Association of Objectively Measured Timing of Physical Activity Bouts With Cardiovascular Health in Type 2 Diabetes

The prevalence of obesity and type 2 diabetes has risen dramatically over recent decades (1). According to the National Diabetes Statistics Report 2020, an estimated 34.2 million people in the U.S. have diabetes, and another 88 million adults have prediabetes (2). Regular physical activity (PA) is an important component of blood glucose management and overall cardiovascular health in individuals with diabetes and prediabetes. Moderate- to vigorous-intensity PA (MVPA), often defined as requiring a moderate to large amount of effort and with a notably to substantial acceleration in heart rate, improves cardiorespiratory fitness and is associated with a substantially lower risk of cardiovascular complications and overall mortality in type 2 diabetes (3,4). While the total amount of MVPA performed is a significant determinant of cardiorespiratory fitness and decreased risk for cardiovascular diseases, few studies have examined the role of the time of day of bout-related MVPA, and it remains unclear whether timing of MVPA associates with these health benefits of MVPA.

Animal studies have shown that metabolic responses to a bout of MVPA vary depending on the time of day, which suggests time-of-day–dependent beneficial effects of MVPA potentially mediated by the endogenous circadian system (5,6). Indeed, some human interventional studies have found that timing of exercise regulates weight loss (7,8), glucose control (9), and acute cardiovascular responses (10,11). However, these studies assessed the effects of time of day of structured exercise in relatively smaller groups of participants or for a short duration. Given that not all free-living MVPA involves exercise, bouts of MVPA can occur in the forms of housework, commuting, or occupation-related PA. Thus, it is important to determine whether the temporal distribution of MVPA accumulated in multiple bouts across the day is associated with health benefits in both healthy participants and patient populations.

Large cohort studies with objectively measured PA provide an essential opportunity to address these research questions. However, there are no epidemiological studies that have examined the association between timing of bout-related MVPA and cardiometabolic risk factors, likely because of the lack of established methods to characterize the temporal distribution of bout-related MVPA. Here, we use the baseline data from the Look AHEAD (Action for Health in Diabetes) study, a landmark clinical trial in which adults with overweight/obesity and type 2 diabetes were randomized to either an intensive lifestyle intervention or diabetes support and education to investigate the effects of the lifestyle intervention on cardiometabolic outcomes. The goals of the current study were to 1) characterize the temporal distribution (timing) of bout-related MVPA in adults with overweight/obesity and type 2 diabetes; 2) define whether the timing of bout-related MVPA varies according to sociodemographic or anthropometric factors at baseline; and 3) determine associations of timing of bout-related MVPA with baseline cardiorespiratory fitness, an independent predictor for cardiovascular complications and overall mortality (12), and risk for coronary heart disease (CHD).

Participants for this study are a subgroup of those who completed baseline testing before being randomized into the multicenter Look AHEAD study, which was a trial examining the effect of intensive lifestyle intervention on the primary and secondary prevention of cardiovascular diseases in adults with type 2 diabetes and overweight/obesity (13). The complete inclusion and exclusion criteria have been previously reported (14). Briefly, participants had been diagnosed with type 2 diabetes, had a BMI ≥25 kg/m² (or ≥27 kg/m² when taking insulin), and were 45–76 years of age. Of the 5,145 participants enrolled in the Look AHEAD study, 2,153 were enrolled at eight clinical sites participating in the accelerometer substudy and had baseline recordings; thus, they are included in the current analyses. Descriptive data for the accelerometer subgroup (compared with the entire Look AHEAD sample) have been reported previously (15). All participants provided written informed consent, and study procedures were approved by each enrollment center’s institutional review board. The analysis was deemed exempt from review by the Brigham and Women’s Hospital Institutional Review Board (Protocol # 2019P002292).

Participants were instructed to wear the hip-mounted RT3 triaxial accelerometer (Stayhealthy, Monrovia, CA) for 7 consecutive days during waking hours, removing it only for periods of sleep, bathing, showering, or other water-based activities (16). Participants were also instructed not to alter their habitual PA pattern while wearing this device. The data collection mode for the accelerometer was set in the three-axis and 1-min epoch mode, and various quality control procedures were implemented.

