Early-Life Predictors of Systolic Blood Pressure Trajectories From Infancy to Adolescence: Findings From Project Viva

Abstract

Childhood blood pressure (BP) is a strong predictor of later risk of cardiovascular disease. However, few studies have assessed dynamic BP trajectories throughout the early-life period. We investigated the relationship between early-life factors and systolic BP (SBP) from infancy to adolescence using linear spline mixed-effects models among 1,370 children from Project Viva, a Boston, Massachusetts-area cohort recruited in 1999–2002. After adjusting for confounders and child height, we observed higher SBP in children exposed to gestational diabetes mellitus (vs. normoglycemia; age 3 years: β = 3.16 mm Hg (95% confidence interval (CI): 0.28, 6.04); age 6 years: β = 1.83 mm Hg (95% CI: 0.06, 3.60)), hypertensive disorders of pregnancy (vs. normal maternal BP; age 6 years: β = 1.39 mm Hg (95% CI: 0.10, 2.67); age 9 years: β = 1.84 mm Hg (95% CI: 0.34, 3.34); age 12 years: β = 1.70 mm Hg (95% CI: 0.48, 2.92)), higher neonatal SBP (per 10-mm Hg increase; age 3 years: β = 1.26 mm Hg (95% CI: 0.42, 2.09); age 6 years: β = 1.00 mm Hg (95% CI: 0.49, 1.51); age 9 years: β = 0.75 mm Hg (95% CI: 0.17, 1.33)), and formula milk in the first 6 months of life (vs. breast milk only; age 12 years: β = 2.10 mm Hg (95% CI: 0.46, 3.74); age 15 years: β = 3.52 mm Hg (95% CI: 1.40, 5.64); age 18 years: β = 4.94 mm Hg (95% CI: 1.88, 7.99)). Our findings provide evidence of programming of offspring SBP trajectories by gestational diabetes, hypertensive disorders of pregnancy, and formula milk intake and of neonatal BP being a potentially useful marker of childhood BP. These factors could be relevant in identifying children who are at risk of developing elevated BP.

High blood pressure (BP) in adulthood is one of the leading risk factors for morbidity and mortality in the world (1) and has been shown to be predicted by childhood high BP (2). Between 1999 and 2012, the prevalence of high BP (defined as systolic or diastolic BP at the 95th percentile or higher) among US youths aged 8–17 years remained relatively unchanged (10.6% in 1999 and 11.0% in 2012) (3). To develop preventive strategies that mitigate adult hypertension as early as possible, a better physiological and epidemiologic understanding of the early determinants of BP is required.

Observational evidence, including our prior work, has shown that intrauterine exposure to adverse conditions, such as older maternal age (4), prepregnancy obesity (5), prenatal smoking (6), hypertensive disorders of pregnancy (HDP) (7), and gestational diabetes mellitus (GDM) (8), could program the development of elevated BP in infancy, childhood, or adolescence. The evidence surrounding associations of birth weight with later BP, however, still remains inconclusive (9). Postnatal experiences and growth, such as rapid infant weight gain, also are associated with BP later in childhood (10, 11). However, these studies did not examine patterns of BP change in relation to these developmental factors. Without sufficient information on dynamic BP patterns throughout the early-life period, the physiological insights into the temporal relationship between early-life risk factors and BP patterns are incomplete. Therefore, identifying developmental factors that are related to BP profiles throughout the early-life period could potentially inform strategies that aim to reduce the level of BP attained later in life.

In order to address these clinically unmet needs, we used data from a Boston, Massachusetts-area prebirth cohort to assess associations of prenatal, perinatal, and postnatal factors with BP trajectories from infancy to adolescence. We hypothesized that the aforementioned factors would be predictive of trajectories of BP from infancy to adolescence.

METHODS

Study population

Children were participants in Project Viva, an ongoing prospective cohort study of prenatal and perinatal influences on maternal, fetal, and child health (12). We recruited eligible pregnant women at clinical visits during the first trimester of pregnancy between April 1999 and November 2002 from 8 obstetrical offices of Atrius Harvard Vanguard Medical Associates, a multisite group medical practice in eastern Massachusetts. We considered participants eligible if pregnant mothers were fluent in English, with less than 22 weeks of gestation at study entry and a singleton pregnancy (12). Of 2,128 live singleton births, we included 1,572 children (73.9%) who had 1 or more systolic BP (SBP) measurements taken between infancy and adolescence. Mothers provided written informed consent at enrollment and each postnatal follow-up visit, and children provided assent at the midchildhood and adolescent visits. The Institutional Review Board of Harvard Pilgrim Health Care approved the project in line with ethical standards established by the Declaration of Helsinki.

