Childhood to Early-Midlife Systolic Blood Pressure TrajectoriesNovelty and Significance
Early-Life Predictors, Effect Modifiers, and Adult Cardiovascular Outcomes
Previous studies examining blood pressure change over time have modeled an average population trajectory. Recent research among older adults suggests there may be subgroups with different blood pressure trajectories. Identifying subgroups at risk of developing adult hypertension early in life can inform effective risk reduction efforts. We sought to identify different systolic blood pressure trajectories from childhood, their correlated risk factors, and early-midlife cardiovascular outcomes. Blood pressure data at ages 7, 11, 18, 26, 32, and 38 years from a longitudinal, representative birth cohort study (n=975) were used to identify 4 distinct trajectory groups via group-based trajectory modeling: normal (21.8%), high-normal (43.3%), prehypertensive (31.6%), and hypertensive (4.2%). The categories refer to blood pressure beginning at the age of 7 years and most recently measured at the age of 38 years. Family history of high blood pressure (odds ratio [OR], 43.23; 95% confidence interval [CI], 5.27–354.65), male sex (OR, 109.48; 95% CI, 26.82–446.96), being first born (OR, 2.5; 95% CI, 1.00–8.69) and low birth weight (OR, 2.79; 95% CI, 2.49–3.09) were associated with hypertensive group membership (compared with the normal group). Higher body mass index and cigarette smoking resulted in increasing blood pressure across trajectories, particularly for the higher blood pressure groups. Prehypertensive and hypertensive trajectory groups had worse cardiovascular outcomes by early midlife. Harmful blood pressure trajectories are identifiable in childhood, associated with both antecedent and modifiable risk factors over time, and predict adult cardiovascular disease risk. Early detection and subsequent targeted prevention and intervention may reduce the lifecourse burden associated with higher blood pressure.
High blood pressure in adulthood is a leading cause of morbidity and mortality, a major modifiable risk factor for cardiovascular disease, and associated with other cardiovascular risk factors, including impaired glucose tolerance, obesity, and dyslipidemia.1 The standard approach of treating high blood pressure in middle and old-age can help mitigate these risks, but considerable burden remains. An approach that identifies those at greatest risk of developing high blood pressure much earlier in life could permit more effective risk reduction via earlier, age-appropriate prevention and intervention strategies.
In this context, universal routine screening to identify children at risk of developing high blood pressure in adulthood has been recommended by some organizations.2,3 However, in 2013 the US Preventive Services Task Force concluded that current evidence was insufficient to assess the balance of benefits and harms of screening for primary hypertension in children and adolescents.4 Proponents of universal screening note that blood pressure levels track from childhood to adulthood and that children with elevated blood pressure are more likely to develop hypertension in adulthood compared with children with low blood pressure.5,6 Importantly, hypertension is often asymptomatic and elevated blood pressure is associated with organ damage and changes in cardiac structure in children.7 However, critics of universal screening argue that (1) the correlations between child and adult blood pressure levels are too weak8 (with blood pressure percentiles for children and adolescents used to define normative values not based on later cardiovascular risk)9 and (2) that further research is needed examining the longitudinal association between childhood blood pressure, adult hypertension, and cardiovascular disease.4
Because of conflicting recommendations, clinicians need better information for medical decision-making around risk assessment and potential intervention options. Prospective-epidemiological studies can provide such information. Recent longitudinal research suggests that there may be subgroups with differential blood pressure trajectories among older populations.5 If there is developmental heterogeneity in blood pressure trajectories over time then (1) it should be theoretically possible to identify early in life, groups of people at greatest risk of developing high or clinically significant blood pressure in adulthood and (2) the investigation of risk factors that differentiate between healthy and unhealthy trajectories is an important objective.