Data reduction criteria for the accelerometer data were similar to what has been used previously (16–18). In short, nonwear time was defined by an interval ≥60 consecutive min of no activity counts, with allowance for 1–2 min of activity ≤1.5 METs. Wear time was determined by subtracting nonwear time from 24 h. A valid day was defined as a that in which the accelerometer was worn for ≥10.5 h. According to Miller et al. (16), this allows >80% of data to be valid. Participants with <4 valid wear days were excluded in the analyses (n = 70). Since this study focuses on individuals with a diurnal lifestyle, we excluded participants with high nocturnal activity (>10% activity counts between 0100 and 0400 h, n = 29). We excluded participants with spurious long total duration of MVPA (≥median + 3.5 × interquartile range [∼180 min MVPA/day, top 0.6%], n = 13). We also excluded participants with abnormal recordings (i.e., no or very few zero-activity recordings throughout, n = 3), which indicates device malfunction. These criteria left us with 2,038 participants qualified for further analyses.

The intensity of PA was expressed in METs by dividing the estimated energy expenditure per minute by estimated resting energy expenditure per minute as computed using the proprietary software provided by Stayhealthy (18). Periods of MVPA were defined as ≥3 METs (19). Bout-related MVPA was defined as any activity ≥3 METs and ≥10 min in duration, allowing for a 1–2-min interruption in MVPA (20). Bout-related METs × min/week was calculated by summing the MET values for each minute identified as part of an MVPA bout and then adjusted for the number of valid days.

A questionnaire was used to assess age, sex, race/ethnicity, current diabetes treatment, education, current smoking status, and history of cardiovascular diseases. History of cardiovascular diseases was based on self-report of myocardial infarction, coronary revascularization, stroke, transient ischemic attack, heart failure, or peripheral arterial revascularization. Depression symptoms were assessed by the Beck Depression Inventory (21). Height, weight, BMI, waist circumference, blood pressure, and other serum measures were assessed using standard procedures (14). Cardiorespiratory fitness was defined as the estimated level of METs (1 MET = 3.5 mL/kg/min of oxygen uptake) achieved on a maximal graded exercise treadmill test (22). In brief, this test was terminated at the point of volitional exhaustion or when the test termination criteria were observed (23). A baseline test was considered valid if the maximal heart rate was ≥85% of age-predicted maximal heart rate (220 − age) for a participant not taking a β-adrenergic–blocking medication (β-blocker). If a participant was taking a β-blocker, the baseline test was considered valid if the rate of perceived exertion was ≥18 at the point of termination. Four-year Framingham risk score (FRS) was derived for each individual using the sex-specific prediction algorithms proposed by D’Agostino et al. (24). The algorithms estimate risk of any CHD event from age, plasma levels of triglycerides, total cholesterol, HDL cholesterol, systolic blood pressure, presence or absence of diabetes, and smoking status by using separate prediction equations for participants with and without a history of CHD or stroke.

To assign categories of timing of bout-related MVPA, we first examined the distribution of bout-related MVPA from the overall study population across the clock hours (Supplementary Fig. 1A). After excluding bouts in the overnight hours (0000–0500 h, the 5 clock hours with the lowest numbers of bouts [<20/h], total excluding 78 of 12,267 bouts from 2,038 participants), we divided the remaining clock hours into four time windows in a way that each window had a quarter of total MVPA bouts: morning (0500–1042 h), midday (1043–1342 h), afternoon (1343–1659 h), and evening (1700–2400 h). For the participants who either did not have bout-related MVPA or only had bout-related MVPA in 1 day, we assigned them into the inactive group. For those who had bout-related MVPA on at least 2 different days, if ≥50% of their bout-related MVPA volume (calculated as METs × min) occurred during the same time window as defined above, we considered them as engaged in temporally consistent bout-related MVPA and classified them into the corresponding timing group. For example, if someone spent ≥50% of his or her bout-related MVPA volume between 0500 and 1042 h, then this individual was categorized into the morning group. For those who had <50% of the bout-related MVPA volume in any of the four time windows, we considered them as showing inconsistent temporal distribution of bout-related MVPA and classified them into the mixed group. Representative examples for each of the active groups are shown in Supplementary Fig. 1B. A detailed breakdown of the percentage of bout-related MVPA in each time window across the different groups is shown in Supplementary Fig. 1C. There were three participants who had ≥50% of their bout-related MVPA volume between 0000 and 0459 h and were excluded from the analyses. See Supplementary Fig. 2 for a Consolidated Standards of Reporting Trials diagram.

