Advertisement
Mayo Clinic Proceedings Home

Dose-Response Effects of Exercise on Glucose-Lowering Medications for Type 2 Diabetes: A Secondary Analysis of a Randomized Clinical Trial

Open AccessPublished:January 30, 2020DOI:https://doi.org/10.1016/j.mayocp.2019.09.005

      Abstract

      Objective

      To investigate whether a dose-response relationship exists between volume of exercise and discontinuation of glucose-lowering medication treatment in patients with type 2 diabetes.

      Patients and Methods

      Secondary analyses of a randomized controlled exercise-based lifestyle intervention trial (April 29, 2015 to August 17, 2016). Patients with non–insulin-dependent type 2 diabetes were randomly assigned to an intensive lifestyle intervention (U-TURN) or standard-care group. Both groups received lifestyle advice and objective target-driven medical regulation. Additionally, the U-TURN group received supervised exercise and individualized dietary counseling. Of the 98 randomly assigned participants, 92 were included in the analysis (U-TURN, n=61, standard care, n=31). Participants in the U-TURN group were stratified into tertiles based on accumulated volumes of exercise completed during the 1-year intervention.

      Results

      Median exercise levels of 178 (interquartile range [IQR], 121-213; lower tertile), 296 (IQR, 261-310; intermediate tertile), and 380 minutes per week (IQR, 355-446; upper tertile) were associated with higher odds of discontinuing treatment with glucose-lowering medication, with corresponding odds ratios of 12.1 (95% CI, 1.2-119; number needed to treat: 4), 30.2 (95% CI, 2.9-318.5; 3), and 34.4 (95% CI, 4.1-290.1; 2), respectively, when comparing with standard care. Cardiovascular risk factors such as glycated hemoglobin A1c levels, fitness, 2-hour glucose levels, and triglyceride levels were improved significantly in the intermediate and upper tertiles, but not the lower tertile, compared with the standard-care group.

      Conclusion

      Exercise volume is associated with discontinuation of glucose-lowering medication treatment in a dose-dependent manner, as are important cardiovascular risk factors in well-treated participants with type 2 diabetes and disease duration less than 10 years. Further studies are needed to support these findings.

      Study Registration

      Abbreviations and Acronyms:

      BMI (body mass index), HbA1c (glycated hemoglobin A1c), IDES (Italian Diabetes and Exercise Study), IQR (interquartile range), T2D (type 2 diabetes)
      Current treatment regimens for type 2 diabetes (T2D) include diet, physical activity, and pharmacologic therapy.
      American Diabetes Association
      6. Obesity management for the treatment of type 2 diabetes.
      The latter is associated with lowered quality of life,
      • Huang E.S.
      • Brown S.E.
      • Ewigman B.G.
      • Foley E.C.
      • Meltzer D.O.
      Patient perceptions of quality of life with diabetes-related complications and treatments.
      interference with the ability to live a normal life,
      • Nicolucci A.
      • Kovacs Burns K.
      • Holt R.I.
      • et al.
      DAWN2 Study Group
      Diabetes Attitudes, Wishes and Needs second study (DAWN2): cross-national benchmarking of diabetes-related psychosocial outcomes for people with diabetes.
      and increased economic costs.
      • Espeland M.A.
      • Glick H.A.
      • Bertoni A.
      • et al.
      Look AHEAD Research Group
      Impact of an intensive lifestyle intervention on use and cost of medical services among overweight and obese adults with type 2 diabetes: the Action for Health in Diabetes.
      Nonsurgical approaches such as lifestyle interventions have had limited efficacy in reducing the need for glucose-lowering medications and glycemic control.
      • Gregg E.W.
      • Chen H.
      • Wagenknecht L.E.
      • et al.
      Look AHEAD Research Group
      Association of an intensive lifestyle intervention with remission of type 2 diabetes.
      ,
      • Lean M.E.
      • Leslie W.S.
      • Barnes A.C.
      • et al.
      Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial.
      Exercise volumes exceeding 150 minutes of exercise per week have been shown to induce superior improvements in glycated hemoglobin A1c (HbA1c) levels.
      • Di Loreto C.
      • Fanelli C.
      • Lucidi P.
      • et al.
      Make your diabetic patients walk: long-term impact of different amounts of physical activity on type 2 diabetes.
      ,
      • Umpierre D.
      • Ribeiro P.A.
      • Kramer C.K.
      • et al.
      Physical activity advice only or structured exercise training and association with HbA1c levels in type 2 diabetes: a systematic review and meta-analysis.
      However, a clear consensus of the existence of a dose-response relationship in relation to exercise volume is currently not established in patients with T2D.
      • Di Loreto C.
      • Fanelli C.
      • Lucidi P.
      • et al.
      Make your diabetic patients walk: long-term impact of different amounts of physical activity on type 2 diabetes.
      • Umpierre D.
      • Ribeiro P.A.
      • Kramer C.K.
      • et al.
      Physical activity advice only or structured exercise training and association with HbA1c levels in type 2 diabetes: a systematic review and meta-analysis.
      • Balducci S.
      • Zanuso S.
      • Nicolucci A.
      • et al.
      Italian Diabetes Exercise Study (IDES) Investigators
      Effect of an intensive exercise intervention strategy on modifiable cardiovascular risk factors in subjects with type 2 diabetes mellitus: a randomized controlled trial: the Italian Diabetes and Exercise Study (IDES).
      • Vancea D.M.
      • Vancea J.N.
      • Pires M.I.
      • Reis M.A.
      • Moura R.B.
      • Dib S.A.
      Effect of frequency of physical exercise on glycemic control and body composition in type 2 diabetic patients.
      • Boule N.G.
      • Kenny G.P.
      • Haddad E.
      • Wells G.A.
      • Sigal R.J.
      Meta-analysis of the effect of structured exercise training on cardiorespiratory fitness in type 2 diabetes mellitus.
      • Umpierre D.
      • Ribeiro P.A.
      • Schaan B.D.
      • Ribeiro J.P.
      Volume of supervised exercise training impacts glycaemic control in patients with type 2 diabetes: a systematic review with meta-regression analysis.
      Most studies to date have used exercise levels at or below current recommendations and therefore the limited efficacy seen in these studies may be a result of insufficient exposure to exercise. In addition, the clinical relevance of existing studies is limited because studies have largely relied on self-reported exercise levels, which are known to be sufficiently inaccurate to risk misclassification.
      • Prince S.A.
      • Adamo K.B.
      • Hamel M.E.
      • Hardt J.
      • Connor Gorber S.
      • Tremblay M.
      A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review.
      Further studies are needed to replicate the nature of any exercise dose-response relationship, if it exists, between objectively measured exercise and glucose metabolism and other cardiovascular risk factors in patients with T2D.
      Recently we reported that 56% (36 of 62) of participants receiving a 12-month intensive lifestyle intervention, which prescribed exercise volumes exceeding current recommendations, achieved discontinuation of treatment with glucose-lowering medications while maintaining optimal HbA1c levels.
      • Johansen M.Y.
      • MacDonald C.S.
      • Hansen K.B.
      • et al.
      Effect of an intensive lifestyle intervention on glycemic control in patients with type 2 diabetes: a randomized clinical trial.
      A unique feature of the study was that all completed exercise was objectively measured throughout the entire intervention. However, the effect of varying exercise volumes on the need for glucose-lowering medications and effects on HbA1c levels and cardiovascular risk factors were not explored.
      We test the hypothesis that a positive association exists between increasing volumes of exercise and the odds of discontinuing glucose-lowering medication treatment and changes in HbA1c levels in patients with T2D diagnosed within 10 years.

