Association of resting heart rate with blood pressure and incident hypertension over 30 years in Black and White adults: the CARDIA study

Abstract

Few studies have assessed the association of resting heart rate (RHR) through young adulthood with incident hypertension by middle age. We investigated the association between RHR measured over 30 years with incident hypertension in a cohort of young Black and White men and women.

A joint longitudinal time-to-event model consisting of a mixed random effects sub-model, quadratic in follow-up time, and a survival sub-model adjusted for confounders, was used to determine hazard ratios (HR) for a 10 beats-per-minute (bpm) higher RHR. Race-sex specific effects were examined in a single joint model that included interactions of race-sex groups with longitudinal RHR. Out of 5115 participants enrolled in Year 0 (1985–86), after excluding prevalent cases of hypertension at baseline, 1615 men and 2273 women were included in the analytic cohort.

Hypertension event rates per 1000 person-years were 42.5 and 25.7 in Black and White men, respectively, and 36.2 and 15.3 in Black and White women, respectively. The HRs for a 10 bpm higher RHR were 1.47 (95%CI: 1.23–1.75), 1.51 (95% CI: 1.28–1.78), 1.48 (95% CI: 1.26–1.73), and 1.02, (95% CI: 0.89–1.17) for Black men, White men, White women, and Black women respectively.

Higher RHR during young adulthood is associated with a greater risk of incident hypertension by middle age. The association is similarly strong in Black men, White men, and White women, but absent in Black women, which may suggest racial differences in the effect of sympathetic nervous activity on hypertension among women.

SUMMARY

The associations of higher resting heart rate with incident hypertension were similar in young Black men, White men, and White women, but absent in young Black women. This finding requires further investigation.

INTRODUCTION

Resting heart rate (RHR) is well recognized as an indicator of cardiovascular health and as prognostic for overall mortality and for cardiovascular disease (CVD) mortality. Several epidemiologic studies published this century, reviewed in Seviiri et al (2018)¹ and Tadic et al. (2018)², show a RHR measurement at a single occasion is associated with overall and CVD mortality. Furthermore, the temporal increase between two RHR measures over a decade span¹, a five-year span³, and a three-year span⁴ were also associated with increased risk for mortality¹, CV related events³ and heart failure³. In relation to blood pressure (BP) and incident hypertension^5–8 RHR measured at a single occasion^5,7,8 as well as at multiple occasions⁶ is a significant predictor of subsequent higher BP and incident hypertension. However, few studies have investigated the long term association, i.e., greater than 10 years, throughout young adulthood, between RHR and BP in later life. Because hypertension is a potent predictor for CVD events⁹, the utility of RHR as an upstream predictor of incident hypertension deserves careful examination. As there are sex and race differences in the long-term trends of BP¹⁰, whereas race and sex differences in the association of RHR during young adulthood with incident hypertension are not known, we have two aims in examining the utility of RHR in predicting hypertension. Using data from the Coronary Artery Risk Development in Young Adults (CARDIA) study, we assessed: (1) the functional form of the longitudinal RHR measurements and its association with incident hypertension by middle age and (2) whether the association differs by race-sex group. We hypothesize that the longitudinal RHR measures will be significantly associated with incident hypertension.

METHODS

Anonymized data and materials have been made publicly available at the National Institutes of Health’s Biologic Specimen and Data Repository Information Coordinating Center (https://biolincc.nhlbi.nih.gov/studies/cardia/).

Study participants

In 1985–1986 (Year 0), the CARDIA study recruited 5,115 Black and White men and women free of clinical CVD at baseline, 18 to 30 years of age at baseline (Year 0), from four field centers. The baseline exam was followed by 8 more clinic examinations in the subsequent 30 years. Sitting resting RHR and BP measurements were made at all nine examinations. Details on the design, recruitment, and cohort examination procedures have been published¹¹and a summary is available in the Online Supplement. All participants gave informed consent, and the CARDIA protocol was approved by the Institutional Review Board at each field center.

