Journal of Physical Activity and Health, 2014, 11, 1614 -1621 http://dx.doi.org/10.1123/jpah.2012-0405 © 2014 Human Kinetics, Inc.
Official Journal of ISPAH www.JPAH-Journal.com ORIGINAL RESEARCH
Physical Activity and Aerobic Fitness are Positively Associated With Heart Rate Variability in Obese Adults Kaisu Marjut Kaikkonen, Raija irmeli Korpelainen, Mikko P. Tulppo, Hannu Sakari Kaikkonen, Marja Liisa Vanhala, Mika Antero Kallio, Sirkka M. Keinänen-Kiukaanniemi, and Juha Tapani Korpelainen Background: Autonomic nervous system (ANS) dysfunction and obesity are intrinsically related to each other. In normal-weight subjects physical activity (PA) and fitness are related to cardiovascular autonomic regulation, providing evidence that aerobic training may improve ANS functioning measured by heart rate variability (HRV). The goal of this study was to investigate the association between lifetime PA, aerobic fitness and HRV in obese adults. Methods: Participants included 107 (87 females) volunteers (mean age 44.5 years, median BMI 35.7) who completed health and lifestyle questionnaires and measurements of maximal aerobic performance, anthropometry and 24 h HRV. Results: In the multivariate linear regression analyses, lifetime physical activity explained 40% of the variance in normal R-R intervals (SDNN). Each 1-category increase in the activity index increased SDNN by 15.4 (P = .009) and 24% of the variance in natural logarithmic value of ultra-low frequency power (P = .050). High measured VO2max explained 45% of the variance in natural logarithmic value of high-frequency power (P = .009) and 25% of the variance in low frequency/high frequency ratio (P < .001). Conclusions: Lifetime physical activity and aerobic fitness may reduce obesity-related health risks by improving the cardiac autonomic function measured by HRV in obese workingage subjects. This research supports the role of lifetime physical activity in weight management strategies and interventions to reduce obesity-related health risks. Keywords: autonomic nervous system, weight management, obesity-related health risks Obesity is a major public health problem with increasing prevalence worldwide. It has been defined as a chronic disease resulting from an interaction of both genetic and environmental factors.1,2 Obesity is associated with an increased risk for development of type 2 diabetes, hypertension, heart diseases, certain cancers and stroke.3 Morbidity and mortality risk are higher among obese individuals than among people with normal weight.4,5 The high rate of mortality in obese individuals is mainly a result of cardiovascular causes and is higher in people with excess fat in intra-abdominal depots.6 Regular physical activity and good aerobic fitness improve a number of health outcomes and reduce all-cause mortality.7,8 A review of 24 prospective studies indicated that health risks associated with overweight and obesity are attenuated by regular physical activity or good cardiorespiratory fitness, independent of the presence of overweight and obesity. Physical activity not only attenuates several health risks of overweight and obesity, but active obese individuals also have lower morbidity and mortality than sedentary normal weight individuals.9 KM Kaikkonen ([email protected]), H Kaikkonen, and Vanhala are with the Dept of Sports and Exercise Medicine, Oulu Deaconess Institute, Oulu, Finland. R Korpelainen is with Oulu Deaconess Institute and Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland. Tulppo is with the Dept of Exercise and Medical Physiology, Verve, Oulu, Finland. Kallio is with the Dept of Clinical Neurophysiology, Oulu University Hospital, Oulu, Finland. Keinänen-Kiukaanniemi is with the Institute of Health Sciences, Center for Life Course Epidemiology University of Oulu and Oulu University Hospital, Unit of Primary Health Care, Oulu, Finland. JT Korpelainen is with Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland. 1614
Many studies have suggested that autonomic nervous system (ANS) dysfunction and obesity are intrinsically related to each other.14–16 A 10% increase in body weight has been associated with a decline in parasympathetic cardiac power.17 Similarly, a 10% decrease in body weight has been related to an increase in the parasympathetic tone.18 Collectively these studies indicate that reduction in parasympathetic cardiovascular activity is associated with elevated weight providing a potential mechanism for the development of arrhythmias and other cardiac problems related to obesity.18,19 Reduced vagal function or increased sympathetic activity may provide the propensity for lethal arrhythmias.20 Impaired cardiac autonomic function has also been associated with an approximately doubled risk of mortality.21 HRV measurement has been used as an index of sympathetic tone and it is a clinically useful noninvasive tool for the assessment of ANS. HRV refers to the beat-to-beat variation in heart rate and is a marker of cardiac autonomic control.22 Reduced HRV has been considered as a marker of autonomic dysfunction and it has been shown to be associated with an increased risk of incident myocardial infarction, cardiovascular mortality and all-cause death.10 Reduced HRV has also been associated with poor prognosis of cardiovascular diseases,11 and it has been shown to be related to risk factors for cardiovascular diseases.12,13 HRV is influenced by various physiological and pathological conditions. A recent study suggests that regular physical exercise has strong beneficial effects on cardiac autonomic nervous function measured by HRV in obese individuals, and that exercise may offset the negative effect of obesity.23 Previous studies also suggest that in normal-weight subjects aerobic fitness is related to cardiovascular autonomic regulation, with regularly active individuals demonstrating better heart rate variability indices than their sedentary counterparts. The results of these studies provide evidence that aerobic
PA and Heart Rate Variability in Obese Adults 1615
training may improve ANS functioning measured by HRV.24–26 We found only few studies investigating the effect of physical fitness or regular physical activity on HRV and no studies examining the role of lifetime physical activity in working-age obese individuals. This cross-sectional study with 107 subjects aimed to investigate the association between lifetime physical activity, aerobic fitness and HRV activity in obese population.
