# paper.pdf ## Metadata - PDFFormatVersion=1.7 - Language=en - IsLinearized=false - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - Author=Nick Perry - Creator=Microsoft® Word for Microsoft 365 - CreationDate=D:20260520172752-04'00' - ModDate=D:20260520172752-04'00' - Producer=Microsoft® Word for Microsoft 365 - pdf:producer=Microsoft® Word for Microsoft 365 - dc:creator=Nick Perry - xmp:creatortool=Microsoft® Word for Microsoft 365 - xmp:createdate=2026-05-20T17:27:52-04:00 - xmp:modifydate=2026-05-20T17:27:52-04:00 - xmpmm:documentid=uuid:36EB8D97-D385-4439-8F2C-A434B61BE9B2 - xmpmm:instanceid=uuid:36EB8D97-D385-4439-8F2C-A434B61BE9B2 ## Contents ### Page 1 Supplementary Online Content Zhang T, Zhang Y, Lee DH, et al. Long-term resistance training and risk of type 2 diabetes.JAMA Netw Open. 2026;9(6):e2619420. doi:10.1001/jamanetworkopen.2026.19420 eMethods. Study Population, Assessments, Ascertainment of T2D, Adiposity-Related Variables, and Analyses eTable 1. Baseline Characteristics by Nonresponse Before the End of the Study eTable 2. Baseline Characteristics by Availability of Resistance Training Information in Each Cohort eTable 3. Association Between Resistance Training and Risk of Type 2 Diabetes in Each Cohort eTable 4. Association Between Resistance Training and Type 2 Diabetes With Additional Adjustment for Baseline Waist Circumference, Intentional Weight Loss, and Time‑Updated BMI in Pooled Cohorts (NHS, NHS II, and HPFS) eTable 5. Stratified Analyses of the Association Between Resistance Training and Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 6. Association Between Resistance Training and Risk of Type 2 Diabetes AmongParticipants Meeting Aerobic Activity Guidelines (≥15 MET-h/wk) in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 7. Association Between Resistance Training and Risk of Type 2 Diabetes AmongParticipants Meeting at Least Twice the Aerobic Activity Guidelines (≥30 MET-h/wk) in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 8. Association Between Resistance Training and Risk of Type 2 Diabetes AmongParticipants Meeting at Least 3 Times the Aerobic Activity Guidelines (≥45 MET-h/wk) in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 9. Joint Association of Resistance Training and Body Mass Index Category With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 10. Joint Associations of Resistance Training and Walking With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 11. Time updated Characteristics During Follow-up According to Joint Categories of Aerobic Activity, Resistance Training, and Sedentary Television Viewing in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) ### Page 2 eTable 12. Joint Associations of Meeting Recommendations for Aerobic Activity and Resistance Training and Sedentary Television Viewing Time With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 13. Joint Associations of Meeting Recommendations for Aerobic Activity and Resistance Training and Categories of Total Sedentary Time With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 14. Joint Association of Upper- and Lower-Limb Resistance Training and Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS and NHS II) eTable 15. Time Updated Characteristics During Follow-up According to Cumulative Mean and Consistency of Resistance Training in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 16. Joint Association of Cumulative Average and Consistency of Resistance Training With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 17. Characteristics of Resistance Training Trajectory Groups at Ages 40, 52, and 60 Years in NHS II eTable 18. Association Between Resistance Training Trajectories and Type 2 Diabetes With Additional Adjustment for Baseline Waist Circumference, Intentional Weight Loss, and BMI in NHS II eTable 19. Sensitivity Analyses of Association Between Resistance Training and Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) eTable 20. Association of Resistance Training With Risk of Type 2 Diabetes in Pooled Cohorts (NHS, NHS II, and HPFS) Estimated Using Marginal Structural Models With Inverse Probability Weighting eFigure 1. Flow Diagram of Participant Inclusion in NHS, NHS II, and HPFS eFigure 2. Dose-Response Relationship Between Resistance Training and Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) # eFigure 3. Resistance Training Trajectories, Ages 40 to 60 Years, NHS II # eReferences # This supplementary material has been provided by the authors to give readers additional information about their work. ### Page 3 # eMethods. Study Population, Assessments, Ascertainment of T2D, Adiposity-Related Variables, and Analyses eMethod 1. Study population The NHS is a prospective cohort study of 121,700 female registered nurses aged 30-55 years at enrollment in 1976. The NHS II was established in 1989 and includes 116,429 younger female registered nurses aged 25-42 years. The HPFS began in 1986 and enrolled 51,529 male health professionals aged 40-75 years. In these cohorts, the cumulative follow‑up rate exceeds 90%,1 defined by the Channing Division of Network Medicine as the number of person-years in the cohort when participants are censored after their last questionnaire response or at death divided by the total number of person years in eachcohort. Thus, participants contribute person‑time until death or the end of follow‑up; if a participant doesnot respond to one or more intermediate questionnaires but later returns a subsequent questionnaire,person‑time is counted through the date of that subsequent response, and if a participant’s lastquestionnaire response is followed by a documented death, person‑time is counted through the date ofdeath. eMethod 2. Assessment of physical activity and sedentary behavior Physical activity was assessed using validated self-administered questionnaires and repeatedly measured during follow-up.2-4 The correlation between questionnaire-based physical activity and the average of four one-week activity diaries collected across different seasons was 0.62.2 Although resistance training validity was not directly assessed in all cohorts, evidence from a subsample of HPFS participants using a comparable resistance training question demonstrated a de-attenuated correlation of 0.79 between the 1992 questionnaire and the average of four seasonal activity diaries.3 In the NHS, resistance training questions were collected in 2000, 2004, 2008, 2012, and 2014, and in the NHS II in 2001, 2005, 2009, 2013 and 2017. In the HPFS, resistance training questions were collected