Sedentary Time and Physical Activity in Older Women Undergoing Exercise Training

Abstract

Older adults have low rates meeting the physical activity (PA) guidelines and high sedentary time. Low PA and excessive sedentary time have been linked to adverse health outcomes. Less is known about whether exercise training influences sedentary time and PA in various intensities.

PURPOSE:

To examine the effects of a 16-week aerobic exercise training on time spent being sedentary and on PA time of light (LPA) and moderate-to-vigorous (MVPA) intensities, and step numbers in older women.

METHODS:

Inactive women (n = 61; age = 65.5 ± 4.3 years) participated in moderate-intensity walking of either a low or a moderate dose (33.6 and 58.8 kJ·kg⁻¹ body weight per week, respectively). They wore a SenseWear Mini Armband at baseline and at end-intervention to determine sedentary, LPA, and MVPA time, and step numbers.

RESULTS:

Time being sedentary, or spent on LPA and MVPA, did not change differently by exercise groups with different doses (all p values for group by time interaction > 0.580). Overall, time being sedentary reduced from baseline to end-intervention by approximately 39 minutes (p < 0.001) and LPA increased by 19 minutes per day (p = 0.003). MVPA time increased (p < 0.001), which was primarily accounted for by the supervised exercise. Interestingly, daily steps increased more in the moderate-dose than the low-dose group (p = 0.023 for group by time interaction; 33.6% and 19.8% median increase in moderate- and low-dose groups, respectively). Also, there were individual differences in these changes.

CONCLUSION:

Results indicated that on average, older women did not reduce time of LPA or MVPA outside of the exercise program, or increase sedentary time as a result of participating in the exercise program.

INTRODUCTION

The beneficial health effects of moderate-to-vigorous-intensity physical activity (MVPA) are well recognized. The US Department of Health and Human Services recommends accumulating at least 150 minutes of MVPA or 75 minutes of vigorous-intensity PA per week for adults, including older adults, for obtaining substantial health benefits of exercise (1). Although the literature identifies regular PA as an important predictor of successful aging, older adults continue to exhibit the lowest rate for meeting the PA guidelines in comparison to all other adult age groups (2,3). Studies have prescribed MVPA in a supervised setting to individuals in order to increase their MVPA levels; however, there is evidence suggesting that exercise participation may lead to compensatory adjustments in non-exercise PA and sedentary time which may lessen the intended healthful effects of exercise (4–6). Further, these compensatory adjustments are suggested to be more in older women (7).

Despite the vulnerability of older adults, there are a limited number of studies in this age group that have objectively examined the impact of an exercise intervention on total PA or non- exercise PA. An earlier study in a small group of older men and women found increased total daily PA counts measured by accelerometry following 6 and 12 weeks of combined aerobic and resistance training (8). In the Dose Response to Exercise in Women (DREW) study, step counts in postmenopausal women outside of the exercise intervention were not different from baseline levels irrespective of exercise group assignment (4, 8, or 12 kcal•kg⁻¹ body weight per week) (9). These studies did not differentiate MVPA or light-intensity PA (LPA) except during the supervised exercise sessions. However, this information is important because MVPA and LPA may offer different degrees of health benefits. The beneficial effects of LPA for mortality were found to be less than MVPA of equal volume (10), even though LPA is also beneficially associated with obesity, lipid and glucose metabolism, and mortality (11).

Examining time spent on MVPA and LPA during an exercise intervention is inherently linked to the investigation of sedentary time, which has significant ramifications given the accumulating evidence of an association between sedentary behavior and all-cause mortality, cardiovascular disease, type 2 diabetes, and metabolic syndrome (12–14). Of note, reducing sedentary behavior has been added to the latest PA guidelines (1). However, older adults not only have the lowest rate for meeting the PA guidelines, but they also spend an average of 9.4 hours (range: 8.5 – 10.7 hours) per day being sedentary (15). Few studies have examined sedentary time in older adults who start participating in an intervention with prescribed MVPA. An analysis of the Lifestyle Interventions and Independence for Elders (LIFE) study showed that participants in the group participating in MVPA had an average of 9 minutes shorter sedentary time than those in the health-education control group (16). On the contrary, a small study did not observe a change in sedentary behavior in older adults participating in a behavioral intervention with a goal to reach 150 minutes of MVPA each week (17). Thus, it remains unclear whether sedentary time changes in older adults as a result of participating in an exercise program.

