Association of Physical Activity and Sedentary Time with Cardio-Autonomic Regulation in Women

Abstract

Background:

Low moderate-to-vigorous-intensity physical activity (MVPA) and high sedentary time (ST) may contribute to cardiovascular disease (CVD) risk in women, perhaps via cardiac autonomic dysregulation. We examined associations of total, leisure, and occupational MVPA and ST with cardiac autonomic regulation in women.

Methods:

Data were from 522 women (age = 37.7 ± 5.7 years; 59%white) who participated in the follow-up study of the Pregnancy Outcomes and Community Health Study (between 2011 and 2014). MVPA and ST (hours/day) were self-reported using the Modifiable Activity Questionnaire. Cardiac autonomic regulation was assessed by calculating heart rate variability (HRV) indices (resting heart rate, natural logarithm standard deviation of normal R-R intervals; lnSDNN [total variability], natural logarithm root mean square of the successive differences; lnRMSSD [cardiac parasympathetic activity]) with Kubios software from a 5-minute, seated electrocardiogram. Progressive generalized linear models evaluated associations of total, leisure, and occupational MVPA and ST with HRV indices while adjusting for confounders (demographics, health-related factors), and then potential mediators (clinical variables). A final model evaluated the relationship between ST and HRV stratified by MVPA level.

Results:

Adjusting for confounders, total and leisure MVPA were associated with favorable lnSDNN (B = 0.027 [p = 0.014] and B = 0.074 [p = 0.009], respectively) and lnRMSSD (B = 0.036 [p = 0.015] and B = 0.075 [p = 0.043], respectively). Adjustment for mediators tended to strengthen the observed significant associations. No associations were found between occupational MVPA or any ST measure with HRV indices (p > 0.05). Neither MVPA nor ST were associated with heart rate. When stratified by MVPA level, leisure ST was associated with unfavorable lnRMSSD (B = −0.041, [p = 0.042]) only among women who did not meet leisure MVPA recommendations.

Conclusion:

Cardiac autonomic dysregulation may be a mechanism through which low leisure MVPA and, among low-active women, high leisure ST contribute to CVD risk among women.

Introduction

Recent evidence indicates that young-to middle-aged women (i.e., ages 20–55 years old) have had blunted improvements in rates of cardiovascular disease (CVD) as compared with declining CVD rates in similarly aged men and older women.^1,2 The ability to reduce CVD risk in these women requires an investigation of contributing factors and mechanisms through which they accumulate CVD risk. Lifestyle factors such as physical inactivity and excessive sedentary time (ST) are suggested as contributing factors for CVD.

Physical inactivity is defined as not engaging in sufficient levels of moderate-to-vigorous-intensity physical activity (MVPA).³ Physical inactivity is prevalent among adults worldwide and is higher among younger women compared with men⁴; this is unfortunate since physical inactivity is a major and modifiable CVD risk factor.⁵ In addition, excessive time spent in sedentary behavior (i.e., any waking activity that occurs in a lying, reclining, or seated posture and has energy expenditure of ≤1.5 metabolic equivalents)⁶ is emerging as a CVD risk factor, independent of physical inactivity.³ Excessive sedentary behavior is also prevalent among women, who spend >55% of their waking time in sedentary behavior.⁷ Taken together, women tend to be physically inactive and accumulate excessive ST, both of which may contribute to CVD risk in this population. Yet, the physiological pathways that relate physical inactivity and excessive ST to CVD in women are not fully understood.

Cardiac autonomic dysregulation is a physiological mechanism that links risk factors, such as hypertension (HTN) and diabetes mellitus (DM), to CVD outcomes.⁸ Cardiac autonomic regulation is commonly assessed by measuring heart rate variability (HRV), a noninvasive method that measures the variation in time intervals between consecutive heartbeats.⁹ Lower variability in the time intervals indicates altered cardiac autonomic balance, regulation, and/or flexibility and suggests vulnerability to CVD.¹⁰ Many studies have consistently found associations between reduced HRV with CVD and mortality.¹¹ Evaluation of HRV in young- to middle-aged women may be especially relevant as reduced HRV is subclinical and has been associated with female sex-specific factors such as adverse pregnancy outcomes.¹² Thus, reduced HRV is likely detectable before progression to overt CVD and may provide insight into the pathway between lifestyle risk factors (e.g., physical inactivity and excessive ST) and elevated CVD risk in these women.