in 1990, 1992, 1994, 1996, 1998, 2000, 2002, 2004, 2006, 2008, 2010, 2012, 2014 and 2016. Resistance training was assessed up to 5 times in NHS, 5 times in NHS II, and 14 times in HPFS. “up to 14 repeatedassessments” refers to the maximum number in HPFS. At each questionnaire cycle, participants reported their average weekly time over the past year spent on specific recreational activities, including resistance training (weight lifting using free weights or weight machines), walking (with pace categories), jogging, running, bicycling, tennis, swimming, other aerobic exercises (e.g., aerobics, dance, skiing, stair machines), other vigorous activities (e.g., rowing, squash, calisthenics), and daily flights of stairs climbed. For each activity, participants selected from predefined categories (none, 1-4 min/week, 5-19 min/week, 20-59 min/week, 1 h/wk, 1-1.5 h/wk, 2-3 h/wk, 4-6 h/wk, 7-10 h/wk, or≥11 h/wk), and the midpoint of each category was used to calculate hours/week. In NHS and NHS II, resistance training time was reported separately for the arms and legs, and the two values were summed to obtain total resistance training time. In the HPFS, the resistance training questionwas phrased as “Weightlifting or weight machine” before 2010. Starting in 2010, the question was revised to distinguish between upper- and lower-limb exercises, aligning with the format used in NHS and NHS II. For aerobic activities, total metabolic equivalents (MET)-h/wk were calculated by summing the MET- h/wk for all reported aerobic activities, with stair climbing first converted from flights/day to hours/week ### Page 4 before applying the corresponding MET value. Resistance training and aerobic activity were calculated as cumulative average across follow-up. Sedentary behaviors were assessed every 2 to 4 years with questions on the average time spent sitting at work or while driving, sitting at home watching television or videos, sitting while reading, and other sitting at home. These components were converted to hours per week and summed to derive total sedentary time. Response options were none, 1-4 h/wk, 5-9 h/wk, 10-14 h/wk, 15-20 h/wk, 21-40 h/wk, and ≥41 h/wk; midpoints were used in analyses. Correlations between inactivity reported in diaries and that reported on questionnaire were 0.41 for men and 0.41-0.44 for women.2,3 eMethod 3. Ascertainment of T2D The disease outcome of this study was incident T2D. If participants reported a T2D diagnosis on any of the follow-up questionnaires, a supplementary questionnaire was mailed to participants regarding symptoms, diagnostic testing, and treatment. Before 1998, the diagnostic criteria for T2D cases were from the National Diabetes Data Group:5 raised glucose concentration (fasting plasma glucose ≥7.8 mmol/L, random plasma glucose ≥11.1 mmol/L, or plasma glucose 11.1 mmol/L ≥2 hours after an oral glucose load) and at least one symptom related to diabetes (excessive thirst, polyuria, weight loss, or hunger); no symptoms, but raised concentrations of glucose on at least two occasions; or treatment with insulin or other hypoglycemic drugs. After 1998, T2D cases were diagnosed according to the ADA criteria, where the cut-off point for fasting plasma concentrations of glucose was lowered to 7.0 mmol/L.6 After January 2010, a glycated hemoglobin (HbA1c) concentration of 6.5% or more was additionally used as a diagnostic criterion for confirming participants with T2D.7 Cases of gestational diabetes and type 1 diabetes were excluded. The validity of the supplementary questionnaire for the diagnosis of diabetes was high, with 97% of diabetes cases confirmed by a review of medical records in a subsample of men (n=59) and 98% of diabetes cases confirmed in a subsample of women (n=62).8-10 Formal validation of non‑cases was not performed.However, because participants are health professionals with good access to medical care and regular healthscreening, including laboratory testing (especially when symptoms are present), long‑term undiagnosedT2D is likely less common than in the general population. In these cohorts, we have also observed relativelyhigh negative predictive values for other physician‑diagnosed conditions, such as periodontal disease, withnegative predictive values of 71–74%.11,12 eMethod 4. Assessment of covariates Biennial questionnaires included information on covariates: age, race, family history of T2D, body mass index (BMI), smoking status, menopausal status (for women only), and hormone therapy use (for women only). Dietary intake and alcohol consumption were assessed every four years. Diet quality was assessed using the Alternative Healthy Eating Index (AHEI)-2010 score. The validity and reproducibility of these self-reported variables (dietary, demographic, and lifestyle) have been well documented in previous studies.2-4,13-17 Multivariable models were adjusted for known and potential confounders, including race, family history of diabetes (yes or no), smoking status (never, past, or current [1-14, 15-24, or ≥25 cigarettes/day]), alcohol consumption (none, 0.1-4.9, 5-14.9, or ≥ 15 g/d), total calorie intake (quintiles), AHEI-2010 score ### Page 5 (quintiles), menopausal status (premenopausal or postmenopausal [never, past, or current hormone use, women only]), and aerobic activity (none, 0.1-7.5, 7.5-14.9, 15-29.9, or ≥30 MET-h/wk). These variables were selected a priori based on existing literature on physical activity and T2D risk 18,19 and on our prespecified causal framework for the association between resistance training and T2D, with the aim of capturing major sociodemographic and lifestyle confounders rather than mediators. Sex was collected as a biological variable at cohort enrollment. Race and ethnicity (White or other [identifying as African American, Asian, Hispanic, Native American, or other origin not captured by the listed categories]) were assessed to describe the study population and to account for potential differences in type 2 diabetes risk. Because these cohorts primarily enrolled White health professionals, most participants were White. eMethod 5. Handling of adiposity‑related variables Within our causal framework, adiposity-related variables may lie on the pathway between resistancetraining and type 2 diabetes (T2D), because long‑term resistance training can improve bodycomposition,20,21 including reductions in adiposity, which in turn may influence T2D risk. At the sametime, adiposity‑related variables also reflect broader healthy lifestyle patterns, such as a healthier diet, thatare related to both