Does frequency or duration of standing breaks drive changes in glycemic response? A randomized crossover trial

Intervention strategies to break up sitting have mostly focused on the modality (i.e., comparing different intensities and/or type of activities) and less on how frequency and duration of breaks affect health outcomes. This study compared the efficacy of different strategies to break up sitting time [i.e., high frequency, low duration standing breaks (HFLD) and low frequency, high duration standing breaks (LFHD)] in reducing postprandial glucose. Eleven sedentary and prediabetic adults (mean ± SD age = 46.8 ± 10.6 years; 73% female) participated in a cross‐over trial. There were six blocks that represented all potential combinations (ordering) of the study conditions and participants were randomly assigned to a block. Each participant underwent three 7.5‐h laboratory visits (1 week apart) where they engaged in either continuous sitting, HFLD, or LFHD condition while performing their usual office‐related tasks. Standardized breakfast and lunch meals were provided. Postprandial mean glucose, area under the curve (AUC), and incremental area under the curve (iAUC) were evaluated using mixed models. Compared with LFHD condition, the HFLD standing breaks condition significantly lowered mean glucose by −9.94 (−14.13, −5.74) mg/dL·h after lunch, and by −6.23 (−9.93, −2.52) mg/dL·h, for the total lab visit time. Overall, the results favor frequently interrupting sitting with standing breaks to improve glycemic control in individuals with prediabetes. Further studies are needed with larger sample sizes to confirm the results.

associated with plasma glucose levels 3,[8][9][10] and that different patterns of accumulating sedentary time can have a differential effect on glucose metabolism. For example, a study that examined the relationship between sedentary accumulation patterns and various cardiometabolic biomarkers of the Australian Diabetes, Obesity, and Lifestyle study (AusDiab), a national population-based cohort study, 11 showed that frequently interrupted sitting (compared to patterns with relatively more prolonged sitting) were beneficially associated with fasting plasma glucose and 2-h post-load glucose. 9 Furthermore, results from experimental studies also suggest that reducing prolonged sitting time (sitting continuously for more than 30 mins) 8 using various modalities (i.e., with light activity breaks, moderate-vigorous activity breaks, or resistance exercise) can significantly attenuate postprandial glucose compared with continuous sitting. [12][13][14][15][16] Interestingly, results from these studies also shows that the benefits of breaking up sedentary time on postprandial glucose through light physical activity (LPA) breaks and moderatevigorous physical activity (MVPA) breaks were not statistically different from one another, suggesting that intensity of breaks may not be important. 17,18 Although these two studies indicated that standing breaks did not affect postprandial glucose, 17,18 other studies have shown that breaking up sitting time through standing reduced glucose iAUC by 5%-30% over the duration of the studies, but only in prediabetic individuals, a more at-risk group. 13,19,20 All of these suggest that prolonged sedentary time and potentially the pattern at which it was accrued may play an essential role in an individual's risk of developing type 2 diabetes, especially in those classified as prediabetic. Considering that prediabetic individuals have an even higher risk for type 2 diabetes, developing simple yet effective strategies to improve glycemic outcomes in this population could significantly improve health in this population.
The effects of different combinations of standing bouts and standing frequency for interrupting sitting time have not been fully explored. Most experimental studies that have evaluated the effects of interrupting sitting time have used different modes of physical activity (e.g., LPA, MVPA, resistance exercises, or chair squats), making it difficult to disentangle the benefits of interrupting sitting time from the benefits of being physically active. A study in 2016 that explored this specific question showed that in a sample of normoglycemic individuals, energy expenditure from short intermittent standing was 296 ± 78 kJ (9.0% ± 2.3%) higher than prolonged standing breaks over 8 h despite standing and sitting time being similar across the conditions. 14 However, the two conditions did not differ in glucose, insulin, or triglyceride concentrations in this sample of normoglycemic individuals. A recent study showed that breaking prolonged sitting by short but frequent bouts of sit-stand transitions (i.e., 10x sit-stand transitions every 20 min) resulted in significant reductions in postprandial insulin levels. 21 These two studies suggest that the frequency and duration of interruptions in prolonged sitting can impact glucose metabolism. Evaluating the optimal combination of frequency and bouts of standing breaks can directly inform interventions that utilize sit-stand workstations, a common intervention to reduce workplace sitting. 22 Thus, this study aimed to determine if interrupting sitting behaviors can improve blood glucose levels by evaluating the difference in acute postprandial glucose response between the continuous sitting condition and two conditions that implemented intermittent standing regimens. Furthermore, the study aimed to assess whether the interruptions' patterning (i.e., frequency vs. duration) drives this effect by evaluating the difference in the acute postprandial glucose response between conditions where participants engaged in high frequency, low duration standing breaks vs. low frequency, high duration standing breaks.