A burgeoning body of evidence points to a range of early-life predictors of high blood pressure in adulthood. These include intrauterine (eg, low birth weight,10 being first born,11 and maternal pregnancy hypertension12), postnatal (eg, not being breastfed),13 familial (eg, family history of high blood pressure),14 and psychosocial factors (eg, childhood low socioeconomic status).15 Males are significantly more likely to have higher blood pressure than females from childhood to midlife.16
Identification of factors that modify high blood pressure development from childhood to adulthood may help clinicians to stage age-appropriate interventions. Excess body weight is an established risk factor for high blood pressure, and increases in body mass index (BMI) have been associated with higher blood pressure tracking from childhood to adulthood.17 Other potentially modifiable risk factors include excess alcohol consumption18 and cigarette smoking19; however, studies have mostly been cross sectional, or among older cohorts.
To date, researchers have commonly resorted to using assignment rules based on subjective categorization criteria (eg, 95th percentile) to test taxonomical theories about normal versus pathological development. In this study, advanced longitudinal methods were applied to test the hypothesis that blood pressure changes are best understood via the investigation of subgroups within the population that differ in their rate of blood pressure increase over time. Although relatively new to health research, advanced longitudinal statistical modeling techniques have been used to examine developmental trajectories for other health states.20 We sought to identify early-midlife cardiovascular correlates of these developmental trajectories to help demonstrate broader clinical use. To the best of our knowledge, this is the first time that this type of modeling has been used to identify blood pressure trajectory groups from childhood to adulthood in a single longitudinal birth cohort study.
The aims of this research were to (1) identify latent groups of individuals sharing systolic blood pressure developmental trajectories from 7 to 38 years, (2) identify early-life predictors and potential modifiers of these trajectories, and (3) describe the association between trajectories and early-midlife cardiovascular risk indicators.
Participants are members of the Dunedin Multidisciplinary Health and Development Study, which has tracked the development of 1037 individuals born in 1972 to 1973 in Dunedin, New Zealand. The study design and inclusion criteria have been described elsewhere21 (for details see the online-only Data Supplement).
This study used systolic blood pressure data measured at ages 7, 11, 18, 26, 32, and 38 years (all ages at which >1 measure of blood pressure was taken). Blood pressure was measured in a quiet room, using a cuff of appropriate size, with the Study member in a seated position, by medically trained assessors. Up to 18 years, a London School of Hygiene and Tropical Medicine blind mercury sphygmomanometer (Cinetronics Ltd, Mildenhall, United Kingdom) was used. Up to 38 years, a Hawksley random-zero sphygmomanometer (Hawksley and Sons Ltd, Sussex, United Kingdom) with a constant deflation valve was used. Systolic blood pressure was assessed as the first Korotkoff sound and based on the mean of either 2 or 3 measures taken at 5-minute intervals according to a standard protocol. Antihypertensive medication use information was collected at ages 26 (n=6), 32 (n=12), and 38 (n=26) years. Participants were coded as having a systolic blood pressure of 140 mm Hg for the assessment age they were on medication. Data were included if participants had blood pressure information at ≥3 ages (n=975), with information at a specific age subsequently excluded if a participant was pregnant (n=30 at the age of 26 years; n=26 at the age of 32 years; and n=8 at the age of 38 years; for details on the inclusion criteria based on blood pressure data see the online-only Data Supplement).
Birth weight (g) and maternal pregnancy hypertension (diastolic blood pressure ≥90 mm Hg) data were taken from antenatal and perinatal hospital records. Birth order data (first born, second born, third born and higher) were obtained from parent report at the age of 3 years. Breastfeeding information was collected at the age 3 assessment and checked against validated health records.22
The collection of family history data about high blood pressure among study members’ biological siblings, parents, and grandparents was conducted in 2003 to 2006 and is described elsewhere23 (for details see the online-only Data Supplement).
Childhood socioeconomic status from birth to 5 years was based on parents’ self-reported occupational status24 (for details see the online-only Data Supplement).
Effect Modifiers From 7 to 38 Years
Height was recorded to the nearest millimeter using a portable stadiometer (Harpenden; Holtain, Ltd). Weight was measured to the nearest 0.1 kg using calibrated scales. Participants were weighed in light clothing. BMI was calculated in kilograms per square meter (weight [kg]/height [m2]).25
Alcohol consumption at 7 and 11 years was coded as zero for all participants. At the age of 18 years, participants were asked how often they usually drank alcohol. At ages 26, 32, and 38 years, participants reported how many drinks (standard units) on average they consumed weekly.