resistance training and lower T2D risk and may therefore serve as markers oftime‑varying confounding. For this reason, BMI, waist circumference, and intentional weight loss werenot included in the primary multivariable model, which aimed to estimate the total effect of resistance training on T2D. To explore the extent to which adiposity‑related factors and their correlates might account for the observed associations, we fitted supplementary Cox models with additional adjustment for time‑updatedBMI, treating it as time‑varying covariate updated at each questionnaire cycle. Because waist circumference and intentional weight loss were assessed at only a limited number of time points (waist circumference in NHS: 1986, 1996, 2000; NHS II: 1993, 2005, 2017; HPFS: 1987, 1996, 2008; and intentional weight loss in NHS: 1992; NHS II: 1993, 1997, 2005; HPFS: 1992, 2008), and the baselines for this analysis were 2002 in NHS, 2003 in NHS II, and 1992 in HPFS, we did not have sufficientlyfrequent or well‑aligned measurements to model time‑varying waist circumference or intentional weightloss without substantially restricting the analytic sample and follow‑up period. Therefore, waist circumference was modeled as a baseline (or pre‑baseline) covariate, categorized into three sex‑specificgroups (for women: <80, 80–<88,≥88 cm; for men: <94, 94–<102,≥102 cm), and intentional weightloss (yes/no) was modeled similarly using the measurement closest to the start of follow‑up. As a result, these analyses do not capture changes in adiposity or weight‑loss intention during follow‑up and shouldbe interpreted as exploratory. Any attenuation of the associations in these models was interpretedcautiously as being compatible with adiposity‑related pathways and improved control for residualconfounding. eMethod 6. 2-year lag We applied a 2‑year lag between exposure and outcome to reduce potential reverse causation. At eachquestionnaire cycle, resistance training reported at time t was related to T2D events occurring during the ### Page 6 subsequent 2‑year interval (t to t+2 years). For example, resistance training reported in 1990 was used tomodel T2D incidence during the 1992–1994 follow‑up period, and the cumulative average of resistancetraining in 1990 and 1992 was used to model T2D incidence during 1994–1996, and so on. Person‑yearswere accrued from baseline (2002 in NHS, 2003 in NHS II, and 1992 in HPFS) until the first diagnosis ofT2D, death, or the end of follow‑up (June 2021), whichever occurred first. eMethod 7. Joint analysis Joint analyses were implemented by creating cross‑classified categories of resistance training and thesecond exposure, and including these combined categories as the main exposure in the same Cox model, with the group in the lowest category of both exposures as the reference. These joint analyses were conducted separately for resistance training with BMI, walking (duration and pace), total aerobic activity, total sedentary time, and sedentary television viewing, as well as for combinations defined by upper vs.lower limb exercises and by long‑term training consistency. This approach is equivalent to modeling an interaction between categorized resistance training and the second exposure. Therefore, we did not includeadditional product (interaction) terms beyond the cross‑classified categories. Joint analyses of upper‑ and lower‑limb resistance training were conducted in NHS and NHS II only,because the resistance training question in HPFS was not consistently limb specific across questionnaire cycles. eMethod 8. Resistance training trajectories from ages 40 to 60 Years in NHS II For trajectory analyses, follow-up began at age 60 years in NHS II, and covariates were adjusted using their baseline values at age 60 years. Because most NHS participants were aged ≥60 years after 2002, making evaluation of ages 40-60 infeasible, and because the resistance training questions in HPFS changed across questionnaire cycles, we restricted trajectory analyses to the NHS II cohort. We included participants who had at least three assessments of resistance training between ages 40 and 60 years and were free of diabetes, myocardial infarction, coronary artery bypass graft, angina, stroke, or cancer at age 60 years. Long-term resistance training patterns at ages 40, 44, 48, 52, 56, and 60 years were classified into five trajectory groups using 0.5 h/wk as the minimum threshold at each assessment: Group 1 (consistently low; maximum <0.5 h/wk, consistently below the threshold), Group 2 (high-to-low; initial ≥0.5 h/wk and final <0.5 h/wk, with exactly one crossing of the threshold), Group 3 (low-to-high; initial <0.5 h/wk and final ≥0.5 h/wk, with exactly one crossing of the threshold), Group 4 (fluctuating; ≥2 crossings of the threshold), and Group 5 (consistently high; minimum ≥ 0.5 h/wk, consistently above the threshold). We then examined the association between these five trajectory groups and incident T2D after age 60 years using Cox models, with Group 1 as the reference. Because the number of incident T2D cases within each diagnosis period was too small to allow reliable stratified estimates, we did not further stratify the trajectory models by diagnosis period. eMethod 9. Marginal structural model with inverse probability weighting ### Page 7 To address time‑varying confounding more formally, we conducted sensitivity analyses using marginalstructural models (MSMs) with inverse probability weighting. For each questionnaire cycle, we estimated stabilized exposure weights for resistance training categories. The numerator of the weight was the probability of the observed resistance training level conditional on prior resistance training and baselinecovariates; the denominator additionally conditioned on time‑varying covariates. Baseline covariatesincluded race (White or non‑White) and family history of diabetes (yes or no). Time‑varying covariates included smoking status, alcohol consumption, total calorie intake, AHEI‑2010 score, menopausal statusand postmenopausal hormone use (women only), aerobic activity, and BMI, all categorized as in the primary Cox models. Stabilized censoring weights were constructed analogously, and the product of exposure and censoring weights was applied. Weighted pooled logistic regression with generalized estimating equations was used to approximate Cox models and obtain robust 95% CIs. Because BMI may act as both a mediator and a confounder, it was included in the weight models but was not additionally adjusted for in the outcome models.22 ### Page 8 # eTable 1. Baseline Characteristics by Nonresponse Before the End of the Study For each cohort, participants were defined as “lost” if their last returned questionnaire during follow‑up occurred before the end of follow‑up andno information on death or incident type 2 diabetes was available at any time during follow‑up. Participants were defined as “retained” if they returned a questionnaire at or after the end of follow‑up or if information on death or incident type 2 diabetes was available at any time duringfollow‑up. Values are means (SDs) for continuous variables and percentages for categorical variables, standardized to the age distribution of the study population except for age. Percentages may not sum to 100% because of rounding. ### Page 9 Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; HPFS, Health Professionals Follow-up Study; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; TV, television; SD, standard deviation. ### Page 10 # eTable 2. Baseline Characteristics by Availability of Resistance Training Information in Each Cohort ### Page 11 Values are means (SDs) for continuous variables and percentages for categorical variables, standardized to the age distribution of the study population except for age. Percentages may not sum to 100% because of rounding. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; HPFS, Health Professionals Follow-up Study; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; TV, television; SD, standard deviation. ### Page 12 # eTable 3. Association Between Resistance Training and Risk of Type 2 Diabetes in Each Cohort Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 NHS Cases 2003 352 170 166 115 Person-years 392765 114712 55234 58366 41208 Model 1 1 (Reference) 0.77 (0.68-0.86) 0.73 (0.63-0.86) 0.62 (0.53-0.72) 0.54 (0.45-0.65) Model 2 1 (Reference) 0.81 (0.72-0.91) 0.81 (0.69-0.95) 0.70 (0.60-0.83) 0.63 (0.52-0.77) Model 3 1 (Reference) 0.84 (0.75-0.95) 0.90 (0.77-1.06) 0.81 (0.68-0.95) 0.76 (0.63-0.93) NHS II Cases 2967 910 407 365 251 Person-years 524297 253063 131243 149782 113973 Model 1 1 (Reference) 0.67 (0.62-0.72) 0.57 (0.51-0.63) 0.44 (0.40-0.50) 0.41 (0.36-0.47) Model 2 1 (Reference) 0.76 (0.70-0.82) 0.69 (0.62-0.77) 0.56 (0.50-0.63) 0.53 (0.47-0.61) Model 3 1 (Reference) 0.82 (0.75-0.88) 0.83 (0.74-0.92) 0.72 (0.64-0.81) 0.77 (0.67-0.88) HPFS Cases 1570 444 150 110 58 Person-years 364693 147186 56611 49661 27903 Model 1 1 (Reference) 0.76 (0.68-0.85) 0.66 (0.55-0.78) 0.56 (0.46-0.68) 0.53 (0.40-0.69) Model 2 1 (Reference) 0.82 (0.73-0.92) 0.72 (0.61-0.86) 0.64 (0.52-0.78) 0.60 (0.46-0.79) Model 3 1 (Reference) 0.87 (0.78-0.97) 0.81 (0.68-0.97) 0.72 (0.59-0.88) 0.69 (0.53-0.90) ### Page 13 Model 1: Adjusted for age (months) and calendar year. Model 2: Model 1 plus race (White or other [identifying as African American, Asian, Hispanic, Native American, or other origin not captured by the listed categories]), family history of diabetes (yes or no), smoking status (never, past, or current [1-14, 15-24, or ≥25 cigarettes per day]), alcohol consumption (none, 0.1-4.9, 5-14.9, or ≥15 g/d), total calorie intake (quintile), AHEI-2010 score (quintile), menopausal status (in women only, premenopausal or postmenopausal [never, past, or current postmenopausal hormone use]). Model 3: Model 2 plus aerobic activity (none, 0.1-7.5, 7.5-14.9, 15-29.9, or ≥30 MET-h/wk). A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task;NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 14 # eTable 4. Association Between Resistance Training and Type 2 Diabetes With Additional Adjustment for Baseline Waist Circumference, Intentional Weight Loss, and Time‑Updated BMI in Pooled Cohorts (NHS, NHS II, and HPFS) Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 Cases 6540 1706 727 641 424 Person-years 1281755 514961 243087 257810 183085 Model a 1 (Reference) 0.84 (0.79-0.89) 0.82 (0.76-0.89) 0.75 (0.69-0.82) 0.78 (0.70-0.87) Model b 1 (Reference) 0.81 (0.77-0.86) 0.78 (0.72-0.85) 0.71 (0.65-0.78) 0.74 (0.66-0.82) Model c 1 (Reference) 0.89 (0.84-0.94) 0.87 (0.80-0.94) 0.80 (0.73-0.87) 0.86 (0.77-0.95) Model a: Additionally adjusted for baseline waist circumference, in addition to the covariates in multivariable model 3 in Table 2. Baseline waistcircumference was categorized into three sex‑specific groups (for women: <80, 80-<88, and≥88 cm; for men: <94, 94-<102, and≥102 cm). Model b: Additionally adjusted for intentional weight loss at baseline (yes/no), in addition to the covariates in multivariable model 3 in Table 2. Model c: Additionally adjusted for time‑updated BMI (<25, 25-29.9, or≥30 kg/m²), in addition to the covariates in multivariable model 3 in Table 2. A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; CI, confidence interval; HR, hazard ratio; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 15 # eTable 5. Stratified Analyses of the Association Between Resistance Training and Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 Age (years) < 65 1 (Reference) 0.81 (0.75-0.87) 0.79 (0.72-0.88) 0.69 (0.62-0.77) 0.74 (0.65-0.84) ≥ 65 1 (Reference) 0.89 (0.81-0.97) 0.92 (0.81-1.04) 0.83 (0.72-0.95) 0.76 (0.64-0.91) BMI (kg/m2) < 29.9 1 (Reference) 0.88 (0.82-0.95) 0.88 (0.79-0.97) 0.79 (0.71-0.88) 0.79 (0.70-0.90) ≥ 30 1 (Reference) 0.93 (0.85-1.02) 0.90 (0.78-1.03) 0.79 (0.68-0.91) 0.85 (0.71-1.02) Alcohol intake < 15 g/d 1 (Reference) 0.83 (0.78-0.88) 0.84 (0.77-0.91) 0.74 (0.67-0.81) 0.75 (0.67-0.84) ≥ 15 g/d 1 (Reference) 0.85 (0.71-1.02) 0.83 (0.63-1.08) 0.70 (0.53-0.94) 0.61 (0.42-0.89) AHEI-2010 score ### Page 16 < median (55) 1 (Reference) 0.81 (0.75-0.87) 0.85 (0.77-0.95) 0.73 (0.65-0.82) 0.79 (0.69-0.91) ≥ median (55) 1 (Reference) 0.88 (0.80-0.96) 0.84 (0.74-0.95) 0.76 (0.67-0.86) 0.71 (0.61-0.83) Family history of diabetes No 1 (Reference) 0.78 (0.73-0.85) 0.81 (0.73-0.91) 0.69 (0.61-0.78) 0.75 (0.65-0.86) Yes 1 (Reference) 0.90 (0.82-0.97) 0.86 (0.76-0.97) 0.80 (0.70-0.90) 0.73 (0.63-0.86) Diagnosis period Before January 2010 1 (Reference) 0.73 (0.67-0.79) 0.74 (0.65-0.84) 0.68 (0.60-0.77) 0.76 (0.66-0.88) January 2010 or later 1 (Reference) 0.89 (0.81-0.97) 0.83 (0.73-0.94) 0.76 (0.66-0.88) 0.65 (0.53-0.79) Menopausal period (NHS and NHS II only) Premenopausal 1 (Reference) 0.68 (0.57-0.80) 0.74 (0.59-0.93) 0.58 (0.46-0.74) 0.78 (0.61-1.00) Postmenopausal 1 (Reference) 0.82 (0.76-0.88) 0.82 (0.74-0.91) 0.77 (0.69-0.85) 0.77 (0.67-0.87) All models were adjusted for the same covariates as multivariable Model 3 shown in Table 2, except for the stratification variable. ### Page 17 A 2‑year