Exposures: pre-, peri-, and postnatal factors

Prenatal factors

Mothers reported their age, prepregnancy weight, height, and smoking history via questionnaires and interviews at recruitment. We calculated prepregnancy body mass index (BMI) as self-reported prepregnancy weight (kg) divided by height squared (m²). We obtained results of a 2-stage clinical glycemic screening (a nonfasting 50-g oral glucose challenge test followed by a 100-g, 3-hour oral glucose tolerance test) to categorize women as having normoglycemia, isolated hyperglycemia, impaired glucose tolerance, or GDM, based on previously detailed criteria (13). We also extracted data on clinical blood pressure, urine protein results, and diagnoses from outpatient and hospital medical records, which we used to identify HDP (normal blood pressure, gestational hypertension, preeclampsia, or chronic hypertension) (7).

Perinatal factors

Mothers reported their child’s race/ethnicity, categorized as white, black, Hispanic, Asian, or other. We extracted data on infant sex and birth weight from hospital medical records and calculated length of gestation from a prenatal ultrasonogram or by subtracting the date of the last menstrual period from the date of delivery (14). We calculated birth weight–for–gestational age (BW-for-GA) z scores using national reference data (15). Within 3 days after delivery, we measured neonatal SBP up to 5 times at 1-minute intervals and averaged all measurements for each newborn (4).

Postnatal factors

At the 6-month visit, we grouped infants into 4 categories on the basis of their extent of breastfeeding or formula feeding in the first 6 months of life (i.e., formula only, mixed feeding, weaned (defined as having initiated breastfeeding but having discontinued it completely before 6 months of age), or breast milk only) (16). We used measures of the child’s weight and length at birth and in infancy to calculate BMI and determined age- and sex-specific BMI z scores (zBMIs) using the World Health Organization growth reference chart (17). We used the change in zBMI from birth to infancy as an indicator of infant weight gain.

Outcome: child SBP

At in-person research visits carried out during infancy (median age, 6.3 months (interquartile range, 5.3–9.0)), early childhood (median age, 3.2 years (interquartile range, 2.9–5.9)), midchildhood (median age, 7.7 years (interquartile range, 6.6–10.6)), and early adolescence (median age, 12.9 years (interquartile range, 12.0–16.4)), trained research assistants recorded SBP on the child’s upper arm up to 5 times at 1-minute intervals using biannually calibrated oscillometric automated monitors (in infancy: Dinamap 8100, Pro 100, or Pro 200 (Dinamap, Tampa, Florida); in childhood: Dinamap Pro-100 (Dinamap); in adolescence: Omron HEM-907XL (Omron, Bannockburn, Illinois)) following standardized procedures (11, 18–20). We averaged all 5 SBP measurements at each research visit, since quantifying between-person differences rather than absolute levels improves precision (21). The proportion of missingness for SBP at each research visit ranged from 22.6% to 35.5%; the structure of missingness, however, was completely random (see Web Table 1, available at https://academic.oup.com/aje). In a subset of 378 children, we also obtained additional SBP data ranging from childhood to adolescence (range, 4.2–18.4 years) from medical records, where pediatricians recorded SBP at routine clinic visits (i.e., well-child visits at which health-care providers would administer scheduled immunizations, monitor growth and development, and make recommendations to parents for optimal child health (22)). We used SBP rather than diastolic BP because it is measured more accurately with our automated instrument (6, 8, 10) and because SBP is more predictive of later cardiovascular disease risk than diastolic BP (23). To exclude implausible measurements, we excluded SBP measurements that were more than 5 standard deviations or less than −5 standard deviations from the mean.

Other covariates

Mothers reported their highest level of education and the fathers’ highest level of education via questionnaires and interviews at recruitment, which we categorized as university-educated or not university-educated. We extracted data on parity from hospital medical records. We used the child’s weight and length/height obtained at research visits and from medical records (14, 18) (which were matched in time with BP recorded at routine clinic visits) to calculate BMI and determined age- and sex-specific height z scores and zBMIs using the World Health Organization growth standards (17).

Statistical analysis

We used linear spline mixed-effects models to examine associations of the aforementioned pre-, peri-, and postnatal factors with SBP from infancy to adolescence. We found that 2 knot points in the early childhood (age 3 years) and midchildhood (age 9 years) periods sufficiently described the trajectory of SBP change, resulting in 3 linear slopes representing the “infancy to early childhood,” “early to midchildhood,” and “midchildhood to adolescence” periods, respectively (see Web Appendix 1, Web Table 2, and Web Figures 1 and 2 for details). We used actual values instead of z scores because there are currently no age-, sex-, and height-specific national reference values available for children below 2 years of age.