Robust evidence indicates that MVPA is associated with greater HRV in adults, including young- to middle-aged women.¹³ However, limited research has examined the association between ST and HRV in adults.^14–18 These studies have yielded inconsistent findings, and no studies have been conducted specifically in young- to middle-aged women. Moreover, most studies had small samples, did not use gold standard HRV assessment procedures (i.e., electrocardiogram [ECG]), and/or did not consider joint associations with MVPA. This final limitation is important as recent evidence indicates that the harmful effects of excessive ST on health outcomes may be attenuated in the presence of high MVPA.¹⁹ Additionally, accumulating data suggest that MVPA and ST across occupational and leisure domains may differently affect CVD risk.^14,15,20 Therefore, associations of CVD risk across domains of these behaviors should be researched.

The primary aim of this study was to examine the independent and joint associations of total and domain-specific MVPA and ST with HRV in young- to middle-aged women. We hypothesized that greater MVPA would be associated with higher (i.e., better) HRV, while greater ST would be associated with lower (i.e., worse) HRV. We further aimed to explore associations across leisure and occupational domains. Lastly, we hypothesized that the adverse effects of ST on HRV would be more apparent among inactive women versus women who meet leisure MVPA recommendations.

Methods

Study setting and population

This study was a secondary, cross-sectional analysis of data from the follow-up study of the Pregnancy Outcomes and Community Health (POUCH) Study.²¹ Briefly, the POUCH Study enrolled 3,019 women during pregnancy to prospectively examine social and biological factors linked to preterm delivery. A subset of women (subcohort n = 1,371) was studied in greater detail. The selection strategy for the subcohort included all women who had a preterm delivery (<37 weeks gestation), women with specific risk factors for preterm delivery, and a random sample of the remaining women. Subcohort women who agreed to be recontacted were invited into the POUCHmoms Study for follow-up 7 to 15 years after delivery (between 2011 and 2014). To be included in POUCHmoms, women could not have been currently pregnant or pregnant in the past 6 months.

Of the 1,371 women invited, n = 91 declined future contact or were deceased. Thus, 1,280 women were eligible for the POUCHmoms Study. Of these women, n = 602 declined, had relocated out of state, or were unable to be located. Therefore, only 678 women completed the POUCHmoms follow-up assessment.²¹ As previously reported, age, clinical risk factors, and rates of adverse pregnancy outcomes were similar between women that did and did not complete the follow-up, although women who completed the follow-up assessment were more educated, less likely to be African American, and less likely to receive Medicaid insurance compared with noncompleters.²²

To be included in the current analysis, women who completed the follow-up assessment additionally had to have complete self-reported ST and MVPA data and HRV measurements of sufficient quality. All participants provided written informed consent. This follow-up study was approved by the Institutional Review Boards of the Michigan State University and the University of Pittsburgh.

Measurements

ST and MVPA

The interview-based Modifiable Activity Questionnaire (MAQ) was used to measure ST and MVPA.²³ This questionnaire assesses daily ST and MVPA during leisure and occupational time during the past 12 months. Leisure ST was evaluated by asking the participants about their typical daily hours spent sitting, including activities such as TV watching, computer use unrelated to work, reading, crafts, and helping children with homework. For occupational ST, the participants were asked about their daily hours spent sitting at work in a job that they had held for at least 1 month over the past year. Leisure and occupational ST were examined as individual domains and as total ST, calculated by summing the two domains. To assess leisure MVPA, the participants were asked to estimate the frequency and average minutes for each time they participated in planned physical exercise that they had performed at least 10 times over the past year. These average minutes were converted into hours/day. Occupational MVPA was measured by asking the participants about their average daily hours spent in physically demanding activities at work in a job that they had held for at least 1 month over the past year. Leisure and occupational MVPA were examined as individual domains and as total MVPA, calculated by summing the two domains.