      Patients and Methods

      Research Design

      The original trial is registered at ClinicalTrials.gov (NCT02417012). Full details of the original trial protocol are published,
      • Ried-Larsen M.
      • Christensen R.
      • Hansen K.B.
      • et al.
      Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.
      and the prespecified Supplementary Statistical Analysis Plan for these particular secondary analyses is available online at http://www.mayoclinicproceedings.org.

      Study Design

      The original study was a randomized (2:1), single-center, assessor-blinded, 2-arm, parallel-group trial designed to test whether an intensive lifestyle intervention was equally effective in maintaining glycemic control compared with standard care in patients with T2D, conducted in the Capital Region of Denmark (April 29, 2015 to August 17, 2016).
      • Ried-Larsen M.
      • Christensen R.
      • Hansen K.B.
      • et al.
      Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.
      The primary results were published in 2017.
      • Johansen M.Y.
      • MacDonald C.S.
      • Hansen K.B.
      • et al.
      Effect of an intensive lifestyle intervention on glycemic control in patients with type 2 diabetes: a randomized clinical trial.
      Guidelines from the Declaration of Helsinki were followed, and the study was approved by the Scientific Ethical Committee at the Capital Region of Denmark. All participants provided oral and written informed consent.
      In this extension study, participants in the U-TURN group were divided into tertiles (lower, intermediate, and upper; described in the Supplementary Statistical Analysis Plan) based on the objective assessment of the accumulated volume of aerobic and resistance exercise completed from baseline to the 12-month follow-up.

      Participants and Eligibility

      Ninety-eight participants were recruited for the original study. Details about randomization and allocation procedures and sample size considerations have been published elsewhere.
      • Ried-Larsen M.
      • Christensen R.
      • Hansen K.B.
      • et al.
      Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.
      Inclusion criteria were a diagnosis of T2D within the last 10 years, 18 years or older, body mass index (BMI; calculated as the weight in kilograms divided by the height in meters squared) of 25 or greater but 40 or less kg/m2, and without severe comorbid conditions, insulin use, or HbA1c level greater than 9%. Only participants attending the 12-month follow-up assessment and with registered exercise on the Polar watch (the latter only applied to the U-TURN group) were included in this secondary analysis.

      Interventions

      The intensive lifestyle intervention for the U-TURN group has previously been described in detail.
      • Ried-Larsen M.
      • Christensen R.
      • Hansen K.B.
      • et al.
      Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.
      All participants received standard care consisting of medical counseling, lifestyle advice, and education in T2D provided by a study nurse at baseline and every third month during the intervention period.
      Prespecified algorithms for glucose-lowering medications were followed during the 12-month intervention in both groups.
      • Ried-Larsen M.
      • Christensen R.
      • Hansen K.B.
      • et al.
      Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.
      The treatment target for glycemic control was HbA1c level of 6.5%. If this target was reached at a medical consultation, the glucose-lowering medication dose was halved. In the case of unchanged values or if an additional reduction in HbA1c level was observed at the following medical consultation, the treatment with medication was discontinued. The endocrinologist’s regulation of all medication was blinded to group allocation.
      In addition to standard care, the U-TURN group was prescribed a high-volume exercise intervention aiming for at least 240 minutes of aerobic and resistance exercise per week in phase 1 (first 4 months) and at least 300 minutes of aerobic and resistance exercise per week in phases 2 and 3 (last 8 months) with concomitant dietary counseling.

      Outcomes and Measurements

      The primary outcome was discontinuation according to the blinded target-driven algorithm of glucose-lowering medication treatment at the 12-month follow-up.
      The mean difference for change in HbA1c level from baseline to the 12-month follow-up between the exercise tertiles and standard care is a key secondary outcome. Additionally, we explored the dose-response relationship of exercise volume and body composition measures, fasting insulin level, fasting glucose and 2-hour glucose levels, fitness, energy intake, blood lipid levels, and lipid-lowering medication, blood pressure, and blood pressure–lowering medication.
      Exercise volume and intensity measurement methods were described in the prepublished protocol.
      • Ried-Larsen M.
      • Christensen R.
      • Hansen K.B.
      • et al.
      Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.
      All exercise and physical activity data were registered through a Polar V800 watch in the U-TURN group. Participants in the U-TURN group were instructed to wear the watch 24 hours a day every day of the intervention. During all exercise sessions, participants used the heart rate monitoring chest band to ensure accurate registration of exercise intensity.
      Study participants were divided into tertiles based on the combined volume of aerobic and resistance exercise accumulated during the 12 months of the intervention. Bouts less than 10 minutes and aerobic exercise with a registered intensity of less than 57% of maximum heart rate, corresponding to light to very light intensity,
      • Garber C.E.
      • Blissmer B.
      • Deschenes M.R.
      • et al.
      American College of Sports Medicine
      American College of Sports Medicine position stand. Quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: guidance for prescribing exercise.
      were not categorized as exercise and were excluded from the analysis. An elaborate description is found in the Supplementary Statistical Analysis Plan.
      Assessments were performed in one laboratory and biochemical analyses were completed at the central laboratory (Rigshospitalet, Denmark) using standard procedures.
      • Ried-Larsen M.
      • Christensen R.
      • Hansen K.B.
      • et al.
      Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.