We included each pre-, peri-, and postnatal factor as a fixed effect and as an interaction with each of the 3 linear slopes in the linear spline models. We also included a random intercept and random slopes to account for repeated SBP observations in each child and to reflect the heterogeneity in the data. Because child length/height is a major determinant of SBP, we included length/height z score in all models. In sensitivity analyses, we further adjusted for child zBMI. We additionally adjusted for maternal and paternal educational level, maternal height, parity, and type of study visit (research or clinic visit) in all models. Details on the construction of these models are provided in Web Appendix 2.

For each pre-, peri-, and postnatal factor, we used the linear spline model to predict SBP from age 6 months to age 18 years and plot its corresponding trajectory, while holding other covariates constant at their mean values. We estimated adjusted differences (and 95% confidence intervals) in predicted SBP between categories of pre-, peri-, and postnatal factors at 3-year intervals using the margins command in Stata (StataCorp LLC, College Station, Texas). We also analyzed terms for the interaction of each factor with each of the 3 linear SBP slopes (in mm Hg/year); these estimates describe whether the association of each pre-, peri-, or postnatal factor with SBP becomes stronger/weaker with increasing age. We restricted all analyses to children without missing data on pre- (n = 1,370), peri- (n = 825), and postnatal factors (n = 662). In additional sensitivity analyses, we further restricted our analyses to 1) children with SBP measured at research visits only and 2) children with complete data for all pre-, peri-, and postnatal factors (n = 662). We performed all analyses using Stata 15.1.

RESULTS

Cohort description

Table 1 shows the characteristics of children in the study. Compared with those excluded, children included in the study were more likely to be white (68% vs. 54%) and exclusively fed breast milk in the first 6 months of life (28% vs. 18%), and they had mothers who were older (32.6 years vs. 30.4 years), had lower prepregnancy BMI (24.6 vs. 25.4), were more likely to be university-educated (71% vs. 52%), and were less likely to have smoked during pregnancy (10% vs. 18%) (Web Table 3). At successive research visits, the mean SBP was 90.2 (standard deviation (SD), 13.4) mm Hg in infancy, 92.3 (SD, 10.9) mm Hg in early childhood, 94.4 (SD, 8.5) mm Hg in midchildhood, and 107.2 (SD, 9.2) mm Hg in early adolescence (Web Table 1). No differences in SBP were observed between research and clinic visits across age categories, except at ages 7–8 years and 8–9 years, where SBP was higher at clinic visits than at research visits (Web Figures 3 and 4). The predicted mean linear SBP velocities were 0.4 (SD, 3.4) mm Hg/year for the infancy–early childhood period, 1.1 (SD, 0.5) mm Hg/year for early childhood–midchildhood, and 2.1 (SD, 0.4) mm Hg/year for midchildhood–adolescence.

Table 1.

Characteristics of Study Participants Recruited for Project Viva in 1999–2002, Boston, Massachusetts

Maternal or Child Factor	No. of Subjects	Mean (SD)	%
Maternal prenatal factors
University educational level	979		71
Nulliparous	651		48
Age, years	1,370	32.6 (4.7)
Height, m	1,370	1.7 (0.1)
Prepregnancy BMIa	1,370	24.6 (5.2)
Smoking history
Never smoked	949		69
Smoked before pregnancy	286		21
Smoked during pregnancy	135		10
Gestational glucose tolerance status
Normoglycemia	1,138		83
Isolated hyperglycemia	114		8
Impaired glucose tolerance	48		4
Gestational diabetes	70		5
Hypertensive disorders of pregnancy
Normal blood pressure	1,212		89
Gestational hypertension or preeclampsia	138		10
Chronic hypertension	20		1
Father university-educated	911		66
Child factors
Sex
Female	667		49
Male	703		51
Race/ethnicity
White	943		68
Black	178		13
Hispanic	60		4
Asian	53		4
Other	136		10
Infant perinatal factors
Birth weight-for-gestational age z score (in SD units)	825	0.2 (0.9)
Neonatal systolic blood pressure	825	72.4 (8.8)
Postnatal factors
Infant feeding type in first 6 months of life
Breast milk only	186		28
Mixed feeding	176		27
Weaned	225		34
Formula only	75		11
Infancy zBMI gain (in SD units)	662	0.03 (1.29)

Abbreviations: BMI, body mass index; SD, standard deviation; zBMI, age- and sex-specific BMI z score.

^a Weight (kg)/height (m)².

