Effects of interrupting daily sedentary behavior on children’s glucose metabolism: A 6-day randomized controlled trial

Abstract

Background:

Metabolic disease risk in youth is influenced by sedentary behaviors. Acute in-lab studies show that, during a single day, interrupting a sedentary period with short bouts of physical activity improves glucometabolic outcomes.

Objective:

To determine if acutely improved glucose metabolism persists after multi-day interruptions of sitting with walking brief bouts. We hypothesized that children who underwent interrupting sitting on multiple days would demonstrate lower insulin area under the curve during an oral glucose tolerance test compared to uninterrupted sitting.

Methods:

Healthy, normoglycemic children (N=109) ages 7-11 years were randomized to one of two conditions: Control (3h of daily Uninterrupted Sitting) or Interrupted Sitting (3-min of moderate-intensity walking every 30min for 3h daily); with dietary intake controlled through provision of foodstuffs for the entire experiment. Participants attended six consecutive daily visits at a research ambulatory unit. The primary outcome was insulin area under the curve during the oral glucose tolerance test on day 6 during interrupted or uninterrupted sitting; secondary outcomes included glucose and c-peptide area under the curve, energy intake at a buffet meal on day 6, and free-living activity.

Results:

Among 93 children (42 uninterrupted sitting, 51 interrupted sitting), daily interrupted sitting resulted in 21% lower insulin (β=0.102 CI:0.032–0.172, p=.005) and a 10% lower C-peptide (β=0.043, CI:0.001-0.084, p=0.045) area under the curve. Matsuda and Glucose Effectiveness Indices were also improved (p’s<0.05).There were no group differences in energy intake or expenditure.

Conclusions:

Sustained behavioral change by interrupting sedentary behaviors is a promising intervention strategy for improving metabolic risk in children.

Introduction

Sedentary behavior (SB), defined as any waking behavior characterized by an energy expenditure ≤1.5 metabolic equivalents, while in a siting, reclining or lying posture,¹ is believed to contribute to metabolic morbidity in adults.^2-6 In youth, SB is associated with metabolic syndrome,⁷ incident type 2 diabetes,⁸ and poor metabolic health.^{9, 10} Despite its detrimental effects, children ages 9-18y spend 6-7 non-sleeping hours/day sedentary.⁷ Thus, investigators and public health advocates have explored the utility of decreasing SB to improve metabolic health.¹¹ Notably, interrupting SB may also be more acceptable and attainable than traditional sustained physical activity goals. Although findings are mixed regarding whether decreasing SB improves weight,^11-13 there is mounting evidence that irrespective of weight/adiposity, decreased SB may provide physiologic benefits for vascular stiffness, blood pressure, and glucose metabolism.^14-17

To date, research has focused on the hypothesis that interrupting SB with brief bouts of activity compared to a session of SB lowers circulating insulin and glucose in both adults and children.^{18, 19} Acute interventions (one-session comparisons) with interrupted SB demonstrate improved glucose metabolic status.^18-22 Although such data are promising, it remains unknown if these acute alterations are transient and wane with continued implementation, or if the improvement may persist.

Thus, we sought to determine if sustained, daily interrupted SB in youth provides persistent improvement in circulating insulin compared to youth with daily SB. We hypothesized that, compared to daily uninterrupted sitting, interrupting sitting with short bouts of walking over 6 days, would result in lower insulin, C-peptide, and glucose incremental area under the curve (AUC) during an oral glucose tolerance test (OGTT).

Methods

Research Design Overview

The National Institutes of Health (NIH) Institutional Review Board approved this trial (NCT03223649). Written parental consent and child assent were obtained prior to procedures. Eligible children were randomized 1:1 for this parallel design trial to either the control condition, executing sedentary behaviors for 3h (the SIT condition) or the active condition (SIT+WALK) during which they were sedentary for the same 3h period except for 3-min moderate intensity walking on a treadmill every 30 minutes (as previously described^{17, 20}), both executed daily for 6 consecutive days. Throughout the study, children were sent home between sessions with food intended to be an individualized, eucaloric diet that supplied all consumption for the experimental period. On the final study day participants completed an OGTT while carrying out their designated intervention. Immediately following the OGTT, participants were given a monitored buffet meal and energy intake was recorded.

