Acute Effect of Height-Adjustable Workstations on Blood Glucose Levels in Women With Impaired Fasting Glucose Levels While Working: A Pilot Study

Abstract

Objective:

To examine the effect on continuously monitored blood glucose (CGM) among participants with impaired fasting glucose (IFG) who used a height-adjustable desk while working.

Methods:

The study was a repeated measures pilot study in overweight or obese women who had IFG (blood glucose [BG] >100 mg/dL) and a sedentary job. Blood glucose was monitored with CGM devices during two 1-week periods at work; 1 week in the seated position and 1 week using alternate bouts of sitting and standing (by adjusting their desks) throughout the workday.

Results:

Ten women completed the study. Sedentary time significantly predicted BG independently of diet and overall physical activity (P=.02). Dietary carbohydrates, protein, and fat were significant predictors of BG (P<.001).

Conclusions:

Sedentary time is a strong predictor of increased BG in women with IFG and a sedentary job.

Introduction

Sedentary behavior is a distinct cardiometabolic risk factor that is independent of habitual physical activity.^1,2 More specifically, prolonged bouts of sedentary time are associated with adverse health outcomes, including type 2 diabetes mellitus and premature death,³ and sedentary behavior is inversely associated with insulin action.^4–6 Breaking up prolonged sedentary time is associated with beneficial effects on cardiometabolic risk markers, such as reducing body mass index, waist circumference, and triglycerides and postprandial glucose levels, regardless of total time spent in sedentary behavior and in moderately vigorous activity.^6–8 In laboratory-based studies, postprandial glucose and insulin concentrations decreased after short activity breaks (eg, low-intensity walking).^3,7,9–11

Working adults spend approximately one-third of their lives at work, and 77% of that time is spent in sedentary behavior.^12,13 Recent studies have shown the feasibility of replacing traditional desks with height-adjustable workspaces to reduce sedentary time.^14–16 These height-adjustable workspaces could have important implications for decreasing sedentary time and improving cardiometabolic risk factors such as impaired fasting glucose (IFG) levels.

IFG affects an estimated 79 million Americans older than 20 years and increases the risk of type 2 diabetes mellitus, cardiovascular disease, and stroke.¹⁷ Weight reduction and increased physical activity are effective strategies for preventing or delaying the onset of type 2 diabetes mellitus,¹⁷ and reducing sedentary time effectively decreases the risk of type 2 diabetes.^3,9 However, in the studies that presented those results, glucose levels were not continuously monitored. With continuous glucose monitoring (CGM), 288 glucose measures can be generated in 1 day (ie, 1 every 5 minutes). CGM technology enables nearly continuous examination of glycemic variability, which can be tracked for up to 7 days.

CGM technology was developed in response to the need for improved glucose control to reduce complications related to diabetes and to decrease the risk of severe hypoglycemic episodes.^18,19 Before CGM, measurement of hemoglobin A_1c (HbA_1c) was the primary tool for assessing glucose variability. One problem was that treatment responses to HbA_1c levels led to episodes of severe hypoglycemia, primarily in patients with type 1 diabetes mellitus. Because HbA_1c can be widely variable and is based on a population mean and not an individual, methods capable of frequently monitoring blood glucose control were needed²⁰. Additionally, diabetes-related complications occur inconsistently in patients with the same HbA_1c levels.²¹ Clinically, CGM is recommended for managing type 1 diabetes and now functions simultaneously with insulin pumps like an artificial pancreas. Clinical recommendations state that patients with type 2 diabetes who have difficulty controlling hypoglycemic episodes should use CGM, but CGM is not generally used daily.¹⁹ To date, CGM technology has not been studied in adults with IFG. In addition, the effect of intermittent standing on glucose control in adults with IFG over an entire workweek has yet to be examined. The observational and laboratory-based studies have examined healthy, overweight, and obese persons; therefore, investigating the effect of a sit-stand workstation on blood glucose levels among person with IFG would address a gap in the literature.

The purpose of the present study was to examine the effect of decreased sitting time and increased intermittent standing time during the workday on blood glucose levels among participants with IFG. Specifically, we sought to compare glucose levels measured with CGM during a 1-week traditional sitting pretest with levels measured during a 1-week sit-stand intervention. We hypothesized that the participants would have lower blood glucose levels during a sit-stand intervention compared with prolonged sedentary time.

