Relationship of Tachycardia With High Blood Pressure and Metabolic Abnormalities
A Study With Mixture Analysis in Three Populations
Abstract Faster resting heart rate has been shown to be associated with a higher risk of developing hypertension and a greater incidence of cardiovascular morbidity and mortality. The aim of this study was to investigate the distribution of heart rate and its relationship with blood pressure and other cardiovascular risk factors in three populations. One European general population (Belgian study), one North American general population (Tecumseh study), and one European hypertensive population (HARVEST trial) were studied. Within each population, mixture analysis was used to investigate whether a mixture of two normal distributions explained the variance in heart rate better than a single distribution. In the men of all populations, mixture analysis identified a larger subpopulation of subjects with normal heart rate and a smaller one with fast heart rate. The subgroups with tachycardia had higher blood pressure and lipid levels than those with normal heart rate. In the populations in which they were measured, fasting insulin and postload glucose were also higher in the men with faster heart rate. A subgroup with tachycardia could also be singled out among the women from Tecumseh, but no relation between heart rate and blood pressure could be found. These findings show that in Western societies, high heart rate pertains to a distinct subgroup of subjects, who are more frequently men and exhibit the characteristic features of the insulin resistance syndrome. Sympathetic overactivity is likely to be the mechanism underlying this clinical condition.
There is a body of evidence to demonstrate that elevated resting heart rate is associated with a greater risk of developing sustained hypertension and with increased cardiovascular morbidity and mortality.1 2 3 4 The reason for this association is not clear; it could be related to observations in some studies of an association between tachycardia and other risk factors, such as high blood pressure (BP), smoking, and alcohol consumption.2 3 4 5 6 7 Moreover, it is not well known whether the cardiovascular risk related to heart rate is uniformly distributed across the heart rate range or is peculiar to a subset of the population with “abnormally” high heart rate levels. Another aspect that needs to be clarified is whether the high heart rate reflects an abnormal intrinsic regulation of the pacemaker or whether it is due to the alarm reaction associated with the doctor’s visit and other environmental stimuli.
To shed light on this controversial issue, we studied the distribution of heart rate and its relationship with BP and other clinical variables in three white populations. The first purpose of our investigation was to ascertain whether the variation of heart rate in these populations could be explained by a single normal distribution or by a mixture of two distributions. To separate out the two subpopulations, we used univariate mixture analysis, a statistical test devised in the Ann Arbor laboratory.8 If two subpopulations with “normal” and “abnormally” high heart rate could be identified, we studied whether they differed in BP, lipids, postload glucose, and insulin, when available. Finally, in two populations in which 24-hour ambulatory heart rate and BP were recorded, we compared the distribution of clinic heart rate with that of heart rate measured in ambulatory conditions.
The data sets of three populations studied in different countries were examined (Table 1⇓). A North European general population (Belgian),9 a North American general population (Tecumseh Blood Pressure Study, United States),10 and a stage I hypertensive population enrolled in a multicenter Italian study (HARVEST trial)11 were analyzed. The age of the subjects ranged from 20 to 88 years in the Belgian population, from 17 to 41 years in the Tecumseh study, and from 18 to 45 years in the HARVEST trial. Detailed information on the clinical characteristics of the three populations has been reported at length elsewhere.9 10 11 12 All studies were approved by the local institutional review ethical committees, and subjects gave informed consent. The procedures used in these studies were in accordance with institutional guidelines.
BP and heart rate in the three studies were assessed according to the recommendations of the International Scientific Societies, but the conditions in which they were measured and the number of readings on which they were calculated differed from study to study (Table 1⇑). In the Tecumseh and the HARVEST studies, BP and heart rate were taken by a doctor, whereas in the Belgian study, they were measured with an automatic device (Dinamap, Critikon Co). The number of readings taken in the three studies varied from two to six (Table 1⇑). BP and heart rate were measured in the lying position in the HARVEST study and in the sitting position in the other two studies.