Less than 1% of the covariate data (including sociodemographic data, anthropometric data, smoking status, and Beck depression score) were missing. We examined the pattern and distribution of complete and incomplete covariate data and found that a single covariate is not predictive of incomplete data. Therefore, we assumed that the data were missing completely at random and used predictive mean matching with R 3.5.1 statistical software (R Foundation for Statistical Computing, Vienna, Austria) using the package “mice” (25) to impute the missing covariate data. Missing data on cardiorespiratory fitness, 4-year FRS, and exposure (timing of bout-related MVPA) were not imputed.

Descriptive statistics (number, mean, and SD) and percentages for categorical variables were calculated for participant baseline characteristics. To test differences among the six timing of MVPA bout groups, we used the Kruskal-Wallis test for continuous variables and χ² test for categorical variables; χ² statistics were also calculated for each cell to determine the most contributing cell.

Our primary outcome was cardiorespiratory fitness. The secondary outcome was 4-year FRS. To determine the association between the outcomes and timing of bout-related MVPA, we used multivariable-adjusted linear regression to estimate the β-coefficient and 95% CIs across the six timing of bout-related MVPA groups. The reference group was defined as the mixed group in all analyses. Model 1 was adjusted for sex, age, race, education, clinic site, current smoking status, diabetes duration, Beck depression score, and β-blocker usage (for cardiorespiratory fitness only). We also explored other covariates, such as employment status (full time/part time vs. unemployed/retired) and marital status (never married, married/marriage-like relationship, widowed, divorced/separated, other). Since including these other variables in the models did not have a major impact on the results, we omitted these other covariates in the main analysis. To determine whether any potential associations could be explained by cumulative weekly bout-related MVPA and bout intensity, we adjusted for bout-related METs × min/week and average intensity of MVPA bouts (METs/min) in model 2. To test whether any potential associations with 4-year FRS could be explained by cardiorespiratory fitness, we further added cardiorespiratory fitness as a covariate to model 2. The relative significance of each group in the models was appraised using the Wald test.

In secondary analyses, given the known sex differences in the circadian system and exercise physiology (26–28), we tested for effect modification in the above models using product terms of sex with timing of bout-related MVPA group and then stratified the analyses by sex. In general, two-sided P values of 0.05 were considered for statistical significance. To reduce type I error for multiple comparisons, Bonferroni correction was applied for stratified analyses. Statistical analyses were conducted with R 3.5.1 statistical software using the package “car” (29).

Baseline characteristics for the 2,035 participants included in the current analyses are shown in Table 1 by timing of bout-related MVPA groups and in Supplementary Table 1 by sex. Sociodemographic data, separated by timing of bout-related MVPA group, showed that the inactive group contributed the most to the group differences in sociodemographic factors, with more women (67.6% vs. 47.7–59.3%), more African American participants (22.3% vs. 12.6–20.7%), and fewer participants with graduate school education (24.2% vs. 29.9–35.0%) compared with the other groups. The highest amount of bout-related MVPA occurred in the mixed group (mean ± SD 1,080.9 ± 902.2 METs × min/week), while the inactive group, by definition, demonstrated lowest bout-related MVPA (67 ± 138.7 METs × min/week). Assessment of lifestyle factors in the different groups indicated a significant difference by group in the Beck depression score, with higher scores seen in the inactive group and lowest scores in the evening group. There were no significant differences in BMI, blood pressure, smoking status, diabetes duration, fasting glucose and HbA_1c, and lipid parameters among the six groups.