Participants

Recruitment and data collection occurred at the NIH Hatfield Clinical Center from July 2017 to March 2020, at which time the protocol ended just beneath randomization target due COVID19 restrictions. Inclusion criteria included: age 7-11.99 years old, a fasting plasma glucose <100 mg/dL, good general health, and a body mass index (BMI) ≥ the 5th percentile, as determined by the CDC age- and sex-specific growth charts.²³ Detailed exclusion criteria included: any signs of significant chronic disease (including increased liver transaminases or hypercholesterolemia >2x the upper limit of normal reference range), any indication of symptoms of diabetes or other endocrine disorders or use of medications that might influence metabolism or cognitive function, precocious puberty, post-menarche for females, a psychiatric disorder that would impede adherence to study, and low cognitive ability as demonstrated by age-adjusted score <85 on the NIH Toolbox Picture Vocabulary Test. Participants were also excluded for food allergies or personal dietary restrictions that would preclude them from consuming the daily provided diet or test meal.

Screening Visit

A physical examination, fasting blood analysis, and 12-lead electrocardiogram were used to assess eligibility. Height was measured in triplicate with a wall-mounted, calibrated stadiometer and weight was measured with an electronic scale. Height and weight were used to calculate BMI (kg/m²). BMI percentiles for sex and age were calculated in accordance with the US Center of Disease Control growth standards.²³

An assessment for pubertal maturation was conducted by a licensed provider via Tanner staging of girls’ breasts and orchidometry for boys’ testicular volumes. A treadmill fitness test via a modified Balke treadmill protocol was done to determine maximal oxygen uptake (VO₂max).²⁴ Ventilatory threshold (VT) achieved was estimated using dual-criteria graphs and the V-slope method.²⁵ Using heart rate and VO₂ at 80% of VT, treadmill speed and grade for the SIT+WALK condition were estimated. Dual-energy x-ray absorptiometry scan (DEXA, Lunar iDXA; GE Healthcare, Madison, WI; analyzed with GE Encore X software) assessed body composition. Children were asked about frequency and duration of physical activity with the International Physical Activity Questionnaire (IPAQ), short self-report form via caregiver proxy.²⁶ Results were used to determine energy requirements for individualized meal calculations and for randomization stratification.

Study Protocol

Participants were randomized using software (randomization.com) that generated randomly permuted blocks with average block size 4 (with the principal investigator generating the complete allocation sequences at study start). Randomization was stratified based on sex (male, female), age (7-9.49 vs. 9.5-11.99y), weight status (BMI<85th percentile vs. ≥85th percentile), and IPAQ level (low, moderate, and high + very high usual physical activity) to promote similar subject characteristics in the two groups. Participants learned of their treatment condition at their first test visit and study team members assisting with study visits were not blinded to treatment condition. The laboratory personnel measuring primary and secondary study outcomes were, however, unaware of group assignments.

Individualized Diet:

Beginning on Test Day 1, all participants were given an individualized diet based on their individual energy needs as estimated by the sex- and age-specific equations for energy requirements of children and adolescents available from the Institute of Medicine of the National Academies (pages 82-84).²⁷ The energy content of the meals and snacks was calculated based on each individual’s total body weight and height, adjusting for weekly physical activity ascertained using the IPAQ. The individualized macronutrient-controlled diet was prepared by the NIH dietary service and distributed daily such that at the conclusion of each daily lab session, each participant consumed an afternoon snack and was sent home with 3 packed meals and one snack (typically dinner, breakfast, mid-morning snack and lunch) to be consumed as their sole food intake until they returned for their next daily clinic visit. Meals consisted of the following proportions for daily total energy intake: breakfast 20%, lunch & dinner 30% each, snacks 20% of total daily requirements. Daily menus provided approximately 15% of energy from protein, 32% energy from fat and 53% energy from carbohydrate. This cycle continued on days 2-5 until they returned for their final fasting visit on day 6. Nutrition staff verified food intake via weigh-back monitoring. Participants were also asked to report any additional food (not provided within the supplied meals) that they mistakenly consumed, and nutrition staff calculated non-compliance when necessary.

Standardized Monitoring:

At the start of Test Day 1, participants were given a wrist accelerometer (Actigraph GT3X+) to wear on the nondominant wrist throughout their 6-day intervention to assess in-study as well as free-living activity (providing insights into potential activity compensation). Furthermore, waist-worn accelerometry was obtained on Test Day 6 for additional protocol fidelity monitoring.