In the Belgian and HARVEST studies, BP and heart rate were recorded also by means of 24-hour ambulatory monitoring. In both studies, only devices validated according to the British Hypertension Society13 and the Association for the Advancement of Medical Instrumentation14 recommendations were used. The methods used in the application of the instrumentations and the analysis of the recordings have been reported extensively elsewhere.15 16
Medical history and anthropometric data were taken in all studies, and fasting blood specimens were drawn for routine biochemistry. In the Belgian study, serum glucose was determined also, after a 75-g glucose load, and in the Tecumseh study, fasting insulin was measured. Other details on the methods used in the studies have been published previously.9 10 11 12
The independent association of heart rate with BP in the three populations was studied with multiple forward stepwise regression analysis, using BP as the dependent variable and heart rate, age, body mass index (BMI), smoking, alcohol intake, and physical activity habits as independent variables.
The distribution of heart rate in the populations stratified by sex was assessed with the Shapiro-Wilk’s test, and if a nonnormal distribution was present, data were inspected with the Q-Q plot. The Q-Q plot plots empirical quantiles against theoretical quantiles for normal distribution.17 When the distribution of the variable under examination has the same shape as the reference distribution, the Q-Q plot is linear (Fig 1c⇓). When the distribution is skewed and/or the kurtosis is different from 0, one or both ends of the plot deflect from the reference line (Fig 1a⇓). To determine more objectively when the pattern of points deviates from the comparison line, 95% confidence limits for the normal Q-Q plot can be estimated.17 For better visual inspection, plots were subsequently untilted, by subtracting the values of the comparison line from the data points.18 With this approach, departures from the reference value are easier to see (Fig 1b⇓ and 1d⇓).
In the populations in which the heart rate distribution was skewed, we used univariate mixture analysis to determine whether the apparently heterogeneous population was composed of more than one homogeneous normal subpopulation.8 Mixture analysis is a technique used in the biological sciences to investigate the likelihood that a mixture of normal distributions better explains the variation of a trait than a single distribution. Typically, overlap between the subpopulations results in observations that can be classified into any one of the groups. Individuals were assigned to the two subpopulations by a classification rule based on a likelihood that minimized the expected total number of misclassifications and allowed identification of a reliable cutoff level between the two groups. In the populations in which the two subpopulations differed by age and BMI, a subsequent mixture analysis was generated after adjusting for the aforementioned variables. Further details on this statistical procedure have been published elsewhere.19
Comparisons between subgroups were performed by Student’s t test for continuous variables and by χ2 for categorical ones. In the subpopulations in which age, BMI, smoking, alcohol intake, and physical activity were significantly different, a general linear model procedure was used to calculate BP levels and biochemical parameters adjusted for the above-mentioned confounders.
Data are expressed as mean±SEM, unless specified otherwise. Significance was accepted at P<.05.
In all populations, mean heart rate was higher in the women than in the men and lower in the Belgian population, in which it had been measured with an automatic device (Table 1⇑). BMI was similar in the three populations and was greater in the male sex.
To assess the association of heart rate with systolic BP, diastolic BP, and mean BP, a series of multivariate regression analyses was performed (see “Methods” for the model). For brevity, only the results for mean BP are reported (Table 2⇓). In all populations, heart rate turned out to be a significant independent predictor of BP in both men and women. However, the BP/heart rate association was much stronger in the men than the women. In the men, heart rate explained 10%, 12.2%, and 4.9% of the variance in mean BP in the Belgian, Tecumseh, and HARVEST populations, respectively. The corresponding values for the women were 3.1%, 3.8%, and 4.3%, respectively.
In the Belgian and HARVEST populations, the relationship of 24-hour BP with 24-hour heart rate could also be studied. In both studies, the BP/heart rate association was weaker for 24-hour than for clinic measurement.
