Inverse Correlation Between Heart Rate Recovery and Metabolic Risks in Healthy Children and Adolescents: Insight from the National Health and Nutrition Examination Survey 1999–2002

The NHANES is a population-based survey designed to collect information on the health and nutrition of the U.S. household population. The NHANES used a stratified and cluster sampling design to obtain a representative sample of the noninstitutionalized civilian U.S. population. The survey data are released every 2 years after 1999. Detailed survey operations manuals, consent documents, and brochures of the NHANES 1999–2002 are available on the NHANES website (11).

In NHANES 1999–2002, a cardiorespiratory fitness test was administered to 6,858 subjects aged between 12 and 49 years without physical functioning limitations, cardiovascular conditions or symptoms, pulmonary conditions or symptoms, pregnancy, and usage of certain medications, including antiarrhythmics, β-blockers, calcium channel blockers, digitalis, nitrates, central nervous system stimulants, and ephedra-based weight loss medications.

We collected data from 3,928 subjects aged between 12 and 19 years for further analysis. Of these subjects, 2,585 participants completed the treadmill test. Among 2,585 subjects, complete information for the five variables needed to assess the metabolic syndrome was available for 1,123 children and adolescents. To maximize the number of young participants available for analysis, we used measurements of glucose and triglycerides from the biochemistry panel, since the reference measurements for these two variables were only conducted for participants of the morning examination (12). Of the remaining 1,123 participants, we excluded 116 without a recorded fasting time or with <6 h of fasting (12) and 9 with hypertension (defined by a self-reported current use of antihypertension medications or average blood pressure >140/90 mmHg). Participants (nine subjects) who were currently using antihyperlipidemia medications or who had a fasting serum total cholesterol >240 mg/dl or triglycerides >200 mg/dl were also excluded. Among the 1,123 subjects, none had diabetes (defined as a fasting serum glucose ≥126 mg/dl or a self-reported current use of antidiabetes medications). Therefore, the final analytic sample was confined to 993 subjects.

Data were collected at all study sites by trained personnel according to standardized procedures. Sociodemographic information such as age, sex, and race/ethnicity was collected during the household interview. Laboratory measurements were performed in a mobile examination center. Weight and height were measured using standard methods and digitally recorded. BMI was calculated as weight in kilograms divided by the square of height in meters. Waist circumference was measured at the iliac crest to the nearest 0.1 cm. Three and sometimes four blood pressure determinations were taken by a physician using a mercury sphygnomanometer. Blood pressure was measured in the right arm unless otherwise specified. Averaged systolic (SBP) and diastolic (DBP) blood pressure were obtained.

Blood specimens were processed locally, then stored and shipped to central laboratories for analysis. Levels of serum total cholesterol and triglycerides were measured enzymatically, and levels of HDL cholesterol were measured using precipitation. Serum glucose levels were processed by using the hexokinase enzymatical method. Serum C-reactive protein (CRP) levels were measured by latex-enhanced nephelometry with a Behring Nephelometer Analyzer system.

The NHANES submaximal treadmill test protocol includes a 2-min warm-up, two 3-min exercise stages, and a 3-min cool-down period. Heart rate was monitored continuously using an automated monitor and was recorded at the end of each stage. HRR was defined as the change from peak heart rate during exercise to that measured during 1, 2, or 3, min of recovery. Maximal oxygen uptake (Vo_2max) can be estimated by using measured cardiovascular responses to known levels of exercise workloads, assuming the relation between heart rate and oxygen consumption is linear during exercise (13). The estimated Vo_2max was further categorized to three fitness levels based on sex-age specific cut points of estimated Vo_2max (14).

Data were expressed as means ± SE. The continuous variables were compared between boys and girls by unpaired Student's t tests, while the category parameters were by χ² tests. The correlation between various metabolic risk factors and each HRR parameter were derived by Pearson correlation analysis. The strength of association between each HRR parameter and other associated metabolic risk factors was estimated by multiple linear regression models. Log transformation was performed for variables with significant deviation from normal distribution before further analyses. To consider various metabolic risk factors as a whole, we also calculated the metabolic Z score. The score was derived by converting each component of the metabolic risk factors, including blood pressure, waist circumference, serum glucose level, serum triglyceride levels, and serum HDL levels, into a Z score based on sex-specific means. The blood pressure Z score was derived by averaging the SBP and the DBP Z scores. The metabolic Z score was calculated by summation of the former four Z scores minus the HDL Z score. A P < 0.05 was considered statistically significant. Sampling weights that account for unequal probabilities of selection, oversampling, and nonresponse were applied for all analyses using the Complex Sample Survey module of SPSS 13.0 for Windows XP (SPSS, Chicago, IL).

The basic characteristics of the study subjects are summarized in Table 1. A total of 509 boys and 484 girls were included for analysis. Mean age of the boys were similar to girls (15.55 ± 0.10 vs. 15.40 ± 0.17 years, P = 0.478). The metabolic risk factors including the SBP (111.64 ± 0.96 vs. 107.41 ± 0.49 mmHg, P < 0.001) and the serum glucose levels (87.17 ± 0.58 vs. 83.44 ± 0.27 mg/dl, P < 0.001) were significantly higher in boys than in girls. The DBP (63.34 ± 0.79 vs. 65.61 ± 0.8 mmHg, P = 0.001) and the serum HDL levels (46.14 ± 0.47 vs. 51.03 ± 0.59 mg/dl, P < 0.001) were higher in girls than in boys. There were no significant differences between boys and girls in race, BMI, waist circumference, serum total cholesterol levels, serum CRP levels, and serum triglyceride levels.