Randomized Condition Execution:

Those assigned to SIT executed 3h of sedentary (seated or recumbent) activity of their choice under the supervision of the study team and confirmed with physical activity monitoring (described above). They could rise only to use the bathroom. SIT+WALK participants were prompted by the study team to walk on the test treadmill for 3-min every 30 minutes for 3h, for a total of 6 walking bouts (18 minutes total) with treadmill settings individualized for each subject to reach 80% VT. The maximum heart rate achieved during each walking bout was recorded.

Throughout both conditions, the participants were provided their metabolic diet snack per their supplied meal plan (described above) to consume during their 3h sessions (remaining food stuffs were weighed back as described but this rarely occurred). On the morning of Test Day 6, participants arrived by 7:00 AM in a fasted state. In addition to engaging in their randomized condition, participants underwent a 3h OGTT (1.75 g/kg dextrose; maximum dose 75g). With an intravenous line, fasting blood samples were collected at −5 and 0 minutes, after which participants consumed the dextrose solution. Blood samples were then collected at 30, 60, 90, 120, 150, and 180 minutes. For the SIT+WALK condition, blood was drawn immediately after each 3-min walk. Participants consumed nothing but small amounts of water during the OGTT.

After the conclusion of the 3h OGTT, a standardized multiple-item buffet test meal (approximately 10,000 kcal) was given in a confined room without interruptions or audience, and for a duration of the participant’s preference. The meal had an array of macronutrients (~12% protein, 51% carbohydrate, and 37% fat) from foods frequently consumed by children.²⁸ Participants were instructed to ‘eat as much as you want’ and the amounts of foods and beverages consumed were calculated by measuring the differences in weights (g) of each item before and after the meal. Caloric and macronutrient consumption was calculated using the USDA National Nutrient Database for Standard Reference (Agricultural Research Service) and nutrient information from the manufacturers.

Assays

Serum was collected at each timepoint for glucose, insulin, C-peptide, cortisol, free fatty acids (FFA), and triglycerides. All samples were assayed by the NIH Clinical Center Department of Laboratory Medicine. Insulin and C-peptide were measured via electrochemiluminescence immunoassay with Roche Diagnostics reagents on a Roche Cobas 6000 instrument (model e601; Roche Diagnostics, Indianapolis, IN). Plasma glucose was collected and kept on ice in tubes with powdered sodium fluoride until centrifugation and was determined using a Roche/Hitachi instrument (model c502; Roche Diagnostics, Indianapolis, IN). For insulin, the analytical sensitivity was 0.2μU/mL, cross-reactivity with proinsulin was 0.05%, average intraassay coefficient of variation (CV) was 1.2%, and average inter-assay CV was 6%. For C-peptide, the analytical sensitivity was 0.01 ng/mL, cross-reactivity with proinsulin was 32.5%, and average intra- and inter-assay CV was 1.3% and 2.8%, respectively. Total cholesterol and triglycerides were also analyzed using the Roche/Hitachi c502 instrument. Cortisol was measured on a Immulite 2000 XPi (Siemens Medical Solutions USA, Inc, Tarrytown, NY) via chemiluminescent enzyme immunoassay. FFA were measured on a Roche/ Hitachi Cobas 6000 instrument (model c501; Roche Diagnostics, Indianapolis, IN) using the Wako enzymatic method. The analytical sensitivity was 0.01mEq/L and average intra- and inter-assay CV were 9.8% and 6.8%, respectively.

Statistical Analyses:

Statistical analyses for descriptive characteristics, biomarker AUCs, and energy intake comparisons were performed using SPSS Statistics 28 (IBM Corporation, Armonk, NY). The power calculation used to determine sample size was based on previous data evaluating the efficacy of acute interrupted sedentary behavior in decreasing insulin AUC during OGTT.^{17, 20} Previous data indicated that within this age group interrupted behavior produced a moderate to large (Cohen’s d=0.24-.8) effect size. Thus, for an independent sample comparison with an 80% power, α=0.05, our sample size required evaluable data from 84 subjects. We aimed to randomize 120 subjects to account for withdrawal and incomplete data. Furthermore, given that the outcomes were measured only at one timepoint (i.e. no baseline) and responses to condition can be assumed to be missing at random, complete-case analyses were employed in place of intention to treat analyses.²⁹