Distribution of Heart Rate in the Populations
In the men of the three populations, clinic heart rate distribution was nonnormal (P<.0001 according to the Shapiro-Wilk’s test), with positive coefficients of skewness in all studies (range, .57 to .82). In the women, heart rate was nonnormally distributed only in the Tecumseh population (P<.0001), with a coefficient of skewness of .49, and was normally distributed in the Belgian and HARVEST studies.
The heart rate Q-Q plot was linear in the women of the Belgian and HARVEST populations. On the contrary, a clear departure from the upper end of the reference (normal) line toward higher values of heart rate was observed in the men of the three populations and the women from Tecumseh. The findings for the HARVEST population are illustrated in Fig 1⇑.
In both the Belgian and HARVEST studies, 24-hour heart rate recorded outside the hospital showed a normal distribution.
Classification by Mixture Analysis
In the men and women in whom the Q-Q plot showed a skewed distribution, the mixture analysis identified two subgroups. In Fig 2⇓ the results related to the Tecumseh population are reported. In all populations, the larger group had lower values of heart rate (“normal” heart rate) and the smaller group had higher values (“high” heart rate). After classification, with the use of the test discussed in Schork and Schork,19 we could reject the hypothesis of a single skewed distribution in favor of a mixture of two distributions (all P<.0001).
The heart rate cutoff point between the two subpopulations varied from population to population and within the Tecumseh population was slightly higher in the female sex (Table 3⇓). As expected, in the males, the lowest cutoff value was found in the Belgian study. The percentage of male subjects with high heart rate ranged from 8.4% (Belgian population) to 19.3% (Tecumseh population).
Age, BMI, and Lifestyle Factors by Heart Rate Group
In the HARVEST study, age tended to be lower in the subjects with high heart rate (Table 3⇑). In the Belgian men, BMI was greater among the subjects with high heart rate. No significant differences in BMI were found in the other two populations. Men with high heart rate were more sedentary than those with normal heart rate in the HARVEST study (P=.004). No significant differences in smoking or alcohol consumption were found according to heart rate levels.
Belgian and HARVEST men with tachycardia on the basis of clinic measurement also had higher values of ambulatory heart rate compared with subjects with normal clinic heart rate. Average 24-hour heart rate was 78.4±1.9 beats per minute (bpm) in the Belgian males with high clinic heart rate and 70.0±0.4 bpm in those with normal heart rate (P<.0001). The corresponding values for the HARVEST males were 76.5±0.8 bpm and 70.5±0.3 bpm, respectively (P<.0001).
BP and Results of Blood Tests by Heart Rate Group
In Table 4⇓, BP adjusted for confounders (see “Methods”) in the subjects with normal heart rate and high heart rate is shown. In the men of all populations, both systolic and diastolic BPs were higher in the subjects with high heart rate. The between-group difference in diastolic BP failed to reach the level of statistical significance in the HARVEST population. No BP differences according to heart rate levels were found in the women from Tecumseh.
Among the men, total cholesterol and triglycerides adjusted for confounders (see “Methods”) were more elevated in the subjects with high heart rate than in those with normal heart rate. The differences were significant in the Tecumseh (4.8±0.1 mmol/L versus 4.6±0.04 mmol/L; P=.03) and HARVEST (5.3±0.1 mmol/L versus 5.1±0.03 mmol/L; P=.02) studies for cholesterol and in the Belgian population for triglycerides (4.2±0.5 mmol/L versus 3.1±0.1 mmol/L; P=.04). No heart rate–related differences in lipids were found in the women from Tecumseh.
Within the Belgian men, postload glucose proved to be much higher in the subjects with high heart rate than in those with normal heart rate, being 5.9±0.2 mmol/L in the former and 5.0±0.1 mmol/L in the latter (P<.0001). Similar results were obtained for fasting insulin in the Tecumseh study (Fig 3⇓): In the subjects with high heart rate, insulin was significantly increased in comparison with those with normal heart rate. The difference was greater in the male sex.
