Relationships of Changes in Physical Activity and Sedentary Behavior With Changes in Physical Fitness and Cardiometabolic Risk Profile in Individuals With Type 2 Diabetes: The Italian Diabetes and Exercise Study 2 (IDES_2)

Both physical inactivity and sedentary behavior are independently associated with increased mortality from any cause and cardiovascular disease (CVD) in the general population (1,2) and also in people with type 2 diabetes (3,4), who are usually well below the recommended level of physical activity (PA) and spend many daily hours as sedentary time (SED-time) (5). Being physically inactive is not the same as being sedentary, as it refers to an individual accumulating an insufficient amount of moderate- to vigorous-intensity PA (MVPA) according to current guidelines (i.e., <150 min ⋅ week⁻¹) (6). Conversely, sedentary individuals are those spending too much time (generally >8 h ⋅ day⁻¹) in any waking behavior characterized by an energy expenditure ≤1.5 MET (6). As the recommended amount of daily MVPA (∼21.5 min) represents <3% of the time spent awake, even a physically active individual may be sedentary if he or she does not accumulate enough light-intensity PA (LPA) during the remaining daily hours in place of SED-time.

These observations represent the basis for the current American Diabetes Association guidelines that recommend that individuals with type 2 diabetes regularly perform the indicated amount of MVPA and also reallocate as much as possible of SED-time to LPA and interrupt prolonged sitting with breaks of LPA (7). Unfortunately, adherence to guideline recommendations is generally poor among individuals with diabetes, especially in the long-term, though only a few studies have investigated maintenance of changes in PA/sedentary behavior over time.

Counseling interventions designed to promote walking through the provision of pedometers in the general population (8) or in individuals with prediabetes (9) or type 2 diabetes (10,11) resulted in up to ∼8 min ⋅ day⁻¹ increases in MVPA after 1 year and smaller (∼4 min ⋅ day⁻¹), or even no, increments after longer follow-up, together with no change in SED-time. However, in these studies, the intervention failed to significantly improve CVD risk factors and other surrogate outcomes (8–10) or to produce sustained increases in cardiorespiratory fitness (12), raising the question of whether increments in MVPA of this extent and/or duration are sufficient to produce clinically meaningful effects. Conversely, in the Italian Diabetes and Exercise Study 2 (IDES_2), a counseling intervention targeting both PA and sedentary behavior, which was effective in promoting a sustained increase in PA and decrease in SED-time, resulted in clinically meaningful improvements in physical fitness and, to a lesser extent, in CVD risk factors and scores over a 3-year follow-up (13). However, while the increase versus control in MVPA was only 6.4 min ⋅ day⁻¹ during the 3-year period and 3.6 min ⋅ day⁻¹ at the end of the study, the reduction in SED-time and the reciprocal increment in time spent in LPA in comparisons with individuals receiving standard care were 0.8 h ⋅ day⁻¹ over time, with the increase in LPA substantially contributing to the 3.3 MET h ⋅ week⁻¹ increase in total PA volume (13). Therefore, it is unclear whether the improvements in physical fitness and CVD risk profile were attributable to the increase in MVPA, despite being as modest as in the walking-based intervention studies, or to the relatively greater changes in SED-time and LPA.

In the present prespecified ancillary analysis of the IDES_2, we investigated the relationships of changes in MVPA and SED-time (and the relative importance of each of them) with changes in physical fitness and CVD risk profile in patients with type 2 diabetes participating in this trial.

Study design and methods have previously been described (13,14) and will be briefly reported here. The IDES_2 was an open-label, assessor-blinded, parallel, superiority randomized clinical trial that assessed the efficacy of a behavioral intervention in increasing daily PA and reducing SED-time over a 3-year follow-up in patients with type 2 diabetes.

The research protocol complies with the Declaration of Helsinki. It was approved by the Ethics Committee of Sant’Andrea University Hospital, and participants provided written informed consent.