The primary outcome of interest was insulin positive AUC (calculated via trapezoidal method with GraphPad Prism v7) during the OGTT. Secondary outcomes included glucose, C-peptide, cortisol, FFA and triglyceride AUCs. Univariate analysis of variance (ANOVA), controlling for percent fat (arcsine square root transformed) and pubertal stage (pre, early, or late), were used to compare the OGTT AUCs, Matsuda Index,³⁰ Insulinogenic Index,³¹ Oral Disposition Index,^{32, 33} and Glucose Effectiveness Index³⁴ as previously described,¹⁷ and each OGTT time point (time 0, 30, 60, 90, 120, 150, 180 minutes) of each log-transformed analyte by condition (insulin, glucose, C-peptide, FFA, cortisol & triglycerides), controlling for basal analyte as appropriate. A priori covariates of age, sex, race, and ethnicity were evaluated and removed from the final models for negligible contribution (p>0.1). ANCOVA was also used to compare total buffet energy intake (kcal, log transformed) and adjusted for age (y), lean mass (kg), fat mass (kg), and macronutrient composition percentage (adjusted for age). Unadjusted independent sample t-tests compared mean in-session & free-living accelerometry and heart rate variables between conditions. All reported activity counts and step counts were derived from the wrist device using the Actilife software (version 6.12.0), with count data reported as the one-minute vector magnitude of the three gravitational axes and all device wear was included.

Secondary, mixed model analyses were performed using SAS v9.4 (SAS Institute, Cary, NC) to assess the effect of experimental condition on changes in all analytes, log transformed. Variance components were estimated using the restricted maximum likelihood estimate methods to control for within-subject correlations and an unstructured covariance structure. A priori time-invariant model covariates were baseline serum analyte, age, sex, percent fat (arcsine square root transformed), and pubertal stage. Covariates were included in final models if p<0.1 and all models included baseline serum value. Models with cortisol included OGTT start time as a covariate. Time-variant variables were time (minutes) and serum values.

Post hoc analysis were mixed models with a time*condition*weight group interaction and an ANOVA to compare the change in HOMA-IR between screening and end of intervention by condition adjusting for pubertal stage and percent fat (arcsine square root transformed). Data are reported as means SD unless otherwise noted, all AUC’s have been calculated as incremental AUC, and all CI’s 95%.

Results

After screening 129 potential participants, 109 were randomized and 93 participants contributed complete data for final analyses (Figure 1 Consort Diagram). The most common reasons for drop out were children changing their mind about consent for an OGTT with an IV in place or having insufficient IV access at time of OGTT. There were no significant demographic or anthropometric differences between those who were ineligible or withdrew prior to randomization and those who were randomized. Participants returned on average 41±22 (Mean ± SD) days after screening to begin the intervention. The children who completed the intervention were 10.2±1.3y, 48% female; 55% did not have overweight or obesity, 21% had overweight (BMI 85^th−<95^th percentile for age), and an additional 24% had obesity (BMI ≥95^th percentile for age) according to US standards. Mean fasting glucose was 87.8±6.3 mg/dL (Table 1). None of the sociodemographic or laboratory variables, including insulin resistance, free-living physical activity and peak VO₂, differed between the randomized groups.

Figure 1. CONSORT Flow Diagram.

Among 109 children randomized to uninterrupted sitting for 3h per day (SIT) or sitting interrupted with 3min of moderate intensity walking every 30min ((SIT+WALK), 93 had evaluable data at the end of the study. VT: Ventilatory threshold

Table 1.