Previous studies have demonstrated that resting clinic heart rate is an independent risk factor for adult cardiovascular disease in general and coronary heart disease in particular.1 2 3 4 However, the pathogenesis of the connection between elevated heart rate and cardiovascular disease remains obscure. A number of mechanisms for this association have been postulated. Data from animal models suggest that the atherogenic action of high heart rate may be related to its effects on blood flow characteristics, which would favor the occurrence of arterial wall lesions.20 21 According to some authors, tachycardia may merely indicate poor physiological fitness and/or subclinical loss of cardiac reserve.3 4 Moreover, it has been postulated that elevated heart rate may reflect a higher consumption of tobacco or alcohol,3 4 which are well-recognized risk factors for cardiovascular disease. A portion of the effect on coronary heart disease has been attributed to high BP, which appeared consistently positively correlated with pulse rate in several studies,5 6 7 but the nature of this relationship remains unclear.
In the present study, the relationship between heart rate and BP has been assessed through the analysis of three populations. To assess whether differences in lifestyle habits could influence the relationship of tachycardia with hypertension and other cardiovascular risk factors, we studied two Western general populations from different geographic areas.9 10 The analysis of the HARVEST dataset11 allowed us to investigate whether the relationship between tachycardia, increased BP, and metabolic abnormalities held true also in a hypertensive population and to compare the results of clinic measurements with those obtained by 24-hour recordings.
The statistical analysis of these populations permitted us to detect whether underlying factors can have a small effect (microphenic factors) or a large one (megaphenic factors) on the overall distribution of the heart rate.22 Most quantitative traits such as heart rate are affected only by microphenic factors, which can be the product of the individual genome, environmental influences, and their interaction.22 Megaphenic factors are rare, but when present, they tend to displace the average value of the affected subgroup from the average value of those people who are not affected. In this case, a mixture of two distributions is likely to explain the variation in the trait better than a single distribution. As the distribution of heart rate in most of our subjects was skewed, we wanted to ascertain whether the skewness resulted from the existence of two statistically separate populations. To do so, we used univariate mixture analysis, which is an entirely objective way to detect the existence of more than one homogeneous subpopulation within an apparently heterogeneous population.8
Prevalence and Clinical Significance of Tachycardia
In this study, we found a close correlation between BP and heart rate in all populations, and the relationship persisted after adjustment for other factors potentially influencing heart rate. The association was stronger in the male sex. However, it should be pointed out that heart rate explained only a small fraction of the variance in BP (4.9% to 12.2% in the men). Thus, although the heart rate/BP association appears to be strong from a statistical standpoint, the clinical relevance of this association is minimal. On the other hand, mixture analysis showed that in the men across all populations, this association was mostly explained by a subpopulation of subjects with “high” heart rate who had higher levels of BP. The percentage of male subjects with tachycardia varied from 8.4% to 19.3%. Among the women, a separation between subjects with high and normal heart rate could be found only in the Tecumseh study, but no BP difference was observed between the two subpopulations. A sex-related difference in the association between heart rate and BP was previously reported by other authors.2 5
Another interesting finding of the present analysis is that men with tachycardia also had high values of cholesterol and triglycerides, high fasting insulin, and increased postload glucose, which are characteristic features of the insulin resistance syndrome.23 This may explain why subjects with elevated heart rate develop sustained hypertension in later life, as documented by prospective studies conducted in either young24 or adult individuals.1 25 A higher BP, overweight, and disturbances of the glucose metabolism are all well-known risk factors for future hypertension. The clustering of these risk factors together with dyslipidemia, referred to as syndrome X,23 found in the present analysis in the subpopulations with high heart rate may explain why cardiovascular morbidity is higher in individuals with tachycardia.