Cardiorespiratory fitness differed by timing of bout-related MVPA group (P_trend < 0.0001) (Table 2 and Fig. 1A). Compared with the mixed group, the inactive group had lower cardiorespiratory fitness (mean β = −0.46 [95% CI −0.67, −0.26] METs), whereas the morning group showed higher cardiorespiratory fitness (0.25 [0, 0.5] METs). Importantly, the group differences in cardiorespiratory fitness remained after further adjustment for bout-related METs × min/week and mean MVPA bout intensity (P_trend = 0.0005). Here, the inactive, morning, and evening groups all had higher cardiorespiratory fitness (0.44 [0.21, 0.68], 0.25 [0.01, 0.49], and 0.29 [0.04, 0.54] METs, respectively) than the mixed group, indicating that the association between timing of bout-related MVPA and cardiorespiratory fitness was independent of the overall volume of bout-related MVPA as well as the intensity of MVPA bouts. Moreover, the association between cardiorespiratory fitness and timing of bout-related MVPA varied by sex (P_interaction = 0.01). In men, cardiorespiratory fitness was lower in the midday group (−0.46 [−0.87, −0.06] METs) and tended to be higher in the morning group (0.36 [−0.03, 0.75] METs) than in the mixed group (P_trend = 0.02), independent of the volume of bout-related MVPA and mean MVPA bout intensity. In women, cardiorespiratory fitness, compared with the mixed group, was lower in the inactive group and higher in the evening group (−0.26 [−0.5, −0.02] and 0.35 [0.04, 0.66] METs, respectively; P_trend = 0.0006). However, such a relationship in women was altered in the model with adjustment for the volume and intensity of bout-related MVPA (P_trend = 0.0005). Here, the inactive group, similar to the evening group and afternoon group, also had higher cardiorespiratory fitness than the mixed group (0.64 [0.36, 0.92], 0.36 [0.03, 0.7], and 0.47 [0.18, 0.77] METs, respectively) (Fig. 1A, red).

We next determined whether the risks for CHD as estimated by 4-year FRS differed by timing of MVPA at baseline in the Look AHEAD participants. In the total population, we found significant group difference in the 4-year FRS at baseline (P_trend = 0.02) (Table 3 and Fig. 1B), although the significant association was lost after further adjustment for overall bout-related MVPA and mean MVPA bout intensity (P_trend = 0.32). We also observed significant sex differences in this association (P_interaction = 0.0008). In women, 4-year FRS was not significantly associated with timing of bout-related MVPA (P_trend = 0.26). Unexpectedly, in men, such an association was significant (P_trend = 0.01), with the inactive and morning groups having a higher 4-year FRS compared with the mixed group (β = 1.94% [95% CI 0.63%, 3.26%] and 2.27% [0.80%, 3.74%], respectively). Further adjustment for bout-related METs × min/week and MVPA bout intensity did not alter the association (P_trend = 0.04), with the morning group having a higher 4-year FRS (2.18% [0.70%, 3.65%]) compared with the mixed group, equivalent to an approximately one-fifth increase of the average CHD risk among all male participants (Supplementary Table 1). Given that cardiorespiratory fitness may affect the CHD risks and vice versa, we also considered cardiorespiratory fitness as a possible confounder. After additional adjustment for cardiorespiratory fitness, the significant association in men persisted (P_trend = 0.004), with the morning group still showing the highest 4-year FRS among all groups (2.37% [0.90%, 3.82%]) and cardiorespiratory fitness negatively associated with FRS (P < 0.0001).