Associations of prenatal factors with child SBP trajectories

Exposure to higher prepregnancy BMI was associated with higher SBP trajectories that were significant at ages 6, 9, 12, and 15 years (Figure 1A and Table 2). Maternal age and smoking history were not associated with child SBP from infancy to adolescence, although children exposed to intrauterine smoking before or during pregnancy appeared to have higher SBP trajectories (Figure 1B). Exposure to gestational hypertension or preeclampsia in utero (vs. normal blood pressure) (Figure 1C) was similarly associated with higher SBP trajectories that were significant at ages 6 years (β = 1.39 mm Hg, 95% confidence interval (CI): 0.10, 2.67), 9 years (β = 1.84 mm Hg, 95% CI: 0.34, 3.34), and 12 years (β = 1.70 mm Hg, 95% CI: 0.48, 2.92). Maternal glucose intolerance (vs. normoglycemia) was associated with higher SBP trajectories (Figure 1D), which were significant at ages 3–6 years for exposure to GDM and at ages 6–18 years for exposure to isolated hyperglycemia (Table 2). No prenatal factors were associated with rate of SBP change across the different periods (Table 3).

Figure 1.

Predicted trajectories of child systolic blood pressure (BP) from ages 6 months to 18 years according to maternal prepregnancy body mass index (BMI; weight (kg)/height (m)²) (A), smoking history (B), hypertensive disorders of pregnancy (C), and glucose tolerance status (D) among participants recruited into Project Viva in 1999–2002, Boston, Massachusetts. The models adjusted for type of study visit, all prenatal factors, and child height z score. GDM, gestational diabetes mellitus; GH, gestational hypertension; IGT, impaired glucose tolerance.

Table 2.

Associations of Prenatal, Perinatal, and Postnatal Factors With Predicted Systolic Blood Pressure in Infancy, Childhood, and Adolescence Among Participants Recruited Into Project Viva in 1999–2002, Boston, Massachusetts