For the present study, we excluded one participant who withdrew study consent, 732 participants who had hypertension as defined by the ACC/AHA 2017 guidelines at Year 0, 136 participants who had no follow-up after baseline, 2 who were missing more than 7 of 9 RHR measurements, 6 missing the baseline RHR measurement, and 350 missing a baseline covariate, most often fasting glucose or insulin due to fasting fewer than 8 hours. This left 3888 available for the analytic cohort.

Exposure Measurement

A 30-second pulse was taken at the radial artery by palpation and counting after the BP cuff had been applied and prior to the start of BP measurement. The recorded value was multiplied by 2 to obtain beats per minute (bpm).

Blood Pressure and Hypertension Outcomes

Because the definition of hypertension was updated in 2017 by the publication of the ACC/AHA Guidelines¹², and a subsequent investigation⁹ found the relations of the revised cutoffs for the BP categories to be quite strong for CVD events and for all-cause mortality in adults younger than age 40 years, we restrict our primary examination to hypertension as defined in that guideline. A sensitivity analysis was also done (Table S2) using the earlier JNC7 criteria using the cutoffs of 140 and 90 mmHg for systolic and diastolic BPs. Details regarding the BP measurement protocol at all examinations – consistent with the 2017 ACC/AHA BP guidelines, “Accurate Measurement of BP in the Office”¹² - are provided in Yano et al.⁹ Hypertension was defined as systolic BP ≥130 mmHg or diastolic BP ≥80 mm Hg or taking an antihypertension medication. Seated BP was measured using a Hawksley random zero sphygmomanometer (Hawksley, Sussex, United Kingdom) for the Year 0 to Year 15 examinations and an Omron model HEM907XL for Y20 to Y30. To eliminate machine bias, a calibration study was conducted and values standardized to the sphygmomanometric measures were used for the Year 20, 25, and 30 BP measurements. The BP was measured on the right arm at three 1-minute intervals in a quiet room after a 5-minute rest; the average of the second and third measurements was used.

Assessment of Covariates

Demographic, anthropometric, and lifestyle measures obtained at the Year 0 examination were used in these analyses. Height and weight were measured with the participant wearing light clothing with no shoes, and body mass index (BMI) was computed. Age, race, years of education, and smoking status were self-reported. Alcohol intake (ml/d) was computed from self-reported frequency of consumption of beer, wine, and liquor per week¹³. Glucose was assayed using hexokinase coupled to glucose-6 phosphate dehydrogenase. Insulin concentrations were measured by a modification of the immunoassay techniques of Herbert et al.(1965)¹⁴ Physical fitness was assessed by treadmill duration using a modified Balke treadmill protocol (Sidney S, 1992)¹⁵.

Statistical Analyses

Descriptive baseline characteristics of participants were tabulated separately for each sex and race group. To assess the association of longitudinal RHR measures with incident hypertension, a joint longitudinal and time-to-event model was implemented with the R package “JM”¹⁶. This model is an extension of the widely used Cox proportional hazards model. The Cox model relates the expected hazard rate at time t to the product of a baseline hazard rate at time t and the exponential function of a linear combination of independent baseline variables: h(t) = h₀(t) exp(b₁×₁ + b₂×₂ + … +b_px_p). In the joint modeling setting, a mixed model function of the true longitudinal measurements, m(t), is added to the linear combination of independent variables: b₁×₁ + b₂×₂ + … +b_px_p + α m(t). The α then quantifies the effect of the true longitudinal biomarkers to the risk for an event. A more rigorous specification of the joint longitudinal – time-to-event model is given in the supplemental material. In this study examination of person-specific plots of RHR over time suggested modeling RHR as a linear function of time might not be adequate. Thus we considered three different mixed model formulations for the longitudinal RHR submodels. Model A, the simplest formulation, specified longitudinal RHR measures as a linear function of follow-up time. To allow for possible curvature in the longitudinal trend, two other models were examined, model B specifying longitudinal RHR measures as a quadratic function of time, and model C specifying a natural cubic spline. Each of the three longitudinal submodels adjusted for Year 0 age and included interactions of the race-sex groups with follow-up time. The survival submodel was adjusted for Year 0 age, BMI, waist-to-hip ratio, alcohol consumption, smoking status, treadmill test duration, years of education, fasting glucose, fasting insulin, parental history of hypertension, systolic BP, and race-sex groups. For the joint longitudinal and survival model, the “piecewise-PH-aGH” method was specified and an interaction factor (via the “interFact” argument) for the race-sex groups with RHR was included. Follow-up for Year 30 was completed by August 31, 2016. Because two of the three models, B and C, are not nested, the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were compared among the three joint models and a final joint model was determined. In these analyses, a smaller AIC and BIC is preferred. Statistical analyses were conducted using SAS for Windows, release 9.4 (SAS Institute Inc., Cary, NC, USA) and the “JM” package in R.3.5.1 (R foundation for statistical computing).