No participants smoked daily at 7 or 11 years, so daily smoking was coded as zero. The current number of cigarettes smoked per day (for at least 1 month in the previous year) was recorded at ages 18, 26, 32, and 38 years.
Age 38 Cardiovascular Risk Indicators
Biomarkers were obtained for measurements of nonfasting total cholesterol (mmol/L), high-density lipoprotein cholesterol level (mmol/L), triglyceride level (mmol/L), and glycohemoglobin at the age of 38 years. Venipuncture was conducted at the same time each day (4:15–4:45 pm). Ninety-five percent of the sample consented to phlebotomy, with pregnant women at this age excluded from the analyses (for further details on biomarker assessment see the online-only Data Supplement).
Waist and hip girth were measured by averaging 2 measurements taken using a steel tape calibrated in centimeters with millimeter gradations. Waist girth was taken as the perimeter at the level of the noticeable waist and hip girth at the widest point of the hips.
A composite measure of metabolic abnormalities (excluding blood pressure) based on the US National Cholesterol Education Program Adult Treatment Panel III guidelines was created (online-only Data Supplement).26 In brief, participants were considered to have met the criteria for the presence of metabolic abnormalities if they had 3 of 4 abnormalities.
Group-based trajectory modeling was used to identify distinctive groups of individual trajectories in the population from ages 7 to 38 years. We used Proc Traj, a Statistical Analysis Software macro to estimate the model parameters27 using a censored normal model appropriate for continuous normally distributed data.28
Several criteria20,28 were used to determine the number of blood pressure trajectory groups and the trajectory shapes (eg, cubic in age; for details see the online-only Data Supplement). To determine the best model, 12 models were fitted.
We examined 2 types of covariates, early-life factors and effect modifiers (time-varying covariates). To examine the association between cardiovascular disease risk indicators at 38 years and blood pressure trajectories, we used analysis of variance for continuous measures and χ2 statistics for categorical measures.
A 4-trajectory group model with cubic specifications for all groups was identified. Figure shows the plotted predicted trajectory lines for each of the 4 groups designated as normal, high-normal, prehypertensive, and hypertensive, with 95% confidence intervals (CIs) based on predicted trajectory means. The means for all 4 groups significantly differed from each other at all ages, beginning at the age of 7 years. The normal group (21.8% of the sample population) and high-normal group (43.3%) had mean blood pressures in adulthood in the normal systolic blood pressure range (90–120 mm Hg). The prehypertensive group (31.6%) had a mean systolic blood pressure within the prehypertensive range (120–139 mm Hg) throughout adulthood. The hypertensive group (4.2%) had the highest mean blood pressure at the age of 7 years and displayed the steepest rise in blood pressure with mean blood pressure in the hypertensive range at the age of 38 years (≥140 mm Hg). Table S1 provides data on predictive values for prehypertension (≥120 mm Hg) and hypertension (≥140 mm Hg) at the age of 38 years based on prehypertensive and hypertensive group membership and blood pressure measured at each assessment age.
Both early-life factors and effect modifiers were examined in relation to blood pressure trajectories (Tables 1–3). Tables 1 and 2 also provide data on the characteristics of participants for early-life factors based on trajectory group membership. Information on the physical characteristics of participants and effect modifiers over time based on group membership is presented in Table S2. Analyses were limited to 913 study members in the multivariable trajectory model because of some missing data among the covariates. There were a number of individuals assigned to different trajectory groups as a consequence of adding all covariates (early-life factors and effect modifiers), resulting in a 4% improvement in model fit (Bayesian Information Criterion=−17577.86 versus −18172.22).