These behavioral changes in non-exercise PA and time spent sedentary are highly individual in that differences have been noted among individuals enrolled in the same exercise intervention (18). The variability in these changes has important implications for identifying factors that may potentially contribute to this response. One possible factor is cardiorespiratory fitness, a physiological measure reflecting a combination of genetic potential, behavioral, and functional health of various organ systems (19,20). Another factor is body weight; one prospective study observed baseline body weight to be associated with a decrease in MVPA during a follow-up approximately 6 years later (21). Therefore, the purpose of this study was to examine the effects of an aerobic exercise training program on time spent on PA at different intensities and being sedentary, and the number of steps in older women, and whether cardiorespiratory fitness and body weight affected any of these changes.

METHODS

Participants

Data used in this study were from the Women’s Energy Expenditure in Walking Programs (WEWALK) study, which was a randomized controlled trial designed to examine energy expenditure responses to 16 weeks of moderate-intensity walking in older women (22). The study was registered at ClinicalTrials.gov (NCT01722136). The research protocol was approved by the University of South Carolina Institutional Review Board, and all participants provided written informed consent.

The participant inclusion and exclusion criteria were described previously (22). Briefly, participants were older women (age 60–75 years), non-obese (body mass index, BMI = 18–30 kg·m⁻²), self-reported weight being stable (± 3%) in the past 3 months, physically inactive (less than 20 minutes × 3 times per week of structured exercise) in the past 3 months, and non-smoking in the past year. They did not have self-reported or signs of serious cardiovascular, metabolic, or respiratory diseases, or any other conditions that might affect adherence to the study protocol, affect exercise safety, or may be aggravated by exercise. Additionally, participants did not use medications known to affect exercise performance or metabolism.

Exercise training

Women were randomized to one of two moderate-intensity walking training protocols at low- or moderate-doses (target exercise energy expenditure of 33.6 and 58.8 kJ·kg⁻¹ body weight weekly, respectively). Both groups were instructed to attend three supervised exercise sessions per week located in the Clinical Exercise Research Center at the University of South Carolina. The target intensity of the exercise was 50–55% of each woman’s heart rate reserve, calculated as [(peak heart rate – resting heart rate] × intensity (50–55%) + resting heart rate], with resting heart rate and peak heart rate obtained before and during the graded exercise test (see below). Exercise energy expenditure was estimated based on walking duration at each speed and grade, and body weight using a standardized formula by the American College of Sports Medicine (23). Exercise progressively increased for both groups from a low intensity and low weekly dose until the assigned exercise intensity and dosage were reached. The low- and moderate-dose groups reached the target exercise within 5 and 8 weeks, respectively. Each exercise session began with a 3-minute warm-up and ended with a 3-minute cool down. Women wore heart rate monitors (FT1, Polar, Lake Success, NY, USA) throughout each exercise session to monitor heart rate.

Tests and measurements

Height and body weight were measured at baseline before randomization, weekly during exercise intervention, and at the end of intervention. Measurements were obtained while women were in standard scrubs and without shoes or outer garments.

Graded exercise test.

A graded exercise test was performed at baseline and upon completion of the exercise training program for determination of cardiorespiratory fitness. The test at baseline was also used to exclude women with cardiopulmonary limitations to exercise. A maximum of 10 minutes of walking on the treadmill prior to the test was allowed for participants to familiarize themselves with the instrumentation. After a brief rest, women walked at a self-selected speed with the grade of treadmill increasing by 2% every 2 minutes throughout the test. Gas analysis was performed by a metabolic measurement system (TrueOne 2400, Parvo Medics, Salt Lake City, UT, USA), and a 12-lead electrocardiogram was monitored by a standard system (Quinton Q-Stress®; Cardiac Science, Bothell, WA, USA) continuously during the entire test. Blood pressure was measured during each 2-minute stage. A valid peak volume of oxygen consumption (${\dot{\text{V}}\text{O}}_{\text{2peak}}$) was obtained when at least 2 of the following criteria were met: 1) plateau in ${\dot{\text{V}}\text{O}}_2$ (a change < 2 ml•kg⁻¹•minute⁻¹) with increasing work rate, 2) heart rate surpasses > 90% of age-predicted maximum heart rate (220 - age), 3) respiratory exchange ratio ≥ 1.10, and 4) a rate of perceived exertion ≥ 17 on the Borg exertion scale.

Physical activity.