| Study participants and recruitment
Sedentary office employees with prediabetes were recruited. Inclusion criteria were: (i) ages 35-65 years, (ii) sedentary work habits (at least 6 h of work-related sitting), (iii) prediabetic (fasting glucose level of 100-125 mg/dL determined at a screening visit), (iv) willing to engage in three 7.5-hour laboratory visits, (v) willing to wear the activPAL and continuous glucose monitor (CGM), (vii) current sit-stand workstation owner, and (viii) BMI 25-45 kg/m 2 . Participants were excluded due to the presence of: (i) chronic mobility limitations, (ii) psychiatric disorders, (iii) cardiometabolic abnormality (i.e., diagnosed with type 1 or 2 diabetes, hypertension, or dyslipidemia), (iv) food allergy/restriction, or (v) being pregnant. All interested participants were initially screened for the above inclusion and exclusion criteria using an online self-report questionnaire. Participants that met the eligibility criteria were invited for a 30-min screening visit to determine their fasting blood glucose concentration. During this visit, they were also asked to wear a physical activity monitor (activPAL device) for seven consecutive days to confirm their sedentary working habits. The institutional review board approved all study procedures, and written consent was obtained prior to participation. The study protocol is registered on Clini caltr ials.gov (NCT04144920).

| Study design
This crossover randomized trial included three conditions: (i) continuous sitting (CS), (ii) high frequency and low duration (HFLD) standing breaks, and (iii) low frequency and high duration (LFHD) standing breaks. Total sitting and standing time was equal in both HFLD and LFHD conditions (see Table 1). The only difference between the two conditions was the pattern by which sitting time was interrupted. A common threshold for the prolonged sitting time in epidemiological studies of sedentary behavior is 30 min of continuous sitting. 2,6,23,24 The HFLD condition used half of this threshold, where participants interrupted their sitting time every 15 min using a 2.5-min standing break. In contrast, participants performed twice this threshold in the LFHD standing breaks where they completed a 10-min standing break every hour of sitting.
Six potential sequences for three study conditions were presented to the participants, with each sequence referred to as a study block (a total of six study blocks). Each eligible participant was randomly assigned to a block that determined the sequence that they would perform the conditions. Conditions were randomized using a computergenerated random number prepared by another researcher not directly involved in the project. Each participant was blinded to the assigned condition during the visit until after their first standardized meal. Because of the nature of this design, participants can potentially infer the condition they would engage in on their last lab visit as soon as they are informed of the condition for their second visit. However, this effect was captured and controlled for in statistical analysis.