Methods Study Design and Ethics The study design was cross-sectional: data from questionnaires were combined with measurements of peak VO2 consumption and HRV. The procedures of the study were in accordance with the Declaration of Helsinki, and formal ethics committee approval was obtained for the study. All participants gave an informed and written consent.
Downloaded by Purdue Univ on 09/17/16, Volume 11, Article Number 8
Subjects The subjects were recruited by advertising in a newspaper, which attracted 489 responses. All candidates were interviewed by phone with a standardized scheme. Ninety-five of them were excluded from the study due to: body mass index (BMI) under 30, pregnancy, medications affecting the heart rate, not able to perform an exercise test due to medical reasons, previously diagnosed diabetes and age under 18 or over 64 years. Thus, 394 subjects fulfilled the inclusion criteria and 121 of them were randomly selected to the study. All 121 were interviewed and examined by a trained nurse and they also were clinically examined by a physician. Four subjects were excluded after these interviews and examinations (1 pregnant, 1 atrial fibrillation, 1 using beta blockers, 1 BMI under 30). Ten subjects were excluded because of technically insufficient HRV measurement (recordings < 18 h and recordings of insufficient quality). Complete data sets were gathered from 107 participants.
Measurements Questionnaire. The subjects filled in a questionnaire including
questions about demographic features, obesity history, health and medication, physical activity and fitness, smoking and alcohol consumption. To assess the frequency and intensity of lifetime and current physical activity, a modified Paffenbarger questionnaire was used.27 The subjects were asked to recall their participation in physical activities during 3 time periods in their life span, corresponding to the ages of 15, 30 and their current age. Exercise intensity was categorized as none, mild (5 METs, MET= metabolic equivalent total), moderate (7.5 METs), or strenuous (10 METs), and the options were given the scores 0, 1, 2, and 3, respectively. Subjects reported the highest level of exercise performed for at least 15 minutes at a time, at least 3 times a week at each evaluated time point. The lifetime exercise score thus represents the sum of the scores, ranging from 0 to 9, at the evaluated ages. In statistical analyses, the lifetime exercise score was divided into 3 categories: low (scores 0 to 2), moderate (3–7), and high (scores 8 and 9). Participation in activities at the current age was used to evaluate the frequency and intensity of current physical activity level of the study subjects. MET values for each intensity category were multiplied with the frequency of weekly activity, and the weekly MET score was used to represent overall level of current activity. The weekly MET value was classified into 6 categories. In statistical analyses 2 categories, low (cat 1) and other (cat 2 to 6), were used.
Self-rated physical fitness was assessed using a standardized 5-scale question with categories “very poor,” “fairly poor,” “satisfactory,” “fairly good,” and “very good.” In the statistical analyses, fitness categories were classified into 2 categories, poor (very poor, fairly poor, satisfactory) and good (very and fairly good). Satisfactory was included in the poor category since according to our experience those who report their physical fitness to be satisfactory are not actually fit when measured. Alcohol consumption was calculated from a 7-day food record. Alcohol use was reported as number of portions (glasses of wine (12 cl), measures of spirits (4 cl), or bottles of beer (0.33 L), each corresponding to 12 g of alcohol). In the statistical analyses, alcohol consumption was divided into 2 categories (less than 6 servings of alcohol per week, 6 or more servings of alcohol per week). Smoking was reported as current smoker or nonsmoker. Those who had stopped smoking over 2 years ago were categorized as nonsmokers. Anthropometric Measurements. Body weight was measured
to the nearest 0.1 kg using a calibrated scale (SOEHNLE S20, Soehnle waagen, Germany) with the subject wearing light indoor clothes without shoes. Height was measured to the nearest 0.1 cm using a right-angle ruler placed on the head against a tape measure secured to the wall. BMI was calculated by dividing the weight (in kilograms) by the height squared (in meters). Waist girth was measured at the midpoint between the iliac crest and the lowest rib (cm).