Participant Characteristics

Variable		SIT Group	SIT+WALK Group	All Subjects
N		42	51	93
Sex, N (%)	Female	18 (43)	27 (53)	45 (48)
Race, N (%)	White	27 (64)	38 (74)	65 (70)
	Asian	3 (7)	2 (4)	5 (5)
	Black or African American	9 (21)	5 (10)	14 (15)
	Multiple Races	2 (5)	3 (6)	5 (5)
	Unknown/Declined/Other	1 (2)	3 (6)	4 (4)
	Pacific Islander	0 (0)	0 (0)	0 (0)
Ethnicity, N (%)	Latino/Hispanic	6 (14)	8 (16)	14 (15)
	Non-Latino/Hispanic	36 (86)	43 (84)	79 (85)
Age (y) ± SD		10.4± 1.3	10.0± 1.2	10.2±1.3
Pubertal Maturation (Median, range)	Male, testis volume (mL)	Median 3 (2-25)	Median 3 (2-10)	Median 3 (2-25)
	Female, Tanner Breast Stage	Median 1 (1-3)	Median 1 (1-3)	Median 1 (1-3)
*Height (cm) ± SD		146.0±9.8	141.2±8.5	143.4±9.4
Weight (kg) ± SD		46.4±19.6	40.3±11.4	43.1±15.8
Fat Mass (kg) ± SD		16.2±12.4	13.5±6.9	14.7±9.8
*Lean Mass (kg) ± SD		28.9±7.9	25.5±4.8	27.0±6.6
BMI (kg/m2) ± SD		21.2±6.6	19.9±3.9	20.5±5.3
BMI z-score± SD		0.8±1.2	0.8±0.9	0.8±1.0
Weight Status, n (%)	Not Overweight	23 (55)	28 (55)	51 (55)
	Overweight + Obesity	19 (45)	23 (45)	42 (45)
Fasting glucose (mmol/L) ± SD		4.82±0.378	4.83±0.30	4.39±0.35
Fasting insulin (pmol/L) ± SD		76.6±57.3	69.6±42.3	74.6±53.1
HOMA-IR		3.4±3.2	2.4±1.7	2.8±2.5
Peak VO2 (ml/kg/min)		38.3±9.1	38.5±7.4	38.4±8.2
†, ‡IPAQ Score		2 (1-4)	2 (1-4)	2 (1-4)

^*Difference between SIT & SIT+WALK group, p=0.01. Otherwise, no statistical differences between groups.

^†International Physical Activity Questionnaire Score 1=Sedentary, 2=Low Active, 3=Active, 4=Very Active.

^‡Median (range)

There were no significant adverse events attributable to the study. Overall, the provided diet was well tolerated, with participants consuming 92% of the provided diet. During the cumulative 3-min walking intervals, participants randomized to SIT+WALK demonstrated greater wrist-obtained acceleration and step count, and peak heart rate, compared to SIT; Table 2 & Supplemental Figure 1.

Table 2.

Summary of Outcomes

Variable	SIT	SIT + WALK	P Value
Wrist Accelerometry & Heart Rate Outcomes
Wrist wear time outside intervention sessions (h/day)	19.6±1.2	19.0 ±1.9	0.07
Free-living activity outside of intervention sessions (3D counts/day x 10−6)	2.1±0.6	2.2±0.5	0.63
Free-living step count outside intervention sessions (steps/min)	6.6±1.5	6.7±1.2	0.70
Mean activity during 3h intervention sessions (3D counts/min)	1576±475	2142±568	<0.001
Mean step count during 3h intervention sessions (steps/min)	5.5±1.8	11.4±2.8	<0.001
Mean heart rate during 3h intervention sessions (bpm)	89±13	126±15	<0.001
Metabolic Outcomes from OGTT *
Insulin AUC (pmol/L·min)	94520.3±64962.8	73275.9±46791	0.005
C-Peptide AUC (nmol/L·min)	426.4±148.4	382.8±148.1	0.045
Glucose AUC (mmol/L·min)	1223.6±161.0	1192.9±136.1	0.358
Free fatty acids AUC (mmol/L·min)	39.7±9.8	38.5±13.1	0.226
Cortisol AUC (nmol/L·min)	34282±7151.2	36685.9±7797.0	0.223
Triglycerides AUC (mmol/L·min)	114.8±72.4	114.6±63.0	0.148
Change in HOMA-IR	1.03±2.3	0.33±1.2	0.055
Calculated Metabolic Indices *
Matsuda Index	4.84±3.6	5.74±3.67	0.033
Insulinogenic Index	2.08±1.97	1.54±1.21	0.056
Oral Disposition Index	7.34±5.41	6.39±3.07	0.775
Glucose Effectiveness Index	5.76±1.00	6.09±0.63	0.011
Buffet Energy Intake after OGTT †
Total energy intake (kcal)	1312.4±478.4	1207.4±443.8	0.648
Carbohydrate intake (%)	52.0±9.0	51.8±8.4	0.550
Fat intake (%)	35.8±7.0	36.2±6.0	0.539
Protein intake (%)	12.2±3.6	12.0±3.2	0.741

Unadjusted means and SD; Accelerometry analyses (wrist worn Actigraph), n=90, heart rate analyses n=64

^*Metabolic outcomes analyses, n=88-93, compare log-transformed analytes adjusted for: fat mass percentage (arcsine sqrt transformed), pubertal stage (pre, early, or late), and basal analyte log transformed. HOMA-IR comparison analysis: change in HOMA-IR from baseline to OGTT controlled for pubertal stage and percentage fat mass (arcsine sqrt transformed).