The main entry criterion was type 2 diabetes (defined according to the American Diabetes Association criteria) of at least 1-year duration. Additional requirements were age 40–80 years, BMI 27–40 kg/m², physically inactivity and sedentary lifestyle for at least 6 months, ability to walk 1.6 km without assistance, and eligibility after cardiologic evaluation.

In three tertiary referral, outpatient diabetes clinics in Rome, patients were randomized 1:1 to either an intervention (INT) group, receiving theoretical and practical exercise counseling, or a control (CON) group, receiving only general physician recommendations. Randomization was stratified by center and, within each center, by age < vs. ≥65 years and noninsulin versus insulin treatment, with use of permuted-block randomization SAS 9.4 software that randomly varied the block size (range 4, 8).

Patients from both groups received the same treatment regimen, including dietary prescription, to achieve glycemic, lipid, blood pressure (BP), and body weight targets, according to current guidelines. Treatment was adjusted at each visit with a prespecified algorithm.

Physicians, exercise specialists, and participants were not blinded, whereas assessors of accelerometer/diary and biochemical parameters were blinded to group assignment.

The intervention in the INT group consisted of one individual theoretical counseling session, held by a diabetologist, plus eight twice-weekly individual theoretical and practical counseling sessions, held by a certified exercise specialist, once a year for 3 years. This approach was based on the Social Cognitive Theory Model and Health Belief Model and used several behavior change techniques. It was designed to promote a two-step behavior change, i.e., 1) decreasing SED-time by substituting it with a wide range of LPAs and/or interrupting prolonged sitting at home or work with brief bouts of LPA and 2) gradually increasing the time spent in purposeful MVPA by reallocating time from sedentary behavior and/or LPA.

Coprimary end points were changes from baseline in total PA volume, LPA, MVPA, and SED-time over the 3-year period. Secondary end points were changes from baseline in physical fitness and modifiable CVD factors and scores. Results for these end points have been presented in a previous publication (13). Here, we report on the relationships between changes in primary and secondary end points.

Levels of total PA volume, time spent in LPA and MVPA, and SED-time were measured with use of an accelerometer (mywellness key; Technogym, Cesena, Italy) and a daily diary for reporting the time spent wearing the instrument, sleeping, and napping and performing nonaccelerometer recordable activities such as swimming, cycling, skiing, etc. Participants were asked to attach the device at the waist and to wear it all day (except if swimming) up to bedtime to avoid the influence of the “time accelerometer worn,” which may cause underestimation of total daily SED-time. In this way, it was possible to assume that the time patients were awake without wearing the accelerometer was spent in routine morning and evening sedentary activities, unless it was spent in activities that cannot be performed while wearing the accelerometer (e.g., swimming). Results were expressed as total SED-time, calculated by addition of the accelerometer nonwear time in the awake state to the accelerometer-based SED-time. Diary-reported time spent in non-accelerometer-recordable activities was added to that recorded by the accelerometer, according to the intensity of each activity. Measurements were obtained at baseline and every 4 months thereafter for seven consecutive days, except for the initial 4 months, during which the device was worn for the entire period to detail the short-term effect of intervention.

At the same time points, the modifiable cardiovascular risk factors BMI, waist circumference, diabetes duration, fasting plasma glucose (FPG), HbA_1c, triglycerides, total, HDL, and LDL cholesterol, systolic and diastolic BP, and hs-CRP were assessed with standard methods, and total and fatal coronary heart disease (CHD) and stroke 10-year risk scores were calculated with the UK Prospective Diabetes Study risk engine (15).

At baseline and every year thereafter, participants were evaluated for physical fitness. Cardiorespiratory fitness was assessed as VO_2max by maximal treadmill exercise test with use of the Balke protocol. Isometric muscle strength was assessed as maximal voluntary contractions performed at a shoulder press (Technogym) for the upper body and a leg extension machine (Technogym) for the lower body and measured by a strain gauge tensiometer (Digimax; Mechatronic GmbH, Darmstadt, Germany). Hip and trunk flexibility was assessed by a standard bending test.