Research Design and Methods

Study Design and Participants

This repeated-measures non-randomized crossover study was designed to compare blood glucose levels during periods of prolonged sitting and during periods when sitting was interrupted with activity. Participants wore a CGM device on 2 occasions: 1) during 1 workweek while using a desk in the traditional seated position (pretest) and 2) during a second workweek while standing for half the workday (intervention). Participants were recruited from 2 workplaces in the upper Midwest of the United States from February through July 2013. Recruitment methods included flyers, advertisements in the electronic company newsletter, and informational sessions at the workplaces. The volunteers in the study were recruited from a larger worksite wellness study in which they were concurrently enrolled. (The worksite wellness study was examining the effect of sit-stand desks, walking, and wellness education on reducing sedentary time in the workplace.) The 2 studies began at the same time.

Inclusion criteria included having IFG (fasting blood glucose >100 mg/dL), taking the same medications for the past 6 months, being a full-time employee (working ≥35 hours per week), working daily while seated (specifically, sitting for ≥75% of the workday), and being able to safely participate in physical activity that involved increased walking, standing, and using the stairs. Exclusion criteria included a history of heart disease, type 1 or type 2 diabetes mellitus, renal disease, peripheral neuropathy, retinopathy, peripheral arterial disease, lower limb amputation, pregnancy, active substance abuse, hospitalization in the past 6 months for a psychiatric disorder, severe visual impairment, enrollment in a physical activity study, and insulin therapy. Informed consent was obtained, and past medical history was collected. The study was approved by the University of Minnesota Institutional Review Board. A portion of these data were reported previously in a dissertation (by ARB).²²

Measures

Anthropometric Measures

Baseline data included height, weight, blood glucose and cholesterol levels, blood pressure, and body fat percentage. Measurements were completed in the fasting state in a private room at the worksite. Height and weight were measured with a stadiometer and calibrated scale (Seca, Chino, California). Fasting blood glucose levels were measured with a glucometer that required a fingerstick (Breeze 2; Bayer Corp, Whippany, New Jersey), and cholesterol levels were measured with the Alere Cholestech LDX system (Abbott Laboratories, Abbott Park, Illinois). Blood pressure was measured with an automatic sphygmomanometer (BPM-100; BpTRU [defunct]). Body fat percentage was measured with dual energy x-ray absorptiometry.

Continuous Glucose Monitoring

The CGM device (iPro2; Medtronic, Dublin, Ireland) consisted of a sensor inserted into the interstitial fluid and a transmitting device attached to the sensor outside the skin. This blinded device uses an electrochemical system. The sensor, which consists of 3 electrodes (reference, working, and counter electrodes) that complete a circuit, contains 3 layers: 1) an outer semipermeable layer that is selectively permeable to oxygen and glucose; 2) a middle layer coated with glucose oxidase (within this layer a chemical reaction produces hydrogen peroxide and gluconic acid); and 3) the innermost layer, an electrode layer. In the second layer, the gluconic acid is absorbed into the body, and the hydrogen peroxide travels to the electrode layer, where the hydrogen peroxide comes in contact with a small sensor (which measures nanoamperes) and a second chemical reaction occurs. This reaction reduces the hydrogen peroxide into hydrogen, oxygen, and 2 electrons. The 2 electrons are measured by the sensor, and the resulting electron current generates an interstitial signal that is sent to the transmitter. Interstitial signal values downloaded from the transmitter are converted to glucose values (in 5-minute intervals). The transmitter is reusable after proper sterilization and downloading of existing data; however, the sensor is designed for single use. The CGM device could be worn continuously for up to 7 days. Accuracy of the device is roughly 80% for detecting an IFG profile.²³

Other Measures

During the trial, participants wore an accelerometer, the Kinetic Activity Monitor (KAM) (Kersh Health, Plano, Texas), to measure physical activity. The KAM measures activity-minutes in 3 zones: life (<2 mph), health (2–4.5 mph), and sport (>4.5 mph). The accelerometer generates KAM points, which correspond to a 1% increase above basal metabolic rate (which the KAM software calculates from height, weight, age, and sex). Participants were asked to record dietary intake (all meals, snacks, and beverages) and physical activity daily in a diary.