Maternal or Child Factor	Child’s Age and Mean Difference in Predicted Systolic BP, mm Hg
	No. of Subjects	6 Months		3 Years		6 Years		9 Years		12 Years		15 Years		18 Years
		β	95% CI	β	95% CI	β	95% CI	β	95% CI	β	95% CI	β	95% CI	β	95% CI
Maternal prenatal factorsa
Age, years (per 5-year increase)	1,370	0.10	−1.13, 1.33	0.54	−0.17, 1.24	0.19	−0.24, 0.62	−0.16	−0.64, 0.32	−0.05	−0.44, 0.35	0.07	−0.47, 0.61	0.18	−0.62, 0.99
Prepregnancy BMIb (per 5-unit increase)	1,370	0.02	−0.99, 1.03	0.11	−0.52, 0.74	0.43	0.05, 0.82	0.76	0.32, 1.19	0.69	0.34, 1.05	0.63	0.14, 1.11	0.56	−0.15, 1.28
Smoking history (vs. having never smoked)	1,370
Smoked before pregnancy		2.35	−0.16, 4.86	1.23	−0.25, 2.70	1.01	0.10, 1.92	0.79	−0.25, 1.83	0.77	−0.08, 1.63	0.76	−0.41, 1.92	0.74	−0.99, 2.47
Smoked during pregnancy		1.41	−2.26, 5.08	1.26	−0.89, 3.42	1.24	−0.10, 2.59	1.22	−0.40, 2.85	1.20	−0.12, 2.51	1.17	−0.68, 3.03	1.15	−1.65, 3.94
Hypertensive disorders of pregnancy (vs. normal blood pressure)	1,370
Gestational hypertension or preeclampsia		2.74	−0.53, 6.01	0.93	−1.14, 3.00	1.39	0.10, 2.67	1.84	0.34, 3.34	1.70	0.48, 2.92	1.55	−0.12, 3.23	1.41	−1.09, 3.90
Chronic hypertension		1.17	−6.71, 9.04	−0.92	−5.78, 3.94	0.04	−2.90, 2.98	1.01	−2.22, 4.23	1.34	−1.42, 4.09	1.67	−2.11, 5.45	2.00	−3.53, 7.53
Glucose tolerance status (vs. normoglycemia)	1,370
Isolated hyperglycemia		1.13	−2.58, 4.84	0.75	−1.34, 2.85	1.34	0.04, 2.63	1.92	0.40, 3.44	2.33	1.08, 3.58	2.74	1.01, 4.46	3.15	0.59, 5.71
Impaired glucose tolerance		3.59	−1.95, 9.12	−1.21	−4.49, 2.07	−1.47	−3.48, 0.54	−1.72	−3.98, 0.54	−1.41	−3.27, 0.45	−1.09	−3.64, 1.45	−0.78	−4.54, 2.98
Gestational diabetes		−0.50	−5.22, 4.22	3.16	0.28, 6.04	1.83	0.06, 3.60	0.50	−1.50, 2.50	0.39	−1.24, 2.02	0.28	−1.91, 2.48	0.18	−3.07, 3.42
Child factorsa
Female sex (vs. male)	1,370	−2.66	−4.67, −0.66	−0.85	−2.02, 0.32	−0.29	−1.01, 0.43	0.27	−0.56, 1.10	−1.72	−2.40, −1.04	−3.71	−4.65, −2.78	−5.70	−7.09, −4.32
Race/ethnicity (vs. white)	1,370
Black		0.71	−2.68, 4.11	0.61	−1.44, 2.66	0.54	−0.68, 1.77	0.47	−0.83, 1.78	−0.42	−1.50, 0.66	−1.32	−2.78, 0.14	−2.22	−4.36, −0.07
Hispanic		−0.17	−5.61, 5.27	−0.51	−3.73, 2.71	−0.37	−2.29, 1.55	−0.23	−2.35, 1.90	−0.81	−2.57, 0.95	−1.39	−3.77, 0.99	−1.97	−5.47, 1.53
Asian		−3.79	−8.96, 1.37	0.40	−2.91, 3.71	−0.49	−2.52, 1.55	−1.38	−3.70, 0.95	0.76	−1.17, 2.69	2.90	0.15, 5.64	5.03	0.93, 9.14
Other		1.14	−2.19, 4.47	−0.51	−2.47, 1.44	0.35	−0.85, 1.55	1.21	−0.15, 2.57	0.77	−0.34, 1.88	0.33	−1.17, 1.83	−0.11	−2.33, 2.10
Infant perinatal factorsc
BW-for-GA (per 1-SD increase)	825	0.34	−1.19, 1.88	−0.45	−1.30, 0.41	−0.65	−1.16, −0.13	−0.85	−1.45, −0.25	−0.51	−1.00, −0.02	−0.17	−0.82, 0.49	0.17	−0.79, 1.13
Neonatal systolic BP (per 10-mm Hg increase)	825	0.71	−0.71, 2.13	1.26	0.42, 2.09	1.00	0.49, 1.51	0.75	0.17, 1.33	0.36	−0.12, 0.85	−0.02	−0.67, 0.63	−0.40	−1.35, 0.55
Postnatal factorsd
Infant feeding type in first 6 months of life (vs. breast milk only)	662
Mixed feeding		−2.18	−5.68, 1.33	1.32	−0.73, 3.37	0.84	−0.44, 2.11	0.35	−1.21, 1.91	0.51	−0.77, 1.80	0.67	−0.99, 2.34	0.83	−1.58, 3.25
Weaned		−1.96	−5.47, 1.55	−0.63	−2.67, 1.41	−0.31	−1.57, 0.94	0.01	−1.48, 1.49	0.39	−0.84, 1.62	0.78	−0.80, 2.37	1.17	−1.11, 3.45
Formula only		−2.51	−7.42, 2.41	1.66	−1.12, 4.43	1.17	−0.52, 2.85	0.68	−1.30, 2.65	2.10	0.46, 3.74	3.52	1.40, 5.64	4.94	1.88, 7.99
Infancy zBMI gain (per 1-SD increase)	662	1.33	0.12, 2.54	0.95	0.24, 1.67	0.67	0.23, 1.11	0.39	−0.16, 0.95	0.21	−0.24, 0.66	0.03	−0.57, 0.63	−0.15	−1.03, 0.73

Abbreviations: BMI, body mass index; BP, blood pressure; BW-for-GA, birth weight for gestational age; CI, confidence interval; SD, standard deviation; zBMI, age- and sex-specific BMI z score.

^a Adjusted for type of study visit, maternal and paternal educational level, maternal height, parity, and child height z score.

^b Weight (kg)/height (m)².

^c Adjusted for type of study visit, maternal and paternal educational level, maternal height, parity, all prenatal factors, child factors, and height z score.

^d Adjusted for type of study visit, maternal and paternal educational level, maternal height, parity, all prenatal and perinatal factors, child factors, and height z score.

Table 3.

Associations of Prenatal, Perinatal, and Postnatal Factors With Rate of Systolic Blood Pressure Change From Infancy to Early Childhood, Early Childhood to Midchildhood, and Midchildhood to Adolescence Among Participants Recruited Into Project Viva in 1999–2002, Boston, Massachusetts