Measurement of Maximal Aerobic Performance. The subjects performed a graded maximal exercise test under physician’s supervision on a bicycle ergometer (Siemens Megacart -Ergoline 900 BP, Siemens-Elema AB, Electrocardiography Division, Sweden), starting at 25 W and following a ramp protocol with the work rate increasing 25 W every 2 minutes28 until voluntary exhaustion or other symptoms limited the test. Symptoms that would terminate the test were: systolic blood pressure > 240, diastolic blood pressure > 130, arrhythmia, dizziness, and severe headache. The Borg rating scale29 was used at each step throughout the test. The reason for ending the test was recorded. At this stage respiratory quotient (RER) was at least 1.0 in all subjects and all study subjects fulfilled the maximum test criteria. Ventilation (VE), gas exchange (Jaeger Oxycon Pro, Hoechberg, Germany), and HR responses (Polar S810, Polar Electro Oy, Kempele, Finland) were monitored continuously during the ramp protocol. Electrocardiogram (ECG) was recorded with standard 12-lead ECG (Siemens Megacart -Ergoline 900 BP, Siemens-Elema AB, Electrocardiography Division, Sweden). VE and gas exchange were calculated on a breath-by-breath basis, but were reported as mean values for 30 seconds. The highest value of oxygen consumption measured during the test was used as peak oxygen consumption (VO2peak).30 The values were categorized into 7 sex- and age-specific fitness groups (very poor to excellent) according to international norms.31 The subjects were not allowed to eat or drink coffee for 2 hours before the exercise test, and heavy physical exercise and alcohol were not allowed during 24 hours before the day of testing. All the tests were performed between 10:00 AM and 3:00 PM. Measurement and Analysis of R-R Intervals/HR Variability The R-R intervals were recorded over 24 h with a Polar R-R recorder (Polar Electro Oy, Kempele, Finland) at an accuracy of 1 ms. The recording was performed during usual everyday activities. The analyses of the HRV were performed using Hearts software (Heart signal Co, Kempele, Finland). The R-R intervals were edited by visual inspection based on ECG portions to exclude ectopic beats
1616 Kaikkonen et al
and artifacts. Measures of R-R interval dynamics were calculated from the entire 24-h recording. Time and Frequency Domain Measures The mean HR and
Downloaded by Purdue Univ on 09/17/16, Volume 11, Article Number 8
the standard deviation (SD) of all normal R-R intervals (SDNN), which is a summary measure of HRV, were used as time-domain measures of HRV. An autoregressive model was used to estimate the power spectral densities of RR interval variability and the following frequency domain variables were calculated: ultra-low frequency (ULF) power (< 0.0033 Hz), very-low frequency (VLF) power (0.0033 to 0.04), low-frequency (LF) power (0.04 to 0.15 Hz), highfrequency (HF) power (0.15 to 0.40 Hz), and the LF/HF ratio. HF power is an index of the parasympathetic modulation of the heart, whereas LF power is an index of the combined parasympathetic and sympathetic modulation of heart rate. LF/HF ratio represents the sympathovagal balance.13 Because of the skewed distribution of frequency domain variables, the HRV values were log transformed for further analyses.
maximal power output. The unfit group included those in the lowest quartile of measured maximal power output, and the combined other quartiles (middle, high and very high) were referred to as the fit group (Table 1). Student’s independent samples t test and the Chi-squared test for 2 independent proportions were used to evaluate the statistical difference between the 2 different fitness groups. If the data were not normally distributed, Mann-Whitney nonparametric test was used. Depending on the distribution of the data, Pearson’s or Spearman’s correlation analysis was used to study intervariable relations between continuous variables. Multiple stepwise linear regression analysis was used to reveal the independent associates of the R-R interval dynamics. All explanatory variables significant in univariate analyses were entered in the models. All models were adjusted for age and heart rate. Level of significance for all tests was set at P < .05.
Fractal Scaling and Complexity Measures. The power-law
Results
Statistical Analyses The data were analyzed using the SPSS statistical package (SPSS 15.0 for Windows, SPSS Inc. Chicago, Illinois, US). All variables were first tested for normality by a Kolmogorov-Smirnov goodness-of-fit test. Results are reported as mean (SD) for normally distributed variables. Distributions of the spectral values of HR variability were highly skewed. The values were therefore transformed by taking the natural logarithms of the absolute values. In tables and figures the log transformed values are presented as geometric means (SE). BMI values were also skewed and reported as medians. For statistical analyses the participants were stratified into 2 age- and gender-adjusted fitness groups according to measured
The characteristics of the 107 participants by the 2 age- and gendermatched fitness categories are presented in Table 2. Eighty-seven (81%) of the study participants were women (mean age 44.5 years, range 26 to 62). The mean age of the men was 46.4 years (27 to 65). The median BMI was 35.7 (30.3 to 53.4). The mean weight of the women was 97.6 kg (72.9 to 150.6) and of the men 104.8 kg (92.7 to 123.4). The mean waist circumference was 108.0 cm (89.0 to 138.0) in women and 114.8 cm (104.0 to 138.0) in men. Smoking and alcohol consumption status were similar in both fitness groups. The proportion of physically inactive participants was significantly higher in the lower fitness group compared with the fit group (37% Age- and gender-adjusted measured maximal physical performance data are shown in Table 3. The mean maximal power output was 153.7 W (SD 40.6) and it was significantly higher compared with the unfit group (162.7 (SD 40.1) vs. 127.2 (SD 29.3), P < .001). The mean maximal oxygen consumption was 21.8 ml/kg/min (SD 5.5) and it was significantly higher in the fit group (22.6 (SD 5.7) vs. 19.6 (SD 4.4), P = .017). Statistically significant differences were found in mean minimum HR beats/minute (61.4 (SD 9.5) vs. 66.5 (SD 10.0), P = .02), and average heart rate reserve (107.4 (SD 18.4) vs. 96.1 (SD 17.7), P = .006), in favor of those in the fit group. The maximum HR was also higher in the fit group (168.9 (SD 17.1) vs. 162.6 (SD 17.8)), but the difference was not statistically significant.