HRV measurement

HRV was measured using an ECG at the POUCHmoms follow-up visit. Participants were instructed to fast for at least 8 hours before the study visit. Upon arriving, several assessments, including blood sample collection and self-reported questionnaires were completed followed by a 45–60-minute snack break. Thereafter, ECG measurements were obtained while participants were seated quietly in a chair with both feet flat on the floor. Two electrodes were placed on the participant's upper chest, and one electrode was placed on the participant's abdomen to record resting ECG signals using the Biopac MP36RWSW System (Goeta, CA). Sampling rate was set at 1,000 Hz. Thereafter, 6 minutes of ECG signals were recorded and were later exported as AQC files (Biopac AcqKnowledge). AQC files were imported into Kubios Premium HRV analysis software (version 3.3.1, MATLAB; The MathWorks, Inc.) for processing and deriving HRV indexes. ECG signals were successfully recorded from 604 participants.

Established guidelines were followed to calculate HRV from ECG signals.⁹ Among participants with at least 5 minutes of data, the automatic correction was employed to detect artifacts. Any files that had >5% artifacts were immediately excluded. Thereafter, files that had ≤5% artifacts underwent further visual evaluation for noise, distortion, missing or premature R waves, ectopic beats, arrythmias, or irregular rhythms; abnormal samples were corrected if possible, according to the guidelines⁹ or otherwise excluded. To account for the potential effects of respiration on HRV, Kubios Premium software estimated the respiration rate from ECG using the amplitude of R waves (ECG-derived respiration rate)²⁴ to use in sensitivity analyses (see analytical method section). We selected HRV indices that have a well-understood physiological and statistical basis and predict CVD outcomes. As such, heart rate, the standard deviation of normal R-R intervals (SDNN; representing the overall variability), and the root mean square of the successive differences (RMSSD; representing cardiac parasympathetic activity) were selected as outcomes of interest. To eliminate inter-rater variability, all HRV processing was completed by one rater. These protocol choices resulted in high reliability with intra-rater intraclass correlations of >0.95.

Covariates

The POUCHmoms follow-up visit linked prospectively collected pregnancy data from the POUCH study and measured confounders and mediators of our hypothesized associations. Demographic, lifestyle, and health-related factors, including age, race (i.e., non-Hispanic white, African American, or other), education (i.e., high school or less, some college, or college degree), working status (i.e., currently working or not working), type of health insurance (i.e., private, Medicaid, or none), and current smoking status (i.e., yes or no) were self-reported. In addition, waist and hip circumferences were measured in triplicate with a Gulick tape measure. The average of hip and waist measurements was used to calculate waist-to-hip ratio. Following 5 minutes of seated rest, systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured three times using an Omron HEM-907 (Omron Healthcare, Inc., Lake Forest, IL, USA) with an appropriately sized cuff. The average of the second and third measurements was calculated as the resting blood pressure.²⁵ Women with SBP ≥140 mm Hg or DBP ≥90 mm Hg or who reported using antihypertensive medications were classified as hypertensive. The presence of DM and/or glucose-lowering medications were also self-reported. Finally, preterm delivery (i.e., yes or no) and pregnancy complications (i.e., gestational HTN, pre-eclampsia, chronic HTN, or pre-eclampsia superimposed on chronic HTN, or none) were collected prospectively via structured interviews and medical record review.

Analytical method

Although the sample size was predetermined for this secondary analysis, a previous effect size distribution analysis suggested that to achieve 80% power in HRV studies, samples of 21, 61, and 233 participants are needed to detect large, medium, and small effect sizes, respectively.²⁶ Given that the number of included participants in our current analyses was 522 (see results section), our study was statistically powered to detect small effect sizes.

Participant characteristics were summarized descriptively as means with standard deviations, medians with 25th and 75th percentiles, or numbers and percentages, as appropriate. Characteristics of included versus excluded women were compared using independent t-tests or Mann–Whitney U tests for normally and non-normally distributed continuous variables, respectively, and chi square test for categorical variables. Outcome variables that were not normally distributed (i.e., SDNN, RMSSD) were natural log transformed. Confounders and/or mediators were defined a priori by constructing a direct acyclic graph (Supplementary Fig. S1). Pearson's correlations between R-R intervals with heart rate, natural logarithm SDNN (lnSDNN), and natural logarithm RMSSD (lnRMSSD) were also checked.²⁷

Generalized linear models examined the cross-sectional relationship between self-reported ST and MVPA with HRV. Model 1 included simultaneous adjustment for ST, MVPA, and confounders, including demographics, lifestyle, and health-related factors. Because women who had preterm delivery or were at higher risk of preterm delivery were oversampled in the POUCH Study subcohort and therefore in POUCHmoms as well, sampling weights were applied to all analyses. Model 2 was further adjusted for clinical factors as potential mediators, including HTN, DM, antihypertensive, and glucose-lowering medications. This progressive covariate adjustment was employed first for total ST and MVPA and then after separation of ST and MVPA into domains (leisure and occupational). Lastly, because prior research suggests that the relationship between ST and health outcomes may differ according to MVPA level,¹⁹ models evaluating relationships between ST and HRV were repeated separately among women who met (i.e., active) and did not meet (i.e., inactive) current leisure MVPA recommendations of 2.5 hours/week.

