Study design

This randomized, controlled, cross-over trial took place at the University of Otago, in Dunedin, New Zealand between April 2019 and July 2019. The trial was approved by the University of Otago Human Ethics Committee (Health), approval number: H18/138, and written informed consent was obtained from all participants before screening. The study is registered prospectively with the Australian New Zealand Clinical Trial Registry (ANZCTR12619000175178).

Participants

Participants were recruited through the distribution of emails and advertisements placed on campus-related social media sites. Eligible participants were 18–40 y of age who self-reported sitting for >5 h per day, did not smoke, were free of cardiovascular disease or diabetes, had a BMI <30 kg/m², were not on medication known to influence endothelial function or postprandial metabolism, were capable of participating in physical activity and standing for long periods (assessed via administration of PAR-Q and self-report), and had a blood pressure at screening of <140/90 mm Hg (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Participant flow diagram (CONSORT).

https://doi.org/10.1371/journal.pone.0244841.g001

Experimental protocol

The experimental protocol is presented in Fig 2. Participants completed three 6-h interventions: Prolonged Sitting, in which participants sat for 6 h; Prolonged Standing, in which participants stood at a standing desk for 6 h; and Regular Activity Breaks, in which sitting was interrupted by 2 min of walking on a treadmill (5 km·h^-1; 10% incline) every 30 min. The sequence in which participants completed the three interventions was random (see below). Each intervention was separated by a minimum of 4 d (median 7 d; 25^th and 75^th percentiles 6 and 20 d, respectively). The only time participants deviated from their prescribed posture during each intervention was during two predetermined bathroom breaks. At each bathroom break participants walked slowly ~20 m down the hall to the bathroom (regardless of whether then needed to use the facilities) and back again. When sitting or standing, participants were permitted to read or use their laptop (for work or leisure). An investigator was in the room with participants throughout each session to ensure adherence to all protocols.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Intervention session timeline.

https://doi.org/10.1371/journal.pone.0244841.g002

Randomization

Participants were randomly assigned to complete the three interventions in one of the six possible orders. The randomization sequence was generated prior to participant recruitment by MCP using Stata software (version 15 for MAC; StataCorp, College Station, Texas), and concealed electronically. The afternoon prior to each participant beginning their first intervention session the next sequential randomization sequence was revealed and assigned.

Standardization of prior diet and exercise

Participants were asked to avoid physical activity for 24 h prior to each intervention session, and record everything they ate and drank. A copy of this record was returned to participants prior to each subsequent intervention and they replicated the food and drink consumed as closely as possible. Participants wore an ActiGraph accelerometer (ActiGraph GT3X+, ActiGraph, Pensacola, Florida) and an interstitial glucose monitor (Freestyle Libre Pro, Abbot Diabetes Care) over this time to monitor compliance.

Meals

All participants consumed the same breakfast (muesli, trim milk, juice, toast, margarine and honey) at baseline, and a snack (chocolate brownie) at four hours. The breakfast and snack combined provided 4697 kJ energy, 170 g carbohydrate (62% energy) and 35 g fat (28% energy). Gluten-free alternatives were provided on request (n = 3) and provided 4716 kJ energy, 180 g carbohydrate and 33 g fat. All food was consumed within 15 min by all participants, and no other food was consumed over this time. Water intake was provided ad libitum during the first session and then replicated in all subsequent sessions.

Measurement of popliteal artery hemodynamics

The popliteal artery was assessed by one of two experienced sonographers (KNT and TDG), and all measurements within individual participants were completed by the same sonographer. Simultaneous popliteal artery diameter and blood velocity were measured using ultrasound (Terason uSmart 3300, MA, USA) with a 15 MHz linear array transducer (bandwidth 4–15 MHz). Hemodynamics were assessed for ~60 s on each occasion. All baseline and final (360 min) measurements were made with participants in a left lateral recumbent position with the knee slightly flexed, as were all measurements during Sitting and Regular Activity Breaks interventions. However, in the Prolonged Standing intervention, hourly measures taken between 0 and 360 min were in a standing position with the knee slightly flexed. In the Regular Activity Breaks intervention measurements were taken immediately prior to an activity break, meaning the participant had been sitting continuously for ~30 min prior to measurement. The position of measurement within the popliteal fossa was marked to ensure consistency of location. This location was measured from bony landmarks and/or the knee skin crease and replicated in subsequent sessions. The location was verified by comparing the recording from the baseline of the previous intervention session prior to the first assessment in a subsequent intervention. Ultrasound depth, focus position, gain and Doppler settings were optimized for each participant for the first measurement and replicated thereafter. Videos were recorded using Camtasia Studio Screen Recording Software (TechSmith, MI, USA). Offline analysis was performed blinded to intervention assignment, using wall-tracking software (Cardiovascular Suite v 3.6, Quipu, Pisa, Italy).