RESULTS

Year 0 characteristics for each sex-race group are shown in Table 1. Mean age was approximately 24 years in Black men and women and 25 years in White men and women. Mean baseline RHR was lowest in Black men and mean systolic BPs were lower in women than men. Mean BMI was highest in Black women and lowest in White women. Black men and women had greater proportions of current smokers and positive histories of parental hypertension than White men and women. The unadjusted rates/1000 person-years for incident hypertension were 42.5, 25.7, 36.2, and 15.3 in Black men, White men, Black women, and White women, respectively.

Table 1.

Year 0 characteristics by sex and race group.

Characteristics	Black men	White men	Black women	White women
N	767	848	1148	1125
# incident HTN cases	495	414	731	395
Rate/1000 person-years (95% CI)	42.5 (39.0, 46.5)	25.7 (23.4, 28.3)	36.2 (33.7, 38.9)	15.3 (13.9, 16.9)
Age, years	23.9 (3.8)	25.3 (3.4)	24.2 (3.8)	25.4 (3.4)
Resting heart rate, bpm	64 (9)	67 (10)	72 (11)	72 (11)
SBP, mmHg	112 (9)	111 (9)	106 (8)	104 (8)
DBP, mmHg	67 (8)	68 (8)	66 (8)	65 (7)
Body mass index, kg/m2	24.2 (3.8)	23.9 (3.2)	25.3 (6.0)	22.9 (4.0)
Waist:Hip ratio	0.81 (0.05)	0.83 (0.04)	0.74 (0.06)	0.72 (0.04)
Alcohol intake, ml	17 (30)	18 (24)	5 (12)	9 (14)
Treadmill duration, min	11.3 (2.1)	12.3 (2.1)	7.3 (1.8)	9.3 (2.1)
Education, years	12.9 (1.9)	14.6 (2.5)	13.1 (1.8)	14.6 (2.3)
Fasting glucose, mg/dl	83 (10)	84 (8)	79 (11)	81 (11)
Fasting insulin, μU/ml	10 (7)	9 (5)	13 (9)	9 (6)
Current smoker, n (%)	283 (37%)	240 (28%)	364 (32%)	295 (26%)
Parental history of hypertension, n (%)	407 (53%)	351 (41%)	647 (56%)	492 (44%)

Values shown are mean and STD unless otherwise noted.

Figure 1 (A) shows unadjusted mean RHR decreasing over 30 years in Black women, and White men and women. For Black males - the divergent group - mean RHR increased steadily the first 10 years and then continued to increase at a more moderate rate in the subsequent 20 years. Mean RHR was higher among those who eventually were diagnosed with incident hypertension in White women and all men, but not in Black women (Figure 1, B and C).