relationship of RR interval variability was quantified by calculating the slope (β) as described previously32 by a regression analysis of power (on a logarithmic scale) and frequency (on a logarithmic scale) plots of the smoothed power spectrum over the frequency range of 10–4 to 10–2 Hz. The detrended fluctuation analysis (DFA) technique was used to detect qualitative changes in HR dynamics.33 The short-term (4 to 11 beats) α1 and long-term (>11 beats) α2 scaling exponents were calculated. Standard deviation of instantaneous beatto-beat R-R interval variability (DFA1) and long-term R-R interval variability (DFA2) measured from Poincare plots were quantified.34 Approximate entropy (ApEn), a measure quantifying the regularity of time series, was calculated as described previously.35
Table 1 Maximal Power Output Values in Fitness Quartiles According to Different Ages and Gender Age (years) 25–35
Gender Men (n = 3) Women (n = 25)
36–46 47–54 55–65
Men (n = 6)
Low (w)
Middle (w)
High (w)
Very high (w)
< 200.1
200.1–212.0
212.1–264.9
> 264.9
125.0–137.5
137.6–150.0
150.1–168.5
168.6–200.0
183.0.0–197.8
197.9–230.5
230.6–261.0
261.1–276.0
Women (n = 21)
110.0–125.0
125.1–150.0
150.1–156.0
156.1–175.0
Men (n = 7)
171.0–192.0
192.1–215.0
215.1–269.0
269.1–285.0
Women (n = 19)
96.0–106.0
106.1–128.0
128.1–150.0
150.1–165.0
Men (n = 4)
156.0–160.8
160.9–182.5
182.6–197.5
197.6–200.0
Women (n = 22)
97.0–111.8
111.9–125.0
125.1–135.5
135.6–159.0
Note. Categories in obese adults (n = 107). Abbreviations: w, watts (maximal power output).
PA and Heart Rate Variability in Obese Adults 1617
Measures of 24-hour R-R Interval dynamics across the 2 different age- and gender-adjusted fitness groups are presented in Table 4. There was a statistically significant difference between the groups in average heart rate during 24-hour recording (73.4 (SD 7.6) vs. 77.0 (SD 9.5), P = .045) in favor of the fit group. The values of all spectral components were higher in the fit group. The differences were statistically significant for SDNN (159.6 (SD 40.0) vs. 137.7 (SD (44.6), P = .018) and natural logarithmic value of ultra-low
frequency power (LnULF) (9.62 (SD 0.52) vs. 9.31 (SD 0.69), P = .016). The LF/HF ratio, the α1 and the β were lower in the unfit group (2.52 (SD 1.31) vs. 2.78 (SD 1.78), 1.27 (SD 0.14) vs. 1.30 (SD 0.16), and –1.27 (SD 0.16) vs. –1.24 (SD 0.16) respectively), but the differences were not statistically significant. Table 5 shows the best predictive multivariate models for 24-h R-R interval dynamics in the pooled study population. Using multiple linear regression analyses and adjusting for all variables
Table 2 Characteristics of the Study Subjects in 2 Different Age- and Gender-Adjusted Fitness Categories (n = 107)
Downloaded by Purdue Univ on 09/17/16, Volume 11, Article Number 8
Fitness category All (n = 107)
Unfit (n = 27)
Fit (n = 80)
Pa
Women, n (%)
87 (81.3)
Men, n (%)
20 (18.7)
22 (81.5)
65 (81.3)
0.979
5 (18.5)
15 (18.7)
Age, years
44.8 (25 to 65))
45.4 (25 to 65)
44.6 (25 to 62)
0.755
Height, cm
164.2 (142.5–182.0)
162.0 (147.0–180.0)
165.0 (142.5–182.0)
0.071 0.662
Weight, kg
98.9 (72.9–150.6)
97.9 (79.9–129.0)
99.3 (72.9–150.6)
Men
104.8 (92.7–123.4)
106.7 (94.5–123.4)
104.1 (92.7–118.8)
Women
97.6 (72.9–150.6)
95.8 (79.9–129.0)
98.2 (72.9–150.6)
BMI
35.7 (30.3–53.4)
36.4 (31.8–50.4)
35.4 (30.3–53.4)
0.401
Men
34.9 (31.5–44.6)
37.1 (33.2–44.6)
34.2 (31.5–40.4)
Women
37.2 (30.3–53.4)
37.5 (31.8–50.4)
37.1 (30.3–53.4)
Waist circumference, cm
109.2 (89.0–138.0)
111.1 (91.0–138.0)
108.6 (89.0–138.0)
Men
114.8 (104.0–138.0)
119.5 (105.0–138.0)
113.2 (104.0–124.0)
Women
108.0 (89.0–138.0)
109.18 (91.0–129.0)
107.6 (89.0–138.0)
SBP, mm Hg
140.8 (107 to 185)
141.0 (117.5–172.0)
140.7 (107.0–185.0)
0.925
DBP, mm Hg
89.4 (68.0 to110.0)
91.8 (80.0–110.0)
88.5 (68.0–110.0)
0.096
0.281
Alcohol > 6 drink/week, n (%)
21 (19.6)
6 (22.2)
15 (18.8)
0.862
Current smokers, n (%)
20 (18.7)
5 (18.5)
15 (18.8)
0.997
Current physical activity level low (>12.5 METh/week), n (%)
19 (17.8)
10 (37.0)
9 (11.3)
0.002
Lifetime activity level high, n (%)
17 (15.9)
3 (11.1)
14 (17.5)
0.552
Self-rated physical fitness level good, n (%)
13 (12.1)
0 (0.0)
13 (16.3)
0.025
Note. Unfit = the lowest quartile of measured maximal power output (low); fit = combined category of all other 3 quartiles of measured maximal power output (middle, high and very high). Values are means (range) or proportions n (%) except BMI values, which are presented in median due to skewed distribution. Abbreviations: BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; METh, a relative energy expenditure of physical activity expressed in metabolic equivalents. a P-value for the t test and for the Chi-squared test (two independent proportions). Mann-Whitney nonparametric test was used to evaluate the statistical difference in BMI values.