| Study protocol
After consent was acquired, participants were scheduled to come in for their first laboratory visit. All three lab visits were 1 week apart and scheduled on the same day of the week. Before each visit, participants were instructed to fast overnight (at least 10 h) and avoid any MVPA for at least 2 days and avoid consuming alcoholic beverages for at least 3 days prior to each visit. A day before the first visit, participants were invited for a 30-min laboratory visit to insert a continuous glucose monitor (CGM) device and attach a new activPAL accelerometer. This is because the CGM sensor needs to be attached for at least 12 h to ensure accurate glucose reading. Participants were instructed to come in at approximately 7:00 am for all their visits. As part of the visit check-in process, participants were asked if they engaged in any structured MVPA in the last 2 days and if they consumed any alcoholic beverages in the previous 3 days. During each visit, investigators took an initial CGM reading to ensure that the CGM sensor was collecting glucose data.
A standardized breakfast meal was then provided with instructions to consume the meal within a 15-min period. All meals were consumed in the same room as the trials to avoid participant activity. Following breakfast, participants were asked to perform their usual desk-based work activities in a private room with a sit-stand workstation and a desktop computer. A study smartphone that displayed the time and posture (i.e., sit or stand) that participants were supposed to engage in was set up right below the computer monitor, which was clearly visible to the participants. Investigators set sound cues 30 s before and when participants needed to change their postures to alert them whenever they were supposed to change their postures. Participants were instructed to avoid any light movement (e.g., swaying, fidgeting, or squatting) and were prompted when to sit or stand-up depending on their study condition on that day using smartphone prompts. Investigators provided a 15-min break to the participants after the 210-min mark, where they consumed their standardized lunch meal. They were also allowed to use the restroom during this period.

| Outcome measures
Continuous Glucose Monitor. The Freestyle Libre Pro (Abbott Laboratories, Chicago, IL) was used in the study. The circular sensor (35 mm in diameter and weighs 5 gm) is designed to be worn continuously for 14 days and is waterproof, lightweight, and minimally obtrusive. It was attached to the lower back brachia of the participant's non-dominant side following the sensor's instruction manual. The sensor is programmed to measure interstitial glucose at 15-min intervals. At the end of the third visit, data from each sensor were acquired using the Libre Pro reader and uploaded to an online patient repository (LibreView). Data were then processed, and 15-minute epoch data were downloaded into a local secure computer drive and used to derive the study outcomes. Continuous glucose data corresponding to each visit date and time were isolated and inspected for quality. The total area under the curve (AUC) for postprandial glucose was calculated using the trapezoidal rule (see equation below). 25 Where S i represents the area of the trapezoid formed for each epoch of time (Δt= 15 min), G i is the glucose measure for a particular moment, and G base is the baseline glucose. For total AUC calculation, G base is equal to zero. In addition, the mean glucose level in the 30 minutes prior their first meal during each visit was used as the baseline at which incremental area under the curve (iAUC) was calculated (both for postprandial breakfast and lunch period). For iAUC, only periods with glucose levels above the baseline were included in the calculations. The glucose AUC is a common index of glucose excursion after glucose loading. It has been widely used for calculating the glycemic index and for evaluating the efficacy of medications for postprandial hyperglycemia. Finally, the mean amplitude of glycemic excursions (MAGE), a measure of glycemic variability, was calculated as the mean of glucose values exceeding one SD from the mean glucose level for the entire visit period. 26 activPAL Device. Objective measures of sitting, standing, and moving time were derived from the activPAL micro accelerometer worn on the midline of the right thigh. Collected data during the visits were processed into events of sitting, standing, or moving (i.e., stepping) using the activPAL software version 7.2.32 (PAL Technologies Ltd). Data specific to each visit were isolated to correspond with the glucose data. All observations measured by the activPAL as lying/seated were considered sedentary. The remaining observations were then classified as either standing or moving events.
Standard Meals. Meals (breakfast and lunch) were provided during each lab visit to control for any dietary influence. Each meal was standardized to provide 33% of the participant's total daily caloric needs based on their basal metabolic rate following a typical American diet (55% carbohydrate, 30% fat, and 15% protein). Basal metabolic rate was estimated using Schofield's equation using a 1.5 activity factor. A typical breakfast consisted of a croissant, ham, cheddar cheese, cereals with milk, a fruit cup, and orange juice while lunch items consisted of a ciabatta ham and cheese sandwich and orange juice. The same meal was provided during all follow-up visits.