Sensitivity analyses excluded women with underlying medical conditions that could have affected cardiac autonomic function and HRV (e.g., hypoglycemia, post-traumatic stress disorder, carpal tunnel syndrome, heart arrythmias, neuropathy, cardiac problems), and then further excluded women who did not meet the ECG-derived respiration rate criteria (9–24 breaths/minute). Because most HRV indices have a positive correlation with heart period such that, as heart period increases, HRV indices also increase, researchers have suggested that HRV should be adjusted for heart period or rate.²⁷ Therefore, adjusted HRV indices were calculated according to the current recommendations using the coefficient of variation (CV) technique as following: _CVHRV index = 100 × HRV index/heart period.²⁷ Sensitivity analyses using the adjusted HRV indices were conducted to evaluate the potential influence of heart period.

Stata version 15.0 (StataCorp, College Station, TX, USA) was used to conduct all statistical analyses. The significance level was set as α < 0.05. Cohen's d was calculated as the β divided by the standard deviation of the dependent variable to examine the magnitude of association as recommended for HRV: d = 0.25 is small; d = 0.5 is medium; and d = 0.9 is large.²⁶

Results

A total of 678 women completed the POUCHmoms follow-up assessment visit. Of them, 604 women had sufficient ECG records for 5 minutes of HRV analysis. Of these women, 82 women had invalid HRV records due to the following reasons that prevented HRV calculation: ECG distortion (n = 49), arrythmia/irregular ECG (n = 20), >5% artifacts (n = 10), excessive noise (n = 2), and file error (n = 1). Thus, following exclusions, 522 women were included in the current analyses (Fig. 1). Compared with included women, excluded women (n = 156) had similar demographic characteristics and lifestyle behaviors, but tended to have higher systolic and diastolic blood pressure values, a higher prevalence of HTN, and more frequent use of antihypertensive and glucose-lowering medications (Supplementary Table S1).

FIG. 1.

Flowchart of women who completed ECG measurement. ECG, electrocardiogram.

Table 1 presents characteristics of the sample. The majority were non-Hispanic white (59.0%), nonsmoking (71.5%), currently working (73.4%), insured by Medicaid (54.8%), had normal delivery term (74.5%), and were without other pregnancy complications (89.1%). On average, systolic and diastolic blood pressures were in the normal range, although some participants had HTN (18.8%) or diabetes (4.8%). Median self-reported total ST was 7 hours/day and total MVPA was 0.82 hours/day. Based on leisure time activity, 48.9% of women met MVPA recommendations (i.e., ≥ 2.5 hours/week).

Table 1.

Characteristics of Participants (N = 522)

Characteristic	Mean (SD), median (25th–75th), or n (%)
Age (years)	37.7 (5.7)
Race
White	308 (59.0)
African American	184 (35.2)
Other	30 (4.7)
Education
High school or less	128 (24.5)
Some college	244 (46.7)
College degree	150 (28.7)
Working status
Currently working	383 (73.4)
Currently not working	139 (26.6)
Insurance
Private	189 (36.2)
Medicaid	286 (54.8)
None	47 (9.0)
Currently smoking
No	373 (71.5)
Yes	149 (28.5)
Waist-to-hip ratio	0.8 (0.1)
Systolic blood pressure (mm Hg)	114.0 (13.7)
Diastolic blood pressure (mm Hg)	75.5 (11.1)
Hypertension
No	424 (81.2)
Yes	98 (18.8)
Using medication	62 (63.3)
Not using medication	36 (37.7)
Diabetes
No	497 (95.2)
Yes	25 (4.8)
Using medication	13 (52.0)
Not using medication	12 (48.0)
Preterm delivery
Yes	133 (25.5)
No	389 (74.5)
Pregnancy complications
Gestational hypertension	19 (3.6)
Pre-eclampsia	15 (2.9)
Chronic HTN or pre-eclampsia superimposed on chronic hypertension	23 (4.4)
None	465 (89.1)
Heart rate (beats/minute)	76.8 (10.1)
lnSDNN	3.6 (0.5)
lnRMSSD	3.4 (0.6)
Total ST (hours/day)	7.00 (4.5–9.5)
Leisure ST (hours/day)	3.00 (2.0–5.0)
Occupational ST (hours/day)	3.00 (1.0–5.0)
Total MVPA (hours/day)	0.82 (0.2–3.5)
Leisure MVPA (hours/day)	0.34 (0.1–0.7)
Meeting recommendations	255 (48.6)
Not meeting recommendations	267 (51.1)
Occupational MVPA (hours/day)	0.00 (0.0–2.6)