      Statistical Analyses

      The original analysis plan specified a repeated-measures mixed model. However, because the model did not converge, we used the simpler model without repeated measures.
      • Laptook A.R.
      • Shankaran S.
      • Tyson J.E.
      • et al.
      Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network
      Effect of therapeutic hypothermia initiated after 6 hours of age on death or disability among newborns with hypoxic-ischemic encephalopathy: a randomized clinical trial.
      The primary analysis was based on the as-observed population. To account for observed confounding, all participants were assigned a propensity of being allocated to an exercise tertile. Because the original design included a standard-care group, participants in standard care were per default assigned a propensity of 0 because they did not exercise; that is, this group will continue to be a randomized control group. The propensity of belonging to a tertile was data driven, based on the baseline variables (prerandomization [ie, not influenced by post hoc elements]); sex and high-density lipoprotein cholesterol level at baseline. The propensity score was computed as a single continuous variable, ranging from 0 to 1 in the primary analyses. Variables to be associated with a high or low level of physical activity were used as a proxy for the propensity of engaging in increased exercise. Following estimation of the propensity score, the primary outcome was analyzed using multivariable logistic regression adjusted for the propensity score and an appropriate continuous baseline score (ie, glucose-lowering medication score at baseline), by including these as independent variables in the regression.
      Statistical tests were conducted between different tertiles and the standard-care group in a hierarchical manner for the predefined primary and secondary analyses. A gatekeeping procedure using serial testing was applied to implicitly adjust for multiplicity.
      • Dmitrienko A.
      • D'Agostino Sr., R.B.
      Multiplicity considerations in clinical trials.
      We first investigated whether there was a trend in the odds ratio for discontinuation of glucose-lowering medication treatment at the 12-month follow-up between standard care and each of the exercise tertiles. If present (P<.10 was considered indicative), formal statistical testing for the between-group comparisons and effect-size estimation was initiated in the following order until one of the end points in the hierarchy failed to be statistically significant at α=.05: (1) standard care vs upper tertile, (2) standard care vs intermediate tertile, (3) standard care vs lower tertile, (4) upper tertile vs lower tertile, (5) upper tertile vs intermediate tertile, and (6) intermediate tertile vs lower tertile.
      The secondary outcomes were analyzed using linear regression models or logistic regression models similar to the model used for the primary analysis of the primary outcome. Performance of the primary model was assessed using C statistics. A value of 0.5 suggests that the model discriminates no better than chance alone, 0.7 to 0.8 is acceptable, 0.8 to 0.9 is excellent, and greater than 0.9 is outstanding discrimination.
      • Hosmer D.W.
      • Lemeshow S.
      Applied Logistic Regression.
      Five sensitivity analyses were performed: (1) including energy intake at baseline and change in energy intake from baseline to the 12-month follow-up; (2) using models without adjustments; (3) including prespecified variables found to predict exercise adherence: BMI, age, fitness level, educational level, and sex
      • Unick J.L.
      • Gaussoin S.A.
      • Hill J.O.
      • et al.
      Four-year physical activity levels among intervention participants with type 2 diabetes.
      • Wing R.R.
      • Hamman R.F.
      • Bray G.A.
      • et al.
      Diabetes Prevention Program Research Group
      Achieving weight and activity goals among diabetes prevention program lifestyle participants.
      • Forechi L.
      • Mill J.G.
      • Griep R.H.
      • Santos I.
      • Pitanga F.
      • Molina M.
      Adherence to physical activity in adults with chronic diseases: ELSA-Brasil.
      • Eaglehouse Y.L.
      • Venditti E.M.
      • Kramer M.K.
      • et al.
      Factors related to lifestyle goal achievement in a diabetes prevention program dissemination study.
      ; (4) adjusted for data-driven propensity score with no adjustment for baseline value; and (5) post hoc analysis excluding participants beginning insulin treatment during the intervention. Furthermore, between-group comparisons and effect size estimation were allowed in post hoc analyses to be continued regardless of statistical significance in the gate-keeping sequence.
      The statistical analysis plan was completed before any analyses were performed. All analyses were performed in the statistical program R (version 3.5.1)
      The R Foundation
      R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing.
      using the packages lme4
      • Bates D.
      • Maechler M.
      • Bolker B.
      • Walker S.
      Fitting linear mixed-effects models using lme4.
      and drc,
      • Ritz C.
      • Baty F.
      • Streibig J.C.
      • Gerhard D.
      Dose-response analysis using R.
      and statistical significance level was set at α<.05 (2-tailed).