^†Buffet analyses, n=89, compare log transformed Kcal adjusted for: age (y), lean mass (kg), fat mass (kg) & macronutrient % analyses adjusted for: age (y)

OGTT Analyses

Figure 2 and Table 2 present the summary of results. For the primary outcome, OGTT insulin AUC was 21% lower in the SIT+WALK vs. SIT condition. After controlling for fat mass, puberty and basal insulin concentrations, the adjusted ANCOVA showed a significant difference in insulin AUC between conditions (β=0.102, CI:0.032-0.172, p=0.005), which yielded a Cohen’s d coefficient of 0.38 (moderate effect). Removing participants with advanced puberty status (testicular volume >15ml and Tanner stage IV breast development) had no impact on the primary outcome of insulin AUC. Mixed model results found no significant main effects for condition on insulin in unadjusted or adjusted models, nor was the condition*time interaction significant (Supplemental Tables 1 & 2. However, mean insulin concentrations were lower at 90 minutes in the SIT+WALK condition (p=0.007; Figure 2B).

Figure 2. Oral glucose tolerance test results (insulin, C-peptide, glucose) by condition.

The effect of sitting interrupted with 3min of moderate-intensity walking every 30min (SIT+WALK; white bar/dotted line) vs. uninterrupted sitting (SIT; black bar/solid line) on hormones and analytes: (A), 3h serum insulin AUC (B) Insulin linear trend, (C), 3h serum C-peptide AUC (D), c-peptide linear trend, (E), 3h plasma glucose AUC and (F) glucose linear trend. Unadjusted mean & SE are shown in A, C, E.

AUC p-values are results of ANOVA comparing log-transformed analytes adjusted for: fat mass (% percentage arcsine sqrt transformed), pubertal stage (pre, early, or late), and basal analyte log transformed.

Linear trend p-values are results of mixed models for effect of time*condition adjusted for fat mass (% percentage arcsine sqrt transformed), pubertal stage (pre, early, or late), and basal analyte log transformed. Error bars reflect SE

C-peptide analyses demonstrated an adjusted 10% lower AUC in the SIT+WALK vs. SIT condition, which was significant in the adjusted ANCOVA model (β=0.043, CI:0.001-0.084, p=0.045). Mixed model results again showed no significant main effect or condition*time interaction on C-peptide concentrations over 3h (Supplemental Table 3). However, mean C-peptide concentrations were lower at 90 minutes in the SIT+WALK condition (p=0.042; Figure 2D).

Glucose analyses demonstrated no significant differences in AUC between conditions (p=0.358, Table 2), nor individual time points within the OGTT or the linear trend throughout the time course (Figure 2, Supplemental Table 3). Similarly, there were no differences in the exploratory outcomes of FFA, cortisol or triglycerides AUC between condition (Supplemental Figure 2). Mixed model results showed no significant main effect or condition*time interaction on these outcomes over 3h (Supplemental Table 3).

There were no significant interactions between weight status (non-overweight vs. overweight/obesity) and condition for the AUC outcomes (all p> 0.05, data not shown). There were no significant differences in the time course of analyte concentrations by condition between weight status groups, with the exception of FFA, where there was a marginally significant condition*time*weight status interaction (p=0.055, CI:0.37-2.10; Supplemental Table 4).

Calculated Indices

Insulin sensitivity during the glucose tolerance test, as estimated by the Matsuda Index, was greater during the SIT+WALK compared to SIT condition (5.74±3.67 vs 4.84±3.6, p=0.033). The Insulinogenic Index showed a trend towards a reduction in insulinemia during the SIT+WALK condition (1.54±1.21vs. 2.08±1.97, p=0.056). The Glucose Effectiveness Index was significantly greater for the SIT+WALK compared to SIT condition (6.09±0.63 vs. 5.76±1.00, p=0.011) and there was no detected difference in calculated Oral Disposition Index between conditions (Table 2; p=0775).