Mean changes over baseline throughout the 3-year follow-up in PA/SED-time, physical fitness, and CVD risk factors/scores were calculated for participants who completed the study. These individuals were then stratified by quartiles of changes in MVPA, SED-time, LPA, and PA volume, and data were expressed as mean ± SD. One-way ANOVA was used for assessing differences among quartiles.

Univariate correlations between changes in MVPA, SED-time, LPA, and PA volume and changes in physical fitness and CVD risk factors/scores were assessed by Spearman ρ. Multivariable linear regression analyses were applied to assess the independent correlates of changes in physical fitness and CVD risk factors/scores over the 3-year period. Age, sex, study arm, the baseline value of the dependent variable, and changes in MVPA and SED-time were included as covariates in model 1. As SED-time was almost entirely reallocated to LPA, which contributed significantly to the increase in total PA volume, LPA was substituted for SED-time in model 2, whereas PA volume was substituted for MVPA and SED-time in model 3.

All of the above analyses were performed in the whole cohort and in the two groups separately. All P values <0.05 were considered statistically significant. Statistical analyses were performed with SPSS, version 20 (SPSS, Chicago, IL).

As previously described (13), 449 patients were assessed for eligibility from October 2012 to February 2014 in the three centers, and after exclusion of 149 patients for various reasons, the remaining 300 were randomized to the CON or INT group (n = 150 each). Of these 300, 267 patients completed the study at the final evaluation (CON = 134 and INT = 133), whereas 33 patients (CON = 16 and INT = 17) dropped out for various reasons; of those in the INT group, >90% attended the counseling sessions.

The baseline features of study participants and results for coprimary and secondary outcomes have previously been reported, together with adverse events, which did not differ between groups and were unrelated to PA (13).

There were no significant differences in age and sex among quartiles of MVPA and SED-time change (not shown). Patients falling in quartile I of MVPA and SED-time change showed a substantial decrease in MVPA and increase in SED-time (on average, −10.8 min ⋅ day⁻¹ and +0.72 h ⋅ day⁻¹, respectively), those in quartile II showed virtually no change, and those in quartiles III and IV showed modest (on average, +4.0 min ⋅ day⁻¹ and −0.60 h ⋅ day⁻¹, respectively) and marked (on average, +14.3 min ⋅ day⁻¹ and −1.53 h ⋅ day⁻¹, respectively) improvements, respectively (Table 1). There was only a partial correspondence between quartiles of MVPA change and quartiles of SED-time change; in fact, though 47.8% and 60.6% of patients assigned to quartiles I and IV of MVPA change fell into quartiles I and IV of SED-time, respectively, there were also a few individuals in the highest quartile of MVPA change who fell in the lowest quartile of SED-time change and vice versa (Table 1). Moreover, most of the participants falling in the highest quartile (IV) of MVPA and SED-time change were from the INT group, whereas those in quartile III were almost in equal number from the two groups and those in the lowest quartiles (I and II) were mainly from the CON group (Supplementary Table 1). Therefore, while >40% of INT participants and almost 40% of CON participants showed a marked improvement and worsening, respectively, in both MVPA and SED-time, ∼30% of INT participants showed no meaningful change or even substantial worsening and ∼30% of CON participants showed modest or even striking improvements in these parameters (Supplementary Table 1).

Improvements with quartiles of mean change over baseline in MVPA were observed for PA/SED-time and physical fitness parameters as well as for HbA_1c, FPG, BMI, triglycerides, and total and fatal CHD 10-year risk (Table 2). Improvements with quartiles of mean change over baseline in SED-time were observed for PA and physical fitness parameters as well as for HbA_1c and total and fatal CHD 10-year risk (Table 3). Similar results were obtained for quartiles of mean change over baseline in LPA (Supplementary Table 2) and PA volume (Supplementary Table 3). Improvements in physical fitness parameters and CVD risk factors/scores with quartiles of changes in accelerometer-based measures were observed also in the two groups separately (not shown).