In the present cross-sectional analysis of a large cohort of adults with type 2 diabetes and overweight/obesity, we demonstrated that timing of bout-related MVPA was associated with cardiorespiratory fitness after controlling for age, sex, race/ethnicity, and other demographic characteristics. Importantly, such an association was independent of overall levels of bout-related MVPA and MVPA bout intensity and varied by sex. Specifically, men who performed the majority of bout-related MVPA in the midday had the lowest cardiorespiratory fitness, while those who were most active in the morning hours tended to have the highest cardiorespiratory fitness. In women, the evening group had higher cardiorespiratory fitness, although this may be largely explained by the group difference in overall bout-related MVPA and MVPA bout intensity. This is because after further adjustment for the volume and intensity of MVPA bouts, while the evening group still had higher cardiorespiratory fitness than the mixed group, its differences with other groups became smaller. Interestingly, we also found a significant association between timing of bout-related MVPA and 4-year FRS in men but not in women. Men who performed the majority of bout-related MVPA in the morning hours had higher 4-year FRSs, indicating a higher chance of developing CHD in the next 4 years independent of their overall levels of bout-related MVPA, MVPA bout intensity, and cardiorespiratory fitness. Our findings underscore the importance of incorporating timing as a crucial dimension in PA-related research (30,31).

To our knowledge, this is the first large cross-sectional study to describe timing of MVPA using objectively measured MVPA and to examine the associations of timing of MVPA with cardiorespiratory fitness and risks for CHD. Previous studies on PA patterns have mostly focused on frequency, intensity, and volume (12,32). Because the adaptive responses to a bout of PA are dose dependent (32), we focused on PA at a relatively high level (i.e., moderate to vigorous intensity) to detect a robust effect of time of day. We excluded light-intensity PA from the analysis because this type of PA tends to occur throughout the waking hours, which obscures any temporal distribution of the more effective PA with higher intensity. The methods used to characterize each individual’s timing of bout-related MVPA are similar to what has been used previously to characterize timing of exercise (7,33). These studies used predetermined time windows for group classification, whereas we used the population’s natural distribution of objectively measured bout-related MVPA across the day to divide the time windows. Our method may be more suitable for large cohort studies since it allows each time-specific group to have a comparable number of participants, thus avoiding misinterpretation as a result of small subgroup size. Moreover, not all bout-related MVPA takes place in the form of structured exercise. The current method allows for capturing the full spectrum of the time-of-day–dependent associations of bout-related MVPA.

Our findings are in line with the growing evidence on the link between the circadian system and exercise physiology. Animal data show that the adaptive metabolic responses to a bout of MVPA vary depending on time of day and are directly under the regulation of the clock genes (5,6). Human exercise intervention trials have found time-specific improvements in glucose control (9) and blood pressure (34). As for cardiorespiratory fitness, previous studies are scarce and inconclusive (35–38). These studies were performed in a small number of young, healthy adults with controlled exercise training sessions; thus, their results might not be directly comparable to our findings with objectively measured MVPA in older individuals with type 2 diabetes. The adjusted 0.82-MET difference between the morning and midday groups in cardiorespiratory fitness in men is considered equivalent to a 2.9 mL/kg/min difference in oxygen consumption. For clinical relevance, this is similar in magnitude to an ∼12% increase in VO_2max in adults with type 2 diabetes after a 20-week moderate-intensity exercise training with an average of 49-min sessions occurring 3.4 times per week (4). Moreover, a difference of 0.82 METs approximates an 8–16% reduction in disease-specific mortality and nonfatal CHD events in middle-aged subjects (39). It is worth noting that although increased MVPA volume and intensity improve cardiorespiratory fitness (40) and the morning group is the second most active, total MVPA level does not explain the observed time-of-day–specific association in men. Neither does the intensity of bout-related MVPA, indicating that the association between timing of bout-related MVPA and cardiorespiratory fitness is not dose or intensity dependent. A potential explanation of these findings could be that the mode of MVPA that the participants engaged in the morning and evening is more effective in improving cardiorespiratory fitness than those performed in other time windows. For example, aerobic training induces more cardiorespiratory adaptation than resistance training or other physical labor (41), although accelerometer-based PA data cannot distinguish types of activity, leaving us unable to test this hypothesis. Another possibility is that the efficacy of bout-related MVPA on cardiorespiratory fitness varies at different times of the day in older men with type 2 diabetes. However, the underlying mechanism remains to be elucidated, which could be related to other behavioral factors (e.g., fasting/postprandial states, sleep-wake cycles) and/or the circadian regulation of the adaptive responses to PA (5,6). The reverse explanation could also be true: Men with less fitness were more active in the midday, while those with higher fitness chose to be active in the morning. Future prospective studies and/or randomized controlled trials are needed to test these hypotheses, which will provide essential insight on PA prescription to optimize its efficacy.