      Results

      Participants

      A total of 98 participants were randomly assigned to either the U-TURN intervention (n=64) or to standard care (n=34; Figure 1). Three participants in standard care did not attend the 12-month assessment. Two participants in the U-TURN group only attended the baseline assessment and 1 did not register any exercise volume (Figure 1). Hence, 92 were included in the analysis (U-TURN, n=61, standard care, n=31). Mean ± SD age at baseline was 54.4±9.0 years, and sex distribution was even (47% [43 of 92] women; Table 1).
      Figure thumbnail gr1
      Figure 1Flow of participants through the study. HbA1c = glycated hemoglobin A1c; BMI = body mass index; StC = standard care.
      Table 1Baseline Characteristics for Each Group
      BMI = body mass index; HbA1c = glycated hemoglobin A1c; HDL = high-density lipoprotein; IQR = interquartile range; LDL = low-density lipoprotein; Vo2max = maximun oxygen consumption.
      ,
      Data are presented as mean ± SD unless otherwise stated. Dichotomous and categorical data are presented as actual number (percent). Tertiles are stratified according to the observed exercise volume during the 1-year trial.
      nStandard care (n=31)nLower tertile (n=21)nIntermediate tertile (n=20)nUpper tertile (n=20)
      Demographic characteristics
       Age (y), mean ± SD3156.8±8.32152.3±8.52053.7±10.12053.8±8.9
       Female sex, No. (%)3114 (45)2114 (67)207 (35)208 (40)
       Type 2 diabetes duration (y), median [IQR]316.0 [3.0-9.0]214.0 [4.0-6.0]204.5 [2.0-7.3]204.5 [2.8-7.3]
      Glycemic control
       HbA1c (%), mean ± SD316.7±0.9216.5±0.66206.7±0.97206.8±0.86
       Fasting glucose (mmol/L), median [IQR]317.80 [6.85-9.00]216.70 [6.20-7.40]207.65 [6.63-9.23]208.25 [6.35-8.70]
       Fasting insulin (pmol/L), median [IQR]31116.0 [68.0-169.0]21107.0 [73.0-171.0]20134.5 [89.0-164.8]20114.0 [69.3-147.8]
       2-h glucose (mmol/L), mean ± SD3016.35±4.162114.59±3.151915.80±3.912015.73±5.02
      Lipids
       Total cholesterol (mmol/L), mean ± SD303.96±0.98214.32±0.56204.02±0.89204.19±1.05
       LDL cholesterol (mmol/L), mean ± SD302.21±0.85212.49±0.56202.41±0.84202.50±0.98
       HDL cholesterol (mmol/L), mean ± SD301.27±0.36211.36±0.40201.06±0.26201.24±0.30
       Triglycerides (mmol/L), median [IQR]301.37 [0.84-1.79]211.21 [1.00-1.75]201.57 [1.17-2.03]201.37 [0.89-1.90]
      Blood pressure
       Systolic (mm Hg), mean ± SD23135.7±8.219130.2±12.319128.8±18.320123.6±9.3
       Diastolic (mm Hg), mean ± SD2384.0±7.91981.6±9.01979.7±9.22076.5±7.0
      Body composition
       Body mass (kg), mean ± SD3197.6±15.22192.5±15.02094.3±13.42099.7±13.6
       BMI (kg/m2), mean ± SD3132.3±4.42131.8±3.62030.8±4.12032.3±4.0
       Fat mass (kg), mean ± SD3135.8±10.82135.9±7.92033.9±9.62037.3±9.3
       Lean body mass (kg), mean ± SD3157.9±10.72152.8±10.82056.2±8.82058.4±10.7
       Abdominal fat mass (kg), mean ± SD314.1±1.3213.9±1.0203.9±1.2204.2±1.2
       Body fat percentage (%), mean ± SD3137.9±9.22140.5±6.52037.3±8.12038.9±8.2
      Physical fitness, physical activity, and diet
       Vo2max (mL O2/min), mean ± SD302668±771212648±859202738±622202807±691
       Relative Vo2max (mL O2/kg/min), mean ± SD3027.4±6.32128.5±7.42029.3±6.72028.2±6.2
       Energy intake (kJ/d), mean ± SD259997±38662010,294±3511208814±2350208767±3072
      a BMI = body mass index; HbA1c = glycated hemoglobin A1c; HDL = high-density lipoprotein; IQR = interquartile range; LDL = low-density lipoprotein; Vo2max = maximun oxygen consumption.
      b Data are presented as mean ± SD unless otherwise stated. Dichotomous and categorical data are presented as actual number (percent). Tertiles are stratified according to the observed exercise volume during the 1-year trial.
      With higher exercise volumes, the odds of discontinuing glucose-lowering medication increased (Table 2; Figure 2). The model showed acceptable or excellent discrimination (C statistic, 0.80; 95% CI, 0.71-0.91). The numbers needed to treat in the lower, intermediate, and upper tertiles were 4, 3, and 2, respectively. No statistically significant trend for intensification of glucose-lowering medication was found (P=.15) (Table 2). Change in HbA1c levels from baseline to the 12-month follow-up in the lower tertile was not statistically significantly reduced (−0.3%; 95% CI, −0.9% to 0.3%) (P=.29) compared with standard care, whereas HbA1c level was reduced in the intermediate (−0.7%; 95% CI, −1.3% to −0.1%) (P=.02) and upper tertiles (−0.7%; 95% CI, −1.2% to −0.2%) (P=.008).
      Table 2Between-Group Comparisons of Changes From Baseline to 12-Month Follow-up
      BMI = body mass index; HbA1c = glycated hemoglobin A1c; HDL = high-density lipoprotein; LDL = low-density lipoprotein; OR = odds ratio; StC = standard care; T = tertile; Vo2max = maximum oxygen consumption.
      ,
      Data are presented as least squares mean (95% CI) unless stated otherwise. Outcomes were analyzed using regression analyses adjusting for baseline value and propensity score. Propensity scores were estimated based on sex and HDL cholesterol level at baseline because these variables predicted group allocation. Analyses were based on a sequential analytic approach; if an overall trend was present (P<.10 was considered indicative), between-group comparisons for effect size estimation were initiated in a specific order, and if no difference was present (P<.05), the statistical inference sequence was terminated.
      nStC (n=31)nLower tertile (n=21)nIntermediate tertile (n=20)nUpper tertile (n=20)ComparisonEstimate (95% CI)P value
      Primary outcome
       Discontinuation of glucose-lowering medications, n (%)315 (16)2110 (48)2012 (60)2014 (70)Overall trend.003
      T3-StCOR, 34.4 (4.1 to 290.1).001
      T2-StCOR, 30.2 (2.9 to 318.5).005
      T1-StCOR, 12.1 (1.2 to 11.9).03
      T3-T1OR, 2.8 (0.7 to 11.7).15
      Key secondary outcomes
       Change in HbA1c (%)310.15 (−0.2 to 0.5)21−0.2 (−0.5 to 0.2)20−0.6 (−0.9 to −0.3)20−0.6 (−0.8 to −0.3)Overall trend.007
      T3-StC−0.7 (−1.2 to −0.2).008
      T2-StC−0.7 (−1.3 to −0.1).02
      T1-StC−0.3 (−0.9 to 0.3).29
       Reduction in glucose-lowering medications, n (%)319 (29)2112 (57)2019 (95)2016 (80)Overall trend.003
      T3-StCOR, 10.6 (1.2 to 91.5).03
      T2-StCOR, 48. (2.5 to 928.7).01
      T1-StCOR, 3.7 (0.3 to 41.4).29
       Intensification in glucose-lowering medications, n (%)3115 (48)214 (19)201 (5)202 (10)Overall trend.15
      Other exploratory outcomes
       Glycemic control
      Change in fasting glucose (mmol/L)29−0.4 (−1.3 to 0.5)21−0.5 (−1.3 to 0.2)20−1.0 (−1.7 to −0.3)20−1.4 (−2.0 to −0.7)Overall trend.16
      Change in fasting insulin (pmol/L)27−21.1 (−52.2 to 10.0)20−19.1 (−43.8 to 5.6)20−30.5 (−55.1 to −5.9)19−55.8 (−78.0 to −33.6)Overall trend.07
      T3-StC−34.7 (−77.6 to 8.1).11
       Change in 2-h glucose (mmol/L)27−0.4(−2.3 to 1.5)21−2.2 (−3.6 to −0.7)19−2.7 (−4.2 to −1.2)20−4.0 (−5.3 to −2.7)Overall trend.02
      T3-StC−3.6 (−6.2 to −1.1).006
      T2-StC−2.3 (−5.2 to 0.6).11
      Lipids
       Change in total cholesterol (mmol/L)280.6 (0.1 to 1.1)210.32 (−0.1 to 0.7)200.6 (0.2 to 1.1)200.46 (0.1 to 0.8)Overall trend.59
       Change in LDL cholesterol (mmol/L)290.20 (−0.24 to 0.65)210.2 (−0.2 to 0.5)200.6 (0.2 to 1.0)200.40 (0.1 to 0.7)Overall trend.22
       Change in HDL cholesterol (mmol/L)280.22 (0.04 to 0.39)210.1 (−0.1 to 0.2)200.2 (0.1 to 0.4)200.2 (0.1 to 0.4)Overall trend.33
       Change in triglycerides (mmol/L)280.18 (−0.16 to 0.52)21−0.2 (−0.4 to 0.1)20−0.3 (−0.6 to −0.0)20−0.5 (−0.8 to −0.3)Overall trend.007
      T3-StC−0.7 (−1.2 to −0.3).003