HTN, hypertension; ln, natural logarithm; MVPA, moderate-to-vigorous-intensity physical activity; RMSSD, root mean square of successive differences; SD, standard deviation; SDNN, standard deviation of normal R-R intervals; ST, sedentary time.

Table 2 displays the independent associations of self-reported total ST and MVPA with heart rate and HRV. Model 1 (i.e., adjusted for confounders) and 2 (i.e., Model 1 with further adjustment for mediators) found no associations between self-reported total ST with heart rate, lnSDNN, or lnRMSSD (each p > 0.05; d range: 0.00–0.02). In contrast, both Model 1 and 2 detected small, favorable, and statistically significant relationships between self-reported total MVPA with lnSDNN (p = 0.011 and 0.006; d = 0.06 and 0.06, respectively) and lnRMSSD (p = 0.011 and 0.011; d = 0.05 and 0.06, respectively), but not with heart rate (p > 0.05; d = 0.04 and 0.05, respectively).

Table 2.

Associations Between Total Sedentary Time and Moderate-to-Vigorous-Intensity Physical Activity with Heart Rate and Heart Rate Variability

Variables	Model	Total ST		Total MVPA
		B ± SE (p-value)	d	B ± SE (p-value)	d
Heart rate (beats/minute)	1	0.068 ± 0.144 (0.637)	0.01	−0.418 ± 0.240 (0.083)	0.04
	2	0.007 ± 0.142 (0.961)	0.00	−0.485 ± 0.251 (0.054)	0.05
lnSDNN	1	−0.011 ± 0.008 (0.197)	0.02	0.028 ± 0.011 (0.011)	0.06
	2	−0.008 ± 0.008 (0.323)	0.02	0.030 ± 0.011 (0.006)	0.06
lnRMSSD	1	−0.010 ± 0.010 (0.292)	0.02	0.032 ± 0.011 (0.011)	0.05
	2	−0.006 ± 0.009 (0.537)	0.01	0.038 ± 0.015 (0.011)	0.06

Bold indicates significant associations (p < 0.05). Model 1 is adjusted for total ST and total MVPA, age, race, education, working status, insurance, smoking, delivery term, pregnancy complications. Model 2 adds adjustment for hypertension, antihypertensive medications, diabetes, glucose-lowering medications, waist-to-hip ratio.

d, Cohen's d; SE, standard error.

Table 3 displays the independent associations after separation of ST and MVPA into domains (leisure and occupational) with heart rate and HRV. No associations were detected between leisure and occupational ST and occupational MVPA with heart rate, lnSDNN, or lnRMSSD (each p > 0.05; d range: 0.00–0.05). However, small, beneficial, and statistically significant relationships were observed between leisure MVPA and lnSDNN and lnRMSSD in fully adjusted models (p < 0.001 [d = 0.21] and p = 0.003 [d = 0.17], respectively). Leisure MVPA was not associated with heart rate (p > 0.05).

Table 3.