Figure 1:

Top panel A shows unadjusted mean resting heart rate (RHR) in entire cohort (N=3888 plotted over 30 years for Black men, White men, Black women, and White women separately. Middle B and bottom C panels show unadjusted mean RHR stratified by ACC/AHA 2017 followup hypertension status for Black and White women, and for Black and White men.

Among the three joint longitudinal – time-to-hypertension models considered, both the AIC and BIC were lowest (best) in model B. Table 2 shows the chief joint model results for model B. The complete joint model results can be seen in supplemental Table S1. The upper half of Table 2 has the estimated coefficients for the longitudinal submodel and at the lower half are the estimated coefficients for the survival submodel. The parameters of key interest in this table are labeled “Association parameter” for α, and “Association: Race-sex, interaction” for the interaction of α with Race-sex group, where “Race-sex” represents White men, Black women, or White women, leaving Black men as the reference group. However, it is first informative to note differences, or lack of differences, in the longitudinal evolutions of RHR in the four race-sex groups at the top of the table. All parameter estimates for the longitudinal submodel, i.e., linear and quadratic year terms and their interactions with race-sex groups, are highly significant (Table 2, top). Figure 2 illustrates the features of these estimates: intercepts for White men, Black women, and White women all significantly higher than the intercept for Black men indicating significantly lower baseline RHR in Black men; the linear year coefficient positive for Black men, but negative in the other race-sex groups, indicating increasing trend for Black men but decreasing for the other groups; and the curvature determined from the coefficient for the quadratic year component for Black men indicating downward concavity in the trend. The differences in the quadratic coefficients for the other three groups from Black men considerably attenuate a trend.

Table 2.

Estimated coefficients for time to incident hypertension by ACC/AHA 2017 criteria from a joint model with quadratic RHR submodel and a Cox survival submodel.

Joint longitudinal and survival submodels	N=3888 with 2035 hypertension events
	Estimate	95% CI	P-value
Submodel
Longitudinal RHR (model B)
Intercept	67.34	65.67, 69.02	<0.0001
Baseline age, years	−0.153	−0.217, −0.088	<0.0001
Black women	8.141	7.307, 8.975	<0.0001
White men	2.767	1.876, 3.657	<0.0001
White women	7.824	6.987, 8.660	<0.0001
Time, years	0.312	0.198, 0.427	<0.0001
Time2/100	−0.790	−1.228, −0.353	0.0004
Black women X Time	−0.520	−0.663, −0.377	<0.0001
White men X Time	−0.366	−0.514, −0,218	<0.0001
White women X Time	−0.540	−0.677, −0.404	<0.0001
Black women X (Time2/100)	0.992	0.447, 1.537	0.0004
White men X (Time2/100)	0.772	0.217, 1.327	0.006
White women X (Time2/100)	0.986	0.476, 1.497	0.0002
Cox incident HTN survival model*
Association parameter, bpm†	0.038	0.021, 0.056	<0.0001
Association:Black women, interaction, bpm‡	−0.036	−0.058, −0.014	0.001
Association:White men, interaction, bpm§	0.003	−0.021, 0.026	0.83
Association:White women, interaction, bpm∥	0.0006	−0.023, 0.024	0.96

bpm: beats per minute

^*Adjusted for Year 0 risk factors: Age, body mass index, waist:hip ratio, alcohol consumption, smoking status, treadmill duration, education, fasting glucose, fasting insulin, indicator of parent with hypertension, systolic blood pressure, sex, and race.

^†Hazard ratio for 10 bpm higher resting heart rate for Black men: 1.47 (95%CI: 1.23 – 1.75);

^‡for Black women: 1.02 (95%CI: 0.89 – 1.17);

^§for White men: 1.51 (95%CI: 1.28 – 1.78); and

^∥for White women: 1.48 (95%CI: 1.26 – 1.73).