The SenseWear Mini Armband (BodyMedia Inc. Pittsburgh, PA, USA) was used to measure PA and sedentary time at baseline, and mid- and end-intervention. Women were instructed to wear the monitor all the time for 14 days during each phase except during water activities. Previous research indicated acceptable agreement between energy expenditure assessed by the SenseWear monitor with the doubly-labeled water method (24–26) and with indirect calorimetry for sedentary behavior and LPA (27). The monitor is a portable, multi-sensor device worn on the upper left arm that incorporates tri-axial accelerometry with measures of heat flux, galvanic skin response, skin temperature and near-body ambient temperature. Data were analyzed using the software provided by the manufacturer (SenseWear Professional 8.0, BodyMedia, Inc). Specifically, the algorithm combines data collected from the monitor’s sensors with individual information (age, sex, height, weight, smoking, and handedness) to give estimates of energy expenditure for each minute of wear time, which are then converted to metabolic equivalents (METs). The SenseWear software classifies activity as either “awake” or “asleep”. Therefore, the time spent being sedentary was calculated by subtracting time designated as “asleep” from time with METs ranging from 1.0 to ≤ 1.5. The software does not provide information about body posture. For example, standing still and sitting are both classified as being sedentary. Time spent on PA was classified by intensity according to the estimated MET level when the participant was awake, based on the following criteria: light, 1.5 to ≤ 3.0 METS; moderate, 3.0 to ≤ 6.0 METs; and vigorous, > 6.0 METs. Due to participants engaging in very low amounts of time on vigorous activities, moderate and vigorous intensities were combined as MVPA. In addition, the MVPA time outside of the exercise program was calculated by subtracting the daily average time spent in the supervised exercise sessions during the period the SenseWear monitor was worn, for each individual at mid- and end-intervention. The SenseWear monitor also provides an estimate of total number of steps. Data from women with at least 5 days including at least one weekend day of ≥ 90% of 24 hours (21.6 hours) of wear time on each of the days were considered valid.

Statistical analysis

Descriptive statistics, including means and standard deviations (SD) and proportions, were calculated. Differences between randomized exercise groups at baseline were determined using t-tests or chi-square tests, as appropriate. A repeated measure analysis including a group by time interaction was performed using a mixed-effects model with a random intercept, which allowed for various baseline values among individuals. When the group by time interaction was not statistically significant, further analyses were conducted with the two groups combined. As appropriate, follow-up analyses were conducted to determine at which time point (baseline, mid-intervention, or end-intervention) the outcome of interest was different. Separate models were run for each variable of interest, including body weight, ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, sedentary time, and PA measures.

Additional analyses were performed to adjust for body weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ in the models testing sedentary time and PA measures. These were conducted in two ways. One was to adjust for body weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ as time-varying variables, and the other was to adjust for weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ at baseline and their respective changes from baseline to end-intervention. In separate models, we examined whether baseline body weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ moderated the changes in sedentary time and PA by including those variables along with their individual interactions with time (baseline weight by time and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ by time). Exploratory analyses were conducted to follow up significant interactions. Analyses were conducted using the SAS software (SAS Institute, Cary, NC). Statistical significance was defined as a p value < 0.05. Bonferroni adjustments were used to account for multiple comparisons as appropriate.

A total of 69 women completed the study and had valid SenseWear data at baseline. Among these women, valid SenseWear data were not available for 4 women at mid-intervention and 4 different women at end-intervention. Therefore, the primary data analysis was performed and reported on the 61 women who had complete SenseWear data at all three time points. Additional analyses were performed including the 8 women with SenseWear data at two of the three time points. Results were similar to those without these women; therefore, these results were not reported.

RESULTS

Participant characteristics

Participant characteristics are included in Table 1 by assignment group. These women were older (age = 65.5 ± 4.3 years, mean ± SD in this manuscript), non-obese (BMI = 25.6 ± 3.6 kg·m⁻²), and mainly white (85.2%). There were no differences between the two groups in age, race and ethnicity distribution, height, weight, or ${\dot{\text{V}}\text{O}}_{\text{2peak}}$.

Table 1.