In the fully adjusted model (including all covariates), males had significantly greater odds of being in the high-normal, prehypertensive, and hypertensive group compared with the normal group (Tables 1 and 2). Participants who had a higher proportion of relatives with high blood pressure had greater odds of being in the prehypertensive (odds ratio [OR], 6.15; 95% CI, 1.39–27.14) or the hypertensive group (OR, 43.23; 95% CI, 5.27–354.65) compared with the normal group. Study members whose mothers had pregnancy hypertension had greater odds of being in the prehypertensive group (OR, 3.59; 95% CI, 1.24–10.34) but not the hypertensive group (OR, 0.92; 95% CI, 0.15–5.51) when compared with the normal group. Participants who were first born had higher odds of being in the hypertensive group (OR, 2.95; 95% CI, 1.00–8.69) compared with the normal group, and study members with lower birth weights had increased odds of being in the hypertensive group (OR, 2.79; 95% CI, 2.49–3.09). No significant associations were found for early socioeconomic status or breastfeeding.
Increasing BMI was significantly associated with an upward shift in all 4 blood pressure trajectory groups (Table 3). Analyses were then undertaken to determine whether BMI had a greater effect on the trajectory levels of the high-normal, prehypertensive, or hypertensive groups compared with those in the normal group (ie, whether coefficients were significantly different by group). The effect of BMI over time was significantly different (ie, stronger) for 2 groups (the prehypertensive, χ2=21.4; P<0.001 and hypertensive group, χ2=9.0; P=0.002) when compared with the normal group. Increasing number of daily cigarettes was also significantly associated with an upward shift in trajectories for the high-normal and prehypertensive groups. No significant associations were found between trajectory groups and alcohol consumption.
Mean values for cardiovascular risk outcomes at the age of 38 years based on trajectory group membership are presented in Table 4. Significant mean differences were found between groups and cardiovascular risk indicators. After controlling for covariates, significant differences were found for waist/hip ratio (F[3, 868]=42.59; P<0.001), total cholesterol level (F[3, 849]=11.08; P<0.001), high-density lipoprotein cholesterol level (F[3, 849]=4.00; P=0.008), glycohemoglobin (F[3, 839]=3.35; P=0.02), triglyceride level (F[3, 849]=16.59; P<0.001), and the presence of metabolic abnormality (χ2=8.22; P=0.04). In general, the higher the risk for high blood pressure, the greater the risk for other cardiovascular risk factors.
The earliest age at which medical practitioners can identify young healthy individuals who are likely to develop high blood pressure in adulthood is not known. This article addresses this question and reports 4 new findings: (1) those at greatest risk of developing high blood pressure in adulthood had elevated blood pressure at 7 years and were distinguishable from those with normal blood pressure trajectories from childhood into early midlife; (2) several early-life risk factors that antedated our first blood pressure measurement at the age of 7 years were identified using trajectory modeling. These risk factors differentiated between groups of individuals on healthy and unhealthy trajectories; (3) BMI was a major effect modifier for all blood pressure trajectories over time, with effects strongest among the highest blood pressure groups. Cigarette smoking also modified trajectories for the high-normal and prehypertensive groups; and (4) blood pressure trajectories leading to prehypertension and hypertension in adulthood, while of concern in and of themselves, also predicted a range of other cardiovascular risk indicators in early midlife, including a composite index of metabolic abnormalities, indicating even greater clinical salience of these high-risk trajectories.
Against a background of conflicting evidence about the pros and cons of population-based screening for blood pressure risk, new evidence aiding clinical risk assessments earlier in the lifecourse is timely. By implementing a trajectory approach, we found that individuals destined to become prehypertensive and hypertensive by early midlife comprised approximately one third of this general population sample. Moreover, by using data collected over time we were able to better predict those individuals who developed prehypertension and hypertension in early midlife compared with measures of blood pressure taken at only 1 time point (eg, 7 years; Table S1), suggesting the importance of routine blood pressure measurements beginning in childhood.