Table 3 Age- and Gender-Adjusted Maximal Aerobic Performance in 2 Different Fitness Categories in Obese Adults (n = 107) Fitness category Power output, w
All (n = 107)
Unfit (n = 27)
Fit (n = 80)
Pa
153.7 (40.6)
127.2 (29.3)
162.7 (40.1)
Official Journal of ISPAH www.JPAH-Journal.com ORIGINAL RESEARCH
Physical Activity and Aerobic Fitness are Positively Associated With Heart Rate Variability in Obese Adults Kaisu Marjut Kaikkonen, Raija irmeli Korpelainen, Mikko P. Tulppo, Hannu Sakari Kaikkonen, Marja Liisa Vanhala, Mika Antero Kallio, Sirkka M. Keinänen-Kiukaanniemi, and Juha Tapani Korpelainen Background: Autonomic nervous system (ANS) dysfunction and obesity are intrinsically related to each other. In normal-weight subjects physical activity (PA) and fitness are related to cardiovascular autonomic regulation, providing evidence that aerobic training may improve ANS functioning measured by heart rate variability (HRV). The goal of this study was to investigate the association between lifetime PA, aerobic fitness and HRV in obese adults. Methods: Participants included 107 (87 females) volunteers (mean age 44.5 years, median BMI 35.7) who completed health and lifestyle questionnaires and measurements of maximal aerobic performance, anthropometry and 24 h HRV. Results: In the multivariate linear regression analyses, lifetime physical activity explained 40% of the variance in normal R-R intervals (SDNN). Each 1-category increase in the activity index increased SDNN by 15.4 (P = .009) and 24% of the variance in natural logarithmic value of ultra-low frequency power (P = .050). High measured VO2max explained 45% of the variance in natural logarithmic value of high-frequency power (P = .009) and 25% of the variance in low frequency/high frequency ratio (P < .001). Conclusions: Lifetime physical activity and aerobic fitness may reduce obesity-related health risks by improving the cardiac autonomic function measured by HRV in obese workingage subjects. This research supports the role of lifetime physical activity in weight management strategies and interventions to reduce obesity-related health risks. Keywords: autonomic nervous system, weight management, obesity-related health risks Obesity is a major public health problem with increasing prevalence worldwide. It has been defined as a chronic disease resulting from an interaction of both genetic and environmental factors.1,2 Obesity is associated with an increased risk for development of type 2 diabetes, hypertension, heart diseases, certain cancers and stroke.3 Morbidity and mortality risk are higher among obese individuals than among people with normal weight.4,5 The high rate of mortality in obese individuals is mainly a result of cardiovascular causes and is higher in people with excess fat in intra-abdominal depots.6 Regular physical activity and good aerobic fitness improve a number of health outcomes and reduce all-cause mortality.7,8 A review of 24 prospective studies indicated that health risks associated with overweight and obesity are attenuated by regular physical activity or good cardiorespiratory fitness, independent of the presence of overweight and obesity. Physical activity not only attenuates several health risks of overweight and obesity, but active obese individuals also have lower morbidity and mortality than sedentary normal weight individuals.9 KM Kaikkonen ([email protected]), H Kaikkonen, and Vanhala are with the Dept of Sports and Exercise Medicine, Oulu Deaconess Institute, Oulu, Finland. R Korpelainen is with Oulu Deaconess Institute and Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland. Tulppo is with the Dept of Exercise and Medical Physiology, Verve, Oulu, Finland. Kallio is with the Dept of Clinical Neurophysiology, Oulu University Hospital, Oulu, Finland. Keinänen-Kiukaanniemi is with the Institute of Health Sciences, Center for Life Course Epidemiology University of Oulu and Oulu University Hospital, Unit of Primary Health Care, Oulu, Finland. JT Korpelainen is with Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland. 1614
Many studies have suggested that autonomic nervous system (ANS) dysfunction and obesity are intrinsically related to each other.14–16 A 10% increase in body weight has been associated with a decline in parasympathetic cardiac power.17 Similarly, a 10% decrease in body weight has been related to an increase in the parasympathetic tone.18 Collectively these studies indicate that reduction in parasympathetic cardiovascular activity is associated with elevated weight providing a potential mechanism for the development of arrhythmias and other cardiac problems related to obesity.18,19 Reduced vagal function or increased sympathetic activity may provide the propensity for lethal arrhythmias.20 Impaired cardiac autonomic function has also been associated with an approximately doubled risk of mortality.21 HRV measurement has been used as an index of sympathetic tone and it is a clinically useful noninvasive tool for the assessment of ANS. HRV refers to the beat-to-beat variation in heart rate and is a marker of cardiac autonomic control.22 Reduced HRV has been considered as a marker of autonomic dysfunction and it has been shown to be associated with an increased risk of incident myocardial infarction, cardiovascular mortality and all-cause death.10 Reduced HRV has also been associated with poor prognosis of cardiovascular diseases,11 and it has been shown to be related to risk factors for cardiovascular diseases.12,13 HRV is influenced by various physiological and pathological conditions. A recent study suggests that regular physical exercise has strong beneficial effects on cardiac autonomic nervous function measured by HRV in obese individuals, and that exercise may offset the negative effect of obesity.23 Previous studies also suggest that in normal-weight subjects aerobic fitness is related to cardiovascular autonomic regulation, with regularly active individuals demonstrating better heart rate variability indices than their sedentary counterparts. The results of these studies provide evidence that aerobic
PA and Heart Rate Variability in Obese Adults 1615
training may improve ANS functioning measured by HRV.24–26 We found only few studies investigating the effect of physical fitness or regular physical activity on HRV and no studies examining the role of lifetime physical activity in working-age obese individuals. This cross-sectional study with 107 subjects aimed to investigate the association between lifetime physical activity, aerobic fitness and HRV activity in obese population.