Measurement of Flow Mediated Dilation (FMD)

Flow mediated dilation was measured in the popliteal fossa following published guidelines [23] at baseline and 360 min, after the participant had rested quietly for >10 min in a left lateral recumbent position. Following a 2-min baseline, transient ischemia was induced by inflating a 12-cm cuff around the mid calf to 200 mm Hg within 2 s (SC12 cuff, TD312 blood pressure cuff inflator; Hokanson, Bellevue WA, USA). Occlusion was maintained for 5 min and recording continued for 3 min following rapid release of the cuff. Baseline diameter (D_base), blood flow velocity (v) and shear rate (SR) were assessed as the mean of the last minute of recording prior to cuff inflation. Blood flow was calculated as the product of half the time-averaged peak envelope velocity and vessel cross-sectional area. Peak diameter post deflation was determined automatically using edge-detection software. Flow-mediated dilation was calculated as the percentage increase in diameter from baseline (FMD = (D_peak-D_base) / D_base x 100), following guidelines utilising allometric scaling to adjust for D_base with a covariate-controlled approach [24].

Muscle mass of left lower limb

Muscle mass of the left lower limb was assessed via bioelectrical impedance (InBody 230, InBody, Seoul, South Korea) at baseline and immediately prior to the bathroom break at 315 min. Changes in muscle mass between the two measurements were assumed to be changes in fluid in the lower limb.

Blood pressure

Arterial blood pressure was measured hourly using an automated sphygmomanometer (OMRON HEM-907) with an appropriately-sized cuff. In all intervention sessions, at baseline (0 min) and 360 min blood pressure was measured in a seated upright position after at least 10 min of sitting. In the Prolonged Sitting and Regular Activity Breaks interventions blood pressure was also measured in a seated upright position for all other measurements. In the Regular Activity Breaks intervention blood pressure measurements were conducted immediately prior to an activity breaks meaning the participant had been sitting for ~30 min prior to each measurement. In the Prolonged Standing intervention blood pressure was measured in a standing position at all other time points. Measurements were taken on the same arm for all conditions, contralateral to the arm with the indwelling venous cannula. Three measurements were taken one minute apart and the mean of the three measurements recorded at each time point.

Blood collection

Participants arrived at the laboratory in a fasted state. Immediately after baseline FMD measurements were completed a cannula was inserted into a vein in the antecubital fossa or forearm. At least 10 min later a fasting blood sample was collected and the session began. Blood samples were collected hourly for 6 h, with additional samples collected 30 and 45 min after both breakfast and the snack; i.e., 11 samples were collected in total. Blood samples were collected into a syringe and transferred immediately into EDTA-containing tubes. At baseline, 180 and 360 min a capillary tube was filled with blood collected from the syringe for analysis of hematocrit. Both the capillary and EDTA containing tubes were stored chilled until they were centrifuged (1100 rpm for 5 min and 300 rpm for 10 min, respectively), within 2 h of collection. Hematocrit was measured using a custom-made Vernier calliper reader immediately after centrifugation, while plasma from the EDTA tube was aliquoted and stored at -80°C for later analysis.

Analytical methods

Plasma glucose concentration was measured using the hexokinase enzymatic method on a Cobas C 311 analyser (Roche Diagnostics; Mannheim, Germany). Plasma insulin concentration was measured using electrochemiluminescence methods on an Elecsys 2010 (Roche Diagnostics; Mannheim, Germany). Kits and calibrators for glucose and insulin analysis were sourced from Roche Diagnostics (Mannheim, Germany). All samples from the same participant were assayed in a single run. Intraassay CVs were 0.7% for glucose and 1.3% for insulin. All analyses were performed by a lab technician who was blinded to intervention allocations.