Associations Between Leisure and Occupational Sedentary Time and Moderate-to-Vigorous-Intensity Physical Activity with Heart Rate and Heart Rate Variability

Variables	Model	Leisure ST		Occupational ST		Leisure MVPA		Occupational MVPA
		B ± SE (p-value)	d	B ± SE (p-value)	d	B ± SE (p-value)	d	B ± SE (p-value)	d
Heart rate (beats/minute)	1	0.406 ± 0.219 (0.065)	0.04	−0.136 ± 0.239 (0.570)	0.01	−0.784 ± 0.597 (0.189)	0.08	−0.391 ± 0.309 (0.207)	0.04
	2	0.302 ± 0.212 (0.155)	0.03	−0.176 ± 0.235 (0.455)	0.02	−1.234 ± 0.633 (0.052)	0.12	−0.363 ± 0.309 (0.240)	0.04
lnSDNN	1	−0.023 ± 0.013 (0.085)	0.05	0.001 ± 0.010 (0.933)	0.00	0.074 ± 0.028 (0.009)	0.15	0.021 ± 0.013 (0.118)	0.04
	2	−0.015 ± 0.012 (0.241)	0.03	0.001 ± 0.010 (0.934)	0.00	0.101 ± 0.024 (<0.001)	0.21	0.017 ± 0.013 (0.254)	0.04
lnRMSSD	1	−0.027 ± 0.015 (0.078)	0.04	0.006 ± 0.013 (0.680)	0.01	0.074 ± 0.037 (0.045)	0.12	0.030 ± 0.018 (0.104)	0.05
	2	−0.019 ± 0.015 (0.207)	0.03	0.005 ± 0.013 (0.695)	0.01	0.104 ± 0.035 (0.003)	0.17	0.025 ± 0.018 (0.155)	0.04

Bold indicates significant associations (p < 0.05). Model 1 is adjusted for leisure and occupational ST and MVPA, age, race, education, working status, insurance, smoking, delivery term, pregnancy complications. Model 2 adds adjustment hypertension, antihypertensive medications, diabetes, glucose-lowering medications, drugs, waist-to-hip ratio.

Table 4 displays the associations of self-reported leisure and occupational ST with heart rate and HRV following stratification of women based on leisure time MVPA recommendations, adjusted for confounders (Model 1). Small, unfavorable, and statistically significant associations were observed between leisure ST with lnRMSSD (p = 0.042; d = 0.08) in women who did not meet leisure MVPA recommendations; no associations were observed among women who did meet leisure MVPA recommendations. Occupational ST did not have statistically significant relationships with heart rate or HRV regardless of leisure MVPA group.

Table 4.

Associations of Leisure and Occupational sedentary time with Heart Rate and Heart Rate Variability in Active (N = 255) Versus Inactive (N = 267) Women

Variables	Leisure MVPA status	Leisure ST		Occupational ST
		B ± SE (p-value)	d	B ± SE (p-value)	d
Heart rate (beats/minute)	Active	0.226 ± 0.283 (0.425)	0.03	0.369 ± 0.307 (0.425)	0.03
	Inactive	0.576 ± 0.302 (0.058)	0.06	−0.076 ± 0.259 (0.770)	0.01
lnSDNN	Active	−0.001 ± 0.012 (0.931)	0.00	−0.013 ± 0.012 (0.291)	0.03
	Inactive	−0.033 ± 0.018 (0.070)	0.08	−0.005 ± 0.018 (0.687)	0.01
lnRMSSD	Active	−0.002 ± 0.016 (0.907)	0.00	−0.024 ± 0.016 (0.145)	0.04
	Inactive	−0.041 ± 0.020 (0.042)	0.08	0.006 ± 0.017 (0.738)	0.01

Bold indicates significant associations (p < 0.05). Active women were defined as those who self-reported accumulating leisure MVPA of ≥2.5 hours/week; inactive women were defined as those who self-reported accumulating leisure MVPA of <2.5 hours/week. All models adjusted for leisure and occupational ST, age, race, education, working status, insurance, smoking, delivery term, pregnancy complications.

Sensitivity analyses that excluded women with potential underlying medical conditions that could have affected HRV and women who did not meet ECG-derived respiration rate yielded similar results. Comparable relationships were also observed when adjusted cvHRV indices were utilized (Supplementary Tables S2 and S3).

Discussion

This study examined the independent and joint associations between total and domain-specific MVPA and ST with HRV in young to middle-aged women. The main findings were that higher total MVPA was independently associated with higher HRV; this association appeared to be primarily driven by leisure MVPA. Yet importantly, the magnitude of these associations was small. In contrast, neither total nor domain-specific ST were independently associated with HRV in the full sample analysis. However, small and adverse associations were observed where higher leisure ST was associated with lower (worse) HRV specifically among women who did not meet leisure MVPA recommendations.