Participant characteristics by assignment group

	Low-dose (n = 32)						Moderate-dose (n = 29)
	Mean ± SD	Min	Q1	Median	Q3	Max	Mean ± SD	Min	Q1	Median	Q3	Max
Age, years	65.7 ± 4.7	60.1	61.6	64.5	68.2	75.0	65.2 ± 4.0	59.4	62.0	64.6	66.9	74.1
Non-White, n (%)	4 (12.5%)	-	-	-	-	-	5 (17.2%)	-	-	-	-	-
Height, cm	162.2 ± 6.4	149.1	158.8	162.4	166.4	172.9	163.0 ± 5.5	153.2	160.5	162.2	166.2	173.7
Weight, kg
Baseline	67.3 ± 10.3	43.6	60.5	66.9	74.6	85.4	67.3 ± 9.0	47.0	61.4	66.1	75.1	84.6
Mid-intervention	66.7 ± 10.2	43.5	59.7	66.9	73.2	85.3	67.0 ± 8.7	48.3	61.9	66.3	75.6	79.4
End-intervention	66.6 ± 10.3	42.7	60.0	66.3	73.3	85.4	66.7 ± 9.0	47.7	61.6	65.7	75.1	79.3
${\dot{\text{V}}\text{O}}_{\text{2peak}}$, ml•kg−1 •min−1
Baseline	20.0 ± 3.7	13.7	17.5	19.8	22.2	28.2	20.1 ± 3.8	13.5	17.3	19.5	22.3	28.8
End-intervention	20.7 ± 4.4	12.9	17.6	21.1	23.0	28.7	22.3 ± 5.0	14.0	20.0	21.6	24.0	35.2

Min, minimum; Q1, lower quartile; Q3, higher quartile; max, maximum.

Women attended 47.0 ± 6.9 sessions and those in the low-dose and moderate-dose groups walked 109 ± 11 minutes and 160 ± 19 minutes per week in our research center, respectively. The adherence to prescribed exercise dose (expressed as actual/prescribed ratio) was 104.6 ± 8.6 % and 99.1 ± 7.9 % for the two groups, respectively. Valid SenseWear data were available from 12.1 ± 2.0 days, 11.9 ± 2.3 days, and 11.6 ± 2.3 days at baseline, and mid- and end-intervention, respectively. The wear time for valid days ranged between 21.65 to 24 hours with an average of 23.60 hours per day.

Changes in body weight and cardiorespiratory fitness with exercise training

The two exercise groups had similar changes in body weight with exercise training (p for group by time interaction = 0.879). In the entire sample, body weight significantly reduced over time (p = 0.007) with weight at baseline being higher than at end-intervention (p = 0.002), but both were not different from mid-intervention (p = 0.025 and 0.392 for comparison with baseline and end-intervention weight, respectively; p < 0.017 signifies statistical significance with Bonferroni adjustments). The average change in body weight over the entire period (values at end-intervention minus at baseline) was −0.7 ± 2.1 kg. The lowest value, lower quartile, median, higher quartile, and the highest value of weight change was −6.4, −1.7, −0.6, 0.7 and 4.9 kg, respectively. ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ did not change differently from baseline to end-intervention between the two groups (p = 0.054 for group by time interaction), and increased significantly in the combined sample (p < 0.001) by 1.44 ± 3.0 ml•kg⁻¹•min^-1. The lowest value, lower quartile, median, higher quartile, and the highest value for ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ changes was −5.8, −0.2, 0.9, 3.2, and 8.8 ml•kg⁻¹•min⁻¹, respectively. Both weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change values were normally distributed around their mean, and the most extreme value was < 2.7 SD and < 2.5 SD away from the mean weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ changes, respectively.

Changes in sedentary time and physical activity with exercise training

The time spent being sedentary, on LPA and MVPA, and number of daily steps are included in Table 2 by group. At baseline, the two groups were similar in each of the time variables and number of daily steps (all p values > 0.430). The two groups did not differ in the changes in the time being sedentary, or time spent on LPA, MVPA, or total PA across the exercise training (all p values for group by time interaction > 0.580). Therefore, the two groups were combined to examine the main effect of exercise training on these time variables.

Table 2.