Since in this study we argue that one main determinant of heart rate distribution in the general population is a megaphenetic factor, it seems appropriate to discuss the nature of this factor and to attempt to clarify the pathophysiological relationship between tachycardia, hypertension, and the metabolic abnormalities. As mentioned above, in all male populations, we found a skewed distribution of clinic heart rate and a highly significant correlation between clinic heart rate and clinic BP. When we studied heart rate and BP measured outside the hospital in ambulatory conditions, heart rate did not show a skewed distribution, and its association with BP was weaker. It is known that BP and heart rate measured in the clinic in part reflect the alarm reaction to the doctor, which may greatly vary from individual to individual.26 Furthermore, it has been demonstrated that BP and heart rate vary directionally in the same manner in response to daily life stressors, suggesting that central influences act consensually on the heart and the arterioles.27 Overall, these findings indicate that the sympathetic nervous system plays a major role in controlling heart rate and BP and suggest that in the subgroups of subjects identified as having tachycardia by mixture analysis, sympathetic overactivity is operative. If one assumes that tachycardia is a marker of abnormal autonomic control, it is easier to understand why it is associated with the classical features of the insulin resistance syndrome and why in the long run it can lead to atherosclerosis and its complications. In fact, it has been shown that sympathetic overactivity may cause insulin resistance through both α and β stimulation. Vasoconstriction mediated by α-adrenergic receptors appears to impair the ability of skeletal muscles to use glucose,28 and α-adrenergic blockade has been shown to improve insulin sensitivity.29 Acute stimulation of β-receptors with epinephrine infusion causes insulin resistance that can be reversed by propranolol.30 Also, chronic β-adrenergic stimulation can lead to insulin resistance, through the conversion from a small to a larger proportion of insulin-resistant fast-twitch fibers in skeletal muscles.31 The relationship between hyperinsulinemia and lipid abnormalities has long been recognized, and the mechanisms responsible for this association have been elucidated.23
The interrelationship between heart rate, BP, and metabolic abnormalities shown by the present analysis in men suggests that although tachycardia may reflect a short-term emotional response to the conditions of measurement, it should not be regarded as being innocuous. Several lines of evidence suggest that the so-called white-coat phenomenon is associated with a greater frequency of target organ damage in hypertension.10 15 32 Thus, the data of the present study call for revision of attitudes toward subjects with high heart rates at clinic examination and suggest that those individuals should not be dismissed as simply being “nervous.” However, we could not provide a general partition value to distinguish between subjects with normal and high heart rate.
The threshold value between tachycardia and normal heart rate identified by mixture analysis varied from 75 to 85 bpm in the three populations. These differences stem from the variability in the measurement of heart rate. In fact, the cutoff point was lower (75 bpm) in the Belgian population, in which an automatic device was used, thereby avoiding the psychological stress related to the presence of the doctor. The International Scientific Societies have established strict rules for the measurement of BP, whereas no specific recommendations have been provided for the assessment of heart rate. And yet, sources of variability are more common with the measurement of heart rate, which can be substantially influenced by the method employed (ECG versus pulse rate) or the position of the body. The calculation of heart rate may be affected also by the number of measurements, which varied from two to six in our populations, the length of resting time before the measurement(s), or the time of the day in which heart rate is measured.
In the present study, we provided substantial evidence for the clinical importance of tachycardia, which should be regarded by clinicians as an important risk factor for cardiovascular diseases. To establish what heart rate levels should be considered hazardous, the methods used to measure heart rate should be carefully standardized in future studies.
Reprints requests to Prof Paolo Palatini, MD, Clinica Medica 1, University of Padova, via Giustiniani, 2, 35126 Padova, Italy.
- Received March 25, 1997.
- Revision received April 29, 1997.
- Accepted May 8, 1997.
Dyer AR, Persky V, Stamler J, Paul O, Shekelle RB, Berkson DM, Lepper M, Schoenberger JA, Lindberg HA. Heart rate as a prognostic factor for coronary heart disease and mortality: findings in three Chicago epidemiologic studies. Am J Epidemiol. 1980;112:736-749.