Procedure

Informational sessions for both studies were conducted at the worksites with question-and-answer sessions after the presentations. Interested attendees contacted a study coordinator by email, and an initial eligibility questionnaire was sent in response. After informed consent was obtained and baseline data were collected, participants were scheduled for the first week of wearing the CGM device. All instructions and expectations were reviewed before the device was inserted. The device was inserted on the first day of the week in a private room at the worksite. For insertion, a location on the side of the lower back or at the top of the gluteus maximus muscle was chosen that would not interfere with normal daily movements; was free of scarring, stretch marks, and hardened tissue; and consisted primarily of fatty tissue. The insertion site was sterilized with alcohol and allowed to dry, the sensor was placed in the inserter, and the protective coverings were removed. The inserter was placed on the skin at a 60° angle, the sensor was inserted with the insertion device, and the inserter was removed from the sensor. The sensor and the site were monitored for bleeding (excessive bleeding can damage the sensor). The sensor was allowed to “wet” for a minimum of 15 minutes (ie, the probe was saturated with interstitial fluid before the probe was attached to the external measurement device, in accordance with the manufacturer’s instructions). During this time the instructions for wearing the device were again reviewed.

The instructions included educating the participant on how and when to take daily self-monitored blood glucose (SMBG; Bayer Breeze 2) measurements, which were required for calibration of the CGM device. Participants were instructed to measure their blood glucose before breakfast, before lunch, before dinner, and before going to bed daily while wearing the device. They were provided with a glucometer and supplies. Participants were given instructions on measuring blood glucose, and they practiced until they felt confident. They were asked to record the SMBG measurements in the daily diary provided and to keep a daily food diary while wearing the CGM device. Details to record in the food diary included meals, beverages, snacks, timing, and portion sizes. Nutrient composition was derived from the US Department of Agriculture National Nutrient Database for Standard Reference, which contains over 8,000 food items. Quantity and type of food were searched, and the nutrient composition for all meals and beverages was recorded. Dietary composition for meals consumed at chain restaurants was cross-checked with the database to ensure accuracy. Participants were also asked to record daily physical activity in the diary and to continue their normal daily activities while wearing the device. The manufacturer recommended avoiding hot tubs or spas given that they could cause the device to malfunction. A phone number and email address were provided for emergencies or questions.

After 15 minutes of “wetting,” the recording device was attached to the sensor. The recording device displayed a green light, signifying that the sensor was adequately placed. After the green light was visible, the device was secured with a transparent, hypoallergenic film dressing that kept the device in place during most activities. Participants were then instructed to record the first SMBG measurement within 1 hour, the second SMBG measurement 3 hours after insertion, and subsequent measurements on the normal schedule. A specialist from the manufacturer of the CGM device was present for device insertion to address any questions or problems with the equipment. During the first week of data collection, all participants were instructed to sit at their desks as usual. At the end of the week, the CGM device was removed at the worksite in a private room. The film dressing was removed, and the transmitter and sensor were taken out. After the transmitter was removed from the sensor, the sensor was discarded in a biohazard container and the transmitter was cleaned and sterilized according to the manufacturers and regulatory guidelines. ARB completed training through the manufacturer on the proper use and insertion of the device and US Food and Drug Administration guidelines for care and cleaning.) The CGM data were downloaded to the manufacturers software, and dietary intake and calibration SMBG measurements were recorded in the software.

After the first week, participants began a 4-week acclimation period in which they had been instructed to gradually increase their standing time, accumulated in intermittent bouts (i.e. starting with 10–30 minute bouts of standing), to half their workday. After those 4 weeks, the second week of data collection began, and participants were instructed to use their sit-stand desk for at least half the workday for the next 5 days. The CGM device was reinserted on the first day of the data collection week. The same procedures for device insertion and wearing instructions were followed during the second week of data collection. Daily standing time was recorded in the diary in addition to dietary intake, SMBG measurements, and physical activity.