lag between the assessment of resistance training and the start of follow‑up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). P values for interaction between resistance training and each stratification variable were all≥0.05. Abbreviations: AHEI‑2010, Alternate Healthy Eating Index 2010; BMI, body mass index; CI, confidence interval; HR, hazard ratio; MET, metabolicequivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow‑up Study. ### Page 18 # eTable 6. Association Between Resistance Training and Risk of Type 2 Diabetes Among Participants Meeting Aerobic Activity Guidelines (≥15 MET-h/wk) in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 Cases 2029 775 419 424 335 Person-years 484173 262457 155563 183431 143465 Model 1 1 (Reference) 0.76 (0.70-0.83) 0.70 (0.62-0.78) 0.61 (0.55-0.68) 0.61 (0.54-0.69) Model 2 1 (Reference) 0.82 (0.75-0.89) 0.77 (0.69-0.86) 0.69 (0.62-0.77) 0.71 (0.62-0.80) Model 1: Adjusted for age (months), calendar year, and cohort. Model 2: Model 1 plus race (White or other [identifying as African American, Asian, Hispanic, Native American, or other origin not captured by the listed categories]), family history of diabetes (yes or no), smoking status (never, past, or current [1-14, 15-24, or ≥25 cigarettes per day]), alcohol consumption (none, 0.1-4.9, 5-14.9, or ≥15 g/d), total calorie intake (quintile), AHEI-2010 score (quintile), menopausal status (in women only, premenopausal or postmenopausal [never, past, or current postmenopausal hormone use]). A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task;NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 19 # eTable 7. Association Between Resistance Training and Risk of Type 2 Diabetes Among Participants Meeting at Least Twice the Aerobic Activity Guidelines (≥30 MET-h/wk) in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 Cases 801 322 190 204 217 Person-years 221347 119035 78352 102973 99861 Model 1 1 (Reference) 0.80 (0.70-0.92) 0.75 (0.63-0.88) 0.63 (0.54-0.74) 0.71 (0.60-0.83) Model 2 1 (Reference) 0.86 (0.74-0.98) 0.82 (0.69-0.97) 0.72 (0.61-0.86) 0.82 (0.70-0.97) Model 1: Adjusted for age (months), calendar year, and cohort. Model 2: Model 1 plus race (White or other [identifying as African American, Asian, Hispanic, Native American, or other origin not captured by the listed categories]), family history of diabetes (yes or no), smoking status (never, past, or current [1-14, 15-24, or ≥25 cigarettes per day]), alcohol consumption (none, 0.1-4.9, 5-14.9, or ≥15 g/d), total calorie intake (quintile), AHEI-2010 score (quintile), menopausal status (in women only, premenopausal or postmenopausal [never, past, or current postmenopausal hormone use]). A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task;NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 20 # eTable 8. Association Between Resistance Training and Risk of Type 2 Diabetes Among Participants Meeting at Least 3 Times the Aerobic Activity Guidelines (≥45 MET-h/wk) in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 Cases 336 143 75 90 114 Person-years 106224 53348 37578 51866 62242 Model 1 1 (Reference) 0.88 (0.71-1.09) 0.72 (0.55-0.94) 0.66 (0.51-0.85) 0.71 (0.56-0.90) Model 2 1 (Reference) 0.90 (0.72-1.12) 0.78 (0.59-1.03) 0.72 (0.55-0.93) 0.79 (0.62-1.01) Model 1: Adjusted for age (months), calendar year, and cohort. Model 2: Model 1 plus race (White or other [identifying as African American, Asian, Hispanic, Native American, or other origin not captured by the listed categories]), family history of diabetes (yes or no), smoking status (never, past, or current [1-14, 15-24, or ≥25 cigarettes per day]), alcohol consumption (none, 0.1-4.9, 5-14.9, or ≥15 g/d), total calorie intake (quintile), AHEI-2010 score (quintile), menopausal status (in women only, premenopausal or postmenopausal [never, past, or current postmenopausal hormone use]). A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task;NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 21 # eTable 9. Joint Association of Resistance Training and Body Mass Index Category With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Group Cases Person-years HR (95% CI) BMI <25 + RT ≥1 h/wk 257 283633 1 (Reference) BMI <25 + RT <1 h/wk 1482 1027753 1.29 (1.13-1.47) BMI 25-29.9 + RT ≥1 h/wk 460 120382 3.85 (3.30-4.48) BMI 25-29.9 + RT <1 h/wk 3705 710655 4.33 (3.80-4.93) BMI ≥30 + RT ≥1 h/wk 348 36880 8.16 (6.93-9.61) BMI ≥30 + RT <1 h/wk 3786 301396 9.20 (8.07-10.5) Body mass index (BMI) categories: <25, 25-29.9, ≥30 kg/m². All models were adjusted for the covariates included in multivariable Model 3 in Table 2. A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; CI, confidence interval; HR, hazard ratio; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study; RT, resistance training; TV, television. ### Page 22 # eTable 10. Joint Associations of Resistance Training and Walking With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Group Cases Person-years HR (95% CI) RT <1 h/wk, walking <3.5 h/wk, nonbrisk pace 4065 548866 1 (Reference) RT <1 h/wk, walking <3.5 h/wk, brisk pace 3280 835467 0.56 (0.53-0.60) RT <1 h/wk, walking ≥3.5 h/wk, nonbrisk pace 685 130460 0.86 (0.78-0.95) RT <1 h/wk, walking ≥3.5 h/wk, brisk pace 883 290830 0.52 (0.47-0.56) RT ≥1 h/wk, walking <3.5 h/wk, nonbrisk pace 255 52858 0.86 (0.75-0.98) RT ≥1 h/wk, walking <3.5 h/wk, brisk pace 487 204603 0.45 (0.40-0.50) RT ≥1 h/wk, walking ≥3.5 h/wk, nonbrisk pace 73 18094 0.87 (0.68-1.12) RT ≥1 h/wk, walking ≥3.5 h/wk, brisk pace 234 119989 0.42 (0.36-0.48) Walking pace was categorized as nonbrisk (<3 mph) and brisk (≥3 mph). All models were adjusted for the same covariates included in the multivariable Model 2 shown in Table 2 plus aerobic activity other than walking (none, 0.1-7.5, 7.5-14.9, 15.0-29.9, or ≥30.0 MET-h/wk). A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; CI, confidence interval; HPFS, Health Professionals Follow-up Study; HR, hazard ratio; MET, metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; RT, resistance training; TV, television. ### Page 23 # eTable 11. Time Updated Characteristics During Follow-up According to Joint Categories of Aerobic Activity, Resistance Training, and Sedentary Television Viewing in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) ### Page 24 ### Page 25 Meeting the recommendation for resistance training was defined as performing≥1.0 h/wk. Meeting the recommendation for aerobic activity was defined as engaging in≥15 MET‑h/wk. Meeting the recommendation for television viewing was defined as