Average daily sedentary time and physical activity at baseline, mid- and end-intervention, by assignment group

Variable	Baseline	Mid-intervention	End-intervention
Sedentary time, minutes•day−1
Low-dose	708 ± 122	683 ± 122	667± 116
Moderate-dose	686± 108	680 ± 99	648 ± 95
Light physical activity, minutes•day−1
Low-dose	272± 115	278± 101	289± 100
Moderate-dose	263 ± 81	264 ± 73	285 ± 82
Total Moderate-to-vigorous physical activity, minutes•day−1
Low-dose	41 ± 30	64 ± 33	58 ± 36
Moderate-dose	47 ± 32	74 ± 35	69 ± 33
Moderate-to-vigorous physical activity outside of the prescribed exercise, minutes•day−1
Low-dose	41 ± 30	48 ± 33	46 ± 34
Moderate-dose	47 ± 32	53 ± 35	51 ± 33
Total physical activity, minutes•day−1
Low-dose	313 ± 140	342±121	347±121
Moderate-dose	310± 104	338 ± 94	354±314
Steps, numbers•day−1
Low-dose	6618±2537	8165±2865	7671±2592
Moderate-dose	6536±2735	8710±2771	8763 ± 2772

Data are mean ± SD.

Time being sedentary significantly reduced (p = 0.001), and LPA, MVPA, and total PA time significantly increased (p = 0.007, <0.001, and <0.001, respectively); however, time of MVPA outside of the exercise program did not change over time (p = 0.126). Time being sedentary was not different between baseline and mid-intervention (p = 0.129), but it was significantly longer at baseline and mid-intervention than at end-intervention by approximately 23 (p = 0.029) and 39 minutes per day (p < 0.001) (from mixed effects model), respectively. Similarly, LPA time was not different between baseline and mid-intervention (p = 0.623), but it was shorter at baseline and mid-intervention in comparison to end-intervention by approximately 16 (p = 0.013) and 19 minutes per day (p = 0.003), respectively. MVPA showed a different pattern of change than sedentary time and LPA. MVPA at baseline was shorter than at mid- and end-intervention by approximately 25 and 20 minutes, respectively (p < 0.001 for both). There was no difference between mid- and end-intervention (p= 0.112). Similarly, total PA time increased by approximately 28 minutes from baseline to mid-intervention (p = 0.001), and 39 minutes from baseline to end-intervention (p < 0.001), but no difference from mid- to end-intervention (p = 0.199).

Interestingly, the number of daily steps changed differently between the two groups (p for group by time interaction = 0.023). Follow-up analyses showed that the only difference between the two groups was between baseline and end-intervention (p for group by time interaction = 0.006). Both groups increased daily steps (p = 0.001 and < 0.001, respectively for the low- and moderate-dose groups), but the degrees of changes were different (increased by 1053 steps and 2227 steps, or a median increase of 19.8% and 33.6%, respectively in low- and moderate-dose groups). Daily steps increased from baseline to mid-intervention similarly in the two groups (p < 0.001 in both low- and moderate-dose groups; p for group by time interaction = 0.200), and did not change in either group from mid- to end-intervention (p = 0.121 and 0.847 in low- and moderate-dose groups, respectively; p for group by time interaction = 0.125).

Influence of body weight and cardiorespiratory fitness

${\dot{\text{V}}\text{O}}_{\text{2peak}}$ was measured at baseline and end-intervention; thus, models adjusting for ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ and body weight did not include mid-intervention. When body weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ at baseline and their respective changes over time were adjusted for in the models, sedentary time remained decreased, MVPA time, total PA time, and daily steps remained increased, although these changes seem to be of smaller degrees compared to without adjustment. LPA and MVPA outside of exercise program no longer changed from baseline to end-intervention (Table 3). When body weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ were adjusted for as time-varying variables in the models, similar results were found (data not shown).

Table 3.

Mixed effects models examining changes in sedentary time and physical activity with covariates of weight and cardiorespiratory fitness at baseline and change during intervention

