolescence. Thirdly, the finding that, although birth weight is related to subsequent blood pressure other more complex measures of fetal growth such as ponderal index at birth or head circumference:length ratio are not related, does not provide support for the programming of blood pressure at critical periods in the second and third trimesters. Finally, the association between birth weight and blood pressure at 8-11 years is much smaller than the effect of current size on blood pressure, suggesting that childhood obesity (currently increasing in prevalence33) remains a more important determinant of blood pressure in children. The association may be amplified with increasing age and may be increasingly concentrated among subjects with the greatest current body mass.34 Our finding that the association was significantly stronger among girls with lower ponderal indices, however, does not favour this suggestion. Longitudinal studies of contemporary children growing into adulthood are needed to examine the consistency of associations with age and to assess the relative importance of fetal factors and current obesity in the determination of blood pressure and in adolescence and early adult life.
Our data confirm the existence of an inverse association between birth weight and blood pressure in singleton born children at 8-11 years of age. There seems, however, a clear sex difference in this association, with the effect being concentrated in the girls and extremely weak in the boys. Only highly correlated, one dimensional measures of size at birth seem to show any association with blood pressure in contemporary children. More sophisticated ratio measures of fetal growth do not seem to be related to subsequent childhood blood pressure. The association between placental weight and placental ratio and blood pressure in children at this age is inconsistent. These findings do not support the suggestion that programming in the second and third trimesters results in identifiable patterns of neonatal size that are associated with an increased risk of raised blood pressure.
We thank the members of our research team (Sally Gassor, Angela Murphy, Catherine Stuart, Louise Went, and Valerie Wilson) and all the children, parents, and schools who participated in the study.
Funding: Wellcome Trust (grant No 038976/Z/93/Z). MW was supported by the British Heart Foundation.
Conflict of interest: None.