The variables recorded during exercise are summarized in Table 2. The warm-up heart rate was significantly lower in boys than in girls (108.83 ± 0.96 vs. 112.31 ± 0.69 per min, P = 0.002). The maximal heart rates were not different between boys and girls at the end of stage 1 (128.12 ± 1.30 vs. 125.91 ± 0.89 per min, P = 0.265) but became higher in boys than in girls (155.55 ± 1.16 vs. 149.11 ± 0.95 per min, P = 0.001) at the end of stage 2. One minute after exercise, the heart rates were not different between boys and girls (141.73 ± 1.15 vs. 139.27 ± 1.19 per min, P = 0.166), but the heart rate of boys became lower than that of girls after 2 min of resting (116.76 ± 1.31 vs. 120.81 ± 0.95 per min, P = 0.028). After 3 min of resting, the heart rates were not different between boys and girls (120.09 ± 0.70 vs. 121.00 ± 0.77 per min, P = 0.346). The HRR parameters were all significantly higher in boys than in girls. Although the estimated Vo_2max was significantly higher in boys, the cardiovascular fitness level was not different between boys and girls.

The correlations between the HRR parameters and other anthropometric and biochemical variables are shown in Table 3. Briefly speaking, the HRR parameters were negatively correlated with most of the metabolic risk factors, including waist circumference, SBP, serum fasting glucose levels, serum triglyceride levels, and serum CRP levels. The HRR parameters were positively correlated with the serum HDL levels. Also noted in Table 3, longer-term HRR (2- and 3-min HRR) were more strongly associated with metabolic risks including waist circumference, serum HDL levels, serum CRP levels, and SBP compared with short-term HRR (1-min HRR).

Multiple linear regression analysis (Table 4) showed that among the five components of metabolic syndrome, waist circumference was the only risk associated with all the HRR parameters in boys. In girls, the 1-min HRR was only inversely correlated with SBP (P = 0.029). The 2-min HRR was inversely correlated with waist circumference (P = 0.001), serum glucose levels (P = 0.021), and serum CRP levels (P = 0.007). The 3-min HRR was only inversely correlated with the waist circumference (P < 0.001). The R² values of the multiple regression models paralleled with the duration of resting (0.082, 0.194, and 0.342 for 1- to 3-min HRR, respectively, in boys and 0.100, 0.317, and 0.432 for 1- to 3-min HRR, respectively, in girls).

To consider the metabolic risks as a whole, we calculated the metabolic Z score to represent the severity of the cluster of the metabolic risks by a single value. The adjusted mean HRR parameters (adjusted for other covariates) based on different metabolic Z-score quartiles were obtained and the association between the metabolic Z-score quartiles and the adjusted mean HRR parameters was illustrated in Fig. 1. For both genders, the adjusted 1-min HRR did not change across four quartiles (Fig. 1A and B). For 3-min HRR, the negative association was significant both in boys and girls (Fig. 1C and D).

HRR after exercise is a well-validated independent predictor of cardiovascular and all-cause mortality (2–5). Slower HRR after exercise was also shown to be associated with higher baseline levels of metabolic syndrome components (6–8). Because the prevalence of obesity and associated metabolic disorders is increasing rapidly in both adults and children in developing countries (15), it is of clinical importance to investigate whether the link between impaired HRR and metabolic risks appears in children and adolescents.

The association between dysregulation of the autonomic nervous system and metabolic syndrome has been widely studied in recent years. Recent studies (16–19) have given a solid background to the hypothesis that the sympathetic nervous system plays a central role in the development of the metabolic syndrome. Associations between HRR and the metabolic syndrome are thought to support the hypothesis (20). However, this hypothesis has been questioned recently by Kizilbash et al. (7), who showed that slower 2-min HRR does not precede development of the metabolic syndrome but appears after metabolic syndrome components are present. One possible reason is that the HRR is not only an indicator of high sympathetic drive but rather a coordinated interaction of parasympathetic reactivation and sympathetic withdrawal during exercise recovery (21–23).

In the present study, multiple linear regression analysis showed that the metabolic risks contribute more to the variation of the 2- or 3-min HRR compared with that of the 1-min HRR (R² increased from the 1- to 3-min HRR). The relative role of parasympathetic reactivation versus sympathetic withdrawal in controlling exercise HRR is dynamic and complicated (23). It has been suggested that short-term HRR indexes (such as 1-min HRR) could be considered as markers of cardiac parasympathetic outflow. In contrast, the second slow heart rate decay phase is thought to be related to the gradual withdrawal of sympathetic activity and to the clearance of stress system metabolites (24). Whether the correlation pattern between metabolic risks and the autonomic nervous system is different in the sympathetic and parasympathetic nervous system needs further study to clarify.

Only one previous report (8) concerning the relation between HRR and metabolic risks has used an analysis stratified by sex. Nilsson et al. (8) found that among 75-year-old subjects, the metabolic syndrome–related components are more strongly correlated to HRR in women than in men. They also found a strong relationship between HRR and waist circumference in women that is not found in men (8). Although we demonstrated that the correlation between HRR and metabolic risks is relatively higher in girls, the waist circumference is highly associated with deteriorated HRR in both sexes.

To maximize our sample size, we decided to use nonreference measurements of concentrations of glucose and triglycerides because these were obtained for examination sessions. However, when we used reference measurements for young participants who attended the morning examination, we obtained very similar results.

Our study has several limitations. First, the cross-sectional design does not permit any causal inference. Second, due to plenty of missing data, we did not include smoking or drinking status in our analysis. Since the inverse correlation between metabolic risks and HRR is strong, it is not likely that incorporation of smoking and drinking status can change the relationship.

In conclusion, metabolic risks, especially waist circumference, are inversely associated with HRR in healthy children and adolescents. Short-term HRR is less significantly associated with metabolic risks compared with longer-term HRR. Our finding suggests that the association between autonomic nervous system dysfunction and metabolic risks operates early in young ages.