      T2-StC−0.5 (−1.0 to 0.0).07
      Blood pressure
       Change in systolic (mm Hg)19−5.4 (−13.2 to 2.4)12−0.5 (−7.0 to 6.0)160.3 (−5.4 to 5.9)20−2.9 (−7.3 to 1.5)Overall trend.74
       Change in diastolic (mm Hg)19−4.1 (−9.9 to 1.7)12−1.7 (−6.6 to 3.2)16−0.0 (−4.2 to 4.2)20−2.0 (−5.2 to 1.3)Overall trend.77
       Body composition
       Change in body mass (kg)270.3 (−4.1 to 4.6)21−3.7 (−7.1 to −0.3)20−7.2 (−10.5 to −3.8)20−10.9 (−13.9 to −7.9)Overall trend<.001
      T3-StC−11.2 (−17.1 to −5.3)<.001
      T2-StC−7.4 (−14. to −0.9).03
      T1-StC−4.0 (−10.6 to 2.68).24
       Change in BMI (kg/m2)27−0.1 (−1.5 to 1.4)21−1.1 (−2.2 to 0.0)20−2.4 (−3.5 to −1.3)20−3.6 (−4.6 to −2.6)Overall trend<.001
      T3 StC−3.5 (−5.5 to −1.6)<.001
      T2-StC−2.3 (−4.5 to −0.2).04
      T1-StC−1 (−3.2 to 1.2).36
       Change in fat mass (kg)27−0.4 (−3.9 to 3.1)21−2.7 (−5.4 to 0.1)20−6.6 (−9.3 to −3.9)20−10.6 (−13.0 to −8.2)Overall trend<.001
      T3-StC−10.2 (−15. to −5.5)<.001
      T2-StC−6.2 (−11.5 to −0.9).02
      T1-StC−2.3 (−7.6 to 3.1).41
       Change in lean body mass (kg)27−1.9 (−5.0 to 1.2)210.9 (−1.5 to 3.4)201.2 (−1.2 to 3.7)20−0.7 (−2.9 to 1.4)Overall trend.46
       Change in abdominal fat mass (kg)270.0 (−0.5 to 0.5)21−0.4 (−0.8 to −0.0)20−0.9 (−1.3 to −0.5)20−1.4 (−1.7 to −1.0)Overall trend<.001
      T3-StC−1.4 (−2 to −0.7)<.001
      T2-StC−0.9 (−1.6 to −0.2).02
      T1-StC−0.4 (−1.1 to 0.3).29
       Change in body fat percentage (%)27−0.4 (−2.9 to 2.1)21−1.9 (−3.9 to 0.0)20−5.1 (−7.1 to −3.2)20−7.7 (−9.4 to −6.0)Overall trend<.001
      T3-StC−7.3 (−10.6 to −3.9)<.001
      T2-StC−4.7 (−8.5 to −0.9).02
      T1-StC−1.5 (−5.4 to 2.4).44
      Physical fitness, physical activity, and diet
       Change in VO2max(mL O2/min)2540 (−257 to 337)21123 (−100 to 346)20437 (214 to 660)19555 (352 to 758)Overall trend.002
      T3-StC514.9 (116.7 to 913.2).01
      T2-StC397.1 (−45.7 to 839.9).08
       Change in relative VO2max (mL O2/kg/min)25−0.2 (−3.5 to 3.2)212.6 (0.0 to 5.1)207.91 (5.4 to 10.5)199.6 (7.3 to 11.9)Overall trend<.001
      T3-StC9.8 (5.2 to 14.3)<.001
      T2-StC8.1 (3. to 13.1).002
      T1-StC2.7 (−2.4 to 7.8).29
       Change in energy intake (kJ/d)2344 (−1820 to 1908)19−1841 (−3247 to −435)19−1462 (−2854 to −71)18−1162 (−2435 to 112)Overall trend.61
      Medication
       Reduction in lipid-lowering medication, n (%)3114 (45)216 (29)208 (40)209 (45)Overall trend.09
      T3-StCOR, 3.6 (0.4 to 35.5).28
       Reduction in blood pressure–lowering medication, n (%)314 (13)213 (14)208 (40)207 (35)Overall trend.02
      T3-StCOR, 11.1 (0.4 to 299.9).15
      a BMI = body mass index; HbA1c = glycated hemoglobin A1c; HDL = high-density lipoprotein; LDL = low-density lipoprotein; OR = odds ratio; StC = standard care; T = tertile; Vo2max = maximum oxygen consumption.
      b Data are presented as least squares mean (95% CI) unless stated otherwise. Outcomes were analyzed using regression analyses adjusting for baseline value and propensity score. Propensity scores were estimated based on sex and HDL cholesterol level at baseline because these variables predicted group allocation. Analyses were based on a sequential analytic approach; if an overall trend was present (P<.10 was considered indicative), between-group comparisons for effect size estimation were initiated in a specific order, and if no difference was present (P<.05), the statistical inference sequence was terminated.
      Figure thumbnail gr2
      Figure 2Bar chart illustrates the proportion of participants within each tertile having discontinued treatment with glucose-lowering medication at the 12-month follow-up for standard care (5/31), lower tertile (10/21), intermediate tertile (12/20), and upper tertile (14/20). Error bars indicate 95% CIs. The primary statistical model (logistic regression adjusted for glucose-lowering medication score and propensity score) showed an overall statistically significant trend (P=.003), as well as statistically significant differences (indicated with lines and asterisks) between upper tertile and standard care (P=.001), intermediate tertile and standard care (P=.005), and lower tertile and standard care (P=.03).
      Sensitivity analysis adjusting for energy intake at baseline and change in energy intake from baseline to the 12-month follow-up did not affect the primary conclusions for the primary outcome materially (Supplementary Table 1, available online at http://www.mayoclinicproceedings.org). Neither did sensitivity analysis with no adjustments (Supplementary Table 2, available online at http://www.mayoclinicproceedings.org), adjusted for prespecified variables (Supplementary Table 3, available online at http://www.mayoclinicproceedings.org), or data-driven propensity (Supplementary Table 4, available online at http://www.mayoclinicproceedings.org) or excluding participants initiating insulin treatment during the intervention (Supplementary Table 5, available online at http://www.mayoclinicproceedings.org).
      The trends in the explorative analysis align with the primary outcome; however, cardiovascular risk factors were only attenuated in the intermediate and upper tertiles when compared with standard care. The overall trend for fasting insulin level was significant (P=.07) and a reduction in 2-hour glucose level in the upper tertile was observed (P=.006). Reductions in body weight, BMI, and measures of fat mass was observed in the upper and intermediate tertiles, with a consistently greater reduction observed in the upper tertile. Triglyceride levels were improved in the upper tertile only. Absolute fitness (maximum oxygen consumption in milliliters of oxygen per minute) only increased in the upper tertile. No trend for change in energy intake was found.
      Post hoc analyses that allowed the primary statistical testing for the between-group comparisons and effect size estimation to be continued regardless of statistical significance are found in Supplementary Table 6 (available online at http://www.mayoclinicproceedings.org). Numbers of adverse events were 5 of 31 (16%), 11 of 21 (52%), 15 of 20 (75%), and 5 of 20 (25%) for standard care and lower, intermediate, and upper tertiles, respectively. Specifically, for musculoskeletal pain, numbers were 0 of 31 (0%), 6 of 21 (29%), 7 of 20 (35%), and 1 of 20 (0.5%) for standard care and lower, intermediate, and upper tertiles, respectively.
      The median volume of exercise across the intervention period in the lower tertile was 178 minutes per week (interquartile range [IQR], 121-213). The median volume of exercise in the intermediate tertile was 296 minutes per week (IQR, 261-310). Median volume of exercise in the upper tertile was 380 minutes per week (IQR, 355-446). The monthly adherence is described in Supplementary Table 7 (available online at http://www.mayoclinicproceedings.org) and Figure 3.
      Figure thumbnail gr3
      Figure 3Bar chart reports the combined volume for each exercise tertile for each month of intervention. Volume is reported as median minutes per week based on all participants within a given tertile for each month of intervention. Error bars indicate interquartile range. The prescribed dosage of exercise is 240 or more minutes of exercise per week in the first phase and 300 minutes of exercise per week in the second and third phases. The upper tertile is compliant with the prescribed dose of combined exercise each month of the intervention, whereas the intermediate tertile is compliant in the first phase and the lower tertile is compliant in the first 3 months of the intervention.