Upon univariate analysis, changes in MVPA and SED-time correlated significantly with those in physical fitness parameters as well as in HbA_1c, FPG, hs-CRP, total and fatal CHD 10-risk, BMI and waist circumference (for MVPA changes only), and total stroke 10-risk score (for SED-time changes only) (Supplementary Table 4). Similar results were obtained for changes in LPA and PA volume (Supplementary Table 4) and in the two groups separately (not shown).

In model 1, changes in MVPA and SED-time were both independent predictors of change in VO_2max (especially MVPA change) and HbA_1c (especially SED-time change) (Table 4). Removal of either one from the model resulted in a stronger association of the other one with VO_2max (especially for 1 h ⋅ day⁻¹ change in SED-time: β −1.608, P < 0.0001) and HbA_1c (especially for 1 min·day⁻¹ change in MVPA: β −0.047, P < 0.0001). Only change in MVPA was an independent predictor of change in upper and lower body strength and inversely of change in waist circumference, triglycerides, and systolic BP (Table 4); when MVPA change was removed from the model, change in SED-time became an independent predictor of change in lower-body strength and triglycerides but not in upper-body strength and waist circumference (not shown). Only change in SED-time was an independent predictor of change in bending, FPG, diastolic BP, and CHD 10-year risk scores and, inversely, of change in HDL cholesterol (Table 4); when SED-time change was removed from the model, change in MVPA became an independent predictor of change in FPG, HDL cholesterol, and CHD 10-year risk scores but not in bending and diastolic BP (not shown). In replacing of SED-time changes with LPA changes (model 2), the latter was an independent predictor of changes in the same parameters as for SED-time, except bending (Supplementary Table 5). Likewise, in replacing of MVPA and LPA changes with PA volume changes (model 3), PA volume changes were an independent predictor of changes in the same variables as for MVPA and/or SED-time (Supplementary Table 6). Results were similar when the multivariable linear regression analyses were conducted separately in CON (Supplementary Table 7) and INT (Supplementary Table 8) participants.

The baseline value of the dependent variable was always significantly associated with its change over the study follow-up, except for VO_2max, and the other covariates (age, sex, and study arm) were also variably associated with changes in the dependent variables (Table 4 and Supplementary Tables 5 and 6).

This prespecified ancillary analysis of the IDES_2 showed that both increases in MVPA and decreases in SED-time contributed to the associated improvements in physical fitness and CVD risk profile. These findings provide important insights into the understanding of the relative roles of changes in PA and sedentary behavior promoted by counseling interventions on these surrogate outcomes, the improvement of which may translate into long-term benefits in terms of morbidity and mortality (16,17). In fact, it has long been debated whether increasing (MV)PA volume below the recommended targets and/or reallocating SED-time mainly to LPA, as occurred in the IDES_2, has a meaningful impact on physical fitness and CVD risk profile. Several studies support the concept that increasing PA volume and/or intensity below a certain threshold (i.e., that recommended by guidelines) would not be effective in improving these outcomes, according to the overload principle of training (18). The current study provides strong support to the concept that this may not apply to physically inactive and sedentary or unfit individuals, including most of those suffering from type 2 diabetes. In particular, our data suggest that even modest increments in MVPA may have a clinically meaningful impact for these people, provided that they are associated with larger decrements in SED-time and reciprocal increments in LPA resulting in substantial increases in total PA volume, which also contribute to improve outcomes.