Free-living physical activity analyses and energy intake

Ninety seven percent of subjects provided free living accelerometry data for analysis; n=3 were not included due to device malfunction or procedural errors. Participants wore the wrist Actigraph on average 97.0±4.8% of the time with a range of 18.6–24.0 hour/day for SIT group and 94.6±7.8% of the time with a range of 14.8–24.0hour/day for the SIT+WALK group. Free-living accelerometry metrics measured outside the 3h monitored research sessions were not significantly different between conditions (Table 2). Additionally, there were no significant differences in total energy intake or proportion of macronutrient intake from the buffet meal by condition (Table 2)

Fasting Insulin Resistance (HOMA-IR)

The change in fasting HOMA-IR between participants’ screening visits and OGTT Test Days suggested a trend towards a greater worsening in the SIT group (+1.03±3.2) than in SIT+WALK group (+0.33±1.18; ANOVA β=.716 CI: −0.17-1.45, p=0.055). Average time between screening visit and Test Day 6 was 40.8±20.8 days inclusive of the 6-day intervention.

Discussion

Children had lower insulin concentrations and greater insulin sensitivity during an OGTT when they engaged in 3-min of moderate intensity walking every 30min for 3h daily for 6 consecutive days compared to children who remained sedentary for 3hrs per day for 6 consecutive days. This finding builds upon previous single-day data that demonstrated acute insulin improvements as a result of interrupted sitting in both adults^{18, 35} and children^{20, 36} and to our knowledge is the first to confirm there is no adaptation that reduces the metabolic benefit of interrupting sedentary activity in youth. Similar to our previous data evaluating impact on glucose in children who had overweight or obesity in a single-day experiment,¹⁷ the current study identified no comparative improvement in glucose as a result of interrupted sitting. Additionally, the data notably demonstrated no adverse compensatory effects on energy intake or free-living physical activity due to intervention.

Although circulating catecholamine concentrations were not measured in the present study, we previously reported¹⁷ that norepinephrine (but not epinephrine or dopamine) increased slightly but significantly during a sedentary breaks paradigm equivalent to that employed herein. The effects of catecholamine excess on glucose homeostasis are complex and include an inhibition of glucose stimulated insulin secretion primarily via pancreatic islet cell α₂ receptors,^{37, 38} suggesting the possibility that a small decrement in insulin secretion might result from the SIT+WALK intervention even though insulin output is simultaneously stimulated via less numerous pancreatic islet cell β₂ adrenergic receptors.^39-42 At the same time, catechols also enhance rates of glucose release (both from glycogenolysis and gluconeogenesis). Consequently, hyperglycemia is the hallmark of this metabolic response and any increases in catecholamines sufficient to suppress insulin production would be expected to have the effect of increasing glycemia during the intervention. As there were nonsignificant decreases in glycemia seen during the study, we do not believe that subtle hypoinsulinemia due to catecholamine secretion explains the observed findings.

The demonstrated impact on insulin demand supports the concept that habitually interrupted SB continues to show measurable, improvement in glucose homeostasis without evidence of adaptation that reduces the benefit. Compared to those in the SIT group, children in the SIT+WALK group had a 21% reduction in prandial insulin requirement, which is similar to our previous acute one-day studies.^{17, 20} Although additional data are ultimately needed, these findings suggest such interventions could be clinically meaningful, given that diabetes develops when the pancreatic demand surpasses an individual specific threshold to maintain euglycemia. Thus, if an intervention can keep the pancreatic demand below an individual’s threshold, diabetes may not develop, or its onset may be delayed. In the context of type 2 diabetes, the pancreatic insulin demand is increased with factors such as diet, activity behaviors and insulin resistance.⁴³ Therefore, interrupted sitting may potentially meaningfully ameliorate the pancreatic demand in this scenario.

Despite the impact on insulin, there was no evidence of benefit in circulating glucose, which is contrary to many acute interventions in adults,^{18, 20, 35, 44} and one acute intervention in adolescents,²² but similar to our previous data in young children.¹⁷ In comparing the known adult data, we hypothesize that the youths studied are relatively healthier and thus simply have better glucose regulation than the adult participants evaluated. Indeed, the current study included only children with healthy fasting glucose, which limited the ability to observe any lowering in glucose from baseline. Future studies are needed to evaluate the impact of such interventions on glucose in children who have an impaired glucose metabolism. Additionally, contrary to the glucose findings reported here, Fletcher et al. demonstrated notable reduction in glucose AUC when adolescents executed interrupted activity with resistance-type activity.²² This indicates that perhaps anaerobic, resistance-type activity which engages large muscle groups may be more beneficial for acute glucose lowering; future studies are needed to determine these relationships.