      Discussion

      The main finding of the study was that exercise was associated with a decreased need for glucose-lowering medication in a dose-dependent manner. In addition, trends in the secondary outcomes align with the primary outcome. However, cardiovascular risk factors were only attenuated in the intermediate and upper tertiles. A lifestyle intervention with volumes of exercise substantially higher than current exercise recommendations was needed to generate improvements in HbA1c levels, cardiorespiratory fitness, triglyceride levels, and glucose tolerance above and beyond those achieved before the intervention with higher levels of medication.
      • Colberg S.R.
      • Sigal R.J.
      • Fernhall B.
      • et al.
      Exercise and type 2 diabetes: the American College of Sports Medicine and the American Diabetes Association: joint position statement.
      • Colberg S.R.
      • Sigal R.J.
      • Yardley J.E.
      • et al.
      Physical activity/exercise and diabetes: a position statement of the American Diabetes Association.
      • Piercy K.L.
      • Troiano R.P.
      Physical Activity Guidelines for Americans From the US Department of Health and Human Services.
      There are only a limited number of studies exploring the dose-dependency of exercise in at-risk populations. Based on the Dose-Response to Exercise in Women studies, metabolic risk factors such as fitness, waist circumference, fasting glucose level, and systolic blood pressure are dose dependent, with greater improvements being achieved with ascending levels of exercise energy expenditure.
      • Church T.S.
      • Earnest C.P.
      • Skinner J.S.
      • Blair S.N.
      Effects of different doses of physical activity on cardiorespiratory fitness among sedentary, overweight or obese postmenopausal women with elevated blood pressure: a randomized controlled trial.
      ,
      • Earnest C.P.
      • Johannsen N.M.
      • Swift D.L.
      • Lavie C.J.
      • Blair S.N.
      • Church T.S.
      Dose effect of cardiorespiratory exercise on metabolic syndrome in postmenopausal women.
      In a secondary analysis, Ross et al
      • Ross R.
      • de Lannoy L.
      • Stotz P.J.
      Separate effects of intensity and amount of exercise on interindividual cardiorespiratory fitness response.
      found that substantial volumes of exercise with high intensity are needed to completely eliminate cardiorespiratory fitness nonresponse.
      We are aware of no studies that have explored the dose dependency of exercise in relation to the need for glucose-lowering medication in patients with T2D. However, there are a limited number of studies that have explored the dose dependency of exercise on HbA1c level. Di Loreto et al
      • Di Loreto C.
      • Fanelli C.
      • Lucidi P.
      • et al.
      Make your diabetic patients walk: long-term impact of different amounts of physical activity on type 2 diabetes.
      reported that at least 10 metabolic equivalent–hours per week are needed to improve HbA1c levels and reduce risk factors for diabetic complications. Furthermore, these authors report that health benefits increase with exercise levels exceeding 10 metabolic equivalent–hours per week. This result is in line with results from our study, which find additional benefits of completing a level of exercise that exceeds current recommendations.
      Conversely, in the Italian Diabetes and Exercise Study (IDES), no association of exercise volume and the probability of having an HbA1c level less than 6.5% was observed at the end of the study.
      • Balducci S.
      • Zanuso S.
      • Nicolucci A.
      • et al.
      Italian Diabetes Exercise Study (IDES) Investigators
      Effect of an intensive exercise intervention strategy on modifiable cardiovascular risk factors in subjects with type 2 diabetes mellitus: a randomized controlled trial: the Italian Diabetes and Exercise Study (IDES).
      This is interesting because participants completing the highest levels of exercise in IDES reported exercise volumes that exceed the high exercise volumes of the upper tertile in our study. However, a major limitation in the IDES data is that the exercise volume was self-reported. In addition, contrary to our study, IDES included participants treated with insulin and with a longer average duration of disease, both of which are indicative of more severe disease states. Results from the Look Action for Health in Diabetes trial and Diabetes Remission Clinical Trial, as well as results from bariatric surgery trials, have established that participants with a short duration of disease are more likely to respond to nonpharmacologic treatment.
      • Gregg E.W.
      • Chen H.
      • Wagenknecht L.E.
      • et al.
      Look AHEAD Research Group
      Association of an intensive lifestyle intervention with remission of type 2 diabetes.
      ,
      • Lean M.E.
      • Leslie W.S.
      • Barnes A.C.
      • et al.
      Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial.
      ,
      • Madsen L.R.
      • Baggesen L.M.
      • Richelsen B.
      • Thomsen R.W.
      Effect of Roux-en-Y gastric bypass surgery on diabetes remission and complications in individuals with type 2 diabetes: a Danish population-based matched cohort study.
      Finally, baseline HbA1c levels in IDES were higher than in our study, which would have made it more difficult for IDES participants to reach the defined treatment target. Interestingly, the 3-year IDES-2 follow-up study has recently reported that an increase of 56 minutes of moderate to vigorous exercise per week was not sufficient to attenuate cardiovascular risk factors.
      • Balducci S.
      • D'Errico V.
      • Haxhi J.
      • et al.
      Italian Diabetes and Exercise Study 2 (IDES_2) Investigators
      Effect of a behavioral intervention strategy on sustained change in physical activity and sedentary behavior in patients with type 2 diabetes: the IDES_2 Randomized Clinical Trial.
      This corresponds to the findings of our lower exercise tertile.
      The clinical benefits of discontinuing or reducing treatment with glucose-lowering medication are unclear. However, factors such as glycemic control, blood lipid profile, and low fitness level have consistently been associated with increased risk for diabetes-related complications.
      • Pedersen B.K.
      • Saltin B.
      Exercise as medicine - evidence for prescribing exercise as therapy in 26 different chronic diseases.
      • Cade W.T.
      Diabetes-related microvascular and macrovascular diseases in the physical therapy setting.
      • Sacks F.
      • Hermans M.
      • Fioretto P.
      • et al.
      Association between plasma triglycerides and HDL-cholesterol and microvascular kidney disease and retinopathy in type 2 diabetes: a global case-control study in 13 countries.
      • Church T.S.
      • LaMonte M.J.
      • Barlow C.E.
      • Blair S.N.
      Cardiorespiratory fitness and body mass index as predictors of cardiovascular disease mortality among men with diabetes.
      Our analysis revealed that a lifestyle intervention including exercise volumes in the lower tertile was sufficient to introduce discontinuation of treatment with glucose-lowering medication in nearly half the participants.
      However, this exercise volume failed to improve any cardiovascular disease risk markers. The data indicate that to achieve discontinuation of treatment with glucose-lowering medications while simultaneously improving clinically important markers associated with risk for T2D-related complications, exercise volumes must approach the level of the intermediate tertile. Specifically, both the intermediate and upper tertiles demonstrated significant improvements in HbA1c levels, fitness, and body composition measures. Although it should be acknowledged that improvements in HbA1c levels are not consistently associated with a reduction in risk for macrovascular complications and mortality,
      • Hemmingsen B.
      • Lund S.S.
      • Gluud C.
      • et al.
      Intensive glycaemic control for patients with type 2 diabetes: systematic review with meta-analysis and trial sequential analysis of randomised clinical trials.
      the improvements in HbA1c levels observed in the intermediate and upper tertiles, if maintained, likely have the potential to reduce the risk for developing microvascular complications, specifically retinopathy and nephropathy.
      • Zoungas S.
      • Arima H.
      • Gerstein H.C.
      • et al.
      Collaborators on Trials of Lowering Glucose (CONTROL) group
      Effects of intensive glucose control on microvascular outcomes in patients with type 2 diabetes: a meta-analysis of individual participant data from randomised controlled trials.
      In addition, only the upper tertile increased absolute fitness and reduced both 2-hour glucose and triglyceride levels. Fitness levels, triglyceride levels, and 2-hour glucose tolerance have all been associated with risk for developing micro- and macrovascular complications.
      • Pedersen B.K.
      • Saltin B.
      Exercise as medicine - evidence for prescribing exercise as therapy in 26 different chronic diseases.
      • Cade W.T.
      Diabetes-related microvascular and macrovascular diseases in the physical therapy setting.
      • Sacks F.
      • Hermans M.
      • Fioretto P.
      • et al.
      Association between plasma triglycerides and HDL-cholesterol and microvascular kidney disease and retinopathy in type 2 diabetes: a global case-control study in 13 countries.
      • Church T.S.
      • LaMonte M.J.
      • Barlow C.E.
      • Blair S.N.
      Cardiorespiratory fitness and body mass index as predictors of cardiovascular disease mortality among men with diabetes.
      Diet is associated with remission of T2D,
      • Lean M.E.
      • Leslie W.S.
      • Barnes A.C.
      • et al.
      Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial.
      and a difference in energy intake could be a possible explanation of the dose-response effect in the odds of discontinuation of glucose-lowering medication and HbA1c level reduction shown in our study. We adjusted for energy intake in the analysis and this did not change the association, indicating that energy intake was not the driving factor of the effects seen on primary and secondary outcomes.
      It should be acknowledged that high levels of exercise are difficult to attain and methods to increase adherence to exercise should be explored. However, a significant proportion of participants in the current study managed to perform and maintain adherence to exercise levels superceeding current recommendations.
      There are some limitations to our study. First, our secondary analyses invariably involve some prognostic imbalance as randomization is lost. Specifically, patients who have a particular exercise volume could be subject to selection bias, which could lead to a difference with respect to their probability of developing the outcome. To account for potential confounding, a propensity score was created to balance out observed confounders across the groups. In addition, post hoc adjustments were applied and none of the sensitivity analyses reduced the robustness of primary results. However, the results may still be biased by different distributions of unknown prognostic factors. In relation to this, the dietary assessment in this study relied on self-reports that are subject to bias, rendering this analysis mainly exploratory. This entails that the sensitivity analysis adjusting for energy intake should be interpreted with caution.
      Second, no objective measure of exercise adherence was obtained for the standard-care group. If participants in the standard-care group increase their exercise levels, this would underestimate the effect of the different exercise doses and thus the reported effects in this study may be conservative. However, because physical fitness did not increase in the standard-care group (Table 2), this indicates that the standard-care group as a whole did not increase aerobic exercise.
      Third, the small number of participants in the tertiles leaves us with reduced statistical power. However, despite this, the primary analysis reached statistical significance, underscoring the notion of a very strong biological signal.
      Fourth, the inclusion criteria in this study limit the generalizability of the results. In addition, there may be some form of reverse causality present, with participants potentially becoming more likely to exercise as their health improves.
      In relation to discontinuations, the titration target for HbA1c level at baseline was also the defined target for initiating reductions in glucose-lowering medication. This resulted in a high percentage of participants having baseline values close to or less than the target for initiating reductions in glucose-lowering medication at the onset of the intervention. This increased the odds of participants being discontinued from treatment with glucose-lowering medication. However, it would not account for the gradual increase in the odds of being discontinued from treatment with glucose-lowering medication parallel to the increase in exercise volume.