Dependent variable	Independent variable	Regression coefficient	SE	P value
Sedentary time, minutes per day
	End-intervention	−29.7	12.4	0.020
	Baseline weight, kg	2.8	1.2	0.027
	Weight change, kg	6.0	4.3	0.164
	Baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, ml•kg−1•min−1	−11.4	3.1	<0.001
	${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change, ml•kg−1•min−1	−3.9	3.0	0.201
Light physical activity, minutes per day
	End-intervention	13.2	7.3	0.075
	Baseline weight, kg	−3.1	1.1	0.008
	Weight change, kg	−4.2	2.8	0.140
	Baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, ml•kg−1•min−1	6.6	2.9	0.025
	${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change, ml•kg−1•min−1	2.1	2.0	0.291
Total moderate-to-vigorous physical activity, minutes per day
	End-intervention	15.5	4.0	<0.001
	Baseline weight, kg	−0.16	0.35	0.653
	Weight change, kg	−3.1	1.6	0.066
	Baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, ml•kg−1•min−1	4.0	0.91	< 0.001
	${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change, ml•kg−1•min−1	1.4	1.2	0.222
Moderate-to-vigorous physical activity outside of prescribed exercise, minutes per day
	End-intervention	0.84	3.9	0.828
	Baseline weight, kg	−0.12	0.35	0.731
	Weight change, kg	−3.2	1.6	0.048
	Baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, ml•kg−1•min−1	3.9	0.90	<0.001
	${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change, ml•kg−1•min−1	1.1	1.1	0.325
Total physical activity, minutes per day
	End-intervention	28.5	9.1	0.003
	Baseline weight, kg	−2.6	1.4	0.057
	Weight change, kg	−7.4	3.5	0.039
	Baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, ml•kg−1•min−1	11.5	3.5	0.002
	${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change, ml•kg−1•min−1	3.6	2.5	0.151
Steps per day
	End-intervention	1366	251	< 0.001
	Baseline weight, kg	36.8	32.6	0.263
	Weight change, kg	−82.5	100.8	0.416
	Baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, ml•kg−1•min−1	288.9	83.6	0.001
	${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change, ml•kg−1•min−1	132.0	71.8	0.071

Baseline is reference time point. The regression coefficient for end-intervention is the difference between end-intervention and baseline in the mean of each outcome, sedentary time or physical activity, adjusting for other variables in the model. The regression coefficients for baseline weight, weight change, baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ change are their respective regression slopes against each outcome variable, adjusting for other variables in the model.

As shown in Table 3, baseline weight was positively associated with sedentary time, negatively associated with LPA time, but was not associated with MVPA or total PA time, or steps. For example, for each kg of greater baseline weight, LPA was 3.1 minutes less, holding all other variables in the model constant. Changes in weight was negatively associated with MVPA outside of exercise program and total PA time. Baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ was negatively associated with sedentary time, positively associated with time spent in LPA, MVPA, and total PA, and number of daily steps. Changes in ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ were not associated with sedentary time or any PA variables.

Separate models including baseline weight, ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, and their interactions with time were used to examine potential moderating effects of baseline weight and ${\dot{\text{V}}\text{O}}_{\text{2peak}}$. LPA time was the only outcome that did not show a significant weight by time (p = 0.237) or ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ by time (p = 0.069) interaction. A significant baseline weight by time interaction was found for MVPA (total and outside of exercise program) and total PA time, as well as number of steps (p = 0.005, 0.005, 0.032, 0.004, respectively). These results indicated differential changes in the outcomes across different baseline weights. To follow up these significant interactions, we categorized into sub- groups of similar sample size using quartiles of baseline weight considering the limited sample size. Changes in MVPA and total PA time, and number of steps within each quartile were subsequently determined (Table 4). The outcomes significantly changed in some but not other quartiles, and the overall trend was that within higher quartiles of baseline weight, MVPA and total PA time and daily steps were more likely to increase.

Table 4.

Changes in sedentary time and physical activity within each quartile categorized based on baseline body weight or baseline cardiorespiratory fitness

Dependent variable	Independent variable	Regression coefficient	SE	P value
Quartiles of baseline weight (low to high)
Total moderate-to-vigorous physical activity, minutes per day
	Quartile 1 (lowest)	11.7	6.8	0.108
	Quartile 2	15.6	9.1	0.110
	Quartile 3	20.2	4.3	< 0.001
	Quartile 4 (highest)	31.0	8.4	0.003
Moderate-to-vigorous physical activity outside of prescribed exercise, minutes per day
	Quartile 1 (lowest)	−3.1	7.0	0.662
	Quartile 2	0.3	8.6	0.972
	Quartile 3	3.9	4.1	0.362
	Quartile 4 (highest)	17.3	8.1	0.050
Total physical activity, minutes per day
	Quartile 1 (lowest)	31.1	20.6	0.153
	Quartile 2	22.8	14.5	0.138
	Quartile 3	30.6	13.3	0.038
	Quartile 4 (highest)	71.2	18.5	0.002
Steps per day
	Quartile 1 (lowest)	1146	554	0.058
	Quartile 2	1339	445	0.009
	Quartile 3	1608	404	0.001
	Quartile 4 (highest)	2370	491	< 0.001
Quartiles of baseline ${\dot{\mathbf{\text{V}}}\mathbf{\text{O}}}_{\mathbf{\text{2peak}}}$ (low to high)
Sedentary time, minutes per day
	Quartile 1 (lowest)	4.8	33.3	0.888
	Quartile 2	−35.4	14.6	0.030
	Quartile 3	−39.2	19.1	0.061
	Quartile 4 (highest)	−84.6	18.5	< 0.001
Total physical activity, minutes per day
	Quartile 1 (lowest)	15.4	12.5	0.237
	Quartile 2	46.4	13.6	0.005
	Quartile 3	30.2	28.6	0.310
	Quartile 4 (highest)	61.1	17.9	0.004
Steps, numbers per day
	Quartile 1 (lowest)	1166	293	0.001
	Quartile 2	1726	463	0.003
	Quartile 3	1483	610	0.030
	Quartile 4 (highest)	2042	480	< 0.001