Stamler J, Berkson DM, Dyer A, Lepper MH, Lindberg HA, Paul O, McKean H, Rhomberg P, Schoenberger JA, Shekelle RB, Stamler R. Relationship of multiple variables to blood pressure: findings from four Chicago epidemiologic studies. In: Paul O, ed. Epidemiology and Control of Hypertension. Miami, Fla: Symposia Specialists; 1975:307-352.
Reed D, McGee D, Yano K. Biological and social correlates of blood pressure among Japanese men in Hawaii. Hypertension. 1982;4:406-414.
Palatini P, Pessina AC, Dal Palù C. The Hypertension and Ambulatory Recording Venetia Study (HARVEST): a trial on the predictive value of ambulatory blood pressure monitoring for the development of fixed hypertension in patients with borderline hypertension. High Blood Press. 1993;2:11-18.
Palatini P, Graniero G, Mormino P, Nicolosi L, Mos L, Visentin P, Pessina AC. Relation between physical training and ambulatory blood pressure in stage I hypertensive subjects: results of the HARVEST Trial. Circulation. 1994;90:2870-2876.
White WB, Berson AS, Robbins C, Jamieson MJ, Prisant LM, Roccella E, Sheps SG. National standard for measurement of resting and ambulatory blood pressure with automated sphygmomanometers. Hypertension. 1993;21:504-509.
Staessen JA, Bieniaszewski L, O’Brien ET, Imai Y, Fagard R. An epidemiologic approach to ambulatory blood pressure monitoring: the Belgian Population Study. Blood Press Mon. 1996;1:13-26.
Chambers JM, Cleveland WS, Kleiner B, Tukey PA. Graphical Methods for Data Analysis. Belmont, Calif: Wadsworth Publishing Company; 1983.
SAS System for Statistical Graphics. In: Ginn JM, West JM, eds. Cary, NC: SAS Institute Inc; 1991:118-141.
Schork NJ, Schork MA. Skewness and mixture of normal distributions. Comm Stat Theoret Methods. 1988;17:3951-3969.
Beere PA, Glagov S, Zarins CK. Retarding effect of lowered heart rate on coronary atherosclerosis. Science. 1984;226:180-182.
Kaplan JR, Manuck SB, Adams MR, Weingand KW, Clarkson TB. Inhibition of coronary atherosclerosis by propranolol in behaviorally predisposed monkeys fed an atherogenic diet. Circulation. 1987;76:1364-1372.
DeFronzo RA, Ferrannini E. Insulin resistance: a multifaceted syndrome responsible for NIDDM, obesity, hypertension, dyslipidemia, and atherosclerotic cardiovascular disease. Diabetes Care. 1991;14:173-194.
Selby JV, Friedman GD, Quesenberry CP Jr. Precursors of essential hypertension: pulmonary function, heart rate, uric acid, serum cholesterol, and other serum chemistries. Am J Epidemiol. 1990;131:1017-1027.
Mancia G, Ferrari A, Gregorini L, Parati G, Pomidossi G, Bertinieri G, Grassi G, di Rienzo M, Pedotti A, Zanchetti A. Blood pressure and heart rate variabilities in normotensive and hypertensive human beings. Circ Res. 1983;53:96-104.
Jamerson KA, Julius S, Gudbrandsson T, Andersson O, Brant DO. Reflex sympathetic activation induces acute insulin resistance in the human forearm. Hypertension. 1993;21:618-623.
Deibert DC, DeFronzo RA. Epinephrine-induced insulin resistance in man. J Clin Invest. 1980;65:717-721.
Zeman RJ, Ludemann R, Easton TG, Etlinger JD. Slow to fast alterations in skeletal muscle fibers caused by clenbuterol, a beta-2-receptor agonist. Am J Physiol. 1988;254:E726-E732.
Weber MA, Neutel JM, Smith DHG, Graettinger WF. Diagnosis of mild hypertension by ambulatory blood pressure monitoring. Circulation. 1994;90:2291-2298.