| Statistical analysis
Participant characteristics were described through frequencies and means (SD). All data processing and statistical analysis were performed in SAS (SAS v9.4, Cary, NC). Statistical significance was set at an alpha level of p < 0.05.
Although this is a pilot study, the sample size calculation was based on effect sizes from previous studies 18,19 that evaluated the effect of interrupting prolonged sitting on glucose iAUC using different modalities. These studies reported a 20%-30% decrease in postprandial glucose iAUC levels in the intervention groups. This study used a conservative estimate of a 15% difference between the two intervention conditions with a 1% population estimate of standard deviation. Using G*Power software (v3.1.9.2), we estimated a required sample of 12 participants allowing for 0.5 correlation coefficients between repeated measurements and an alpha of 0.05 to obtain a power of 80%. Considering a 20% attrition rate, we planned to recruit 15 participants. Random effects mixed model analyses 27,28 were used to examine differences in acute postprandial glucose response between the different study conditions. This method allowed for the clustering of data within an individual (for repeated measures) and for potential missing data at random. All models met assumptions of linearity and normality of residuals. Glucose AUC, iAUC, mean glucose, and MAGE on postprandial periods were evaluated as outcomes. Random person effects were included in the models to account for the clustering of the observations within the individual. In addition, all models were also controlled for period (i.e., visit number), sequence (i.e., block assignment), and treatment×period interaction effects. The HFLD and LFHD conditions were jointly compared with the all-day sitting conditions to address aim 1. For aim 2, a comparison between HFLD and LFHD was conducted. All data from randomized participants were included in the analysis in accordance with the intent-totreat principle.

| Recruitment and baseline characteristics
The study consort diagram (Figure 1) summarizes the study recruitment and data collection flow. A total of 52 participants were invited to the laboratory for fasting blood glucose screening. Fifteen participants met the eligibility criteria and consented to participate in the study. Four participants were then excluded from the study due to unresponsiveness or not being interested in participating. Overall, 11 participants were randomized and included in the analysis although one participant only completed two out of three laboratory visits. The full participant characteristics are summarized in Table 2. The participants were mainly adults (mean age = 46.8 ± 10.6 years; 27% male; mean BMI = 34.6 ± 5.4 kg/m 2 ) with prediabetes (mean fasting glucose = 109.0 ± 9.8 mg/dL). These participants were highly sedentary, averaging 626.9 ± 135.7 min/day of sitting time.

| Physical activity
During the past 6 days prior to their study visits, on average, the participants engaged in 642.7 ± 144.5 min, 625.4 ± 155.5 min, and 694.7 ± 111.1 min of sitting per day for the CS, HFLD, and LFHD conditions, respectively. Furthermore, the average time spent in MVPA activities was similar across study conditions with a mean of 19.5 ± 11.6 min per day for CS, 24.8 ± 15.9 min per day for HFLD, and 17.8 ± 11.4 min per day for LFHD condition. The participants' objectively measured physical activity data during each visit is summarized in Table 2, which verifies their overall compliance to the study protocol. There were some slight deviations in the prescribed regimens though on average, participants closely followed through the sitting and standing protocol for all conditions. In the continuous sitting (CS) condition, participants accumulated their total sitting time in 3.9 ± 1.6 bouts of sitting with very minimal standing. This suggests that participants engaged in few short bouts of standing during the CS condition resulting in a slightly lower total sitting time. As expected, participants performed about 60 min of standing and 360 min of sitting in both HFLD and LFHD conditions. The only difference between the two groups was in the manner in which they accumulated sitting time. In the HFLD condition, participants performed 26.0 ± 1.4 bouts sitting. In contrast, the LFHD condition averaged 5.7 ± 2.4 bouts of sitting.