Maternal or Child Factor	Average Rate of Systolic BP Change, mm Hg/year
	No. of Subjects	From Infancy to Early Childhood		From Early Childhood to Midchildhood		From Midchildhood to Adolescence
		β	95% CI	β	95% CI	β	95% CI
Maternal prenatal factorsa
Age, years (per 5-year increase)	1,370	0.15	−0.34, 0.64	−0.12	−0.26, 0.02	0.04	−0.07, 0.15
Prepregnancy BMIb (per 5-unit increase)	1,370	0.03	−0.38, 0.44	0.11	−0.02, 0.23	−0.02	−0.12, 0.08
Smoking history (vs. having never smoked)	1,370
Smoked before pregnancy		−0.37	−1.38, 0.63	−0.07	−0.37, 0.23	−0.01	−0.24, 0.23
Smoked during pregnancy		−0.05	−1.52, 1.42	−0.01	−0.46, 0.44	−0.01	−0.39, 0.37
Hypertensive disorders of pregnancy (vs. normal blood pressure)	1,370
Gestational hypertension or preeclampsia		−0.60	−1.95, 0.74	0.15	−0.27, 0.57	−0.05	−0.39, 0.29
Chronic hypertension		−0.70	−3.90, 2.51	0.32	−0.64, 1.29	0.11	−0.62, 0.84
Glucose tolerance status (vs. normoglycemia)	1,370
Isolated hyperglycemia		−0.13	−1.59, 1.34	0.19	−0.23, 0.62	0.14	−0.21, 0.48
Impaired glucose tolerance		−1.60	−3.83, 0.63	−0.09	−0.74, 0.57	0.10	−0.40, 0.61
Gestational diabetes		1.22	−0.70, 3.13	−0.44	−1.02, 0.14	−0.04	−0.48, 0.41
Child factorsa
Female sex (vs. male)	1,370	0.61	−0.19, 1.41	0.19	−0.05, 0.42	−0.66	−0.85, −0.48
Race/ethnicity (vs. white)	1,370
Black		−0.03	−1.41, 1.34	−0.02	−0.43, 0.38	−0.30	−0.59, −0.01
Hispanic		−0.11	−2.30, 2.08	0.05	−0.60, 0.69	−0.19	−0.67, 0.28
Asian		1.40	−0.73, 3.53	−0.30	−0.97, 0.37	0.71	0.16, 1.27
Other		−0.55	−1.88, 0.78	0.29	−0.11, 0.68	−0.15	−0.45, 0.15
Infant perinatal factorsc
BW-for-GA (per 1-SD increase)	825	−0.26	−0.87, 0.35	−0.07	−0.24, 0.11	0.11	−0.02, 0.24
Neonatal systolic BP (per 10-mm Hg increase)	825	0.18	−0.39, 0.76	−0.09	−0.26, 0.08	−0.13	−0.25, 0.00
Postnatal factorsd
Infant feeding type in first 6 months of life (vs. breast milk only)	662
Mixed feeding		1.17	−0.24, 2.57	−0.16	−0.60, 0.27	0.05	−0.27, 0.38
Weaned		0.44	−0.96, 1.84	0.11	−0.32, 0.53	0.13	−0.18, 0.44
Formula only		1.39	−0.56, 3.33	−0.16	−0.74, 0.41	0.47	0.06, 0.88
Infancy zBMI gain (per 1-SD increase)	662	−0.13	−0.61, 0.36	−0.09	−0.25, 0.06	−0.06	−0.18, 0.06

Abbreviations: BMI, body mass index; BP, blood pressure; BW-for-GA, birth weight for gestational age; CI, confidence interval; SD, standard deviation; zBMI, age- and sex-specific BMI z score.

^a Adjusted for type of study visit, maternal and paternal educational level, maternal height, parity, and child height z score.

^b Weight (kg)/height (m)².

^c Adjusted for type of study visit, maternal and paternal educational level, maternal height, parity, all prenatal factors, child factors, and height z score.

^d Adjusted for type of study visit, maternal and paternal educational level, maternal height, parity, all prenatal and perinatal factors, child factors, and height z score.

Associations of perinatal factors with child SBP trajectories

Compared with males, females had lower SBP trajectories which were significant in infancy (β = −2.66 mm Hg, 95% CI: −4.67, −0.66) and adolescence (age 12 years: β = −1.72 mm Hg (95% CI: −2.40, −1.04); age 15 years: β = −3.71 mm Hg (95% CI: −4.65, −2.78); age 18 years: β = −5.70 mm Hg (95% CI: −7.09, −4.32)) (Table 2) and had a slower rate of SBP change from midchildhood to adolescence (β = −0.66 mm Hg/year, 95% CI: −0.85, −0.48) (Table 3). We observed no differences in SBP trajectories by child’s race/ethnicity, except in adolescence, where Asian children (vs. white children) had higher adolescent SBP and a faster rate of SBP change from midchildhood to adolescence (Tables 2 and 3).

