aObserved participants in each group; trajectories estimated from sample with risk factors imputed (N = 4779).
eMethods. Statistical Analysis
eTable 1. Model Fit Statistics for Multilevel Models
eTable 2. Descriptive Statistics for Selected Early-Life Factors and Smoking Status
eTable 3. Estimated Difference According to Selected Early-Life Factors: Estimates Derived From Multilevel Models Adjusted for Each Early-Life Factor and Adjusted for Age and Sex
eTable 4. Association Between Early-Life Factors of Interest and Life-Course Resting Heart Rate Trajectory: Estimates From Multilevel Models Adjusted for Age and Sex in Complete Case Sample
eFigure 1. Observed Mean Resting Heart Rate Across the Life Course According to Early-Life Factors
eFigure 2. Estimated Resting Heart Rate Trajectory Across the Life Course According to Early-Life Factors in Women
Customize your JAMA Network experience by selecting one or more topics from the list below.
O’Hare C, Kuh D, Hardy R. Association of Early-Life Factors With Life-Course Trajectories of Resting Heart Rate: More Than 6 Decades of Follow-up. JAMA Pediatr. Published online April 01, 2018172(4):e175525. doi:10.1001/jamapediatrics.2017.5525
Are prospectively measured early-life factors associated with trajectories of resting heart rate across the life course (measured from age 6 to 69 years)?
In this population-based cohort study, multiple readily identifiable and modifiable early childhood factors are associated with higher life-course resting heart rate. Higher birth weight and better childhood growth were associated with lower resting heart rate across the life course; later walking age was associated with a higher childhood resting heart rate, while associations between socioeconomic position and breastfeeding and the resting heart rate trajectory emerged in adulthood.
Early life is a key period associated with resting heart rate across the life course.
Higher resting heart rate (RHR) is associated with increased risk of cardiovascular and all-cause mortality. Limited attention has been paid to early-life determinants of life-course RHR.
To describe trajectories of RHR in the same individuals from age 6 to 69 years.
Design, Setting, and Participants
Data were from the Medical Research Council National Survey of Health and Development population-based cohort of individuals who were followed up from birth in 1946 until 2015. Analysis was conducted between September 2016 and June 2017. Multilevel models were used to estimate life-course mean RHR trajectory by sex and to investigate associations with early childhood factors. The maximal sample included participants who had at least 1 measure of RHR since study entry and a measure of birth weight (N = 4779; observations = 26 182).
Information on early-life factors was ascertained prospectively: childhood socioeconomic position, birth weight, body mass index (calculated as weight in kilograms divided by height in meters squared) change from age 2 to 6 years (conditioned on body mass index at age 2 years), duration of breastfeeding, and markers of neurodevelopment (age at first walking independently and cognitive ability at age 8 years).
Main Outcomes and Measures
Resting heart rate measured on 8 occasions between age 6 and 69 years (3 occasions in childhood and 5 in adulthood).
Of 4779 participants, 2492 (52.1%) were women, and 2287 (47.9%) were men. Mean estimated RHR decreased with increasing age and plateaued in adulthood. In sex-adjusted analyses, higher birth weight and conditional body mass index change were associated with lower RHR at age 6 years and across the life course (–0.56 bpm [95% CI, –0.95 to –0.17] per 1 kg higher birth weight and –0.30 bpm [95% CI, –0.48 to –0.13] per 1 kg/m2 change in body mass index). Associations between socioeconomic position and breastfeeding on RHR trajectory emerged in adulthood such that by age 69 years, RHR was 1.48 bpm (95% CI, 0.45 to 2.51) higher in participants from a disadvantaged vs advantaged background and –1.34 bpm (95% CI, –2.39 to –0.29) lower for those who were breastfed for 8 months or longer vs never. A later age at first walking was associated with higher RHR at age 6 years (1.49 bpm [95% CI, 0.39 to 2.59] higher for those 18 months or older vs those younger than 12 months) but with lower RHR in adulthood (–1.18 bpm [95% CI, –2.75 to 0.39] at age 69 years).
Conclusions and Relevance
Early life is a key period in determining future RHR trajectories with associations with potentially modifiable factors persisting into the seventh decade of life.