Baseline is reference time point. The regression coefficient is the difference between end-intervention and baseline in the mean of each outcome.

Similarly, significant ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ by time interaction was found for sedentary time, total PA and number of daily steps (p = 0.004, 0.035, and 0.039, respectively). Within each quartile of baseline ${\dot{\text{V}}\text{O}}_{\text{2peak}}$, changes in sedentary time, total PA are also shown in Table 4. Sedentary time decreased and total PA time increased in the second and highest quartile, but not in the lowest or the third quartiles. Daily steps increased in all quartiles but differed in magnitude.

Distribution of individual changes in sedentary and physical activity time

The changes from baseline to end-intervention in time being sedentary, and LPA, MVPA, and total PA time were calculated for each woman. Figure 1 shows the distribution of time changes and the percentage of women in each successive range of distribution for sedentary, LPA, MVPA, and MVPA time outside of the exercise program. The majority of women (78.7 %) reduced their sedentary time, and more women (60.7 %) increased LPA time than those who reduced (39.3 %). For MVPA time, 82.0 % of women increased; however, once time spent in the exercise program was removed from MVPA, the distribution shifted to only 54.1 % of women having an increase. Twenty-eight (45.9 %) women reduced sedentary time and increased both LPA and MVPA time (no compensatory changes in sedentary or PA time), and 3 (4.9 %) women increased sedentary time, and reduced both LPA and MVPA time (had compensatory changes in all three).

Figure 1.

Percentage of women in each successive range of time changes of sedentary and physical activity. Changes were calculated using values at end-intervention minus at baseline. Each column represents the percentage of women in the range of time that is less than the labeled value and greater than the previous labeled value.

DISCUSSION

This is one of the few studies that have objectively determined time being sedentary and time spent on PA by intensity, and number of daily steps, in older women who participated in moderate-intensity aerobic exercise training. Our primary results are that sedentary time reduced, time spent on LPA, MVPA, and total PA, and number of daily steps increased from baseline to end-intervention in these older women. We also showed individual differences in the changes, and that weight and cardiorespiratory fitness could affect these changes.

As with many previous exercise trials, the center-based moderate-intensity exercise training was monitored in our study (22). The increase in MVPA time was primarily accounted for by the center-based exercise sessions. The exercise training reached the full target volumes at moderate-intensity by mid-intervention; as such, MVPA time did not further increase from mid- to end-intervention. Women also did not have changes in the time of MVPA outside of the center-based exercise program.

There was an increase in LPA time, which primarily occurred after mid-intervention. The literature suggests greater LPA has a stronger influence on mortality among those performing the least MVPA (10). Considering the time many of our participants spent on MVPA was below or just meeting the PA guidelines, the increase in LPA time is encouraging. The fact that the increase in LPA time primarily occurred after mid-intervention may be because women were still adapting, physiologically and behaviorally, to the increasing exercise dosage in the first few weeks. In line with this, a review of studies examining the effect of exercise training on non-exercise PA suggested that compensatory changes in non-exercise PA would decrease over time as fitness level increases and lifestyle changes develop (7).

Reciprocally, sedentary time decreased by an average of 39 minutes per day from baseline to end-intervention. In the LIFE study, the PA intervention group also decreased sedentary time 6 months after being in the intervention group; however, the decrease in comparison to the health education control group was small (group difference: 9 minutes) (16). The PA intervention involved a combination of center-based and home-based activities of moderate-intensity walking. The change in MVPA time was not reported in this analysis, but another analysis of the LIFE study reported participants in the PA intervention participated in 213 minutes per week of MVPA with an average increase of 15 minutes per week from baseline to after 6 months (28). The health education control group, however, had an average decrease of MVPA by 25 minutes; resulting in a 40-minute per week difference between the two groups. In another study, older adults were randomized to a Get Active group or a Sit Less group receiving consultations with a goal to increase MVPA or reduce sedentary time, respectively (17). The Get Active group increased MVPA by an average of 67 minutes per week, but neither group decreased sedentary time assessed by the SenseWear monitor. In our study, MVPA increased by a larger degree (an average of 20 minutes per day) compared to these two studies in older adults (16,17) which may have contributed to the larger decrease in sedentary time over time.