Children with higher BW-for-GA had lower SBP trajectories during childhood (Figure 2A), significant at ages 6 years (β = −0.65 mm Hg, 95% CI: −1.16, −0.13), 9 years (β = −0.85 mm Hg, 95% CI: −1.45, −0.25), and 12 years (β = −0.51 mm Hg, 95% CI: −1.00, −0.02) (Table 2). Children with higher neonatal SBP had higher SBP trajectories throughout childhood (Figure 2B) (age 3 years: β = 1.26 mm Hg (95% CI: 0.42, 2.09); age 6 years: β = 1.00 mm Hg (95% CI: 0.49, 1.51); age 9 years: β = 0.75 mm Hg (95% CI: 0.17, 1.33)) for every 10-mm Hg increase in neonatal SBP (Table 2), which gradually weakened in adolescence.

Figure 2.

Predicted trajectories of child systolic blood pressure (BP) from ages 6 months to 18 years according to child birth weight for gestational age (BW-for-GA) (A), neonatal BP (B), infant feeding type in the first 6 months of life (C), and infancy gain in age- and sex-specific body mass index (BMI; weight (kg)/height (m)²) z score (zBMI) (D) among participants recruited into Project Viva in 1999–2002, Boston, Massachusetts. Models for BW-for-GA and neonatal BP adjusted for prenatal factors; models for infant feeding type and infancy zBMI gain adjusted for prenatal and perinatal factors. All models additionally adjusted for type of study visit and child height z score. SD, standard deviation.

Associations of early postnatal factors with child SBP trajectories

Children who were fed formula milk only (vs. breast milk only) in the first 6 months of life had higher SBP trajectories in adolescence (Figure 2C) (age 12 years: β = 2.10 mm Hg (95% CI: 0.46, 3.74); age 15 years: β = 3.52 mm Hg (95% CI: 1.40, 5.64); age 18 years: β = 4.94 mm Hg (95% CI: 1.88, 7.99)) (Table 2) and a faster rate of SBP change from midchildhood to adolescence (β = 0.47 mm Hg/year, 95% CI: 0.06, 0.88) (Table 3). Children who were weaned or were on mixed feeding in the first 6 months of life showed no associations with SBP trajectories. Children with greater zBMI gain during infancy had higher SBP trajectories in early childhood that were significant between 6 months and 6 years of age (Table 2) but gradually weakened into adolescence (Figure 2D).

Sensitivity analyses

Additional adjustment for child zBMI did not appreciably change the associations of isolated hyperglycemia/GDM, HDP, BW-for-GA, neonatal SBP, or infant feeding type with child SBP trajectories (Web Figures 5 and 6). The observed associations for prepregnancy BMI and infancy zBMI change with child SBP, however, were attenuated to the null (Web Tables 4 and 5). Furthermore, similar results were observed when the analyses were restricted to children with SBP measured at research visits only (Web Tables 6 and 7) or to those with complete data for all pre-, peri-, and postnatal factors (Web Tables 8 and 9)—albeit with wider 95% confidence intervals due to the smaller number of SBP measurements.

DISCUSSION

In this study, we characterized SBP trajectories in 3 distinct periods, namely infancy to early childhood, early childhood to midchildhood, and midchildhood to adolescence, and demonstrated its associations with pre-, peri-, and postnatal factors. In general, children with intrauterine exposure to isolated hyperglycemia/GDM or HDP had higher SBP trajectories than unexposed children. Children with higher neonatal SBP exhibited higher SBP trajectories in childhood, while those with higher BW-for-GA exhibited lower SBP trajectories in childhood. Additionally, children who were fed formula milk only in the first 6 months of life exhibited higher SBP trajectories in adolescence.

Few longitudinal studies have assessed follow-up measures of BP from infancy to later childhood or adolescence. We assessed BP trajectories from infancy onward rather than from birth onward, since other investigators had reported poor tracking of BP from birth to infancy (24). Zinner et al. (25) previously reported that BP measured in the first days of life was weakly correlated with BP at 6–24 months. Levine et al. (26) also noted that BP measured 2 days after birth was not significantly correlated with BP at 1, 3, 6, or 12 months, which is consistent with our findings of a null association between SBP at birth and in infancy. However, we found that higher neonatal SBP was persistently associated with higher SBP during childhood; thus, neonatal BP might be a potentially useful marker for identification of children who are likely to experience elevated BP trajectories throughout childhood.

Consistent with previous findings, we found that maternal prepregnancy BMI (27) and infant zBMI gain (10, 11, 28) were significantly associated with higher SBP during childhood and adolescence. These associations diminished to the null, however, after adjustment for child zBMI, which could reflect that children of mothers with high prepregnancy BMI or children who experienced rapid infant growth were likely to have high BMI themselves, and thus have higher BP. The observed sex differences in BP are consistent with previous studies and are probably explained by sex differences in body composition and the relationships of individual body compartments to BP (29). The observed ethnic differences in BP are also consistent with a recent study showing Asians to have a higher prevalence of hypertension than whites in similar weight and socioeconomic status categories (30), suggesting that there may be other environmental or genetic factors driving the racial/ethnic disparities in BP.