Figure 2:

Fitted average 30-year trends in resting heart rate (RHR), clockwise from lower left, for Black men, White men, White women, and Black women based on a joint model with linear mixed effects submodel quadratic in year for the longitudinal RHR measures and a piecewise-constant baseline risk function for the relative risk model. Race-sex specific equations for longitudinal RHR are: RHR(Black men) = 67.344 – 0.153*baseline age + 0.312*year – (0.790/100)*year²; RHR(White men) = 70.111 – 0.153*baseline age - 0.054*year – (0.018/100)*year²; RHR(White women) = 75.168 – 0.153*baseline age - 0.228*year + (0.196/100)*year²; RHR(Black women) = 75.485 – 0.153*baseline age - 0.208*year + (0.202/100)*year².

While the RHR longitudinal patterns for Black men differed with those of the other race-sex groups, the association of RHR with incident hypertension was similarly positive and significant for Black and White men and White women, but null for Black women. The RHR association parameter for Black men (estimate=0.0385, p<0.0001) is highly significant, yielding a hazard ratio of 1.47 for a 10 bpm higher RHR. The differences in the association parameters for White men and White women from that for Black men are not significant (estimate=0.003, with p=0.83, and estimate=0.0006 with p=0.96, respectively). These estimates yield for a 10 bpm higher RHR, hazard ratios for incident hypertension of 1.51 (95%CI: 1.18, 1.62) for White men and 1.48 (95% CI: 1.26, 1.73) for White women. For Black women the difference in the RHR association parameter is highly significant (estimate=−0.036, p=0.001) and it determines a null association for incident hypertension for Black women (hazard ratio=1.02).

To illustrate the risk for incident hypertension as provided by these joint models, we show in Figure 3 the predicted survival-from-hypertension curves (solid lines) with 95% confidence limits (dashed lines) for hypothetical participants who are similar on all observed values, including RHR, in the two submodels, except for race and sex. The upper two are for a White man (green lines) and Black man (Black lines), respectively, and the lower two are for a White woman (green lines) and Black woman (Black lines), respectively. The points plotted on the left side of the two panels from 0 to 15 years represent a hypothetical observed sequence of RHR for the four hypothetical participants known to be free of hypertension through year 15. The declining curves from 15 to 30 years are the predicted probabilities of remaining free of hypertension. For the White male, his predicted probability of remaining free of hypertension at 29.7 years is 0.66 (95% CI: 0.54, 0.74), and for the Black male, his predicted probability at the same time point is 0.50 (95% CI: 0.40, 0.60). The corresponding quantities for the White and the Black woman at 29.7 years are 0.77 (95%CI: 0.69, 0.83) and 0.57 (95%CI: 0.53, 0.61), respectively. It should be noted that while the predicted probability of remaining hypertension free is lower for the Black woman, the RHR hazard ratio computed for Black women (1.02, 95% CI: 0.89, 1.17) suggests little or no contribution of RHR to incident hypertension in Black women.

Figure 3:

Top A and bottom B panels on right show predicted probabilities of remaining hypertension free for hypothetical male (top) and female (bottom) participants who have the same average characteristics on all covariates, except race and sex. Solid black and green lines represent Black and White participants, respectively. Dashed lines represent 95% confidence intervals. Points plotted on the left side of the panels represent the resting heart rate (RHR) measures observed from Year 0 to Year 15 for the hypothetical participants. For both upper and lower plots, the hypothetical observed RHR values are: 64, 66, 70, 70, 64, and 74. All hypothetical participants depicted are assumed to be hypertension free at Year 15.

DISCUSSION

In this prospective cohort of Black and White young adults free of hypertension and CVD at baseline, our principal finding is that higher RHR during young adulthood is associated with a greater risk of incident hypertension by middle age in Black men and in White men and women. The association is absent in Black women. In an earlier report, a trend of increasing RHR over 10 years was evident for Black men, but decreasing trends for Black women and White men and women⁵. We found in the subsequent 20 years, the increasing trend in RHR persisted in Black men, but the steepness of the increase was attenuated and reached a plateau at 20 years. These temporal characteristics in RHR are illustrated more clearly in the plots of the fitted longitudinal models (Figure 2).