Statistical Analyses

Data were analyzed with Microsoft Excel (2010 Microsoft Corp, Redmond, Washington) and SAS version 9.3 (SAS Institute Inc., Carey, North Carolina). A linear mixed model regression analysis for a crossover design was used in the SAS software to analyze differences in mean blood glucose values between the pretest and intervention weeks to account for daily variations in glucose metabolism throughout the day. Data collected during working hours (9 am to 5 pm) were analyzed to examine the effect of the sit-stand workstation on blood glucose. Dietary (ie, carbohydrate, fat, protein, and fiber intake in grams), physical activity, and sedentary time covariates were controlled for in this model. Results are reported as mean (SD or SE). P values less than .05 were considered significant.

Results

Of the 68 persons who expressed interest in the study (62 females and 6 males), 27 females and no males met the eligibility criteria. Data were collected from February through July, 2013. Of the 27 females who met the eligibility criteria, 17 did not complete the study: 4 were ineligible; 3 withdrew; 5 were not willing to comply with study requirements; and 5 experienced CGM device malfunction during the pretest, so the second test week was not completed. Participants had IFG and, on average, were overweight or obese along with 50% meeting criteria for the diagnosis of metabolic syndrome²⁴. Table 1 summarizes the baseline characteristics of the sample.

Table 1.

Participant Characteristics (N=10)

Variable	Mean (SD)
Age, y	50.6 (8.5)
Height, m	66.2 (2.8)
Weight, kg	91.9 (15.6)
BMI	32.9 (3.9)
Body fat %a	48.7 (4.0)
SBP	122.3 (11.0)
DBP	81.9 (7.5)
Fasting blood glucose, mg/dL	108.4 (5.6)
Total cholesterola, mg/dL	207.1 (37.6)
LDL-Ca, mg/dL	121.3 (40.0)
HDL-Ca, mg/dL	50.9 (14.8)
Triglycerides, mg/dL	176.6 (84.1)

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); DBP, diastolic blood pressure; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood pressure.

^an=8.

The self-reported standing time resulted in adherence to standing for approximately half the workday (mean [SD], 3.65 [0.4] hours). The results of the first linear mixed model analysis (controlling for the effect of the covariates carbohydrates, fat, protein, and fiber) indicated a nonsignificant effect of standing on blood glucose levels (P=.55) during working hours compared with the seated condition (Table 2). Although the P value was not significant, the sign of the coefficient is in line with the hypothesis that blood glucose would be lower when using the sit-stand desk compared with sitting. In this model, blood glucose during the pretest week was 2.5 mg/dL higher, on average, than during the intervention week. The Supplemental Figure shows mean blood glucose levels during the workday for participants who were sitting and for participants who were using a sit-stand workstation. Carbohydrates, fat, and protein (P<.001) showed a significant effect, but fiber did not (P=.099). Consequently, a 1-g increase in carbohydrates increased blood glucose by 0.08 mg/dL and a 1-g increase in protein increased blood glucose by 0.20 mg/dL, but a 1-g increase in fat decreased blood glucose by 0.19 mg/dL. The Supplemental Table shows the daily average for blood glucose, macronutrients, KAM points, and sedentary minutes.

Table 2.

Linear Mixed Model Analysis With Dietary Covariates

Effect	Estimate	SE	tdf	P Value
Intercept	81.66	3.49	23.4216	<.001
Pretest week	2.55	4.16	0.6116	.55
Intervention week	0
Carbohydrates	0.08	0.01	6.744,255	<.001
Fat	−0.19	0.02	−8.634,255	<.001
Protein	0.20	0.02	8.734,255	<.001
Fiber	0.12	0.07	1.654,255	.099

When the model was run to include the accelerometer data (KAM points and overall physical activity), the results were similar. The effect of the intervention increased slightly compared with the model in Table 2, although the difference was still not significant. In this model, overall physical activity was a significant predictor of blood glucose. With each 1-point increase in KAM points, blood glucose decreased by 0.06 mg/dL (P=.02).

When the statistical model was run to include sedentary minutes as a covariate (Table 3), the effect on the intervention was nearly identical to the previous model with overall physical activity. For every 1-minute increase in sedentary time, blood glucose increased by 0.11 mg/dL (P=.001).

Table 3.