Our results corroborate previous findings that have shown associations of intrauterine exposure to GDM or HDP with childhood BP (31–34), although BP was measured only on a single occasion in those prior studies. Our study extends those findings by showing that the associations may persist as the child ages. These associations could be driven by shared genetic factors between the mother and child (35). For example, Howe et al. (36) reported that genetic variants of adult BP were associated with child BP at age 6 years, with the genetic effect on BP remaining similar from age 6 years to age 17 years. We accounted for potential shared environmental factors such as parental socioeconomic status (37) and other drivers of child BP (i.e., height and BMI), but it is possible that unmeasured factors could have contributed to the findings. The physiological mechanisms by which these prenatal factors might influence child BP are still unclear. We speculate that exposure to HDP or GDM may have an indirect effect through changes in maternal vasculature, thereby affecting placenta formation and blood flow (38–40). Animal and human studies also suggest a direct influence of exposure to hyperglycemia in utero on the kidneys of offspring, leading to altered BP regulation (41, 42).

Our observations of infant feeding type in the first 6 months of life also corroborate previous findings, where infants who were never breastfed were shown to have higher SBP in childhood and adolescence than those who were exclusively breastfed (43, 44). It is possible that other factors, such as being born small for gestational age or having accelerated infant growth, could explain these findings; that is, infants who are growing rapidly due to a period of intrauterine growth restraint may be formula-fed because of a higher demand for milk (45). However, the association between infant feeding type and SBP was not altered in our study after we accounted for birth weight and infant weight gain. We speculate that these findings could be explained by the high protein content in formula milk, which might further stimulate insulin-like growth factor 1 secretion and, in turn, raise BP through a posterior adrenergic response in the sympathetic nervous system (46).

Taken together, our findings suggest evidence of programming of offspring SBP trajectories by isolated hyperglycemia/GDM, HDP, and exclusive formula feeding. These factors could be informative for identification of children who are at risk of developing high BP between infancy and adolescence. Additionally, the higher SBP conferred through exposure to these adverse conditions may have cardiovascular implications in adulthood, as evident in recent studies that showed associations between higher childhood SBP and markers of adult preclinical cardiovascular disease (47, 48). Ongoing efforts to tailor effective interventions in changing dynamic BP patterns among at-risk children are warranted to reduce future cardiovascular disease risk.

Strengths of our study include its relatively large sample size, its prospective design, multiple measures of child BP, and the availability of data on a wide range of maternal and child factors obtained by highly trained staff using standardized protocols. One limitation of the study is that we estimated SBP trajectories using linear-spline models, which assumes a piecewise linear relationship between BP and age. This assumption may be biologically implausible, but results derived under it are often more interpretable than those from other types of models such as fractional polynomials (49). Despite the implausible linearity assumption, our 2-knot linear spline model appeared to fit the BP data equally well to a 2° fractional polynomial model, as evidenced through the residual plots (see Web Figure 2). Another limitation was the different methods of BP measurement employed during research and clinic visits; we used automated oscillometric devices to measure child BP at research visits, while pediatricians were likely to have used manual auscultative methods to measure BP at clinic visits. However, our analytical results were additionally adjusted for type of study visit—a potential proxy for method of BP measurement—which probably would have accounted for any variability that may have arisen from the use of different measurement methods at research and clinic visits. Finally, differences between children with and without follow-up measures of SBP might conceivably have led to selection bias. In addition, our study findings may not be generalizable to other populations from different settings, since a majority of our participants were white and university-educated.

In summary, this study characterized the trajectories of systolic BP from infancy to adolescence. Our findings provide evidence of programming of offspring SBP trajectories by important pregnancy complications (isolated hyperglycemia, GDM, HDP) and early-life nutrition (exclusive formula feeding) and of neonatal BP’s being a potentially useful marker of childhood BP. These factors could be relevant for identifying children who are at risk of high BP. Further studies are warranted to identify interventions that are effective in changing BP in these children and, if so, to determine whether such interventions can reduce the risk of future cardiovascular disease.

Abbreviations

BMI

body mass index

BP

blood pressure

BW-for-GA

birth weight for gestational age

CI

confidence interval

GDM

gestational diabetes mellitus

HDP

hypertensive disorders of pregnancy

SBP

systolic blood pressure

SD

standard deviation

zBMI

age- and sex-specific BMI z score