Higher resting heart rate (RHR) is associated with increased risk of cardiovascular and all-cause mortality.1,2 Elevated RHR even in childhood may increase risk of future mortality, and this risk may increase with age at RHR measurement.2,3 Resting heart rate is perhaps the simplest and cheapest cardiovascular risk marker to measure4; however, to our knowledge, no study to date has described trajectories of age-related changes in RHR in the same individuals followed up from childhood to later life. Studies that have reported variation in RHR with age are limited by use of age-heterogeneous cohorts and/or by duration of longitudinal follow-up.5-7
Cardiovascular risk may track from early life across the life course, and early life is a key period during which future cardiovascular disease (CVD) risk may be determined8; however, limited attention has been paid to early-life determinants of RHR. We hypothesize that early-life markers of better early childhood health and development confer a life-course advantage with beneficial effects persisting into later life and reflected in lower RHR across the life course. We aim to describe the average RHR trajectory from childhood until age 69 years by sex and to examine the extent to which RHR trajectories are associated with selected early-life factors.
The Medical Research Council National Survey of Health and Development (NSHD) is based on a nationally representative sample of 5362 births of all single births to married mothers that occurred in 1 week in March 1946 in England, Scotland, and Wales. The most recent data collection was conducted between 2014 and 2015 when participants were aged 68 to 69 years. Analysis was conducted between September 2016 and June 2017. Ethical approval was obtained from the Queen Square Research Ethics Committee (14/LO/1073) and Scotland A Research Ethics Committee (14/SS/1009). All participants provided written informed consent.
Resting heart rate was first measured in 1952 when participants were aged 6 years, at which time 4603 individuals provided data (excluded: 232 deaths, 1 refusal, 271 living abroad, and 255 failure to contact). The maximal sample for the current analyses was identified as those participants who had at least 1 measure of RHR and measurement of birth weight (N = 4779).
Resting heart rate, in beats per minute (bpm), was measured on 8 occasions. At age 6, 7, and 11 years, RHR was measured twice by palpation at the radial artery by a physician during physical examination at the participant’s school. Measurements were made in the seated position at the beginning and end of the examination. For these analyses, the second measure was used, which was on average the lower of the 2. At age 36 and 43 years, RHR was measured once at the radial artery by palpation by a trained nurse. At age 53, 60 to 64, and 69 years, RHR was measured using an automated blood pressure device (Omron); 2 consecutive measures were taken at age 53 years, and 3 measures were taken at subsequent ages. In analyses, the second measure was used except where missing, when the first measure was used.
Relevant early-life factors were those that have been shown to be associated with CVD in adult life.8-11 Childhood socioeconomic position (SEP) was measured by father’s occupational class, reported when the study participant was aged 4 years (or aged 7 or 11 years, if missing) and defined according to the Registrar General’s classification into 3 groups (classes I and II, professional and managerial; class III, skilled nonmanual and manual; and IV and V, semiskilled or unskilled manual occupations). Birth weight to the nearest quarter of a pound was extracted from hospital records within a few weeks of delivery and converted to kilograms (to convert pounds to kilograms, multiply by 0.45). When the participants were aged 2 years, duration of breastfeeding in months was reported by their mothers. This was grouped into 3 categories: never, 1 to 7 months, and 8 months or longer. Age at first walking, a gross motor milestone and conceptualized as the initiation of physical activity and as a marker of neurodevelopment,12 was reported by the mother when the participant was aged 2 years and classified as early (<12 months), normal (12-17 months), or late (≥18 months). Weight and height were measured by health visitors using standardized protocols at age 2 and 6 years, and body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Height increase (in centimeters) from age 2 to 6 years was derived by subtracting height at age 2 from height at age 6 years. Change in BMI was calculated in the same manner. At age 8 years, children underwent cognitive testing using a battery of tests measuring reading comprehension, pronunciation, vocabulary, and nonverbal reasoning, which were administered by participants’ school teachers. Cognitive scores from each of the tests were standardized, summed, and then restandardized to the maximum sample to have a mean of 0 and an SD of 1.
Smoking status was considered a potential confounder as it has been associated with almost all adult cardiovascular risk factors,13 and a recent large mendelian randomization meta-analysis supported a possible causal association between smoking and higher heart rate (but not hypertension).14 Information on current smoking habits were available at each of the RHR measurements in adult life. Current smokers at each age were coded as 1 and nonsmokers as 0. This was allowed to vary over time with childhood smoking status set to 0 for all study participants.