Linear Mixed Model Analysis With Sedentary Time Covariate

Effect	Estimate	SE	tdf	P Value
Intercept	71.90	4.67	15.3814	<.001
Pretest week	3.12	5.31	0.5914	.57
Intervention week	0
Carbohydrates	0.10	0.01	7.323,596	<.001
Fat	−0.26	0.03	−8.763,596	<.001
Protein	0.23	0.03	9.033,596	<.001
Fiber	0.05	0.08	0.683,596	0.50
Sedentary minutes	0.11	0.03	3.403,596	<.001

The final model included both overall physical activity and sedentary time (Table 4). The treatment effect remained nonsignificant; however, the mean blood glucose was 3.1 mg/dL higher in the pretest week than in the intervention week. The difference between dietary covariates was statistically significant. However, with physical activity and sedentary time included in the same model, overall physical activity was no longer significantly predictive of the blood glucose level. Sedentary time was significantly related to increasing blood glucose, regardless of physical activity (P=.02).

Table 4.

Linear Mixed Model Analysis With Physical Activity and Sedentary Time Covariates

Effect	Estimate	SE	tdf	P Value
Intercept	71.50	5.16	13.8514	<.001
Pretest week	3.12	5.31	0.5914	.57
Intervention week	0
Carbohydrates	0.10	0.01	7.323,595	<.001
Fat	−0.26	0.03	−8.753,595	<.001
Protein	0.23	0.03	9.003,595	<.001
Fiber	0.05	0.08	0.673,595	.50
KAM points	0.01	0.04	0.183,595	.86
Sedentary minutes	0.11	0.05	2.433,595	.02

Abbreviation: KAM, Kinetic Activity Monitor

Discussion

The novel finding of the present study is that sedentary time was a significant predictor of blood glucose level (measured by CGM) independently of diet and physical activity. Additionally, the findings suggest that blood glucose is lower during the workday when prediabetic adults use a sit-stand workstation to increase standing time instead of sitting as usual during the workday. While this finding was not statistically significant in just one week, over a longer period of time this has the potential to result in marked improvements in blood glucose. Nonetheless, this study needs to be followed up with a larger cohort and more sophisticated modeling techniques given the tendencies in the data. The epidemiologic literature describes associations of increased sedentary time with risks of blood glucose dysfunction regardless of physical activity levels.^3,6,25 The present study objectively shows the deleterious effect of sedentary time on blood glucose level. In the final analysis model, sedentary time was the significant predictor of blood glucose level while overall physical activity was no longer a predictor of blood glucose level.

The overall mean of blood glucose measurements over 5 days may not be a good estimate of blood glucose, but the literature does not provide a standard or preferred method of analysis for CGM data; therefore, the present analysis was straightforward. The treatment effect is a comparison of the average blood glucose over 5 days, which considerably reduces the degrees of freedom for the effect estimate and considerably increases the SE magnitude. For this reason, even though the estimated difference in the mean daily blood glucose between conditions was 3.1 mg/dL, this effect was not statistically significant in the present models. Because the glucose level fluctuates dramatically throughout the day, this may not be the most appropriate analysis technique. Persons with IFG have more moderate excursions in blood glucose compared with persons who have type 1 or type 2 diabetes. These excursions may increase the difficulty of finding significant differences, especially when the mean is broadly estimated. A similar, albeit laboratory based study, by Thorp and colleagues found an effect of standing on postprandial glucose, however, no significant difference in fasting plasma glucose between the sitting and interrupted sitting conditions as we found as well.²⁶

Various levels of glucose intolerance among participants likely influenced the results as well. Glucose intolerance ranged from very mild (at the threshold criterion for impaired glucose tolerance) to more severe (near the diagnostic criteria for type 2 diabetes). This wide range further reduces the ability of the present analysis to detect significant differences. If the inclusion criteria threshold had been set at a more conservative 110 mg/dL, the sample would have been more metabolically homogeneous and perhaps the likelihood of detecting significant differences would have increased. Sophisticated analysis techniques are needed to explore these CGM data further. Indeed, modest reductions in blood glucose levels can be clinically meaningful in the treatment and management of prediabetes, and the prevention of frank diabetes. Regression to normal glucose regulation, even if transient, significantly reduces the risk of future diabetes.²⁷ Nevertheless, the blood glucose findings in the present pilot study will likely be useful for designing a full-scale study on this and related topics.

The dietary covariates were also significant predictors of blood glucose level. Increased intake of carbohydrates and protein significantly increased blood glucose; as expected, increased fat intake reduced blood glucose. Whereas the effects of carbohydrates and fats were as expected, the positive effect of protein (increased blood glucose) was unexpected because protein ingestion normally does not increase peripheral glucose.²⁸ Further examination of the dietary logs is needed to determine the type, quality, and quantity of protein consumed because these can greatly affect the rate of metabolism of protein or glucose (or both).^29,30 It is also likely that the protein results may be artifacts or inaccuracies of the dietary log data or analysis, and this will be explored in more detail in future analyses.

Strengths of the study were the novel use of CGM technology in women with prediabetes and the natural setting of the study. CGM technology provides a large amount of data for each person and provides valuable data on glucose control. The data can be used to identify problems, such as the postprandial period (2–3 hours after eating), and to develop interventions to reduce disease risk and progression. The study provided pilot data to guide future use of CGM in patients with impaired fasting glucose. To our knowledge, the present study is the first to continuously monitor blood glucose responses to intermittent standing among women with IFG and the first to collect data over an entire workweek. Additional strengths were the objectively measured physical activity and sedentary time. Few studies have measured these variables in a natural setting while monitoring blood glucose.

This study has several limitations. The inclusion criterion for fasting blood glucose was greater than 100 mg/dL. A higher threshold of greater than 110 mg/dL would result in a metabolically similar sample. Although 6 men volunteered for the study, they did not meet the inclusion criteria, so all the participants were women. This limitation is possibly 2-fold in that women are more likely to volunteer to participate in research and the worksites consisted primarily of female employees. The study recruited a relatively large sample size; however, several participants were lost to follow-up. Some participants dropped out because use of the CGM device involved an invasive procedure in addition to monitoring blood glucose daily, others were lost to follow-up because the monitor malfunctioned, and 5 participants had no downloadable data after the first week of the study (because the device was blinded, the malfunction was not known until downloading was attempted). The manufacturer determined that the data loss was due to bad sensors. Another limitation was self-reported dietary intake which may not be accurate due to recall bias and specific meals were not provided. Nonetheless, the participants were instructed to consume a similar diet during the weeks they wore the monitor.

Future studies should recruit more participants, minimize the number lost to follow-up, and recruit from several workplaces to increase the likelihood of recruiting male participants. The sample size of the present study is fairly consistent with similar studies in the literature, considering that this area of research is novel.^14,31,32 Future studies should examine the long-term effect of reducing sedentary time on blood glucose levels in participants with prediabetes. Lifestyle interventions (including weight loss, regular physical activity, and dietary modifications, such as calorie restriction, increased fiber intake, and limited carbohydrate intake) have prevented or delayed progression from prediabetes to type 2 diabetes.³³ Therefore, reducing sedentary time may be an additional component to consider in lifestyle interventions to prevent disease progression. Finally, the effect of the sit-stand desk on the postprandial period should be examined. Rather than analyzing the entire workday, reducing the analysis period to a specific time point could be used to examine the effect of the intervention during a specific time of the workday.

We found that sedentary time was a significant predictor of blood glucose levels during the workday among women with prediabetes independently of diet and physical activity. Research indicates that sit-stand desks are effective for reducing sedentary time and lowering postprandial blood glucose levels.^14,32 However, additional research is needed to determine whether sit-stand desks are directly effective tools for controlling blood glucose and preventing further disease progression in this population.

Clinical Significance.

The prevalence of impaired fasting glucose is high among adults, and many of them work more than 35 hours per week in highly sedentary jobs. The workplace provides a unique platform for intervention to reduce this cardiometabolic risk factor. A height-adjustable workstation is a strategy that may reduce sedentary time.

Abbreviations

CGM

continuous glucose monitoring

HbA_1c

hemoglobin A_1c

IFG

impaired fasting glucose

KAM

Kinetic Activity Monitor

SMBG

self-monitored blood glucose