This offers a possible explanation for the discrepancies between the IDES_2 and studies specifically designed to increase MVPA by promoting walking through the provision of pedometers, most of which failed to show significant improvements in surrogate outcomes, despite similar increases in MVPA but with no change in SED-time. No effect was in fact reported on HbA_1c or other CVD risk factors (9,10) or pain, depression, anxiety, or health-related quality of life (8), regardless of whether increases in MVPA were maintained, albeit to a lesser extent (8), or not (9), in the long-term. In addition, at variance with the IDES_2, the initial improvement in cardiorespiratory fitness was not maintained in the Look AHEAD (Action for Health in Diabetes) trial, as after 4 years it was approximately one-fourth of that observed after 1 year (12), just like the increment in MVPA (11). Taken together, these findings indicate that a behavioral counseling targeting both MVPA and SED-time, such as in the IDES_2, is more effective in promoting a sustained behavior change than interventions focusing only on PA (19), which appear inadequate to reduce SED-time and might even trigger compensatory sedentary behavior (20). However, it is unclear whether the additional contribution of decreasing sedentary behavior is related to the reduction and/or interruption of SED-time per se. On the one hand, these actions were in fact shown to provide metabolic benefits independent of one another, in addition to time spent in MVPA (21); furthermore, prolonged sitting time was associated with increased risk of CVD events independent of leisure-time PA in postmenopausal women without a history of CVD (22), and not only the amount but also the pattern of SED-time was associated with mortality (23). On the other hand, change in LPA was independently associated with change in the same variables as change in SED-time and change in PA volume was independently associated with change in the same variables as change in MVPA and LPA. This suggests that the additional contribution of reduced SED-time might be attributed to the reciprocal ∼0.8 h⋅day⁻¹ increment in LPA, which impacted more than the ∼6 min⋅day⁻¹ increment in MVPA on the increase in total PA volume and energy expenditure. There is in fact growing evidence that higher LPA is also associated with better outcomes. In healthy sedentary older adults, VO_2max improved significantly at 35–50% heart rate reserve and continued improving at a greater rate up to heart rate reserve of 75–80% (24). An analysis of 18 studies in healthy subjects showed that exercising intensities below ∼45% VO₂ reserve were consistently ineffective at increasing VO_2max in individuals with mean initial VO_2max >40 mL ⋅ min⁻¹ ⋅ kg⁻¹ but not in those with mean initial VO_2max <40 mL ⋅ min⁻¹ ⋅ kg⁻¹, though the lowest intensities examined were ∼30% VO₂ reserve (25). Though a systematic review of 33 intervention studies failed to detect a significant effect of LPA on CVD risk factors (26), other reports showed a favorable impact (27–30). Moreover, two meta-analyses showed that LPA was associated with improved cardiometabolic health and reduced total and CVD mortality risk (31,32). Finally, the curvilinear relationship between LPA/SED-time and mortality risk was attenuated, but not abolished, by MVPA, except for very high levels (33), and an inverse association was reported between number but not intensity of daily steps and all-cause death (34,35) and CVD events (36).

Another plausible, not mutually exclusive explanation for the discrepancies between the IDES_2 and walking-based intervention studies is that the relative extent of increases over baseline in MVPA is as relevant as the absolute increase for the effect on clinical outcomes. In fact, in the IDES_2, time spent in MVPA was only ∼12 min ⋅ day⁻¹ at baseline and, hence, increased by >50% throughout the 3-year follow-up and was still 30% over baseline at end of study (13). Conversely, in walking-based intervention studies, percent increments were much smaller as MVPA ranged from ∼25 to ∼45 min ⋅ day⁻¹ at baseline and increased by up to ∼8 min ⋅ day⁻¹ after 1 year and by one-half or less after longer follow-up (8–11). This interpretation is consistent with a recent UK Biobank longitudinal study showing that, in inactive people, a 6 min ⋅ day⁻¹ difference in brisk walking (corresponding to ∼500 steps in 5 min [37]) was associated with significant health benefits (∼4 life years gained) (38). Likewise, in adults at high risk for type 2 diabetes and CVD, every 2,000 step ⋅ day⁻¹ increment in ambulatory activity at baseline was associated with a 10% lower risk of CVD events and each 2,000 step ⋅ day⁻¹ change from baseline to 12 months was associated with an additional 8% difference (36). Moreover, among older women, as few as ∼4,400 steps ⋅ day⁻¹ were significantly related to lower mortality rates compared with ∼2,700 steps ⋅ day⁻¹ (35) and, among sedentary, overweight postmenopausal women, one-half the currently recommended PA volume was sufficient to improve cardiorespiratory fitness (39). A pronouncement of the American College of Sports Medicine concluded that the benefits of PA on CVD events and all-cause and CVD mortality start with very modest amounts (approximately one-third the amount recommended by guidelines) (1).