| Standardized Meals
The macronutrient content of the meals is summarized in Table S1. In terms of compliance with the standardized meals, fasting states were confirmed before the start of each visit. All participants consumed their meals within 15-20 min. Seventy percent of the participants (8/11) were able to consume 100% of the provided meals. For the two participants that were unable to consume the entirety of their meals during the first visit, the meals on their succeeding visits were adjusted to match what they were consumed on the first visit. Figure 2 depicts the postprandial glucose curve. Breakfast meals were given at minute zero, while lunch meals were served at minute 225. On average, the baseline glucose level during the 30-minute period prior to the breakfast meal was 95.6 ± 25.0 mg/dL, 95.7 ± 30.2 mg/dL, and 96.5 ± 18.7 mg/dL for the CS, HFLD, and LFHD, respectively. The average glucose, AUC, iAUC, and MAGE is summarized in Table 3. During the iAUC calculation, about 30% of the observations fell below the baseline glucose level and were excluded from the calculation of the iAUC.

| HFLD versus LFHD conditions
The comparison of the HFLD and LFHD conditions revealed small to medium effect size (Cohen's d ranged from 0.02 to 0.58), with the largest effect size occurring during the postprandial lunch period. The result revealed that AUC and iAUC during the postprandial lunch and total time were consistently lower in the HFLD condition. However, the analysis of differences in means between conditions did not reach significance levels. Analysis of the mean postprandial glucose revealed similar postprandial glucose levels during

| DISCUSSION
The current study demonstrates that different standing patterns for interrupting sitting time can differentially impact postprandial glucose. Overall, the results from this study indicated that breaking up continuous sitting can be beneficial to postprandial glucose responses. Specifically, using short frequent bouts of standing to interrupt prolonged sitting resulted in better postprandial glycemic response (5%-8% lower mean glucose) compared to engaging in longer duration, but less frequent, standing breaks. Although the analysis of the AUC, iAUC and MAGE outcomes was not statistically significant, the patterns of the results are consistent with previous studies on this topic. For example, two meta-analyses on interrupting sitting time found that frequent interruptions of sitting behavior through LPA or MVPA effectively reduced postprandial glucose. 29,30 In addition, a similar study by Henson et al. in a sample of prediabetic women that evaluated the effects of interrupting prolonged sitting through frequent standing breaks (5-min standing breaks every 30 min of sitting) also showed improvements in postprandial glucose. 19 It should be noted, however, that although the total time spent by participants in seated and standing positions was comparable between this study and the study by Henson et al. (i.e., 420 min of sitting and 60 min of standing), the standing breaks protocol that they used entails a more frequent sit-stand transitions that did not allow for sitting bouts to be accumulated for more than 30 min at a time. This result suggests that frequent interruptions in sitting time, even with simple activities such as standing for at least every 30 min interval, a common threshold for identifying prolonged sitting, can significantly negatively impact the postprandial glucose response of dysglycemic individuals. In contrast, two other studies that utilized a protocol allowing for the accumulation of sitting time through bouts longer than 60 min 13,20 also showed significant improvement in postprandial glucose response with standing breaks. However, their study protocols also elicited significantly longer standing periods (up to 30 min every hour of sitting) resulting in longer total standing time (150-240 min of standing time compared to 60 min in the current study). These studies suggest that in addition to reducing total sitting time to a certain level, the pattern of frequent interruptions to sitting could be an effective strategy to combat the deleterious effects of prolonged sitting on the postprandial glucose level. Unfortunately, the current study was not designed to determine any doseresponse relationship between the number and duration of standing breaks and postprandial glucose response nor any interaction effect between these two variables. Thus, these questions should be investigated in future studies.  The underlying physiological mechanisms driving the benefits of reducing sitting time are poorly understood. A potential explanation for the results is the increase in total energy expenditure and carbohydrate substrate utilization associated with skeletal muscle activation during frequent intermittent standing. 31,32 A recent study found that compared to prolonged sitting and interrupted sitting using longer but less frequent standing breaks (15 min of standing every 30 min of sitting), interrupted sitting through frequent although shorter standing breaks (1.5 min every 2 min of sitting) increased the 8-h total energy expenditure by 20% (617 ± 76 kJ) and 9% (296 ± 78 kJ), respectively. 14 This suggests that the difference in postprandial glucose response of the two strategies in interrupting sitting time may be partially accounted for by the discrepancy in total energy requirement between the two conditions that seemed to be driven by the frequent muscle contractions from the change in posture. However, results from other studies showing no significant differences in postprandial improvement in glucose level between using LPA and MVPA to break up sedentary time suggest that differences in energy expenditure alone may not fully explain this phenomenon. 18 This study also has several limitations. Postprandial glucose levels were measured using a continuous glucose monitor that estimates blood glucose levels from interstitial glucose. Although this method allowed for an unobtrusive way of measuring glucose levels frequently, the values may not be directly comparable to the glucose measurement from venous samples used in prior studies. Diets were not controlled outside the laboratory visits, so the results presented in this study were limited to the data collected during the laboratory visits. Several studies have shown evidence of how these types of interventions can potentially impact glycemic profile up to a day after the visits. 13,19 Gaining information on glycemic profiles outside of the laboratory could lead to insights into the temporality of the observed benefits that resulted from the intervention. However, this approach was outside the scope of this study and should be explored in future studies. Another limitation is the lack of control over menstrual cycle of female participants. It has been previously shown that blood glucose concentrations can fluctuate depending on the stage of the menstrual cycle. To minimize this bias, we randomly allocated participants to different blocks to determine the order that they received the intervention. Lastly, to make sure that other external variables are adequately controlled to improve internal validity, the study implemented strict experimental manipulation to limit other movement behaviors (e.g., MVPA and fidgeting). This control, combined with the small sample size that only included individuals with prediabetes limits the study's generalizability. A study that addresses these limitations and is adequately powered to detect the differences observed in the two interrupted sitting conditions is needed to confirm the results of this study.
This study also makes unique contributions to the field by experimentally testing the different combinations of frequency and bouts to break up sitting time. By focusing on standing as a mode for interrupting sitting time, this study allowed the researchers to isolate the benefits of interrupting sitting time from the benefits of engaging in other higher intensity physical activity behaviors. Thus, the results provided novel insights on potential ways to efficiently break up sitting time that investigators can be leverage in designing future interventions.