Previous blood pressure research has linked risk factors measured early in life with high blood pressure at a single point in time later in life. This research goes beyond that by using group-based trajectory modeling to link risk factors to trajectories capturing the entire developmental course of blood pressure from ages 7 to 38 years. These included several established risk factors such as having a family history of high blood pressure, male sex, maternal pregnancy hypertension, and lower birth weight. Current data strongly support male sex and a family history of hypertension as important early-life risk factors and these are recommended as part of the US Preventive Services Task Force risk assessment for primary hypertension in children. The ORs for male sex and having a family history of high blood pressure were high but the CIs for the ORs were wide, possibly reflecting the small number of participants in these groups, particularly in the hypertensive group. Identifying the effects of pregnancy and in utero factors, such as low birth weight, can help to determine the development of cardiovascular risk indicators, including high blood pressure.29 An association between maternal pregnancy hypertension and offspring’s blood pressure most likely results from both a genetic transmission from parent to child plus in utero effects. A more recent finding was also replicated in our study; that is, first born children had higher odds of being in a trajectory group, resulting in hypertension compared with the normotensive group. Although there are no obvious mechanisms underlying the link between being first born and having higher blood pressure, physical changes occur in the uterus during implantation and placentation in the first pregnancy. During later pregnancies, these changes are more apparent with increased efficiency of placental invasion of the uterine wall and subsequent improved nutrient flow to the fetus.30 Higher blood pressure in later life may, therefore, be associated with a degree of in utero nutrient restriction of first borns compared with later borns.31 Neither breastfeeding duration nor lower early childhood socioeconomic status were significantly associated with blood pressure trajectory group membership.
Knowledge about modifiable risk factors that alter blood pressure trajectories can help to determine intervention foci. This study showed that increasing BMI predicted an upward shift in blood pressure trajectory levels across the entire population. However, the impact of higher BMI was greatest among the higher blood pressure groups, suggesting excess body weight may be particularly problematic for individuals already on a trajectory toward developing higher blood pressure. Our results support recommendations for blood pressure screening of children and adolescents who are overweight and obese32 and population-based strategies that focus on maintaining a healthy weight. Similar to previous longitudinal studies,19 we found that cigarette smoking was a significant risk factor and associated with an upward shift in trajectories for those in the high-normal and prehypertensive groups (≈75% of population). No statistically significant association was found for the hypertensive group; this may be because of small numbers in that group. Alcohol was not associated with shifts in trajectory levels in our study participants as they entered early midlife. Future research is required to ascertain if this lack of association persists at older ages.
To determine the predictive validity of the developmental blood pressure trajectory groups, early-midlife cardiovascular correlates of these trajectories were examined. Individuals in the higher blood pressure trajectory groups were more likely to have higher levels of known cardiovascular risk factors at the age of 38 years. These findings suggest that high systolic blood pressure in early life can serve as a generic risk indicator for poorer cardiovascular health more generally. These findings also support recommendations that routine screening for increased blood pressure in childhood may help to prevent subsequent cardiovascular disease in later life via age-appropriate intervention.3,33 Further follow-up is needed to determine whether this will translate into decreased life expectancy and cardiovascular disease end points (eg, cardiac arrest and stroke).
There are several strengths and some limitations in this study. Few prospective studies exist with the breadth of data to investigate the range of determinants of blood pressure risk factor trajectories in this way. This representative population study has experienced low rates of attrition over almost 4 decades giving confidence in the observed associations. Early-life information was validated using hospital records and measures were created via multiple sources (eg, family history of high blood pressure). The limitations of this study must also be considered. First, despite having BMI information, we could not examine the association between physical activity and blood pressure because activity information was not collected at all time points. Second, although we were able to examine the effect that alcohol consumption and cigarette smoking had over time, frequency of use data were restricted to information collected at ages ≥18 years.
Millions die annually because of diseases directly associated with hypertension, and this number is rising. Intervening in early life to reduce high blood pressure may help increase life expectancy worldwide and reduce the associated burden of disease. The findings from this study suggest that groups of individuals who will develop high systolic blood pressure in early to mid-adulthood have elevated blood pressure in childhood, and that these individuals are more likely to be affected by cardiovascular disease–related comorbidity by age 38 years. In other words, estimating risk for future hypertension (and cardiovascular disease more generally) can be undertaken among children, long before systolic blood pressure rises to clinically detectable levels of prehypertension or hypertension in adulthood. Importantly, this research reinforces the importance of encouraging lifestyle changes including supporting the maintenance of a healthy body weight, particularly for those already on a high-risk blood pressure trajectory into adulthood.