To estimate the mean RHR trajectory as a function of age, multilevel models were fitted. Age was centered at 6 years, the age at first RHR measurement. Models included a random intercept and random slope, which allowed each individual’s intercept and slope to vary around the mean. The change in RHR was modeled with quadratic and cubic age terms to reflect nonlinear change in mean RHR with increasing age. Differential effects of age according to sex were then tested by adding interactions between sex and all age terms. Where there was evidence of a sex difference, these terms were retained in all further analyses. Separate multilevel linear models were first fitted for each putative early-life factor, allowing both the intercept and the slope in RHR to vary according to each factor. Interactions with nonlinear terms for age were added, and the most parsimonious model was chosen by removing the nonsignificant higher order interaction terms (P > .10). The Bayesian Information Criterion and Akaike Information Criterion were used to assess the model fit and aid the model choice (eTable 1 in the Supplement).
Early-life factors that were associated with RHR trajectories in sex-adjusted models were then added 1 at a time into a series of adjusted models. Socioeconomic position was entered first to reflect our hypothesis that SEP acts both prior to birth and across early childhood. The remaining early-life variables were then entered sequentially based on their chronological ordering (ie, birth weight, breastfeeding, walking age, growth at age 2-6 years, and cognitive ability at age 8 years). Conditional early childhood growth was examined by entering both BMI and height at age 2 years and change in BMI and height between age 2 and 6 years simultaneously into models. Life-course smoking status was added to the final multivariable model as a time-varying covariate.
To maintain sample size and minimize bias introduced by missing data, we employed a multiple imputation procedure (eMethods in the Supplement) to impute missing data on childhood explanatory factors. We also conducted analyses in the complete case sample and the maximum sample available for each factor. We additionally conducted several sensitivity analyses (eMethods in the Supplement).
There were 4779 participants in the final sample of whom 4420 (92.6%) had at least 3 observations of RHR across time, 239 participants (5.0%) had 2, and 120 participants (2.5%) had 1. eTable 2 in the Supplement describes the maximal sample according to the selected early-life factors. Mean observed RHR decreased with increasing age and plateaued from early adulthood; estimated mean RHR remained faster in women, although the gap tended to lessen with age (Figure 1). For example, at age 6 years, the mean estimated difference in RHR between males and females was 1.99 bpm (95% CI, 1.43 to 2.55), reducing to 0.48 bpm (95% CI, –0.16 to 1.11) at age 53 years. eFigure 1 in the Supplement displays the observed RHR across the life course according to each early-life factor of interest.
There was no association between childhood SEP and RHR in childhood; however, SEP was associated with linear change with age (Figure 2A, Table 1, and eFigure 2A in the Supplement). Those from a more disadvantaged childhood SEP had a slower linear decline in RHR (0.021 bpm/year [95% CI, 0.0003-0.04] slower for lowest vs highest, P = .05) and thus a faster RHR in adult life than those from more advantaged backgrounds. At age 36 years, those in lowest social class had an estimated RHR of 0.78 bpm (95% CI, 0.20-1.36) higher than those in the highest social class, and this increased to 1.48 bpm (95% CI, 0.45-2.51) at age 69 years (eTable 3 in the Supplement).
A linear inverse association between RHR and birth weight was evident at age 6 years (–0.56 bpm [95% CI, –0.95 to –0.17] slower per 1 kg higher birth weight). Those with lower birth weight had a consistently higher mean RHR across the life course (Figure 2B, Table 1, and eFigure 2B in the Supplement), but there was no evidence of an association with rate of change.
Duration of breastfeeding in infancy was not associated with RHR at age 6 years. There was weak evidence of an association with linear change in RHR with those exposed to breastfeeding for 8 months or longer tending to have a faster linear decline in RHR with age compared with those never breastfed (–0.021 bpm [95% CI, –0.04 to 0.0003; P = .054]) (Figure 2C, Table 1, and eFigure 2C in the Supplement). Resting heart rate was –0.65 bpm (95% CI, –1.25 to –0.05) slower at age 36 years and –1.34 bpm (95% CI, –2.38 to –0.29) slower at age 69 years for those who were breastfed 8 months or longer vs never (eTable 3 in the Supplement).