We also found a significant association between timing of bout-related MVPA and estimated risks for CHD, which was primarily driven by men. As expected, the inactive group in both men and women had higher 4-year FRSs, although this could be mostly explained by the volume and the intensity of MVPA bouts. However, in men, the significant association between timing of bout-related MVPA and FRS was independent of the volume and intensity of MVPA bouts. In particular, the male morning group had the highest 4-year FRSs, showing a 2.18% higher adjusted probability of developing CHD than the mixed group. Such a difference is clinically relevant since it is about one-fifth of the average 4-year FRS (12.2%) among all male participants. Our cross-sectional analysis cannot determine the direction of this association. Given that the male morning group also had relatively high cardiorespiratory fitness, which has a known protective effect on CHD (39), it is unlikely that morning bout-related MVPA contributes to a higher CHD risk, at least through changes in cardiorespiratory fitness. Indeed, we confirmed that the association in men persisted in the further adjusted model, and cardiorespiratory fitness as an added covariate is associated with reduced risk of CHD. Another possibility is that men with higher risk were more likely to perform MVPA in the morning, although it remains a question why this preference existed in men with type 2 diabetes, which may be addressed by future studies. In addition, other factors that were not available in this analysis (e.g., timing of food intake, sleep duration, types of occupation, interactions between the timing of PA and the circadian system on cardiovascular risk factors [10,11,42]) may play a role in associations between morning bout-related MVPA and CHD risks. To address the seemingly paradoxical associations of morning bout-related MVPA with higher cardiorespiratory fitness and CHD risks in men, future longitudinal studies are needed to investigate the prospective association between timing of bout-related MVPA with improvement in cardiorespiratory fitness and incidence of cardiovascular events and the potential mechanisms. Understanding such relationships not only is scientifically important but also may inform PA interventions in clinical populations.

Another interesting observation of the current study is the sex differences in the aforementioned associations: both more prominent in men than in women. This finding is novel and warrants confirmation. While women were generally less physically active than men in the current study population (18), it does not explain the observed sex differences since further adjustment for activity level and intensity did not change most of the results. Other potential explanations may be related to sex-specific physiology and risk of CHD. For example, a meta-analysis found that VO_2max is improved to a greater extent in men than in women in response to a sex-matched dose of endurance training (43). Another contributing factor may be the overall lower 4-year FRS in women (4.5%) versus men (12.2%) in this cohort. Future studies are needed to explore whether the observed sex differences in these associations persist with matching male and female populations.

This study has several strengths, including its large sample size and the availability of objectively measured PA in a well-characterized cohort of Look AHEAD participants. The study also has limitations. First, the presented data are cross-sectional, and no conclusions about the directionality or causality of the findings can be drawn. Yet, the current findings warrant further assessment of the time-specific effects of MVPA on cardiovascular health in the follow-up years of the Look AHEAD study. Second, we used clock time to define timing of bout-related MVPA. Given the individual differences in circadian timing relative to clock hour, it may be more physiologically relevant to use timing of bout-related MVPA relative to endogenous circadian timing (44), which was not available for this data set. Third, we used accelerometer epoch length recordings of 60 s, which was a consensus at the time of data collection. Another study suggested that shorter epoch length can retain more information that may allow for differentiating activity types (45), which could have allowed us to answer the question about mode of MVPA. Finally, sleep patterns, nutritional intake, and timing of food intake may be potential confounders, but these data were not available for the current analyses.

In conclusion, we developed a novel, quantitative method to characterize the timing of bout-related MVPA in participants of the Look AHEAD study. We determined that timing of bout-related MVPA was associated with cardiorespiratory fitness, especially in men. We also identified an association between morning bout-related MVPA and higher risk of CHD in men compared with later timing of PA. These associations are independent of the volume and the intensity of MVPA bouts, although these findings warrant further investigation in future studies with in-depth information on sleep patterns and nutrient intake. Being physically active remains one of the cornerstones in the prevention and treatment of CHD, obesity, and type 2 diabetes. Taken together, these data may contribute to developing individualized exercise prescriptions for people with and without type 2 diabetes.