Methods Study Design and Ethics The study design was cross-sectional: data from questionnaires were combined with measurements of peak VO2 consumption and HRV. The procedures of the study were in accordance with the Declaration of Helsinki, and formal ethics committee approval was obtained for the study. All participants gave an informed and written consent.
Downloaded by Purdue Univ on 09/17/16, Volume 11, Article Number 8
Subjects The subjects were recruited by advertising in a newspaper, which attracted 489 responses. All candidates were interviewed by phone with a standardized scheme. Ninety-five of them were excluded from the study due to: body mass index (BMI) under 30, pregnancy, medications affecting the heart rate, not able to perform an exercise test due to medical reasons, previously diagnosed diabetes and age under 18 or over 64 years. Thus, 394 subjects fulfilled the inclusion criteria and 121 of them were randomly selected to the study. All 121 were interviewed and examined by a trained nurse and they also were clinically examined by a physician. Four subjects were excluded after these interviews and examinations (1 pregnant, 1 atrial fibrillation, 1 using beta blockers, 1 BMI under 30). Ten subjects were excluded because of technically insufficient HRV measurement (recordings < 18 h and recordings of insufficient quality). Complete data sets were gathered from 107 participants.
Measurements Questionnaire. The subjects filled in a questionnaire including
questions about demographic features, obesity history, health and medication, physical activity and fitness, smoking and alcohol consumption. To assess the frequency and intensity of lifetime and current physical activity, a modified Paffenbarger questionnaire was used.27 The subjects were asked to recall their participation in physical activities during 3 time periods in their life span, corresponding to the ages of 15, 30 and their current age. Exercise intensity was categorized as none, mild (5 METs, MET= metabolic equivalent total), moderate (7.5 METs), or strenuous (10 METs), and the options were given the scores 0, 1, 2, and 3, respectively. Subjects reported the highest level of exercise performed for at least 15 minutes at a time, at least 3 times a week at each evaluated time point. The lifetime exercise score thus represents the sum of the scores, ranging from 0 to 9, at the evaluated ages. In statistical analyses, the lifetime exercise score was divided into 3 categories: low (scores 0 to 2), moderate (3–7), and high (scores 8 and 9). Participation in activities at the current age was used to evaluate the frequency and intensity of current physical activity level of the study subjects. MET values for each intensity category were multiplied with the frequency of weekly activity, and the weekly MET score was used to represent overall level of current activity. The weekly MET value was classified into 6 categories. In statistical analyses 2 categories, low (cat 1) and other (cat 2 to 6), were used.
Self-rated physical fitness was assessed using a standardized 5-scale question with categories “very poor,” “fairly poor,” “satisfactory,” “fairly good,” and “very good.” In the statistical analyses, fitness categories were classified into 2 categories, poor (very poor, fairly poor, satisfactory) and good (very and fairly good). Satisfactory was included in the poor category since according to our experience those who report their physical fitness to be satisfactory are not actually fit when measured. Alcohol consumption was calculated from a 7-day food record. Alcohol use was reported as number of portions (glasses of wine (12 cl), measures of spirits (4 cl), or bottles of beer (0.33 L), each corresponding to 12 g of alcohol). In the statistical analyses, alcohol consumption was divided into 2 categories (less than 6 servings of alcohol per week, 6 or more servings of alcohol per week). Smoking was reported as current smoker or nonsmoker. Those who had stopped smoking over 2 years ago were categorized as nonsmokers. Anthropometric Measurements. Body weight was measured
to the nearest 0.1 kg using a calibrated scale (SOEHNLE S20, Soehnle waagen, Germany) with the subject wearing light indoor clothes without shoes. Height was measured to the nearest 0.1 cm using a right-angle ruler placed on the head against a tape measure secured to the wall. BMI was calculated by dividing the weight (in kilograms) by the height squared (in meters). Waist girth was measured at the midpoint between the iliac crest and the lowest rib (cm).