      Conclusion

      Our results indicate that exercise was associated with discontinuation of glucose-lowering medication in a dose-dependent manner, as were important cardiovascular risk factors in well-treated participants with T2D and disease duration less than 10 years. However, discontinuation of glucose-lowering medication treatment is debatable; insofar as remission is a goal, our data suggest that prescribed exercise volumes may need to exceed current recommendations to simultaneously reduce cardiovascular risk factors such as HbA1c levels, fitness, body composition measures, 2-hour glucose levels, and triglyceride levels. However, the feasibility of such high volumes of exercise can be questioned. In our study, a large portion of participants managed to achieve exercise volumes exceeding current recommendations. Future high-quality randomized trials should be performed to confirm these findings and explore what could be gained by adding an even more controlled diet.

      Acknowledgments

      The authors thank all the patients for their participation in the study.

      Supplemental Online Material

      Supplemental material can be found online at http://www.mayoclinicproceedings.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data.

      References

        • American Diabetes Association
        6. Obesity management for the treatment of type 2 diabetes.
        Diabetes Care. 2016; 39: S47-S51
        • Huang E.S.
        • Brown S.E.
        • Ewigman B.G.
        • Foley E.C.
        • Meltzer D.O.
        Patient perceptions of quality of life with diabetes-related complications and treatments.
        Diabetes Care. 2007; 30: 2478-2483
        • Nicolucci A.
        • Kovacs Burns K.
        • Holt R.I.
        • et al.
        • DAWN2 Study Group
        Diabetes Attitudes, Wishes and Needs second study (DAWN2): cross-national benchmarking of diabetes-related psychosocial outcomes for people with diabetes.
        Diabet Med. 2013; 30: 767-777
        • Espeland M.A.
        • Glick H.A.
        • Bertoni A.
        • et al.
        • Look AHEAD Research Group
        Impact of an intensive lifestyle intervention on use and cost of medical services among overweight and obese adults with type 2 diabetes: the Action for Health in Diabetes.
        Diabetes Care. 2014; 37: 2548-2556
        • Gregg E.W.
        • Chen H.
        • Wagenknecht L.E.
        • et al.
        • Look AHEAD Research Group
        Association of an intensive lifestyle intervention with remission of type 2 diabetes.
        JAMA. 2012; 308: 2489-2496
        • Lean M.E.
        • Leslie W.S.
        • Barnes A.C.
        • et al.
        Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial.
        Lancet. 2018; 391: 541-551
        • Di Loreto C.
        • Fanelli C.
        • Lucidi P.
        • et al.
        Make your diabetic patients walk: long-term impact of different amounts of physical activity on type 2 diabetes.
        Diabetes Care. 2005; 28: 1295-1302
        • Umpierre D.
        • Ribeiro P.A.
        • Kramer C.K.
        • et al.
        Physical activity advice only or structured exercise training and association with HbA1c levels in type 2 diabetes: a systematic review and meta-analysis.
        JAMA. 2011; 305: 1790-1799
        • Balducci S.
        • Zanuso S.
        • Nicolucci A.
        • et al.
        • Italian Diabetes Exercise Study (IDES) Investigators
        Effect of an intensive exercise intervention strategy on modifiable cardiovascular risk factors in subjects with type 2 diabetes mellitus: a randomized controlled trial: the Italian Diabetes and Exercise Study (IDES).
        Arch Intern Med. 2010; 170: 1794-1803
        • Vancea D.M.
        • Vancea J.N.
        • Pires M.I.
        • Reis M.A.
        • Moura R.B.
        • Dib S.A.
        Effect of frequency of physical exercise on glycemic control and body composition in type 2 diabetic patients.
        Arq Bras Cardiol. 2009; 92: 23-30
        • Boule N.G.
        • Kenny G.P.
        • Haddad E.
        • Wells G.A.
        • Sigal R.J.
        Meta-analysis of the effect of structured exercise training on cardiorespiratory fitness in type 2 diabetes mellitus.
        Diabetologia. 2003; 46: 1071-1081
        • Umpierre D.
        • Ribeiro P.A.
        • Schaan B.D.
        • Ribeiro J.P.
        Volume of supervised exercise training impacts glycaemic control in patients with type 2 diabetes: a systematic review with meta-regression analysis.
        Diabetologia. 2013; 56: 242-251
        • Prince S.A.
        • Adamo K.B.
        • Hamel M.E.
        • Hardt J.
        • Connor Gorber S.
        • Tremblay M.
        A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review.
        Int J Behav Nutr Phys Act. 2008; 5: 56
        • Johansen M.Y.
        • MacDonald C.S.
        • Hansen K.B.
        • et al.
        Effect of an intensive lifestyle intervention on glycemic control in patients with type 2 diabetes: a randomized clinical trial.
        JAMA. 2017; 318: 637-646
        • Ried-Larsen M.
        • Christensen R.
        • Hansen K.B.
        • et al.
        Head-to-head comparison of intensive lifestyle intervention (U-TURN) versus conventional multifactorial care in patients with type 2 diabetes: protocol and rationale for an assessor-blinded, parallel group and randomised trial.
        BMJ Open. 2015; 5: e009764
        • Garber C.E.
        • Blissmer B.
        • Deschenes M.R.
        • et al.
        • American College of Sports Medicine
        American College of Sports Medicine position stand. Quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: guidance for prescribing exercise.
        Med Sci Sports Exerc. 2011; 43: 1334-1359
        • Laptook A.R.
        • Shankaran S.
        • Tyson J.E.
        • et al.
        • Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network
        Effect of therapeutic hypothermia initiated after 6 hours of age on death or disability among newborns with hypoxic-ischemic encephalopathy: a randomized clinical trial.
        JAMA. 2017; 318: 1550-1560
        • Dmitrienko A.
        • D'Agostino Sr., R.B.
        Multiplicity considerations in clinical trials.
        N Engl J Med. 2018; 378: 2115-2122
        • Hosmer D.W.
        • Lemeshow S.
        Applied Logistic Regression.
        2nd ed. John Wiley & Sons, Inc, New York, NY2000
        • Unick J.L.
        • Gaussoin S.A.
        • Hill J.O.
        • et al.
        Four-year physical activity levels among intervention participants with type 2 diabetes.
        Med Sci Sports Exerc. 2016; 48: 2437-2445
        • Wing R.R.
        • Hamman R.F.
        • Bray G.A.
        • et al.
        • Diabetes Prevention Program Research Group
        Achieving weight and activity goals among diabetes prevention program lifestyle participants.
        Obes Res. 2004; 12: 1426-1434
        • Forechi L.
        • Mill J.G.
        • Griep R.H.
        • Santos I.
        • Pitanga F.
        • Molina M.
        Adherence to physical activity in adults with chronic diseases: ELSA-Brasil.
        Rev Saude Publica. 2018; 52: 31
        • Eaglehouse Y.L.
        • Venditti E.M.
        • Kramer M.K.
        • et al.
        Factors related to lifestyle goal achievement in a diabetes prevention program dissemination study.
        Transl Behav Med. 2017; 7: 873-880
        • The R Foundation
        R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing.
        https://www.R-project.org/
        Date: 2018
        Date accessed: September 20, 2019
        • Bates D.
        • Maechler M.
        • Bolker B.
        • Walker S.
        Fitting linear mixed-effects models using lme4.
        J Stat Softw. 2015; 67: 1-48
        • Ritz C.
        • Baty F.
        • Streibig J.C.
        • Gerhard D.
        Dose-response analysis using R.
        PLoS One. 2015; 10: e0146021
        • Colberg S.R.
        • Sigal R.J.
        • Fernhall B.
        • et al.
        Exercise and type 2 diabetes: the American College of Sports Medicine and the American Diabetes Association: joint position statement.
        Diabetes Care. 2010; 33: e147-e167
        • Colberg S.R.
        • Sigal R.J.
        • Yardley J.E.
        • et al.
        Physical activity/exercise and diabetes: a position statement of the American Diabetes Association.
        Diabetes Care. 2016; 39: 2065-2079
        • Piercy K.L.
        • Troiano R.P.
        Physical Activity Guidelines for Americans From the US Department of Health and Human Services.
        Circ Cardiovasc Qual Outcomes. 2018; 11: e005263
        • Church T.S.
        • Earnest C.P.
        • Skinner J.S.
        • Blair S.N.
        Effects of different doses of physical activity on cardiorespiratory fitness among sedentary, overweight or obese postmenopausal women with elevated blood pressure: a randomized controlled trial.
        JAMA. 2007; 297: 2081-2091
        • Earnest C.P.
        • Johannsen N.M.
        • Swift D.L.
        • Lavie C.J.
        • Blair S.N.
        • Church T.S.
        Dose effect of cardiorespiratory exercise on metabolic syndrome in postmenopausal women.
        Am J Cardiol. 2013; 111: 1805-1811
        • Ross R.
        • de Lannoy L.
        • Stotz P.J.
        Separate effects of intensity and amount of exercise on interindividual cardiorespiratory fitness response.
        Mayo Clin Proc. 2015; 90: 1506-1514
        • Madsen L.R.
        • Baggesen L.M.
        • Richelsen B.
        • Thomsen R.W.
        Effect of Roux-en-Y gastric bypass surgery on diabetes remission and complications in individuals with type 2 diabetes: a Danish population-based matched cohort study.
        Diabetologia. 2019; 62: 611-620
        • Balducci S.
        • D'Errico V.
        • Haxhi J.
        • et al.
        • Italian Diabetes and Exercise Study 2 (IDES_2) Investigators
        Effect of a behavioral intervention strategy on sustained change in physical activity and sedentary behavior in patients with type 2 diabetes: the IDES_2 Randomized Clinical Trial.
        JAMA. 2019; 321: 880-890
        • Pedersen B.K.
        • Saltin B.
        Exercise as medicine - evidence for prescribing exercise as therapy in 26 different chronic diseases.
        Scand J Med Sci Sports. 2015; 25: 1-72
        • Cade W.T.
        Diabetes-related microvascular and macrovascular diseases in the physical therapy setting.
        Phys Ther. 2008; 88: 1322-1335
        • Sacks F.
        • Hermans M.
        • Fioretto P.
        • et al.
        Association between plasma triglycerides and HDL-cholesterol and microvascular kidney disease and retinopathy in type 2 diabetes: a global case-control study in 13 countries.
        Circulation. 2014; 129: 999-1008
        • Church T.S.
        • LaMonte M.J.
        • Barlow C.E.
        • Blair S.N.
        Cardiorespiratory fitness and body mass index as predictors of cardiovascular disease mortality among men with diabetes.
        Arch Intern Med. 2005; 165: 2114-2120
        • Hemmingsen B.
        • Lund S.S.
        • Gluud C.
        • et al.
        Intensive glycaemic control for patients with type 2 diabetes: systematic review with meta-analysis and trial sequential analysis of randomised clinical trials.
        BMJ. 2011; 343: d6898
        • Zoungas S.
        • Arima H.
        • Gerstein H.C.
        • et al.
        • Collaborators on Trials of Lowering Glucose (CONTROL) group
        Effects of intensive glucose control on microvascular outcomes in patients with type 2 diabetes: a meta-analysis of individual participant data from randomised controlled trials.
        Lancet Diabetes Endocrinol. 2017; 5: 431-437