watching <2 h/d of television. Values are means (SDs) for continuous variables and percentages for categorical variables, and are standardized to the age distribution of the study population, except for age. Percentages may not sum to 100% owing to rounding. All variables are time‑updated unless otherwise specified. Abbreviations: AHEI‑2010, Alternate Healthy Eating Index 2010; BMI, body mass index; HPFS, Health Professionals Follow‑up Study; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; TV, television; SD, standard deviation. ### Page 26 # eTable 12. Joint Associations of Meeting Recommendations for Aerobic Activity and Resistance Training and Sedentary Television Viewing Time With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Group Cases Person-years HR (95% CI) None met recommendation 1982 299358 1 (Reference) RT met recommendation 98 18254 0.89 (0.72-1.09) Aerobic met recommendation 899 231050 0.66 (0.61-0.72) Aerobic + RT met recommendations 193 62746 0.57 (0.49-0.67) TV met recommendation 3864 749589 0.80 (0.76-0.85) TV + RT met recommendations 200 59574 0.57 (0.49-0.66) TV + Aerobic met recommendations 2228 759805 0.52 (0.49-0.56) All three met recommendations 574 300321 0.38 (0.34-0.42) Meeting the recommendation for RT was defined as performing≥1.0 h/wk. Meeting the recommendation for aerobic activity was defined as engaging in≥15 MET-h/wk. Meeting the recommendation for TV viewing was defined as watching television for <2 h/d. All models were adjusted for the same covariates included in the multivariable Model 2 shown in Table 2. A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; CI, confidence interval; HR, hazard ratio; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study; RT, resistance training; TV, television. ### Page 27 # eTable 13. Joint Associations of Meeting Recommendations for Aerobic Activity and Resistance Training and Categories of Total Sedentary Time With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Group Cases Person-years HR (95% CI) None met recommendation 1982 299358 1 (Reference) RT met recommendation 188 43096 0.73 (0.63-0.85) Aerobic met recommendation 1251 355536 0.62 (0.58-0.66) Aerobic + RT met recommendations 407 164441 0.48 (0.43-0.53) Total sedentary time met recommendations 2116 355663 0.87 (0.82-0.92) Total sedentary time + RT met recommendations 102 25608 0.69 (0.57-0.85) Total sedentary time + Aerobic met recommendations 1972 546656 0.57 (0.53-0.62) All three met recommendations 352 162454 0.42 (0.37-0.47) Meeting the recommendation for RT was defined as performing≥1.0 h/wk. Meeting the recommendation for aerobic activity was defined as engaging in≥15 MET-h/wk. Meeting the recommendation for total sedentary time was defined as spending less than the median level of sedentary time (23 h/wk; 3.3 h/d). All models were adjusted for the same covariates included in the multivariable Model 2 shown in Table 2. A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; CI, confidence interval; HR, hazard ratio; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study; RT, resistance training; TV, television. ### Page 28 # eTable 14. Joint Association of Upper- and Lower-Limb Resistance Training and Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS and NHS II) Group Cases Person-years HR (95% CI) No RT 5082 971186 1 (Reference) Upper limb RT only 534 185290 0.75 (0.69-0.83) Lower limb RT only 97 20570 1.16 (0.95-1.43) Upper and lower limb RT 1844 655817 0.83 (0.78-0.88) All models were adjusted for the covariates included in multivariable model 3 in Table 2. Upper limb indicates participation in upper-limb RT; lower limb, participation in lower-limb RT. A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task;NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 29 # eTable 15. Time Updated Characteristics During Follow-up According to Cumulative Mean and Consistency of Resistance Training in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Consistency <50% 50-74.9% ≥75% Cumulative average <0.5 h/wk 0.5-0.9 h/wk ≥1 h/wk <0.5 h/wk 0.5-0.9 h/wk ≥1 h/wk 0.5-0.9 h/wk ≥1 h/wk Person-years 1794914 109930 39169 27460 106000 127992 31207 278486 Age, years 63.4 (11.2) 65.2 (9.9) 66.4 (10.7) 60.7 (10.8) 61.4 (10.4) 61.4 (10.5) 61.6 (11.1) 59.9 (10.4) White, % 96.6 97 96.5 97.3 97 96.9 97.1 97.2 BMI, kg/m2 25.9 (4.5) 25.5 (4.2) 25.7 (4.4) 24.6 (4.1) 24.8 (4) 25.1 (4.1) 24 (3.7) 24 (3.5) Resistance training, h/wk 0.1 (0.1) 0.7 (0.1) 1.7 (1.1) 0.4 (0.1) 0.7 (0.1) 1.8 (1.2) 0.8 (0.1) 2.5 (1.8) Aerobic activity, MET- h/wk 19.6 (20) 28.2 (21.6) 32.5 (25.5) 23 (18) 28.3 (21.5) 33.6 (25.8) 29.1 (20.6) 40.4 (30.5) Sedentary TV viewing, h/wk 10.4 (7.8) 9.3 (6.5) 9.4 (6.6) 10.4 (7.5) 9.7 (7) 9.4 (6.9) 10.1 (7.7) 9 (6.8) Smoking status ### Page 30 ### Page 31 Consistency was defined as the proportion of returned questionnaires on which participants reported engaging in resistance training for≥0.5 h per week. Cumulative average resistance training was defined as the average weekly hours of resistance training across all questionnaires up to eachfollow‑up cycle. The joint category “consistency≥75% and cumulative average <0.5 h/wk” is not shown because of the extremely small number of person‑years (n=50) and zero incident type 2 diabetes cases, which precluded reliable estimation of the hazard ratio. Values are means (SDs) for continuous variables and percentages for categorical variables, standardized to the age distribution of the study population except for age. Percentages may not sum to 100% because of rounding. All variables are time‑updated unless otherwise specified. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; BMI, body mass index; HPFS, Health Professionals Follow-up Study; MET,metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; TV, television; SD, standard deviation. ### Page 32 # eTable 16. Joint Association of Cumulative Average and Consistency of Resistance Training With Risk of Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Group Cases Person-years HR (95% CI) Consistency <50% + Cumulative average <0.5 h/wk 8447 1794914 1 (Reference) Consistency 50-74.9% + Cumulative average <0.5 h/wk 75 27460 0.68 (0.54-0.85) Consistency <50% + Cumulative average 0.5-0.9 h/wk 371 109930 1.06 (0.96-1.19) Consistency 50-74.9% + Cumulative average 0.5-0.9 h/wk 302 106000 0.77 (0.68-0.86) Consistency≥75% + Cumulative average 0.5-0.9 h/wk 76 31207 0.66 (0.53-0.83) Consistency <50% + Cumulative average≥1 h/wk 114 39169 1.04 (0.86-1.25) Consistency 50-74.9% + Cumulative average≥1 h/wk 405 127992 0.91 (0.83-1.01) Consistency≥75% + Cumulative average≥1 h/wk 574 278486 0.65 (0.60-0.72) Consistency was defined as the proportion of returned questionnaires on which participants reported engaging in resistance training for≥0.5 h per week. Cumulative average resistance training was defined as the average weekly hours of resistance training across all questionnaires up to eachfollow‑up cycle. The joint category “consistency≥75% and cumulative average <0.5 h/wk” is not shown because of the extremely small number of person‑years (n=50) and zero incident type 2 diabetes cases, which precluded reliable estimation of the hazard ratio. All models were adjusted for the covariates included in multivariable model 3 in Table 2. A 2-year lag between resistance training assessment and start of follow-up was applied in all models. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task;NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 33 # eTable 17. Characteristics of Resistance Training Trajectory Groups at Ages 40, 52, and 60 Years in NHS II ### Page 34 ### Page 35 ### Page 36 Participants who had at least three assessments of resistance training between ages 40 and 60 years and were free of diabetes, myocardial infarction, coronary artery bypass graft, angina, stroke, or cancer at age 60 years were included. Groups 1-5 represent long-term resistance training trajectories between ages 40 and 60 years (assessed at 40, 44, 48, 52, 56, and 60 years) using 0.5 h/wk as the threshold: Group 1, consistently low; Group 2, high-to-low; Group 3, low-to-high; Group 4, fluctuating; and Group 5, consistently high. Values are means (SDs) for continuous variables and percentages for categorical variables, standardized to the age distribution of the study population except for age. Percentages may not sum to 100% because of rounding. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; COPD, chronic obstructive pulmonary disease;MET, metabolic equivalent oftask; NHS II, Nurses’ Health Study II. ### Page 37 # eTable 18. Association Between Resistance Training Trajectories and Type 2 Diabetes With Additional Adjustment for Baseline Waist Circumference, Intentional Weight Loss, and BMI in NHS II Group 1 (consistently low) Group 2 (high-to- low) Group 3 (low-to- high) Group 4 (fluctuating) Group 5 (consistently high) Cases 1155 120 151 171 71 Person- years 301336 52421 71513 64652 60196 Model A 1 (Reference) 0.86(0.71-1.05) 0.79(0.67-0.95) 0.98(0.83-1.16) 0.63(0.49-0.81) Model B 1 (Reference) 0.83(0.69-1.01) 0.79(0.66-0.94) 0.94(0.80-1.11) 0.60(0.47-0.77) Model C 1 (Reference) 0.84(0.69-1.02) 0.87(0.73-1.03) 0.96(0.82-1.14) 0.71(0.55-0.91) For trajectory analyses, follow-up began at age 60 years in NHS II, and covariates were adjusted using their baseline values at age 60 years. Model a: Additionally adjusted for baseline waist circumference, in addition to the covariates in multivariable model 3 in Table 3. Baseline waist circumference was categorized into three groups (for women: <80, 80-<88, and≥88 cm). Model b: Additionally adjusted for intentional weight loss at baseline (yes/no), in addition to the covariates in multivariable model 3 in Table 3. Model c: Additionally adjusted for baseline BMI (<25, 25-29.9, or≥30 kg/m²), in addition to the covariates in multivariable model 3 in Table 3. Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task;NHS II, Nurses’ Health Study II. ### Page 38 # eTable 19. Sensitivity Analyses of Association Between Resistance Training and Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 Simple updated resistance training Cases 7887 784 181 567 619 Person-years 1649111 269157 78105 215689 268636 Model 1 1 (Reference) 0.61 (0.56-0.65) 0.49 (0.42-0.57) 0.54 (0.49-0.59) 0.48 (0.44-0.52) Model 2 1 (Reference) 0.67 (0.62-0.72) 0.57 (0.49-0.66) 0.62 (0.57-0.68) 0.57 (0.52-0.62) Model 3 1 (Reference) 0.71 (0.65-0.76) 0.65 (0.56-0.75) 0.72 (0.66-0.79) 0.70 (0.64-0.76) Cumulative average resistance training with no time lag Cases 7514 2048 856 718 474 Person-years 1471135 595676 273974 291254 211115 Model 1 1 (Reference) 0.72 (0.68-0.76) 0.63 (0.58-0.67) 0.49 (0.45-0.53) 0.45 (0.41-0.49) Model 2 1 (Reference) 0.79 (0.75-0.83) 0.73 (0.68-0.78) 0.59 (0.54-0.63) 0.55 (0.50-0.60) Model 3 1 (Reference) 0.84 (0.80-0.88) 0.85 (0.79-0.92) 0.72 (0.67-0.78) 0.73 (0.67-0.81) Excluding participants with aerobic activity >30 MET-h/wk Cases 5773 1412 540 441 199 Person-years 1040565 388283 158478 145704 71306 Model 1 1 (Reference) 0.72 (0.67-0.76) 0.65 (0.59-0.71) 0.55 (0.50-0.61) 0.50 (0.43-0.58) Model 2 1 (Reference) 0.78 (0.73-0.83) 0.74 (0.67-0.81) 0.64 (0.58-0.71) 0.59 (0.51-0.68) Model 3 1 (Reference) 0.82 (0.77-0.87) 0.83 (0.75-0.91) 0.74 (0.67-0.82) 0.70 (0.60-0.80) Excluding participants with baseline COPD and arthritic disease Cases 4172 1172 457 410 263 Person-years 806609 348501 167585 179836 127758 Model 1 1 (Reference) 0.67 (0.63-0.72) 0.53 (0.48-0.58) 0.45 (0.40-0.50) 0.41 (0.36-0.46) Model 2 1 (Reference) 0.73 (0.68-0.79) 0.62 (0.56-0.68) 0.53 (0.48-0.59) 0.50 (0.44-0.57) Model 3 1 (Reference) 0.79 (0.73-0.84) 0.72 (0.65-0.80) 0.66 (0.59-0.73) 0.67 (0.58-0.76) ### Page 39 Model 1: Adjusted for age (months), calendar year, and cohort. Model 2: Model 1 plus race (White or other [identifying as African American, Asian, Hispanic, Native American, or other origin not captured by the listed categories]), family history of diabetes (yes or no), smoking status (never, past, or current [1-14, 15-24, or ≥25 cigarettes per day]),alcohol consumption (none, 0.1-4.9, 5-14.9, or ≥15 g/d), total calorie intake (quintile), AHEI-2010 score (quintile), menopausal status (in women only, premenopausal or postmenopausal [never, past, or current postmenopausal hormone use]). Model 3: Model 2 plus aerobic activity (none, 0.1-7.5, 7.5-14.9, 15-29.9, or ≥30 MET-h/wk). Unless otherwise indicated, values are hazard ratios (HRs) with 95% confidence intervals (CIs). Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; COPD, chronic obstructive pulmonary disease; HR,hazard ratio; MET, metabolic equivalent of task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 40 # eTable 20. Association of Resistance Training With Risk of Type 2 Diabetes in Pooled Cohorts (NHS, NHS II, and HPFS) Estimated Using Marginal Structural Models With Inverse Probability Weighting Resistance training, h/wk 0 >0 to <0.5 0.5 to <1.0 1.0 to <2.0 ≥2.0 Cases 6540 1706 727 641 424 MSM Model 1 (Reference) 0.87 (0.76-1.01) 0.71 (0.60-0.84) 0.54 (0.45-0.65) 0.56 (0.41-0.77) MSM model: Estimated using marginal structural models with stabilized inverse probability weights. Data are hazard ratios (HRs) and 95% confidence intervals (CIs), unless otherwise indicated. Abbreviations: AHEI-2010, Alternate Healthy Eating Index 2010; CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task; MSM, marginal structural model; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 41 # eFigure 1. Flow Diagram of Participant Inclusion in NHS, NHS II, and HPFS Abbreviations: NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health ProfessionalsFollow-up Study. ### Page 42 # eFigure 2. Dose-Response Relationship Between Resistance Training and Type 2 Diabetes in Pooled Prospective Cohorts (NHS, NHS II, and HPFS) Restricted cubic spline models with three knots at the 50th, 75th, and 90th percentiles were fitted, adjusting for the same covariates as in multivariable Model 3 in Table 2. The solid line indicates hazard ratios, and the dashed lines indicate 95% confidence intervals. The reference level is 0 h/wk of resistance training. A 2‑year lag was applied between the assessment of resistance training and the start of follow-up for incident type 2 diabetes. P for linearity <0.001; P for nonlinearity <0.001. Abbreviations: CI, confidence interval; HR, hazard ratio; MET, metabolic equivalent of task; NHS,Nurses’ Health Study; NHS II, Nurses’ Health Study II; HPFS, Health Professionals Follow-up Study. ### Page 43 # eFigure 3. Resistance Training Trajectories, Ages 40 to 60 Years, NHS II Trajectories of resistance training levels from age 40 to 60 years are shown for 5 distinct groups. Lines represent mean resistance training level (h/wk) within each trajectory group, and shaded areas represent 95% CIs. Abbreviations: CI, confidence interval; NHS II, Nurses’ Health Study II. ### Page 44 # eReferences 1.Knudsen MD, Wang K, Wang L, et al. Development and validation of a risk prediction model for post-polypectomy colorectal cancer in the USA: a prospective cohort study. EClinicalMedicine. Aug 2023;62:102139. doi:10.1016/j.eclinm.2023.102139 2. Wolf AM, Hunter DJ, Colditz GA, et al. Reproducibility and validity of a self-administered physical activity questionnaire. International journal of epidemiology. Oct 1994;23(5):991-9. doi:10.1093/ije/23.5.991 3. Chasan-Taber S, Rimm EB, Stampfer MJ, et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. Jan 1996;7(1):81-6. doi:10.1097/00001648-199601000-00014 4. Al-Shaar L, Pernar CH, Chomistek AK, et al. Reproducibility, Validity, and Relative Validity of Self- Report Methods for Assessing Physical Activity in Epidemiologic Studies: Findings From the Women's Lifestyle Validation Study. Am J Epidemiol. Mar 24 2022;191(4):696-710. doi:10.1093/aje/kwab294 5. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. National Diabetes Data Group. Diabetes. Dec 1979;28(12):1039-57. doi:10.2337/diab.28.12.1039 6. Expert Committee on the D, Classification of Diabetes M. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes Care. Jan 2003;26 Suppl 1:S5-20. doi:10.2337/diacare.26.2007.s5 7. Standards of medical care in diabetes--2010. Diabetes Care. Jan 2010;33 Suppl 1(Suppl 1):S11- 61. doi:10.2337/dc10-S011 8. Hu FB, Leitzmann MF, Stampfer MJ, Colditz GA, Willett WC, Rimm EB. Physical activity and television watching in relation to risk for type 2 diabetes mellitus in men. Archives of internal medicine. Jun 25 2001;161(12):1542-8. doi:10.1001/archinte.161.12.1542 9. Manson JE, Stampfer MJ, Colditz GA, et al. Physical activity and incidence of non-insulin- dependent diabetes mellitus in women. The Lancet. 1991;338(8770):774-778. doi:10.1016/0140- 6736(91)90664-B 10. Field AE, Coakley EH, Must A, et al. Impact of Overweight on the Risk of Developing Common Chronic Diseases During a 10-Year Period. Archives of internal medicine. 2001;161(13):1581-1586. doi:10.1001/archinte.161.13.1581 %J Archives of Internal Medicine 11. Joshipura KJ, Douglass CW, Garcia RI, Valachovic R, Willett WC. Validity of a self-reported periodontal disease measure. Journal of public health dentistry. Summer 1996;56(4):205-12. doi:10.1111/j.1752-7325.1996.tb02437.x 12. Joshipura KJ, Pitiphat W, Douglass CW. Validation of self-reported periodontal measures among health professionals. Journal of public health dentistry. Spring 2002;62(2):115-21. doi:10.1111/j.1752- 7325.2002.tb03431.x 13. Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. Jul 1985;122(1):51-65. doi:10.1093/oxfordjournals.aje.a114086 14. Yuan C, Spiegelman D, Rimm EB, et al. Validity of a Dietary Questionnaire Assessed by Comparison With Multiple Weighed Dietary Records or 24-Hour Recalls. Am J Epidemiol. Apr 1 2017;185(7):570-584. doi:10.1093/aje/kww104 ### Page 45 15. Al-Shaar L, Yuan C, Rosner B, et al. Reproducibility and Validity of a Semiquantitative Food Frequency Questionnaire in Men Assessed by Multiple Methods. Am J Epidemiol. Jun 1 2021;190(6):1122-1132. doi:10.1093/aje/kwaa280 16. Colditz GA, Martin P, Stampfer MJ, et al. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol. May 1986;123(5):894- 900. doi:10.1093/oxfordjournals.aje.a114319 17. Troy LM, Hunter DJ, Manson JE, Colditz GA, Stampfer MJ, Willett WC. The validity of recalled weight among younger women. International journal of obesity and related metabolic disorders : journal of the International Association for the Study of Obesity. Aug 1995;19(8):570-2. 18. Grøntved A, Rimm EB, Willett WC, Andersen LB, Hu FB. A prospective study of weight training and risk of type 2 diabetes mellitus in men. Archives of internal medicine. Sep 24 2012;172(17):1306-12. doi:10.1001/archinternmed.2012.3138 19. Grontved A, Pan A, Mekary RA, et al. Muscle-strengthening and conditioning activities and risk of type 2 diabetes: a prospective study in two cohorts of US women.PLoS Med. Jan2014;11(1):e1001587. doi:10.1371/journal.pmed.1001587 20.Al-Mhanna SB, Franklin BA, Jakicic JM, et al. Impact of resistance training on cardiometabolic health-related indices in patients with type 2 diabetes and overweight/obesity: a systematic review and meta-analysis of randomised controlled trials. Br J Sports Med. May 2 2025;59(10):733-746. doi:10.1136/bjsports-2024-108947 21. Mekary RA, Grøntved A, Despres JP, et al. Weight training, aerobic physical activities, and long- term waist circumference change in men. Obesity (Silver Spring). Feb 2015;23(2):461-7. doi:10.1002/oby.20949 22. Hernan MA, Brumback B, Robins JM. Marginal structural models to estimate the causal effect of zidovudine on the survival of HIV-positive men. Epidemiology. Sep 2000;11(5):561-70. doi:10.1097/00001648-200009000-00012