Consistent evidence indicates that regular MVPA can reduce the risk of CVD.²⁸ These cardiovascular benefits gained by MVPA are hypothesized to be attributed, in part, to improved cardiac autonomic regulation.²⁹ This hypothesis is supported by a meta-analysis of experimental clinical trials reporting beneficial associations between MVPA and HRV.¹³ Building on this evidence, we also found that higher MVPA was associated with better cardiac autonomic regulation (i.e., higher HRV) in women. These data indicate that cardiac autonomic dysregulation may be a mechanism linking physical inactivity and CVD in these women.

Of note, the favorable relationships observed between MVPA and HRV in the current study were apparently due to leisure MVPA, whereas occupational MVPA was found not to be associated with HRV. This finding is of particular interest because emerging evidence indicates the existence of a “Physical Activity Health Paradox” effect on HRV.²⁰ This effect would suggest that leisure physical activity is usually MVPA, performed over short durations with sufficient recovery, and, thus, bring about cardiovascular benefits, including enhanced HRV.^20,30 On the other hand, occupational physical activity is usually performed over long durations with short recovery periods, causing constant cardiovascular overload and potentially cardiac autonomic dysregulation (i.e., reduced HRV).^20,30 This phenomenon was supported by a cross-sectional study in Danish blue-collar workers (n = 514, 40% female) that reported an interaction effect where leisure versus occupational physical activity was more favorably associated with nocturnal HRV.²⁰ Our findings are consistent with this theory in that leisure MVPA was more favorably associated with HRV as compared with occupational MVPA, by both magnitude of effect and statistical significance. However, the direction of the association was that higher occupational MVPA was nonsignificantly associated with higher HRV, whereas the “Physical Activity Health Paradox” theory would hypothesize an inverse association. These studies differed regarding population (mixed gender vs. all female), instruments used to estimate occupational MVPA (i.e., accelerometry vs. questionnaire), and nocturnal versus waking HRV assessment. Further research addressing this phenomenon is warranted.

Strong evidence also suggests that excessive ST may increase the risk of CVD incidence and mortality, independent of physical inactivity.¹⁹ Cardiac autonomic dysregulation is also hypothesized to partially explain this deleterious relationship.³¹ Yet, the current limited literature displays ambiguous findings, with studies reporting no associations^14,15,17,18 or correlations between higher total, leisure, or occupational ST with worse HRV.^14–16 When adjusting for MVPA, one small study, including only young men, found a negative association between total ST and HRV.¹⁶ Two other studies, including mixed sex samples, reported no independent relationship between total ST and HRV,^17,18 similar to our findings. The source of the differing results is not entirely clear, and could be a result of age, sex, or assessment methodology. Future research that considers these differences is warranted.

Although both leisure and occupational ST were hypothesized to associate with unfavorable HRV, these associations were expected to be stronger for leisure ST. This is because adults tend to spend greater time in prolonged bouts of ST (≥30 minutes) during leisure, which could have exaggerated the deleterious impacts on health outcomes.^32–34 Only two previous studies have evaluated the associations of leisure and occupational ST with HRV, both of which were in blue-collar workers. The first study reported no correlations,¹⁵ whereas the other observed a negative relationship between occupational ST and nocturnal HRV.¹⁴ Herein, our stratified analysis detected an inverse association between leisure ST (but not occupational ST) with diurnal HRV only among women who did not meet the leisure MVPA guidelines. These disparities may be explained by MVPA status (i.e., active vs. inactive), timing of HRV measurement (i.e., nocturnal vs. diurnal), and type of occupation (i.e., only blue-collar vs. blue- and white-collar). Regardless, our results are in agreement with a recent harmonized meta-analysis that revealed the strongest dose/response effect of leisure ST on CVD mortality was in adults who did not meet leisure MVPA guidelines.¹⁹ Importantly, meeting leisure MVPA recommendations may eliminate the deleterious impacts of leisure ST on HRV in these women. These differential effects could contribute to the discrepant results in previous research, and future research should consider that the effects of ST may differ dependent on MVPA levels.