At age 6 years, those who first walked independently when they were aged 18 months or older had a higher RHR (1.49 bpm [95% CI, 0.39 to 2.59]), relative to those who took their first steps when they were younger than 12 months. Midlife saw a crossover such that by later life, these patterns tended to the reverse association (Figure 2D, Table 1, and eFigure 2D in the Supplement) (P for age interaction = .01). For example, at age 43 years, there was no estimated difference between categories, while at age 69 years, those who walked later in infancy were estimated to have a lower RHR (–1.18 bpm [95% CI, –2.75 to 0.39]) (eTable 3 in the Supplement).
A greater change in BMI between age 2 and 6 years (conditioned on BMI at 2 years) was associated with a lower RHR at age 6 years and overall lower RHR trajectory across the life course (–0.30 bpm [95% CI, –0.48 to –0.13] per 1 kg/m2 change in BMI). Conditional change in height was not associated with RHR. Cognitive ability at age 8 years was not associated with RHR. Results were similar in the complete case sample (n = 2892; eTable 4 in the Supplement).
After mutual adjustment for all early-life factors, coefficients for most early-life factors were only slightly attenuated from sex-adjusted models. However, entry to the model of conditional early childhood growth substantially reduced the estimate of the association of birth weight with RHR (Table 2, model 5). Resting heart rate was higher among smokers than nonsmokers at all ages, and following adjustment for smoking status associations with SEP were slightly attenuated; however, substantive conclusions regarding the size and direction of other associations remained unchanged (Table 2, model 6).
Examination of linear trajectories from age 6 to 11 years and 36 to 69 years yielded results in line with the main analysis, including after adjustment for adult life self-reported physical activity. Models excluding those with documented heart trouble in childhood and those with any history of illness recorded from birth to age 5 years yielded similar results to the primary findings. Results were similar before and after adjustment for RHR measurement technique.
To our knowledge, we are the first to describe life-course trajectories of RHR in the same individuals followed up from early childhood until the end of the seventh decade. The RHR trajectories are broadly comparable with those inferred from prior cross-sectional population-based studies5,15 and/or age-heterogeneous longitudinal cohorts with shorter periods of follow-up.6 Resting heart rate decreases from childhood to adult life, plateaus earlier in men than in women, and is, on average, higher in females than males across the life course. The NSHD data provide some evidence that early-life factors are associated with RHR across life. Higher birth weight and conditional BMI change were associated with lower RHR at age 6 years and across the life course. Although not associated with childhood RHR, inverse associations with childhood SEP and duration of breastfeeding emerged in adulthood. Childhood cognitive ability was not associated with RHR, but later age at walking independently was associated with higher RHR at age 6 years but slower RHR in adulthood.
We are aware of no prior studies investigating childhood SEP associated with change in RHR from childhood into later life. The rate of change in RHR from childhood to adulthood is less steep in those from more disadvantaged backgrounds, leading to a clear inverse socioeconomic gradient in RHR in adult life. Consistent with our findings pointing to socially determined exposures earlier in life being translated into higher RHR in adulthood, a French population-based cohort reported that a small percentage of the SEP gradient in adult RHR was attributable to differences in leg length.16 A primary mechanism via which SEP may translate into higher RHR is via health behaviors. We therefore adjusted for a strong determinant of adult CVD and RHR (ie, smoking status). Although reduced after this adjustment, some evidence of an SEP gradient persisted; therefore, other pathways may also explain how a disadvantaged childhood translates into higher RHR.17 However, in sensitivity analyses, SEP coefficients were similar before and after adjustment for adult life self-reported physical activity.
Phillips and Barker18 were the first to report an inverse association between birth weight and RHR when measured at a single time in a middle-aged Hertfordshire cohort. A Japanese study reported the same direction of association in a convenience sample of boys aged 12 to 13 years.19 More recently, the age-heterogeneous Bogalusa Heart Study reported that lower birth weight was associated with a higher RHR but only in adults examined at age 50 years and not in children or adolescents.7We show that an inverse association between birth weight and RHR is apparent in childhood and persists across the life course. However, our findings point to a downstream association between childhood growth and RHR, which mediates the association with birth weight. It is of note that a change in BMI, rather than change in height, was more strongly associated with RHR, as this suggests an impact of greater relative adiposity20 and/or timing of the adiposity rebound.21
In the NSHD data, there was a weak association between breastfeeding and the rate of change of RHR in adult life. Given the social patterning of breastfeeding in this cohort, as in others,9,22 our findings may reflect residual confounding by SEP. Alternatively, as recent investigations incorporating detailed physiological assessment point to associations between breastfeeding and subclinical markers of vascular health as early as age 6 years,23 our findings may provide clues as to when in the life course such effects become clinically manifest. Moreover, although a recent comprehensive systematic review concluded that there was no association between breastfeeding and systolic blood pressure nor cholesterol levels, there was an association with risk of diabetes and overweight.24 Notably, both diabetes and overweight have been associated with increased sympathetic tone, which may also be reflected in higher RHR.25,26
Higher childhood cognitive ability may be reflected in lower adult CVD risk27; however, we found no association with RHR. An earlier marker of neurodevelopment, gross motor milestone attainment was associated with RHR. Both later walking age28 and higher RHR may reflect poor health, thus potentially explaining the association observed in childhood. It is important to note that in the NSHD data, walking age is socially patterned—participants from a less advantaged background on average walked earlier—thus, the reverse association emerging in adult life may instead reflect residual confounding by SEP and/or greater loss to follow-up (due to death or withdrawal from the study) of less advantaged participants.