We are indebted to Phil Silva, the founder of the Dunedin Study, and to the Study members and their families for their long-term involvement.
Sources of Funding
The Dunedin Multidisciplinary Health and Development Study is supported by the New Zealand Health Research Council (HRC) and the New Zealand Ministry of Business, Innovation, and Employment (MBIE). This research also received support from the United Kingdom Medical Research Council (Grant G0100527) and the National Institute of Aging (Grants R01AG032282 and R01AG048895). R. Theodore was supported by a University of Otago Health Sciences Postdoctoral Fellowship and an HRC Erihapeti Rehu-Murchie fellowship (Grant 13/579).
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.115.05831/-/DC1.
- Received July 23, 2015.
- Revision received August 3, 2015.
- Accepted August 21, 2015.
- © 2015 American Heart Association, Inc.
- Chobanian AV,
- Bakris GL,
- Black HR,
- Cushman WC,
- Green LA,
- Izzo JL Jr.,
- Jones DW,
- Materson BJ,
- Oparil S,
- Wright JT Jr.,
- Roccella EJ
- 2.↵Expert Panel of Integrated Guidelines for Cardiovascular Health and Risk Reduction in Children and Adolescents. Expert Panel on integrated guidelines for cardiovascular health and risk reduction in children and adolescence: summary report. Pediatrics. 2011;128:S213–S256.
- Pickering TG,
- Hall JE,
- Appel LJ,
- Falkner BE,
- Graves JW,
- Hill MN,
- Jones DH,
- Kurtz T,
- Sheps SG,
- Roccella EJ
- Chen X,
- Wang Y.
- Law CM,
- Barker DJ,
- Bull AR,
- Osmond C.
- Martin RM,
- Ness AR,
- Gunnell D,
- Emmett P,
- Davey Smith G
- Wang X,
- Poole JC,
- Treiber FA,
- Harshfield GA,
- Hanevold CD,
- Snieder H.
- DeFrank RS,
- Jenkins CD,
- Rose RM.
- Ostbye T,
- Malhotra R,
- Landerman LR.
- Sears MR,
- Greene JM,
- Willan AR,
- Taylor DR,
- Flannery EM,
- Cowan JO,
- Herbison GP,
- Poulton R.
- Melchior M,
- Moffitt TE,
- Milne BJ,
- Poulton R,
- Caspi A.
- Elley WB,
- Irving JC.
- Belsky DW,
- Caspi A,
- Goldman-Mellor S,
- Meier MH,
- Ramrakha S,
- Poulton R,
- Moffitt TE.
- Alberti KG,
- Eckel RH,
- Grundy SM,
- Zimmet PZ,
- Cleeman JI,
- Donato KA,
- Fruchart JC,
- James WP,
- Loria CM,
- Smith SC Jr.
- Nagin DS.
Novelty and Significance
What Is New?
In a longitudinal birth cohort study of 1000 participants, we conducted trajectory modeling to identify 4 distinct developmental systolic blood pressure trajectory groups from ages 7 to 38 years.
Groups at greatest risk of developing high blood pressure and related cardiovascular comorbidities (eg, higher total cholesterol) at the age of 38 years had higher blood pressure beginning at the age of 7 years compared with individuals in normal blood pressure trajectory groups.
What Is Relevant?
Individuals in harmful developmental blood pressure trajectory groups were more likely to be male, have a family history of hypertension, be first born, and born lower birth weight; information that is useful for screening purposes.
Higher body mass index and cigarette smoking over time resulted in increasing blood pressure levels, particularly for those individuals in higher blood pressure groups.
Encouraging lifestyle changes (eg, weight reduction, the maintenance of a healthy body weight, and smoking cessation) may help to lower blood pressure levels over time, particularly for those individuals on a trajectory to developing hypertension.
Our findings suggest that it is possible to identify early in life, harmful blood pressure trajectories that are associated with both antecedent and modifiable risk factors over time, and predict cardiovascular disease risk in adulthood.