The race- and sex-specific temporal characteristics of RHR in CARDIA can be compared with other population based studies. The normative RHR data from NHANES, 1999–2008¹⁷ included both Black and White participants for comparison. While both CARDIA and NHANES report race differences in RHR temporal trends among men, the race differences in CARDIA, as already noted, are more forewarning than those reported in NHANES. Whereas NHANES reported significantly lower mean RHR among Black men than White men, no trends across age groups were reported for either race. In contrast, in CARDIA, we found temporal trends in opposite directions, that is, increasing and detrimental in Black men and decreasing in White men. In NHANES, decreasing trends across age groups were present among females, with Black females having the significantly lower mean RHR. In CARDIA, temporal differences between Black and White females were absent, but notably, the association of RHR with incident hypertension was present in White women and absent in Black women.

The Medical Research Council National Survey of Health and Development (NSHD) of the United Kingdom¹⁸ is another source for population based data, although the cohort did not include Black participants. A unique feature of NSHD is that the RHR trajectories depicted for males and females span six decades, from ages 6 through 69 years. The mean RHR trajectories for both males and females declined over the six decades with females having the higher mean from age six until approximately age 50, where the male and female trajectories converged. The trajectories during adult years were in general consistent with the trends seen in White men and women in the present study.

The Jackson Heart Study¹⁹ reported on the prognostic ability of RHR with respect to mortality and heart failure hospitalization in their Black cohort, finding higher RHR was indeed prognostic for these outcomes. This was an important finding because most previous RHR studies were conducted in White populations. It is important to note that the Jackson Heart Study did not report on any sex-specific associations in Blacks. In view of our findings (i.e. a hazard ratio not significantly different from one for incident hypertension in Black women and a curious difference in risk between Black men and Black women), it would be informative to assess the sex-specific prognostic ability of RHR for incident hypertension as well as for mortality and heart failure in the Jackson Heart study.

The results of the joint RHR-incident hypertension model raise some considerations for future research. The finding of longitudinal RHR as an upstream predictor of hypertension in healthy young adults suggests future studies could move towards determining specific RHR ranges that might indicate a young adult should implement lifestyle changes potentially to delay or altogether avoid the onset of hypertension. Indeed, various proposals have been made since the latter part of the twentieth century to establish normal RHR values in adults^20–22. Spodick et al. (1992)²⁰ cited several large epidemiologic studies conducted since the late 1970’s to support his proposal to modify the accepted 60 to 100 bpm limits for normative heart rate at that time. Subsequent recommendations were made in 1999²³ and 2018^21,22. The primary evidence supporting these proposals was the consistent and robust measures of risk for CVD events or mortality provided by RHR measurements. Summary RRs per 10 bpm higher RHR for several outcomes downstream from hypertension were reported in a meta-analysis²⁴ that included 87 studies published through 29 March 2017: coronary heart disease (RR: 1.07, 95%CI: 1.05–1.10), sudden cardiac death (RR: 1.18, 95% CI: 1.10–1.27), heart failure (RR: 0.97, 95% CI:0.92–1.02), atrial fibrillation (RR: 1.06, 95% CI: 1.02–1.10), total stroke (RR: 1.15, 95% CI: 1.11–1.18), and cardiovascular disease (RR: 1.14, 95% CI: 1.06–1.23). In the current study, we document for a 10 bpm increase in the true RHR, relative risks for incident hypertension of approximately 1.5 for three of four race-sex groups. In another phase of research an obvious consideration would be the appropriate sampling frequency of RHR. The data currently available in existing cohort studies are not useful as the RHR measures are taken in intervals from 2 to 5 years apart. In the present study among the men and women who were hypertension free at baseline, 83 men and 53 women became hypertensive by exam year 2. A goal would be to identify such participants early on and thwart the potential elevation of risk. In the competitive and elite sports world, the monitoring of heart rate indices, including RHR and heart rate variability, for improving athletic performance is an active area of research²⁵. Given the global popularity of wrist worn activity trackers (e.g., Fitbit devices or the Apple Watch) – which measure heart rate most reliably when the wearer is in the resting state^26,27, establishing alert ranges of RHR for incident hypertension would benefit the average person.