Statistical analysis

Data were analysed using Stata software (version 16.0 for PC; StataCorp, College Station Texas). The primary outcome for this study was the difference in change in FMD from baseline to 360 min between interventions. A sample size of 18 participants provided 80% power to detect a 0.85 reduction in D_base-adjusted FMD (α = 0.05) assuming a sample size of 1.0, using a test of paired comparisons, and allowing three participants to complete the three interventions in each of the 6 possible intervention orders. This sample size also allowed 80% power to detect a 10% difference in glucose area under the curve (AUC) and a 15% difference in insulin AUC. The effect of the different interventions on raw and allometrically scaled FMD at 230 min was evaluated using linear mixed models with adjustment for baseline diameter.

The effect of the different interventions on the hemodynamic variables (diameter, shear rate and blood flow) was assessed at three time points of interest using linear mixed models with adjustment for values measured at 0 min: 1) at 60 min (to assess the very acute effects of the posture/activity); 2) the mean of the measurements taken at 240 and 300 min (to assess the posture/activity after it had been sustained for several hours while incorporating the effects of a second feeding) and; 3) at 360 min (to assess if changes as a result of the posture/activity were maintained after returning to a seated position). For ease of interpretation, differences in flow between interventions were converted to percentages post-analysis.

Mean blood pressures for each intervention were quantified as the mean of all time points after baseline. The effect of the different interventions on mean systolic and diastolic blood pressure was evaluated using mixed model regression.

The total AUCs for glucose and insulin were calculated using concentrations measured from each of the 11 time points, using the command integ in Stata. This calculation fits a cubic spline curve through the time points on which blood samples were collected and calculates the area under that curve. Incremental AUC (iAUC) were also calculated by subtracting baseline concentrations from each subsequent concentration over the 6 h period and calculating the area under the curve using the same command as above. When samples were missing due to cannula malfunction (n = 4) the area under the curve was calculated without the missing time point (no baseline or 360 min samples were missing). The effect of different interventions on total AUC and iAUC for glucose, and insulin was evaluated using linear mixed models. For ease of interpretation, differences in AUC between interventions are presented as percentages which were calculated post-analysis.

For all mixed models, period and order affects were assessed individually and found not to have an important effect, however, for completeness order was included as a predictor in all final models. The assumptions of regression models were checked for every model and found to be met for all models other than insulin (in which the residuals were skewed). Log transformation addressed the non-normality to a large extent, but robust standard errors were used to allow for small deviations from normality. For consistency, robust standard errors were then used in all models. No adjustment was made for multiple comparisons.

Hemodynamic effects

Few studies have examined vascular effects of prolonged sitting and/or activity breaks. This, combined with differences in methodology between studies, make comparisons difficult. Other studies have examined from 90 min to 6 h of sitting, using a variety of ‘breaking’ interventions (e.g., standing, fidgeting, short walks, longer walks, callisthenics), and in various arteries (superficial femoral, popliteal or brachial). Furthermore, the current study was performed with participants in a fed state. On balance, most studies have demonstrated reduced blood flow and net shear rate during and following sitting for 3–6 h, and most interventions that interrupt sitting mitigate these reductions [15, 19, 25–27]. The results of the current study support that blood flow and net shear rate are lower with prolonged sitting, and that activity breaks ameliorate this response. Blood flow and shear rates in the popliteal artery during walking are also going to be significantly higher, so the measurement ~30 min post-walking underestimates the extent of the favourable hemodynamic stimulus. However, the finding of elevated blood flow between walking bouts is important for understanding the effect that walking has during the seated periods. Standing appears to prevent the decline in blood flow and shear rate initially, but this protection is not maintained past 4 h; a time effect that is perhaps caused by the slow development of edema in the leg that opposes the effects of gravity on arterial blood pressure. The absence of this fluid shift seen with Regular Activity Breaks may contribute to the elevation in blood flow and shear rate observed for the entirety of the 6-h periods. The 6-h standing period was longer than used in previous studies; for example, Morishima et al. [21] demonstrated increased blood flow and shear rate throughout 3 h of standing.