We also attempted to examine the mechanisms underlying the observed improvements in insulinemia and insulin sensitivity during the SIT+WALK test by calculating a measure of glucose effectiveness that has been validated in youth with overweight and obesity.³⁴ In this cohort, we saw an improvement in glucose effectiveness from the SIT+WALK intervention. Muscle contraction increases glucose effectiveness, and it is also likely that muscle contraction enhances muscle insulin sensitivity, both of which lead to decreased insulin demand.^45-49 Thus, interrupted activity may improve glucose homeostasis through increased glucose effectiveness and/or increased insulin sensitivity associated with muscle contraction.

The Oral Disposition Index did not differ between the groups. Disposition indices do not tend to change too much among individuals until they develop impaired glucose homeostasis,³³ and it is unsurprising that this metric showed no significant difference between groups. Weiss et al³⁴ found no significant differences in oral disposition index among children who were lean and those who had overweight, moderate obesity, or severe obesity.

Finally, post-hoc analyses sought to determine if measurable change in fasting insulin resistance (before any activity) was detected. The data here demonstrated only a nonsignificant trend towards an effect on change of HOMA-IR by condition, which suggests that there is little immediate effect from SB interruption on insulin resistance. Longer duration studies are warranted to assess SB interruptions given that SB in youth is associated with insulin resistance and exercise improves HOMA-IR.^{7, 50, 51}

An important finding in this study is that 6 continuous days of interrupting SB did not significantly impact energy balance for either intake or free-living expenditure compared to 6 days sedentary. Thus, there is potential added benefit of small but cumulative energy deficits with short activity breaks. The summative ~70kcal/day of energy expenditure with the 18 minutes of moderate-vigorous walking (per mean participant weight)⁵² without a compensating increase in energy intake calculates to an approximate 420kcal deficit over 6 days. In terms of free-living physical activity, children who were randomized to SIT+WALK did not significantly decrease their overall activity in compensation and thus overall maintained a favorable energy balance despite intervention. Therefore, this strategy may be effective over a longer duration for both improving metabolic outcomes as well as improving energy balance.

Strengths & Limitations

Strengths of the study include a rigorous randomized design accounting for many potential confounders, including most notably, fixed dietary intake during study participation, monitored physical activity, DXA-obtained body composition, and thorough assessment of pubertal development, which is significantly related to insulin sensitivity in children. Furthermore, participants represented diverse weight strata, increasing generalizability, and participant retention was good, with 85% providing complete data. There are additionally limitations to consider. First, due to practical limitations, a baseline OGTT was not obtained, obviating the ability to observe pre-post changes. Thus, the possibility of an effect on glucose homeostasis due to the cumulative effect of repeated differences in interruption of sedentary activity could not be studied. Rather, these result show only that there is no metabolic adaptation abrogating the acute impact of interrupting sedentary time with short activity bouts. Another potential limitation is that the study was designed as a comparison between independent randomized groups of children rather than using a cross-over design; thus, individual differences in participants might not necessarily be counterbalanced by group assignment. Indeed, there were nonsignificantly greater weight, BMI, and fat mass in the SIT group and some participants in each group had (non-pathologically) advanced pubertal development, even though the median breast stage for girls and testis volume for boys were not significantly different and removing participants with outlying advanced puberty status had no impact on the primary outcome of insulin AUC. Additionally, although foodstuffs were provided to children with instructions to only eat the provided foods, without 24-hour direct monitoring, we cannot be certain of study diet adherence in our participants. Similarly, although we monitored physical activity outside of study sessions with accelerometry, we cannot completely account for energy expenditure because there was some non-wear time and device malfunctions (although this was minimal). Since this study was limited to a 6-day intervention, it could not determine what impact longer interventions might have on glucose tolerance or body composition. Whether 3-minute bouts of activity every half-hour would be feasible to implement in a classroom or after-school setting has also not been studied in this protocol and is another limitation. Finally, participant eligibility was limited to children who had relatively healthy glucose metabolism which precludes generalizability to those with impaired glucose homeostasis.

Conclusion

Interrupting SB with short bouts of moderate activity during 3h sessions over 6 consecutive days did not induce any evidence of metabolic compensation, such that the benefits found in single-day cross-over studies of glucose metabolism in metabolically healthy youth continued to be detectable after a 6-day intervention. These data support the paradigm of interrupted SB as a plausible intervention to maintain healthy glucose metabolism in youth, including youth with excess weight who may be at risk for impaired glucose metabolism or type 2 diabetes.