This analysis of the IDES_2 also showed differences in the relative importance of increases in MVPA and decreases in SED-time depending on the variable tested. Changes in both MVPA and SED-time were independently associated with changes in VO_2max, though the relationship was stronger for the former, which was the only independent predictor of muscle strength, in keeping with the concept that PA of higher intensity is required for a significant impact on cardiorespiratory and muscle fitness (18). Conversely, the association of bending with SED-time indicates the beneficial effect of reducing sitting (and increasing the total amount of PA) on flexibility. The much stronger association of glycemic control with SED-time than with MVPA may be related to the greater amount of energy expenditure produced the associated larger increase in LPA with reduction of SED-time, compared with the modest increase in MVPA. Moreover, while it is plausible that waist circumference, triglycerides, and systolic BP were preferentially associated with MVPA (28), it is difficult to explain the findings that HDL cholesterol and diastolic BP were independently predicted by SED-time only, though they are in keeping with previous observations (30,40). Finally, the independent association of CHD 10-year risk scores with both MVPA and SED-time reflects their relationships with parameters used for calculating these scores, with the stronger relationship with SED-time likely depending on its higher strength of association with HbA_1c compared with MVPA.

The main strengths of this study concern the trial design, i.e., the application of an intervention targeting both PA and SED-time, based on solid theoretical grounds, and using several behavioral change techniques, the specific training of investigators, the long study duration, and the large sample size. Other strengths include the objective (accelerometer-based) measurement of PA/SED-time and the concurrent assessment of physical fitness and CVD risk factors. However, this study has some limitations. First, generalizability requires further investigation and validation in different cohorts or settings. Second, this trial was not designed to assess the impact of changes in MVPA and SED-time on physical fitness and CVD risk factors/scores, which were secondary end points, and specific studies should be designed to assess the effect of different extents of increases in MVPA and/or decreases in SED-time on these variables. Third, results might have been affected by unmeasured confounders, e.g., diet, which were not considered in data analysis, though patients received dietary prescriptions and adherence to diet was verified at intermediate visits.

In conclusion, this analysis of the IDES_2 indicates that, in physically inactive and sedentary patients with type 2 diabetes, even modest increments in MVPA may have a clinically meaningful impact on physical fitness and cardiovascular risk profile and that reallocating SED-time to LPA may also contribute to improve these outcomes, possibly by increasing total energy expenditure. This implies that, in these individuals, the behavior change produced by counseling interventions targeting both MVPA and SED-time/LPA may be sufficient to ensure long-term benefits in terms of morbidity and mortality.

Acknowledgments. The authors thank the patients and the IDES_2 Investigators for participating in this study.

Funding. This work was supported by the Metabolic Fitness Association, Monterotondo, Rome, Italy.

The sponsor had no role in design or conduct of the study; collection, management, or interpretation of data; or preparation, review, or approval of the manuscript.

Duality of Interest. S.B. reported personal fees from AstraZeneca, Eli Lilly, Novo Nordisk, and Takeda. S.Z. is an employee of Technogym. A.N. reported grants from AstraZeneca, Eli Lilly, Novo Nordisk, and Sanofi and personal fees from Eli Lilly and Novo Nordisk. G.P. reported personal fees from AstraZeneca, Boehringer Ingelheim, Eli Lilly, Merck Sharp & Dohme, Mylan, Sigma-Tau Pharmaceuticals, and Takeda. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. S.B. and J.H. conceived and designed the study, researched data, and reviewed and edited the manuscript. M.S., G.O., P.C., M.V., L.M., C.I., L.B., F.C., and S.Z. researched data and contributed to the discussion. A.N. conceived and designed the study, performed statistical analysis, and reviewed and edited the manuscript. G.P. conceived and designed the study, researched data, performed statistical analysis, and drafted the manuscript. G.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.