Despite these relatively consistent hemodynamic profiles across sitting / breaking studies, effects on FMD are more variable, with a number of studies [15, 21, 25, 26] documenting a reduction in FMD with sitting and protection by breaking, and others showing no change following any intervention [19, 20, 22]. It is unclear as to why the 6-h sitting intervention did not reduce FMD in the current study. Perhaps in this young, otherwise healthy, homogeneous population, the stimulus was insufficient to evoke an impairment, or the walk to the toilet on two occasions (performed by every participant regardless of need)was enough to mitigate the effects of sitting for long periods. However, from the observations of reduced blood flow and shear stress, it is possible that over time, i.e., repeated exposures, this might precipitate endothelial dysfunction.

Metabolic effects

Performing regular activity breaks reduced postprandial insulin response when compared to both prolonged sitting and prolonged standing. The reduction in insulin response when compared to prolonged sitting is in line with previously summarised literature [8]. However, the fact that prolonged standing did not seem to offer any benefit to glucose metabolism when compared to prolonged sitting provides support to the idea that standing is less likely to result in beneficial metabolic effects in healthy participants [28]. The lack of reduction in postprandial glucose is in contrast to the results of several other studies [8], however, many of the earlier studies provided very high levels of carbohydrate. It is possible that–particularly in healthy individuals–it is harder to observe reductions in glucose response when meals are closer in composition to what would be consumed in real life. However, it is important to keep in mind that while the glucose response is not significantly different, the insulin response is reduced with regular activity breaks indicating an acute improvement in insulin sensitivity. It seems likely that the maintained increase in blood flow in the lower limb seen with regular activity breaks is helping to facilitate increases in non-insulin mediated glucose uptake [29, 30], which if maintained with habitual behaviour may have important implications on overall cardio-metabolic health.

The results of the current study indicate that even healthy individuals benefit from interrupting sedentary time with brisk walking. Perhaps instead of providing standing desks to individuals who sit for prolonged periods at work, we should be redesigning the physical and psychosocial occupational landscapes to provide employees with opportunities to regular interrupt periods of prolonged sitting with short bouts of physical activity [31].

Limitations and delimitations

Consideration should be given to the following aspects of the study design that may have influenced the results. Our study assessed acute responses to a 6-h period of sitting, standing and performing regular activity breaks. While this duration of standing is one of the longer assessed in this context, the chronic impact of these patterns of activity on vascular function and glycemia response are yet to be fully elucidated. The study included only young, healthy participants. There has been suggestion that participants who are less metabolically healthy may have larger positive responses to interventions that interrupt prolonged sitting [28], and therefore, the magnitude of the responses reported here may underestimate the response in the population as a whole. We did not adjust for sex in our analysis, nor did we control for menstrual cycle. While the cross-over design somewhat eliminated the need to adjust for sex, non-adjustment or control for menstrual phase may have introduced variability [32], future studies should consider evaluating any effects of menstrual cycle phase. We used changes in muscle mass measured via bioelectrical impedance to represent fluid shifts in the lower limb. Future studies should consider more direct measurements of limb edema and consider assessing leg symptoms. We did not adjust for multiple comparisons in the statistical analysis. While some arguments support this approach [33], small differences between interventions should be interpreted with caution. Leg movement was not controlled across any of the interventions, nor was it quantified so that it could be included in statistical modelling. It is possible that if we limited fidgeting of the lower limb, or at least adjusted for it statistically that this would have removed some of the noise in our measurements [25]. However, it would have also meant that the results were less likely to represent a real life situation where people are free to move around to a small degree even if their occupational activities mean they are constrained to one posture. The use of an uninterrupted standing intervention, while representing some occupational settings that require continuous standing, does not represent recommendations for the use of standing workstations, which suggest that regular changes in posture are important [10].

Additionally, the acute nature of the study does not consider the long term conditioning that may occur when this behaviour is maintained over days or weeks. However, the results of this study, when combined with others that have investigated the effect of intermittent bouts of standing, reiterate the suggestion that it may be the transitions from sitting to standing that provide the most benefit, rather than the standing itself.