Another obvious objective for future research studies would be to elicit a better understanding of the underlying mechanisms that give rise to hypertension as well as what instigates race and sex differences. Sympathetic nervous overactivity is one suspected component mechanism behind hypertension²⁸. In the past, higher RHR was viewed as a surrogate for sympathetic overactivity^28,29, although recent views are much more temperate^30,31. Still, findings reported in small mechanistic studies allude to possible differential roles for sympathetic nervous activity by race. In a study comparing healthy, young Black and White men, the authors concluded that the Black men exhibited exaggerated sympathetic vascular transduction at rest³². This would be consistent with the present’s study finding of increasing RHR in the Black men. Another study in Black and White women concluded that nonsympathetic mechanisms had a greater role in hypertension in Black women³³. Given the small size of these studies – with 35 men³² and 42 women³³ – the conclusions should not be regarded as unequivocal but they should spawn larger, population based studies to better elucidate the roles of sympathetic and parasympathetic nervous activity in hypertension and the potential interactions with race and sex.

It is beyond the scope of this study to delve into the complex roles of the sympathetic and parasympathetic divisions of the autonomic nervous system in regulating heart rate³⁴. Evidence that there are sex differences and age-related changes in the relative contributions of these systems to BP regulation^35,36 can only allow us to speculate what mechanisms may be at play in our findings. However, findings in other CARDIA studies might provide areas for further study. Gabriel (2018)³⁷ reported that Black men, although being the most physically active at a baseline assessment in 2005–06, had the greatest reductions in physical activity over the subsequent 10 years. This is one possibility that might contribute to the trend of increasing RHR for Black men. Furthermore, if it is the balance of the sympathetic and parasympathetic nervous systems that is relevant, other findings suggest psychosocial stress may be an avenue worth additional study as a contributor to increased sympathetic activity^38–40.

Strength and Limitations

A particular strength of this study is that we modeled the survival process for hypertension and the mixed model evolution of RHR simultaneously. This approach reduces the effect of measurement error in the RHR observations¹⁶ that would have occurred had a Cox regression model with time dependent RHR measures been used. Referring to the online supplement, it is the true value, m_i(t) rather than the observed value, y_i(t), that is inserted into the survival model. While modeling the true repeated measures of RHR in the survival model is a strength, a constraint is that we did not assess the associations of other longitudinal covariates, such as BMI, simultaneously with hypertension in this study. The possibility of including multiple longitudinal covariates in a joint model is a recent development⁴¹. Additionally, the computational complexity and the technical challenges of the method are significant,⁴¹ therefore we will pursue the method in a separate future study.

CONCLUSIONS

We identified a significant association of longitudinal RHR during young adulthood with incident hypertension by middle age. RHR measurements during young adulthood may help identify young adults at high risk for hypertension by middle age: future studies might try to determine the threshold for RHR associated with hypertension risk. Given RHR is an easily measured index of health, future research studies could benefit by including RHR when opportunities arise.

Novelty and Significance.

WHAT IS NEW?

• In a population based cohort study, young adults with higher resting heart rate had higher risk for incident hypertension by middle age.

WHAT IS RELEVANT?

• Resting heart rate is an easily measured index and may inform young adults of higher risk of future hypertension.

PERSPECTIVES.

The prognostic ability of RHR for adverse cardiovascular disease events is well known. Knowledge that RHR is an upstream predictor of hypertension for young adults may encourage self-monitoring for temporal increases in RHR and thus provide a signal for heightened cardiovascular risk.