Measurement of Maximal Aerobic Performance. The subjects performed a graded maximal exercise test under physician’s supervision on a bicycle ergometer (Siemens Megacart -Ergoline 900 BP, Siemens-Elema AB, Electrocardiography Division, Sweden), starting at 25 W and following a ramp protocol with the work rate increasing 25 W every 2 minutes28 until voluntary exhaustion or other symptoms limited the test. Symptoms that would terminate the test were: systolic blood pressure > 240, diastolic blood pressure > 130, arrhythmia, dizziness, and severe headache. The Borg rating scale29 was used at each step throughout the test. The reason for ending the test was recorded. At this stage respiratory quotient (RER) was at least 1.0 in all subjects and all study subjects fulfilled the maximum test criteria. Ventilation (VE), gas exchange (Jaeger Oxycon Pro, Hoechberg, Germany), and HR responses (Polar S810, Polar Electro Oy, Kempele, Finland) were monitored continuously during the ramp protocol. Electrocardiogram (ECG) was recorded with standard 12-lead ECG (Siemens Megacart -Ergoline 900 BP, Siemens-Elema AB, Electrocardiography Division, Sweden). VE and gas exchange were calculated on a breath-by-breath basis, but were reported as mean values for 30 seconds. The highest value of oxygen consumption measured during the test was used as peak oxygen consumption (VO2peak).30 The values were categorized into 7 sex- and age-specific fitness groups (very poor to excellent) according to international norms.31 The subjects were not allowed to eat or drink coffee for 2 hours before the exercise test, and heavy physical exercise and alcohol were not allowed during 24 hours before the day of testing. All the tests were performed between 10:00 AM and 3:00 PM. Measurement and Analysis of R-R Intervals/HR Variability The R-R intervals were recorded over 24 h with a Polar R-R recorder (Polar Electro Oy, Kempele, Finland) at an accuracy of 1 ms. The recording was performed during usual everyday activities. The analyses of the HRV were performed using Hearts software (Heart signal Co, Kempele, Finland). The R-R intervals were edited by visual inspection based on ECG portions to exclude ectopic beats
1616 Kaikkonen et al
and artifacts. Measures of R-R interval dynamics were calculated from the entire 24-h recording. Time and Frequency Domain Measures The mean HR and
Downloaded by Purdue Univ on 09/17/16, Volume 11, Article Number 8
the standard deviation (SD) of all normal R-R intervals (SDNN), which is a summary measure of HRV, were used as time-domain measures of HRV. An autoregressive model was used to estimate the power spectral densities of RR interval variability and the following frequency domain variables were calculated: ultra-low frequency (ULF) power (< 0.0033 Hz), very-low frequency (VLF) power (0.0033 to 0.04), low-frequency (LF) power (0.04 to 0.15 Hz), highfrequency (HF) power (0.15 to 0.40 Hz), and the LF/HF ratio. HF power is an index of the parasympathetic modulation of the heart, whereas LF power is an index of the combined parasympathetic and sympathetic modulation of heart rate. LF/HF ratio represents the sympathovagal balance.13 Because of the skewed distribution of frequency domain variables, the HRV values were log transformed for further analyses.
maximal power output. The unfit group included those in the lowest quartile of measured maximal power output, and the combined other quartiles (middle, high and very high) were referred to as the fit group (Table 1). Student’s independent samples t test and the Chi-squared test for 2 independent proportions were used to evaluate the statistical difference between the 2 different fitness groups. If the data were not normally distributed, Mann-Whitney nonparametric test was used. Depending on the distribution of the data, Pearson’s or Spearman’s correlation analysis was used to study intervariable relations between continuous variables. Multiple stepwise linear regression analysis was used to reveal the independent associates of the R-R interval dynamics. All explanatory variables significant in univariate analyses were entered in the models. All models were adjusted for age and heart rate. Level of significance for all tests was set at P < .05.
Fractal Scaling and Complexity Measures. The power-law
Results
Statistical Analyses The data were analyzed using the SPSS statistical package (SPSS 15.0 for Windows, SPSS Inc. Chicago, Illinois, US). All variables were first tested for normality by a Kolmogorov-Smirnov goodness-of-fit test. Results are reported as mean (SD) for normally distributed variables. Distributions of the spectral values of HR variability were highly skewed. The values were therefore transformed by taking the natural logarithms of the absolute values. In tables and figures the log transformed values are presented as geometric means (SE). BMI values were also skewed and reported as medians. For statistical analyses the participants were stratified into 2 age- and gender-adjusted fitness groups according to measured
The characteristics of the 107 participants by the 2 age- and gendermatched fitness categories are presented in Table 2. Eighty-seven (81%) of the study participants were women (mean age 44.5 years, range 26 to 62). The mean age of the men was 46.4 years (27 to 65). The median BMI was 35.7 (30.3 to 53.4). The mean weight of the women was 97.6 kg (72.9 to 150.6) and of the men 104.8 kg (92.7 to 123.4). The mean waist circumference was 108.0 cm (89.0 to 138.0) in women and 114.8 cm (104.0 to 138.0) in men. Smoking and alcohol consumption status were similar in both fitness groups. The proportion of physically inactive participants was significantly higher in the lower fitness group compared with the fit group (37% Age- and gender-adjusted measured maximal physical performance data are shown in Table 3. The mean maximal power output was 153.7 W (SD 40.6) and it was significantly higher compared with the unfit group (162.7 (SD 40.1) vs. 127.2 (SD 29.3), P < .001). The mean maximal oxygen consumption was 21.8 ml/kg/min (SD 5.5) and it was significantly higher in the fit group (22.6 (SD 5.7) vs. 19.6 (SD 4.4), P = .017). Statistically significant differences were found in mean minimum HR beats/minute (61.4 (SD 9.5) vs. 66.5 (SD 10.0), P = .02), and average heart rate reserve (107.4 (SD 18.4) vs. 96.1 (SD 17.7), P = .006), in favor of those in the fit group. The maximum HR was also higher in the fit group (168.9 (SD 17.1) vs. 162.6 (SD 17.8)), but the difference was not statistically significant.
relationship of RR interval variability was quantified by calculating the slope (β) as described previously32 by a regression analysis of power (on a logarithmic scale) and frequency (on a logarithmic scale) plots of the smoothed power spectrum over the frequency range of 10–4 to 10–2 Hz. The detrended fluctuation analysis (DFA) technique was used to detect qualitative changes in HR dynamics.33 The short-term (4 to 11 beats) α1 and long-term (>11 beats) α2 scaling exponents were calculated. Standard deviation of instantaneous beatto-beat R-R interval variability (DFA1) and long-term R-R interval variability (DFA2) measured from Poincare plots were quantified.34 Approximate entropy (ApEn), a measure quantifying the regularity of time series, was calculated as described previously.35
Table 1 Maximal Power Output Values in Fitness Quartiles According to Different Ages and Gender Age (years) 25–35
Gender Men (n = 3) Women (n = 25)
36–46 47–54 55–65
Men (n = 6)
Low (w)
Middle (w)
High (w)
Very high (w)
< 200.1
200.1–212.0
212.1–264.9
> 264.9
125.0–137.5
137.6–150.0
150.1–168.5
168.6–200.0
183.0.0–197.8
197.9–230.5
230.6–261.0
261.1–276.0
Women (n = 21)
110.0–125.0
125.1–150.0
150.1–156.0
156.1–175.0
Men (n = 7)
171.0–192.0
192.1–215.0
215.1–269.0
269.1–285.0
Women (n = 19)
96.0–106.0
106.1–128.0
128.1–150.0
150.1–165.0
Men (n = 4)
156.0–160.8
160.9–182.5
182.6–197.5
197.6–200.0
Women (n = 22)
97.0–111.8
111.9–125.0
125.1–135.5
135.6–159.0
Note. Categories in obese adults (n = 107). Abbreviations: w, watts (maximal power output).