| PERSPECTIVE
Sedentary behavior is a public health problem and interventions that maximizes the effect of reducing sitting time are needed. The use of sit-stand workstations has been shown to be effective in reducing workplace sitting. However, it remains unclear how to effectively reduce and/or interrupt sitting time to maximize benefits from such interventions. Studies that explored the health benefits of interrupting sitting time have mostly focused on using different modalities (i.e., comparing walking vs standing breaks). However, gaining a full understanding on how different patterns of interrupting time can shed light to potential mechanisms on how sitting can be detrimental to health and contribute to the development of efficient interventions to reduce sitting time especially in settings where individuals are limited to the type of activity in which they can engage in (e.g., office employees). This study provides preliminary evidence favoring the use of short, frequent interruptions in sitting time to improve glycemic control of prediabetic individuals.

| CONCLUSION
Overall, our results suggest that frequency of standing breaks to interrupt prolonged sitting can significantly reduce postprandial glucose response of individuals with prediabetes. This study provides evidence favoring the use of frequent interruptions (2.5 min of standing breaks every 15 min of sitting) in sitting time to improve glycemic control of prediabetic individuals. In contrast, less frequent, although longer bouts of standing breaks (10 min of standing breaks every 60 min of sitting) resulted in similar postprandial glucose levels compared to the continuous sitting condition. Future studies should employ larger sample size and explore potential dose-response relationships between the number of bouts and the potential for interaction between bout duration and frequency.

ACKNO WLE DGE MENTS
This project was partially supported by the Arizona State University Graduate and Professional Student Association's Graduate Research Support Program, which is funded through ASU's Graduate and Professional Students Association, The Graduate College, and the Office of the Vice Provost for Research. The authors acknowledge the contribution of Dr. Paddy Dempsey for providing a framework for the meal plans used in this study. The authors declare no conflict of interest. All listed authors contributed to the conceptualization of the study design. M.T. collected the study data and drafted the manuscript. All authors reviewed and edited the manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.