Acknowledgments. The authors thank the other investigators, the staff, and the participants of the Look AHEAD study for valuable contributions.

Funding. This study was supported by the National Heart, Lung, and Blood Institute (NHLBI) (K99-HL-148500, J.Q., principal investigator [PI]). The Look AHEAD study was supported by the U.S. Department of Health and Human Services through the following cooperative agreements from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK): DK-57136, DK-57149, DK-56990, DK-57177, DK-57171, DK-57151, DK-57182, DK-57131, DK-57002, DK-57078, DK-57154, DK-57178, DK-57219, DK-57008, DK-57135, and DK-56992. The following federal agencies have contributed support: NIDDK, NHLBI, National Institute of Nursing Research, National Institute on Minority Health and Health Disparities, Office of Research on Women’s Health, and the Centers for Disease Control and Prevention. This research was supported in part by the Intramural Research Program of the NIDDK. The Indian Health Service provided personnel, medical oversight, and use of facilities. Additional support was received from The Johns Hopkins Medical Institutions Bayview General Clinical Research Center (M01-RR-02719), the Johns Hopkins-University of Maryland Diabetes Research and Training Center (P60-DK-079637), the Massachusetts General Hospital Mallinckrodt General Clinical Research Center (M01-RR-01066), the University of Colorado Health Sciences Center General Clinical Research Center (M01-RR-00051), the University of Colorado Health Sciences Center Clinical Nutrition Research Unit (P30-DK-48520), the University of Tennessee at Memphis General Clinical Research Center (M01-RR-0021140), the University of Pittsburgh General Clinical Research Center (M01-RR-00005644), NIDDK grant DK-046204, and the University of Washington/VA Puget Sound Health Care System Medical Research Service, Department of Veterans Affairs, Frederic C. Bartter General Clinical Research Center (M01-RR-01346). Other funding for individual investigators includes the National Institute on Aging (RF1-AG-059867 and RF1-AG-064312, K.H., PI), NHLBI (R01-HL-140574, F.A.J.L.S. and K. Tavakkoli, PIs), and NIDDK (K23-DK-114550, R.J.W.M., PI).

The opinions and views expressed in this article are those of the authors and do not necessarily represent the views of the Indian Health Service or other funding sources, the NIDDK, the National Institutes of Health, or the U.S. Department of Health and Human Services.

Duality of Interest. The following organizations have committed to make major contributions to Look AHEAD: Federal Express, Health Management Resources, Johnson & Johnson, LifeScan Inc., Optifast-Novartis Nutrition, Roche Pharmaceuticals, Ross Product Division of Abbott Laboratories, SlimFast Foods Company, and Unilever. J.M.J. is on the scientific advisory board for WW International, Inc., and Naturally Slim and serves as an advisor to Spark360. F.A.J.L.S. received speaker fees from Bayer Healthcare, Sentara Healthcare, Philips, Kellogg Company, Vanda Pharmaceuticals, and Pfizer. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.Q. designed the study, conducted the data analysis, and drafted the manuscript. M.P.W., S.-H.C., P.H.B., D.S.B., and P.A.R. critically revised the manuscript for important intellectual content. J.M.J., K.H., F.A.J.L.S., and R.J.W.M. assisted with the study design, provided input on the data analysis and interpretation of results, and reviewed and edited the manuscript. J.Q. and R.J.W.M. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 80th Scientific Sessions of the American Diabetes Association, Virtual, 12–16 June 2020.

Data Sharing. The data used for analysis during the current study were housed at the data coordinating center and are not available for public distribution. Baseline data on Look AHEAD participants have been supplied to the NIDDK Central Repository and are publicly available at https://repository.niddk.nih.gov/studies/look-ahead/?query=Look%20AHEAD.