Our data demonstrated individual differences in responses to exercise training in these older women. The majority of women reduced their sedentary time, and more than half increased LPA and MVPA time. In a previous study in postmenopausal women who participated in 13 weeks of moderate-intensity walking, about 56% of them increased while the rest reduced MVPA time (29). However, in our study, slightly more than half of the sample had at least one undesirable change (increased sedentary time and/or reduced LPA and/or MVPA). We attempted to examine whether body weight and cardiorespiratory fitness, the two factors that are believed to be associated with sedentary behavior and PA participation, contributes to these individual differences.

Our results supported a moderating role of body weight and cardiorespiratory fitness in their responses to exercise intervention. It appeared women with higher weight (note all women were non-obese) at baseline were more likely to increase PA time than those in the lower weight categories. The findings for cardiorespiratory fitness were not very straight forward in that women in the second and fourth (highest) quartile of ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ were more likely to reduce sedentary time and increase total PA time, which may be because the ${\dot{\text{V}}\text{O}}_{\text{2peak}}$ is a function of both absolute peak oxygen consumption and body weight. A possible explanation is that women who have optimal fitness level and weight may be better able to engage in more activities and be less sedentary outside of the center-based exercise training.

On the other hand, the disappearance of significant changes in LPA time and MVPA time outside of the exercise program, and attenuated changes in sedentary time, and MVPA and total PA time after adjusting for weight and cardiorespiratory fitness (either baseline and change values or time-varying), suggest that body weight and cardiorespiratory fitness account for, at least in part, the changes in sedentary and PA variables. In our study, body weight and cardiorespiratory fitness both changed with small degrees at the mean level with variations among individuals. These physiological characteristics can be the result of habitual behavioral factors including sedentary behavior and PA participation. They may also change as a result of the intervention and affect the changes in the behavioral factors in response to an intervention. These associations are therefore intricate, and may be affected by genetic, metabolic, and endocrine factors.

Another interesting finding of our study was that the increase in daily steps from baseline to end-intervention in the moderate-dose group was almost two folds of that in the low-dose group. In the DREW study, step counts outside of the exercise intervention were not different between the three exercise training groups of varying energy expenditures (4, 8, and 12 kcal∙kg⁻¹ body weight per week) among postmenopausal women, but the step counts during exercise sessions were different as expected (9). In our study, the monitor was not removed during exercise sessions, thus the step counts included PA of any intensity during and outside of the exercise program. Considering that changes in both LPA and MVPA time were not statistically different between the two groups, their different changes in number of steps indicate that step counts and PA time do not always agree in terms of statistical inferences. Step counts and PA time thus represent different goals that can be used to prescribe PA that may lead to different results.

Strengths of this study include supervised exercise intervention, objectively determined PA and sedentary time, and long wear time each day and the number of days women wearing the SenseWear device. The following limitations should be considered when interpreting our findings. First, our participants were non-obese, and thus these findings may not be generalized to obese individuals. Second, our data were obtained when women were still participating in the center-based exercise program. Whether there were changes in sedentary time and PA participation after they stopped the exercise program were not determined. Third, there was not a no-exercise control group, which limited our ability to exclude other reasons which may influence the findings. Additionally, it should be noted that these results cannot be directly compared with studies examining changes in PA energy expenditure during exercise interventions due to the various amounts of energy cost associated with the range of PA. A review on this topic highlights the complexity of this issue (4).

In conclusion, sedentary time reduced, time spent on LPA, MVPA, and total PA, and number of daily steps increased during exercise training in this group of older women. The increased MVPA was primarily due to the center-based exercise program. The encouraging findings of this study are that these older women did not reduce time on LPA or MVPA outside of the exercise program, and did not increase sedentary time as a result of participating in the exercise program. However, there were individual differences in these changes. Body weight and cardiorespiratory fitness partly mediated the changes and their baseline levels also moderated the responses to exercise training.