The main limitation to this work is the lack of observed RHR measures between age 11 and 36 years. Sensitivity analyses modeling trajectories in childhood and at age 36 years and older separately suggest similar patterns across each of the early-life factors investigated. We also addressed potential reverse causality by repeating analyses excluding those with contemporaneously documented childhood ill health. Although early-life risk factors have also been associated with blood pressure in adult life,29 which is correlated with RHR, blood pressure was not measured in this cohort until participants were aged 36 years. However, RHR has been associated with poor CVD outcomes independently of blood pressure,1 and higher RHR may precede hypertension.30
Missing data and loss to follow-up may have introduced bias but are limitations common to long-running birth cohort studies such as the NSHD. Dropout may have been greater among participants with higher RHR and among those who were more disadvantaged, which may have led to an underestimate of the associations in later life. We dealt with missing data in covariates by using multiple imputation and found that similar associations were seen in the imputed and complete case analyses. Given changing secular trends, for example increasing obesity and survival of greater numbers of babies with very low birth weight, our findings may also be less applicable to contemporaneous cohorts. Further, NSHD is ethnically homogenous, thus limiting generalizability of our findings to more diverse populations. The early-life factors examined are indirect markers for the hypothesized underlying physiological processes and in part reflect the historical nature of some of the data collection. For example, the assessment of RHR has varied as collection methods have modernized. However, RHR measured by palpation is highly correlated with automated measures of RHR.3,31 The main strength of the study is thus the prospective ascertainment of exposures and multiple assessments of RHR in a large sample representative of the British-born population that has been followed from birth during 7 decades into later life.
In line with associations with other markers of adult cardiovascular risk, early life appears also to be a key period in determining life-course RHR trajectories. Childhood SEP and socially patterned factors (eg, breastfeeding, walking age) influence the rate of change in RHR leading to important differences in RHR more than 6 decades later. Critically higher RHR and increasing RHR in midlife3 are independently associated with CVD risk; thus, future work should aim to better understand the type and timing of appropriate interventions to improve outcomes.
Accepted for Publication: December 7, 2017.
Published Online: February 12, 2018. doi:10.1001/jamapediatrics.2017.5525
Correction: This article was corrected on August 13, 2018, to indicate that it is open access.
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2018 O’Hare C et al. JAMA Pediatrics.
Corresponding Author: Celia O’Hare, MB, BCh, BAO, Medical Research Council Unit for Lifelong Health and Ageing at University College London, 33 Bedford Pl, London WC1B 5JU, United Kingdom (email@example.com).
Author Contributions: Dr O’Hare had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: O’Hare, Hardy.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: O’Hare.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: O’Hare.
Obtained funding: Kuh.
Study supervision: Hardy.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by the UK Medical Research Council, which provides core funding for the Medical Research Council National Survey of Health and Development and supports the authors (MC_UU_12019/1, MC_UU_12019/2. and MC_UU_12019/3).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We thank National Survey of Health and Development study members for their lifelong participation and past and present members of the National Survey of Health and Development study team who helped to collect the data. Data used in this publication are available to bona fide researchers on request to the National Survey of Health and Development Data Sharing Committee via a standard application procedure. Further details can be found at http://www.nshd.mrc.ac.uk/data; https://doi.org/10.5522/NSHD/Q101; https://doi.org/10.5522/NSHD/Q102; and https://doi.org/10.5522/NSHD/Q103.
Create a personal account or sign in to: