Journal Logo


Associations of Physical Activity Intensities with Markers of Insulin Sensitivity


Author Information
Medicine & Science in Sports & Exercise: December 2017 - Volume 49 - Issue 12 - p 2451-2458
doi: 10.1249/MSS.0000000000001381


The increasing prevalence and associated social and economic burden of type 2 diabetes mellitus (T2DM) is widely acknowledged (39). The predicted impact this is due to have on public resources has made prevention of T2DM a priority nationally and internationally (33). In England, this has been demonstrated by the launch of the National Health Service (NHS) Diabetes Prevention Programme (32). A major focus of prevention programs implemented within routine care pathways is to encourage people to increase their levels of physical activity (PA), to improve their diet, and to lose weight (12).

Physical activity is a powerful therapeutic agent in the promotion of insulin sensitivity (7) and favorable blood plasma glucose levels (8). Evidence ranging from epidemiological data to randomized control trials consistently demonstrates strong associations of moderate to vigorous PA (MVPA) with cardiovascular health and glycemic control in individuals with T2DM (2,8). The globally accepted recommendation that all adults should engage in 150 min of moderate-intensity PA (MPA) or 75 min of vigorous-intensity PA (VPA) per week is therefore encouraged in people with and at high risk of T2DM (9).

Progression of objective measures of activity monitoring technology has facilitated research into PA programs encompassing the entire intensity spectrum, from replacing sedentary time with light-intensity PA (LPA) (21) to investigating the minimum volume of maximal-intensity PA required to improve health. However, accelerometer-measured PA intensity has generally been presented using population-dependent counts-per-epoch cut points categorized as sedentary, light, moderate, or vigorous PA, with MPA and VPA usually combined and analyzed as MVPA. These cut points have been calibrated against energy expenditure, confirming their “construct validity,” but the “external validity” such as the dose–response relationship between incremental accelerometer-measured PA intensity in counts per minute and health outcomes has been less rigorously tested.

The aim of this study was therefore to determine the association of time spent in objectively assessed incremental PA intensities with markers of insulin sensitivity, a precursor to T2DM, in free-living environments in individuals at high risk of T2DM. We hypothesized that time spent in higher intensities of PA would be associated with greater insulin sensitivity in a dose-dependent manner.


This study reports a cross-sectional analysis of baseline data from the Walking Away from T2DM randomized control trial, the methods for which have been published previously (38).


Participants (N = 833) were recruited from 10 primary care practices in Leicestershire, UK, during 2010–2011. Individuals who were identified as being at high risk (individuals scoring within the 90th percentile within each practice) of dysglycemia, impaired glucose tolerance (2-h glucose ≥7.8 and <11.1 mmol·L−1), and/or impaired fasting glycemia (fasting glucose ≥6.1 and <7.0 mmol·L−1) or undiagnosed T2DM, using the Leicester Practice Risk Score, were invited to take part. The Leicester Practice Risk Score has been shown to have good reliability in predicting prevalence of dysglycemia (16,18) and includes questions about anthropometry, ethnicity, family history, and antihypertensive therapy, each weighted based on epidemiological evidence. Participants were unaware of their diabetes risk before entering into the study and were excluded if they had known T2DM or were unable to take part in any walking activity. Baseline measurements were performed before treatment allocation by trained staff, who were blinded to study outcomes and followed standard operating procedures. Following this, practices were randomly assigned to either a control or an intervention arm. Participants in the control practices were given a lifestyle advice leaflet, and participants in the intervention practices were invited to take part in a pragmatic evidence-based structured education program designed to promote PA and a healthy lifestyle. The program involves attendance at a 3-h group education session with two annual follow-on maintenance sessions. The primary aim was to promote walking activity by targeting perceptions and knowledge of diabetes risk and PA self-efficacy as well as facilitating self-regulation such as goal setting and self-monitoring (using pedometers). All participants provided written informed consent, and ethical approval was obtained from a local NHS research ethics committee.

Objective measurement of PA

Participants were instructed to wear a waist-worn accelerometer (ActiGraph GT3X, Pensacola, FL) for seven consecutive days, removing for water-based activities and nonwaking hours. Data were recorded in 15-s epochs and reintegrated into 60-s epoch files for this analysis. Activity intensity was generated in increments of 500 counts per minute from 0 to 4499 counts per minute. Any counts recorded higher than 4500 counts per minute were grouped together to account for a lack of power at intensities above this. Using a regression equation from a similar population (17), we estimated the number of METs equated to the midpoint of these increments. These can be found in Table, Supplemental Digital Content 1, Estimation of number of METs each 500 counts per minute intensity increment equates to (

The number of minutes spent within each intensity band per day was calculated. For a day to be considered valid, at least 600 min of wear time had to be recorded; nonwear time was defined as more than 60 min of continuous zero counts (13). All accelerometer-derived variables were computed by summing the values over all valid days and calculating the mean value per valid day. Data were analyzed using a commercially available software package (KineSoft V3.3.76; Kinesoft, New Brunswick, Canada;

Demographic, anthropometric, and biochemical

On the participants’ first visit, information regarding ethnicity, smoking status, and antihypertensive medication were collected by a health care professional. Body weight (Tanita TBE 611; Tanita, West Drayton, UK), waist circumference (midway between the lower costal margin and the iliac crest), and height were measured to the nearest 0.1 kg and 0.5 cm, respectively.

A standard oral glucose tolerance test using a 75-g glucose load was administered to all participants. Individuals were asked to avoid caffeine and strenuous exercise in the preceding 24 h and to consume only water from 10:00 PM the evening before their visit. Fasting and 2-h postchallenge plasma glucose samples were measured using the glucose oxidase method (Beckman Auto Analyzer, Beckman, High Wycombe, UK) at the Leicester Royal Infirmary. Plasma samples were frozen at −80°C and analyzed for fasting and 2-h insulin at the end of baseline data collection using enzyme-linked immunoassay (80-INSHU-E01.1; Alpco Diagnostics, Salen, NH) within a specialist laboratory (Unilever R&D, Bedfordshire, UK). Analysis was conducted by individuals blinded to the patient’s identity and using stable methodologies, standardized to external quality assurance values.

Measures of insulin sensitivity

The homeostatic model assessment of insulin sensitivity (HOMA-IS) and Matsuda indices were used to estimate insulin sensitivity (27,30):

where G0 is the fasting plasma glucose, I0 is the fasting plasma insulin, G120 is the 2-h glucose, and I120 is the 2-h insulin. These models have been shown to correlate reasonably with gold standard measures of insulin sensitivity (31). Matsuda-IS is more likely to reflect factors related to insulin release and peripheral insulin resistance, whereas HOMA-IS may be a better measure of hepatic insulin resistance (19).

Data inclusion

Individuals at high risk of or with dysglycemia or previously unknown T2DM, a minimum of four valid days of accelerometer data and fasting glucose and insulin data to allow for an assessment of HOMA-IS, were included (28). Valid accelerometer data were available for 727 participants (87%). Cessation of bleeding or insufficient plasma volumes for the fasting insulin analysis meant that 569 (68%) participants met the inclusion criteria. Of these individuals, 508 participants (61%) also had complete 2-h glucose and 2-h insulin results. The number of participants included and excluded can be found in Figure, Supplemental Digital Content 2, Schematic representation of participants included and excluded, Those who were excluded from the HOMA-IS analysis (n = 178) tended to be younger (61.7 yr for missing vs 63.8, P < 0.001) and more likely to be female (42% vs 34%; P = 0.021). However, there was no difference in body mass index (BMI), fasting, or 2-h glucose.

Statistical analysis

Data were analyzed using Stata V.14 (Stata Statistical Software release 14; StataCorp, College Station, TX).

Log-linear regression was used to assess the association of PA intensity with fasting and 2-h glucose and insulin levels and insulin sensitivity (3). Dependent variables were log transformed as they displayed nonparametric distributions. Time spent in each of the PA intensity increments was entered into models separately because of the correlation between bands (see Table, Supplemental Digital Content 3, Time spent in each intensity band and the correlations between each band and total number of counts per day, Sensitivity analyses assessed whether the associations were modified by diagnosis of IGR, sex, or age (<65 or ≥65 yr). Adjustment was not made for multiple comparisons; therefore, data were viewed with caution and in relation to the overall pattern of results. Model 1 was adjusted for age, sex, ethnicity, smoking status, beta-blocker medication, and accelerometer wear time. To examine the extent to which adiposity may attenuate these relationships, further adjustment for BMI was made in model 2. Data were not adjusted for overall PA volume (counts per day) because of collinearity (see Table, Supplemental Digital Content 3, Time spent in each intensity band and the correlations between each band and total number of counts per day, Coefficients were back transformed and represent the factor by which the outcome is multiplied by (95% confidence interval [CI]) for a given unit of time spent at each intensity. Data in the text are presented as the percentage difference (95% CI) in the strength of the association between 10 min of PA and the outcome.


Characteristics of participants included in the analysis are displayed in Table 1.

Characteristics of included participants.

Physical activity

Time spent in each 500 counts per minute intensity banding is shown in Figure 1 and Table, Supplemental Digital Content 3, Time spent in each intensity band and the correlations between each band and total number of counts per day, Physical activity intensity ranged from 0 to 6000 counts per minute. Total PA volume (counts per day) tended to be correlated more strongly with time spent between 1000 and 3000 counts per minute, indicating most activity was undertaken at the lower end of the MPA range.

Number of minutes spent in each intensity band. Box plots indicate the median and interquartile range, minimum and maximum values.

Biochemical outcomes

The regression coefficients (95% CI) presented here and used to generate Figure 2 are displayed in Table, Supplemental Digital Content 4, Associations of PA intensity with markers of insulin sensitivity,

(Above panel A) Model 1 adjusted for age, sex, ethnicity, smoking status, and β-blocker medication and accelerometer wear time. (Above panel B) Model 2 adjusted for age, sex, ethnicity, smoking status, and β-blocker medication accelerometer wear time and BMI. A. Association between time spent within each PA intensity and fasting glucose levels (n = 569). B. Association between time spent within each PA intensity and fasting glucose levels (n = 569). C. Association between time spent within each PA intensity and 2-h glucose levels (n = 567). D. Association time spent within each between PA intensity and 2-h glucose levels (n = 567). E. Association between time spent within each PA intensity and fasting insulin levels (n = 569). F. Association between time spent within each PA intensity and fasting insulin levels (n = 569). G. Association between time spent within each PA intensity and 2 h insulin levels (n = 508). H. Association between time spent within each PA intensity and 2-h insulin levels (n = 508). I. Association between time spent within each PA intensity and HOMA-IS score (n = 569). J. Association between time spent within each PA intensity and HOMA-IS score (n = 569). K. Association between time spent within each PA intensity and Matsuda-IS score (n = 508). L. Association between PA time spent within each intensity and Matsuda-IS score (n = 508). (Below panels K and L) Percentage change in biochemical outcomes associated with a 10-min increase in time spent in bands of 500 counts per minute of PA intensities ranging from 0 to ≥4500 counts per minute. Coefficients are plotted at the midpoint of the intensity band. Dotted lines represent commonly used accelerometer cut points for light (100 counts per minute) and moderate PA (1952 counts per minute) (14). HOMA-IS, homeostatic model assessment of insulin sensitivity; ISI, insulin sensitivity index.

Fasting and 2-h glucose

Physical activity intensity was not associated with fasting glucose levels (Fig. 2A and 2B). Time spent in PA intensities less than 500 counts per minute were associated with higher 2-h glucose levels (0.91%, 95% CI = 0.49%–1.33%). Time spent in PA at intensities between 500 and 2499 counts per minute was associated with a linear change in the strength of the association with 2-h glucose (Fig. 2C). After this level, there was no clear increase in the strength of association, although the error around each regression coefficient increased. This relationship was attenuated slightly after adjusting for BMI, with associations observed up to 2499 counts per minute (−3.98%, 95% CI = −7.79% to −0.01%; Fig. 2D).

Fasting and 2-h insulin

Every 10 min spent at intensities lower than 500 counts per minute was associated with 1.66% (95% CI = 0.86%–2.47%) higher fasting insulin levels. Time spent in intensities of 500 counts per minute and above were associated with lower fasting insulin for each increment of 500 to 3999 counts per minute, ranging from −2.67% (−4.46% to −0.85%) to −20.47 (−29.84% to −7.60%) per 10 min of PA. Above this, the difference in the association remained similar (Fig. 2E). Adjusting for BMI largely attenuated the results (Fig. 2F).

Time spent in PA intensities less than 500 counts per minute was associated with 2.96% (95% CI = 1.80%–4.14%) higher 2-h insulin levels per 10 min of PA, an association also observed after controlling for BMI (2.61%, 1.41%–3.82%). Per 10 min of PA spent in each increment of 500 counts per minute in PA intensity between 500 and 3999 counts per minute, the difference in 2-h insulin changed from −5.00% (−7.52% to −2.42%) to −25.38% (−37.86% to −10.40%; Fig. 2G). The time spent in intensities of 4000 counts per minute and higher was still significantly associated with 2-h insulin levels, but there was no further rise in the strength of the association. The relationship when controlling for BMI was attenuated but still significant and followed a similar pattern (Fig. 2H).

Insulin sensitivity

The time spent less than 500 counts per minute PA intensity was associated with lower scores of both HOMA and Matsuda measures of insulin sensitivity (Fig. 2I and 2K). Times spent in PA intensities of 500 counts per minute and above were linearly associated with differences in HOMA-IS from 2.64% (0.64%–4.68%) to 26.75% (10.99%–44.74%) per 10 min of PA up to 3999 counts per minute. Similarly, differences in the association with Matsuda-IS ranged from 5.32% (2.76%–7.94%) to 34.66% (13.85%–59.26%) per 10 min of PA. Associations with time spent in intensities higher than 4000 counts per minute were statistically significant but of a smaller magnitude than those between 3000 and 3999 counts per minute.

After controlling for BMI, associations with HOMA-IR were largely attenuated (Fig. 2J). However, the relationship with Matsuda was maintained with differences between 500 and 999 counts per minute and between 3500 and 3999 counts per minute 3.91% (1.38%–6.50%) and 23.02% (4.05%–45.46%), respectively (Fig. 2L).

Sensitivity analyses

Subgroup analysis revealed that there were no differences in any of the outcomes when the data were stratified by IGR status, except for the relationship between PA intensities between 500 and 999 counts per minute and fasting glucose (P = 0.037). This analysis indicated that the relationship between PA and fasting glucose was stronger in those with IGR. Fasting glucose levels were 0.7% (0.1%–1.2%) lower per 10 min of PA than those with normal glucose tolerance; 0.1% (0.0%–0.3%). There were no sex or age–intensity interactions.


In this cohort of individuals identified as being at high risk of IGR or with undiagnosed T2DM, time spent in increments of 500 counts per minute of objectively assessed PA intensity were linearly associated with lower levels of fasting and 2-h insulin and 2-h glucose and higher scores of indexes of insulin sensitivity. Differences in insulin sensitivity per 10 min of PA time increased sharply when moving up in bands of 500 counts per minute from 500 to 3999 counts per minute; for example, 10 min of PA between 500 and 999 counts per minute (approximately 2.6 METs [17]) was associated with a 5.32% difference in insulin sensitivity (Matsuda-IS), whereas 10 min spent between 3500 and 3999 counts per minute (approximately 4.3 METs) was associated with a 34.66% difference in insulin sensitivity. There did not appear to be any additional change in the strength of the association for time spent at intensities higher than 4000 counts per minute (approximately 4.6 METs). Results for 2-h glucose and 2-h insulin and Matsuda-IS remained largely unchanged after adjustment for BMI, whereas fasting measures were attenuated. Based on the accelerometer cut points validated by Freedson et al. (14), our results suggest that, in individuals at risk of or with dysglycemia, both LPA and MPA are sufficient to gain improvements in insulin sensitivity; the strength of which is proportional to time spent in PA intensities up to 3999 counts per minute (approximately 4.3 METs), above which no further strength in the association was gained.

It has long been accepted that there is a dose–response relationship between PA volume (broadly defined as intensity–duration interaction) and health up to a certain level, with the greatest benefits occurring when moving from a sedentary inactive lifestyle (1). It shows that the strength of association between time spent in PA and insulin sensitivity increases proportionally to the level of intensity up to a point, which is consistent with this theory. Interestingly, we found no additional benefit in this cohort for time spent higher than 4000 counts per minute, which equates to approximately 4.5 METs (17). This is in contrast to studies that that have shown additional benefits of performing VPA compared with MPA on markers of insulin sensitivity (22,23). Our observation that there was no additional benefit in performing PA at the upper end of the MPA range could be attributed to a lack of power in these intensity bands resulting from the progressively smaller amount of time spent in each intensity increment. Another important finding from our study was that undertaking any activity above the sedentary threshold (0–499 counts per minute) was associated with some degree of greater insulin sensitivity, whereas increased time spent sedentary was associated with worse insulin sensitivity. This finding is consistent with the growing body of evidence showing that greater time spent sedentary is independently associated with the development of T2DM and other morbidity and mortality outcomes (6,37), whereas greater time spent in LPA is associated with a reduced risk of mortality, particularly in older adults. Our findings are also consistent with another recent study, which showed that while higher levels of PA volume are associated with the greatest risk of cardiovascular disease, engaging in any intensity of PA or walking short distances are still associated with a reduced risk (35). Controlled interventions have also demonstrated that breaking sitting time with short bouts of standing and LPA and produces benefits in glucose regulation and insulin sensitivity, particularly in higher risk adults (11). Results from training intervention studies have been more equivocal with some demonstrating that low-intensity exercise training leads to reductions in HOMA insulin resistance (25), whereas others have not (34). Nevertheless, taken together, these studies support the hypothesis that engaging in PA that are below the intensity or duration thresholds of the current guidelines are associated with health benefits, particularly in older adults or those with a high risk of chronic disease, and that PA prescriptions need to be tailored to individual characteristics (36).

Our study also has implications for traditional accelerometer intensity thresholds. The cut points developed for ActiGraph accelerometers by Freedson et al. (14) are widely used and have been well validated (20). They are based on calibration with oxygen consumption during treadmill walking/running and equate to 1.5, 3, and 6 METs for LPA, MPA, and VPA respectively. Frequently used cut points are set at <100 counts per minute for sedentary behavior (29), 1952–5724 counts per minute for MPA, and ≥5725 counts per minute for VPA. Virtually none of the participants from our cohort, who were on average 40 yr older than those who were used to calibrate these cut points, performed PA at more than 5725 counts per minute, and differences in insulin sensitivity per 10 min of activity within the MPA range (1952–5724 counts per minute) varied from 12.37% to 26.75% and from 21.98% to 34.66% for HOMA and Matsuda, respectively. This indicates that large differences in health outcomes could be expected from participation within the moderate range. In line with this, accelerometer cut points specifically for older individuals, which may reflect more accurately associations between PA intensity and health in this population, have recently been investigated (15). Narrower cut point ranges at lower intensities have been proposed. Our results suggest that the lower intensity bands proposed by the Generation 100 study (40), which are dependent on sex and cardiorespiratory fitness, could be used to more precisely assess PA levels and their associations with health in older people.

There are several well-established mechanisms explaining the link between MVPA and improved metabolic health and insulin sensitivity, all of which are based on relative as opposed to absolute intensities. Skeletal muscle is the primary tissue responsible for glucose disposal (10) and affects insulin signaling in response to both acute and chronic bouts of PA. Assuming the PA levels observed here reflect regular behavior patterns, we speculate that the higher the intensity of PA engaged, the more superior the skeletal muscle adaptations that increase insulin sensitivity and drive lower circulating insulin levels (24). Even very small amounts of VPA may be important as they can offset the deleterious effects sedentary time itself has on insulin action (11). Recently, mechanisms elucidating how LPA may affect health have also been proposed including stimulation of the contraction-mediated glucose uptake pathway as well as inducements of alterations to the insulin signaling pathway (4).

To the best of our knowledge, only one other study to date assesses PA using incremental accelerometer count per minute cut point categories (5). The traditional method of reducing data into time spent in these relatively broad categories removes the ability to more accurately define the dose–response relationship with markers of health. Our study extends the findings of Berkemeyer et al. (5) by quantifying associations between participation in PA and health, as well as bringing together recent findings that reducing sedentary time and increasing LPA is also beneficial. Processing accelerometer data in increments of 500 counts per minute intensity covers the whole range of intensities, enabling identification of the minimum intensity at which benefits occur as well as a quantifiable dose–response relationship; all information that will facilitate the development of more achievable interventions with greater improvements. Another major strength of our study is the objective measurement of PA in a free-living environment, which is likely to be a more accurate representation of day-to-day behavior than would be demonstrated by a self-report tool. The deployment of a widely used device (ActiGraph GT3X) means the results are directly comparable to other data sets. In addition, the study population were at high-risk of T2DM and were recruited from primary care. Typically, this is a group that is hard to reach in lifestyle research despite representing the type of population referred to diabetes prevention programs. We also present a comprehensive selection of diabetes-related biochemical outcomes that were measured using robust laboratory techniques.

Limitations include the cross-sectional nature of the analysis, meaning we are unable to draw causal inferences. Although more accurate than self-report, accelerometers may underestimate overall PA because they are unable to accurately quantify nonstep based or weight bearing activities such as swimming or cycling. Moreover, because of individual differences in fitness levels, which were not measured in the original study, performing activity at the same number of counts per minute may reflect different relative intensities within our population (40). It should also be noted that although statistically significant differences in markers of insulin sensitivity and glucose regulation were observed, how these translate to the prevention and management of T2DM remains unclear. However, it is well established that those with greater insulin sensitivity are less likely to develop T2DM (26). Finally, as we were not able to adjust for PA volume because of multiple collinearity, we were unable to establish whether results for higher intensities of PA were simply due to greater PA volumes.

In conclusion, our study provides additional evidence that, for older individuals at high risk of T2DM, any PA may produce tangible benefits, but up to 3999 counts per minute (approximately 4.3 METs), the higher the intensity, the greater the potential improvement for a given PA duration. The nature of this study does not allow us to confirm a causational effect; nevertheless, these data stimulate an interesting line of future research investigating the accelerometer-measured intensity at which health outcomes are affected in different populations. As has been strongly advocated in recent years (36), this could then lead to specific targets being set for specific diseases or risk factors, which may increase participation in those who would benefit most for increasing PA because of having more realistic goals and expectations.

The analysis reported in this article was supported by the National Institute for Health Research Collaboration for Leadership in Applied Health Research and Care–East Midlands (NIHR CLAHRC-EM), the Leicester Clinical Trials Unit, and the National Institute for Health Research (NIHR) Leicester Biomedical Research Centre. The authors thank Duncan Talbot of Unilever (Unilever R&D, Bedfordshire, UK) for his help analyzing the insulin samples. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health.

The Walking Away trial was funded by the NIHR CLAHRC for Leicestershire, Northamptonshire, and Rutland. Unilever (Unilever R&D, Bedfordshire, UK) provided the facilities and equipment required to analyze the insulin samples. The authors declare that there is no conflict of interest associated with this manuscript. The results of the present study do not constitute endorsement by the American College of Sports Medicine.

A. R. and T. Y. had the original idea for the secondary analysis. C. E. processed the accelerometry data. C. J. performed the statistical analysis and prepared the first draft of the article. All authors contributed to the development and writing of the article, revised for important intellectual content, and approved the final manuscript.


1. Arem H, Moore SC, Patel A, et al. Leisure time physical activity and mortality a detailed pooled analysis of the dose–response relationship. JAMA Inten Med. 2015;175(6):959–67.
2. Bassuk SS, Manson JE. Epidemiological evidence for the role of physical activity in reducing risk of type 2 diabetes and cardiovascular disease. J Appl Physiol. 2005;99(3):1193–204.
3. Benoit K. [Internet]. London School of Economincs; [cited 2017 Feb 16]. Available from:
4. Bergouignan A, Latouche C, Heywood S, et al. Frequent interruptions of sedentary time modulates contraction- and insulin-stimulated glucose uptake pathways in muscle: ancillary analysis from randomized clinical trials. Sci Rep. [Internet]. 2016 [cited 2017 Feb 16];6. Available from:
5. Berkemeyer K, Wijndaele K, White T, et al. The descriptive epidemiology of accelerometer-measured physical activity in older adults. Int J Behav Nutr Phys Act. 2016;13(2).
6. Borghouts LB, Keizer HA. Exercise and insulin sensitivity: a review. Int J Sports Med. 2000;21(1):1–12.
7. Boule NG, Kenny GP, Haddad E, Wells GA, Sigal RJ. Meta-analysis of the effect of structured exercise training on cardiorespiratory fitness in Type 2 diabetes mellitus. Diabetologia. 2003;46(8):1071–81.
8. Boule NG, Weisnagel SJ, Lakka TA, et al. Effects of exercise training on glucose homeostasis. Diabetes Care. 2005;28(1):108–14.
9. Colberg SR, Sigal RJ, Yardley JE, et al. Physical activity/exercise and diabetes: a position statement of the American Diabetes Association. Diabetes Care. 2016;39(11):2065–79.
10. Defronzo RA. The triumvariate—beta-cell, muscle, liver—a collusion responsible for NIDDM. Diabetes. 1988;37(6):667–87.
11. Dempsey PC, Owen N, Yates TE, Kingwell BA, Dunstan DW. Sitting less and moving more: improved glycaemic control for type 2 diabetes prevention and management. Curr Diab Rep. 2016;16(11):114.
12. Dunkley AJ, Bodicoat DH, Greaves CJ, et al. Diabetes prevention in the real world: effectiveness of pragmatic lifestyle interventions for the prevention of type 2 diabetes and of the impact of adherence to guideline recommendations: a systematic review and meta-analysis. Diabetes Care. 2014;37(4):922–33.
13. Evenson KR, Terry JW. Assessment of differing definitions of accelerometer nonwear time. Res Quart Exerc Sport. 2009;80(2):355–62.
14. Freedson PS, Melanson E, Sirard J. Calibration of the computer science and applications, inc. accelerometer. Med Sci Sports Exerc. 1998;30(5):777–81.
15. Gorman E, Hanson HM, Yang PH, Khan KM, Liu-Ambrose T, Ashe MC. Accelerometry analysis of physical activity and sedentary behavior in older adults: a systematic review and data analysis. Eur Rev Aging Phys Act. 2014;11(1):35–49.
16. Gray L, Davies M, Hiles S, et al. Detection of impaired glucose regulation and/or type 2 diabetes mellitus, using primary care electronic data, in a multiethnic UK community setting. Diabetologia. 2012;55(4):959–66.
17. Hall KS, Howe CA, Rana SR, Martin CL, Morey MC. METs and accelerometry of walking in older adults: standard versus measured energy cost. Med Sci Sports Exerc. 2013;45(3):574–82.
18. Hansen AL, Carstensen B, Helge JW, et al. Combined heart rate- and accelerometer-assessed physical activity energy expenditure and associations with glucose homeostasis markers in a population at high risk of developing diabetes the ADDITION-PRO study. Diabetes Care. 2013;36(10):3062–9.
19. Hawley JA, Lessard SJ. Exercise training-induced improvements in insulin action. Acta Physiologica. 2008;192(1):127–35.
20. Hendelman D, Miller K, Baggett C, Debold E, Freedson P. Validity of accelerometry for the assessment of moderate intensity physical activity in the field. Med Sci Sports Exerc. 2000;32(9 Suppl):S442–9.
21. Henson J, Davies MJ, Bodicoat DH, et al. Breaking up prolonged sitting with standing or walking attenuates the postprandial metabolic response in postmenopausal women: a randomized acute study. Diabetes Care. 2016;39(1):130–8.
22. Howard B, Winkler EA, Sethi P, et al. Associations of low- and high-intensity light activity with cardiometabolic biomarkers. Med Sci Sports Exerc. 2015;47(10):2093–101.
23. Jelleyman C, Yates T, O’Donovan G, et al. The effects of high-intensity interval training on glucose regulation and insulin resistance: a meta-analysis. Ob Rev. 2015;16(11):942–61.
24. Kahn SE, Hull RL, Utzschneider KM. Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature. 2006;444(7121):840–6.
25. Kodama S, Shu M, Saito K, et al. Even low-intensity and low-volume exercise training may improve insulin resistance in the elderly. Intern Med. 2007;46(14):1071–7.
26. Martin B, Warram J, Krolewski A, Soeldner J, Kahn C, Bergman R. Role of glucose and insulin resistance in development of type 2 diabetes mellitus: results of a 25-year follow-up study. Lancet. 1992;340(8825):925–9.
27. Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care. 1999;22(9):1462–70.
28. Matthews CE, Ainsworth BE, Thompson RW, Bassett DR Jr. Sources of variance in daily physical activity levels as measured by an accelerometer. Med Sci Sports Exerc. 2002;34(8):1376–81.
29. Matthews CE, Chen KY, Freedson PS, et al. Amount of time spent in sedentary behaviors in the United States, 2003–2004. Am J Epidemiol. 2008;167(7):875–81.
30. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia. 1985;28(7):412–9.
31. Otten J, Ahrén B, Olsson T. Surrogate measures of insulin sensitivity vs the hyperinsulinaemic–euglycaemic clamp: a meta-analysis. Are there not some surrogate indexes lost in this story? Reply to Bastard JP, Rabasa-Lhoret R, Laville M and Disse E [letter]. Diabetologia. 2015;58(2):416–7.
32. Public Health England. National Procurement for the Provision of Behavioural Interventions for People with Non-Diabetic Hyperglycaemia. London (England): Available from: NHS England, Department of Health; 2015.
33. Schwarz PE, Greaves CJ, Lindstrom J, Yates T, Davies MJ. Nonpharmacological interventions for the prevention of type 2 diabetes mellitus. Nat Rev Endocrinol. 2012;8(6):363–73.
34. Segerstrom AB, Glans F, Eriksson KF, et al. Impact of exercise intensity and duration on insulin sensitivity in women with T2D. Eur J Int Med. 2010;21(5):404–8.
35. Soares-Miranda L, Siscovick DS, Psaty BM, Longstreth WT, Mozaffarian D. Physical activity and risk of coronary heart disease and stroke in older adults: the cardiovascular health study. Circulation. 2016;133(2):147–55.
36. Warburton DE, Bredin SS. Reflections on physical activity and health: what should we recommend?. Can J Cardiol. 2016;32(4):495–504.
37. Wilmot EG, Edwardson CL, Achana FA, et al. Sedentary time in adults and the association with diabetes, cardiovascular disease and death: systematic review and meta-analysis. Diabetologia. 2012;55(11):2895–905.
38. Yates T, Henson J, Edwardson C, et al. Objectively measured sedentary time and associations with insulin sensitivity: importance of reallocating sedentary time to physical activity. Prev Med. 2015;76:79–83.
39. Zimmet P, Alberti KG, Shaw J. Global and societal implications of the diabetes epidemic. Nature. 2001;414(6865):782–7.
40. Zisko N, Carlsen T, Salvesen O, et al. New relative intensity ambulatory accelerometer thresholds for elderly men and women: the Generation 100 study. BMC Geriatr. 2015;15:97.


Supplemental Digital Content

© 2017 American College of Sports Medicine