Several direct and indirect physiological mechanisms have been proposed to explain how MVPA and ST influence HRV. The consensus is that MVPA improves HRV primarily through increasing cardiac vagal tone.³⁵ This vagal enhancement may be attributed to increased nitric oxide bioavailability and/or suppressed angiotensin II, both of which can exert direct effects on the vagal nerve.³⁵ Furthermore, MVPA may indirectly improve HRV by boosting oxytocin concentration,³⁶ which reacts at the cardiovascular centers in the brainstem, causing vagal enhancement.³⁷ Our findings are most consistent with the proposed mechanisms where higher MVPA was associated with higher resting vagal tone (i.e., higher lnRMSSD). On the other hand, ST may directly affect HRV by chronic reductions in shear stress and, eventually, decreased nitric oxide bioavailability,³⁸ leading to attenuated vagal tone. This hypothesized mechanism was partially supported by our findings where higher leisure ST related to lower resting vagal tone (i.e., lower lnRMSSD) in the absence of sufficient MVPA. Yet, more research into potential mechanisms is needed to advance our understanding of how MVPA and ST may affect HRV.

A strength of our current study is that we separately evaluated the associations of total and domain-specific MVPA and ST with HRV specifically in young- to middle-aged women. Previous studies included either mixed sex samples^{14,15,17,18,20} or young men only¹⁶ and examined association of only total^16–18 or domain-specific^14,15,20 physical activity. Thus, our study provides more comprehensive data on the relationships between MVPA and ST with HRV. Moreover, our unique sample was a cohort of multiracial women, which enhances the generalizability of our results. Furthermore, the POUCHmoms study oversampled women with pregnancy complications such as preterm birth or hypertensive disorders of pregnancy. These adverse pregnancy outcomes are known to increase the long-term risk of CVD and death.^39,40 Therefore, another strength of our study is that it uniquely examined the associations of MVPA and ST with cardiac parasympathetic regulation in this higher risk population. Lastly, we also utilized the gold standard measure of HRV (i.e., ECG) and carefully implemented the robust guidelines to process the ECG data.

Yet, several limitations should be considered when interpreting our findings. Our study was observational and cross-sectional, making it susceptible to biases such as residual confounding and reverse causality. Studies with longitudinal designs that establish temporality or use experimental manipulation of ST are needed. In addition, the reported estimates may only be applied to relatively healthy women. Furthermore, although the MAQ is valid and provides useful estimates of total and domain-specific activity,²³ it is susceptible to measurement error and self-report bias that might influence our estimates. Therefore, future studies should consider using gold standard measures of MVPA (i.e., accelerometers) and ST (i.e., thigh-worn monitors) in conjunction with self-report instruments. Although the respiration rate was estimated from ECG using the amplitude of R waves,²⁴ this feature has not yet been validated in Kubios software. Thus, respiration rate could have significantly influenced our estimates⁹ and should also be considered in future research. Lastly, HRV only reflects overall and cardiac parasympathetic activity when resting. As such, the association of ST with cardiac sympathetic overactivation, a mechanism that has been specifically hypothesized as the primary autonomic pathway linking ST with CVD,³¹ remains to be evaluated.

Conclusion

Altogether, our results suggest that leisure MVPA, among inactive women, and leisure ST may be determinants of cardiac autonomic regulation. Importantly, these findings contribute mechanistic insight into the pathway between low MVPA, high ST, and CVD risk development among women. These data are especially important as MVPA and ST are major modifiable risk factors for CVD. Furthermore, among women with significant barriers to achieving MVPA, lowering ST is potentially an additional strategy to improve HRV and mitigate CVD risk in women. Future longitudinal studies and experimental trials with manipulation of ST are needed to confirm associations with cardiac sympathetic activity.

Authors' Contributions

J.M.C. and C.H. designed the study and collected data. AA and BG analyzed, interpreted, and wrote the article. A.B.A., B.B.G., J.M.C., J.R.J., C.E.K., E.N., and C.H. reviewed, revised, and approved the final article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This POUCHmoms Study was supported by the National Heart, Lung, and Blood Institute [R01-HL103825]. The POUCH Study was supported by the Perinatal Epidemiological Research Initiative Program Grant from the March of Dimes Foundation [20FY01-38 and 20FY04-37], the Eunice Kennedy Shriver National Institute for Child Health and Human Development, the National Institute of Nursing Research [R01-HD34543], the Thrasher Research Foundation [02816-7], and the Centers for Disease Control and Prevention [U01-DP000143-01].

Supplementary Material

Supplementary Figure S1

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3