PA and Heart Rate Variability in Obese Adults 1617
Measures of 24-hour R-R Interval dynamics across the 2 different age- and gender-adjusted fitness groups are presented in Table 4. There was a statistically significant difference between the groups in average heart rate during 24-hour recording (73.4 (SD 7.6) vs. 77.0 (SD 9.5), P = .045) in favor of the fit group. The values of all spectral components were higher in the fit group. The differences were statistically significant for SDNN (159.6 (SD 40.0) vs. 137.7 (SD (44.6), P = .018) and natural logarithmic value of ultra-low
frequency power (LnULF) (9.62 (SD 0.52) vs. 9.31 (SD 0.69), P = .016). The LF/HF ratio, the α1 and the β were lower in the unfit group (2.52 (SD 1.31) vs. 2.78 (SD 1.78), 1.27 (SD 0.14) vs. 1.30 (SD 0.16), and –1.27 (SD 0.16) vs. –1.24 (SD 0.16) respectively), but the differences were not statistically significant. Table 5 shows the best predictive multivariate models for 24-h R-R interval dynamics in the pooled study population. Using multiple linear regression analyses and adjusting for all variables
Table 2 Characteristics of the Study Subjects in 2 Different Age- and Gender-Adjusted Fitness Categories (n = 107)
Downloaded by Purdue Univ on 09/17/16, Volume 11, Article Number 8
Fitness category All (n = 107)
Unfit (n = 27)
Fit (n = 80)
Pa
Women, n (%)
87 (81.3)
Men, n (%)
20 (18.7)
22 (81.5)
65 (81.3)
0.979
5 (18.5)
15 (18.7)
Age, years
44.8 (25 to 65))
45.4 (25 to 65)
44.6 (25 to 62)
0.755
Height, cm
164.2 (142.5–182.0)
162.0 (147.0–180.0)
165.0 (142.5–182.0)
0.071 0.662
Weight, kg
98.9 (72.9–150.6)
97.9 (79.9–129.0)
99.3 (72.9–150.6)
Men
104.8 (92.7–123.4)
106.7 (94.5–123.4)
104.1 (92.7–118.8)
Women
97.6 (72.9–150.6)
95.8 (79.9–129.0)
98.2 (72.9–150.6)
BMI
35.7 (30.3–53.4)
36.4 (31.8–50.4)
35.4 (30.3–53.4)
0.401
Men
34.9 (31.5–44.6)
37.1 (33.2–44.6)
34.2 (31.5–40.4)
Women
37.2 (30.3–53.4)
37.5 (31.8–50.4)
37.1 (30.3–53.4)
Waist circumference, cm
109.2 (89.0–138.0)
111.1 (91.0–138.0)
108.6 (89.0–138.0)
Men
114.8 (104.0–138.0)
119.5 (105.0–138.0)
113.2 (104.0–124.0)
Women
108.0 (89.0–138.0)
109.18 (91.0–129.0)
107.6 (89.0–138.0)
SBP, mm Hg
140.8 (107 to 185)
141.0 (117.5–172.0)
140.7 (107.0–185.0)
0.925
DBP, mm Hg
89.4 (68.0 to110.0)
91.8 (80.0–110.0)
88.5 (68.0–110.0)
0.096
0.281
Alcohol > 6 drink/week, n (%)
21 (19.6)
6 (22.2)
15 (18.8)
0.862
Current smokers, n (%)
20 (18.7)
5 (18.5)
15 (18.8)
0.997
Current physical activity level low (>12.5 METh/week), n (%)
19 (17.8)
10 (37.0)
9 (11.3)
0.002
Lifetime activity level high, n (%)
17 (15.9)
3 (11.1)
14 (17.5)
0.552
Self-rated physical fitness level good, n (%)
13 (12.1)
0 (0.0)
13 (16.3)
0.025
Note. Unfit = the lowest quartile of measured maximal power output (low); fit = combined category of all other 3 quartiles of measured maximal power output (middle, high and very high). Values are means (range) or proportions n (%) except BMI values, which are presented in median due to skewed distribution. Abbreviations: BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; METh, a relative energy expenditure of physical activity expressed in metabolic equivalents. a P-value for the t test and for the Chi-squared test (two independent proportions). Mann-Whitney nonparametric test was used to evaluate the statistical difference in BMI values.
Table 3 Age- and Gender-Adjusted Maximal Aerobic Performance in 2 Different Fitness Categories in Obese Adults (n = 107) Fitness category Power output, w
All (n = 107)
Unfit (n = 27)
Fit (n = 80)
Pa
153.7 (40.6)
127.2 (29.3)
162.7 (40.1)
