A Randomized Controlled Trial of Resistance Exercise Training to Improve Glycemic Control in Older Adults With Type 2 Diabetes

The researchers tested whether supervised, high-intensity progressive resistance training (PRT) — essentially weightlifting — could improve blood sugar control in older adults who already had type 2 diabetes. The intervention was a structured program of machine-based exercises (leg press, chest press, seated row, knee extension, knee flexion, biceps curl, triceps pushdown, and abdominal crunch) performed three times per week for 16 weeks. Each session lasted approximately 45 minutes and involved three sets of 8–10 repetitions at 75–85% of each person's one-repetition maximum (1RM), meaning they lifted weights heavy enough that they could only complete 8–10 reps before muscle failure.

The comparator was a control group that received standard medical care and was instructed to maintain their usual physical activity levels and dietary habits. They did not receive any exercise intervention or placebo exercise.

The primary outcome was glycemic control, measured by glycosylated hemoglobin (HbA1c), which reflects average blood sugar over the previous 2–3 months. Secondary outcomes included fasting plasma glucose, fasting insulin, muscle glycogen stores (measured via muscle biopsy), body composition (lean mass, fat mass, trunk fat), blood pressure, lipid profile (total cholesterol, LDL, HDL, triglycerides), and diabetes medication dosage.

The study enrolled 62 Latino older adults (40 women, 22 men) living in the Boston, Massachusetts area. All participants were aged 60 years or older (mean age 66 ± 8 years, range 60–80) and had a confirmed diagnosis of type 2 diabetes for at least 3 years. All were overweight or obese (mean BMI approximately 30 kg/m²). Participants were required to be sedentary — defined as not having engaged in regular exercise for at least 6 months prior to enrollment — and to have stable diabetes medication for at least 3 months before the study began. Exclusion criteria included insulin use, severe diabetic complications (proliferative retinopathy, nephropathy with creatinine >2.0 mg/dL, neuropathy affecting balance), uncontrolled hypertension (>160/95 mmHg), cardiovascular disease within the past 6 months, musculoskeletal conditions that would prevent resistance training, and current participation in any other exercise program.

**Glycemic control:** Glycosylated hemoglobin (HbA1c) measured by high-performance liquid chromatography (normal range 4–6%, diabetic target typically <7%). Fasting plasma glucose measured by glucose oxidase method. Fasting insulin measured by radioimmunoassay.

**Muscle glycogen:** Percutaneous muscle biopsy from the vastus lateralis (thigh muscle), with glycogen content measured biochemically and expressed as mmol glucose per kg of wet muscle weight.

**Body composition:** Dual-energy X-ray absorptiometry (DXA) scanning, which provides precise measurements of total lean mass, total fat mass, and regional fat distribution (trunk fat, leg fat, arm fat).

**Blood pressure:** Seated resting blood pressure measured by mercury sphygmomanometer after 5 minutes of quiet rest, with three readings averaged.

**Lipid profile:** Total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides measured from fasting blood samples using standard enzymatic methods.

**Strength:** One-repetition maximum (1RM) testing on each exercise machine at baseline and after 16 weeks, defined as the maximum weight that could be lifted once with proper form.

**Medication tracking:** Diabetes medication type and dosage recorded at baseline and weekly throughout the study. Medication changes were made by participants' primary care physicians who were blinded to group assignment.

**Study design:** This was a randomized controlled trial (RCT) — the gold standard for establishing cause-and-effect relationships. Participants were randomly assigned to either the resistance training group (n=31) or the control group (n=31) using a computer-generated random number sequence. Randomization was stratified by sex to ensure balanced representation of men and women in both groups.

**Blinding:** This was a single-blind study. The outcome assessors (laboratory technicians performing blood analyses, DXA technicians, and muscle biopsy analysts) were blinded to group assignment. However, participants obviously knew whether they were exercising or not, and the exercise trainers were also unblinded. This is a common limitation in exercise trials — you cannot easily give a placebo exercise. The lack of participant blinding means that expectation effects (placebo effects) could influence some outcomes, particularly subjective measures. However, the primary outcome (HbA1c) is an objective blood biomarker, which is less susceptible to expectation bias.

**Duration:** The intervention lasted 16 weeks, with exercise sessions three times per week (48 total sessions). This is a moderate-to-long duration for an exercise trial. Sixteen weeks is sufficient to see meaningful changes in HbA1c (which reflects the previous 8–12 weeks of blood sugar), muscle mass, and strength. It is not long enough to assess long-term adherence, sustainability of effects, or hard outcomes like cardiovascular events or mortality.

**Supervision and adherence:** All exercise sessions were supervised one-on-one by exercise trainers. This ensured proper form, safety, and progressive overload. Adherence was tracked — participants completed 89% of scheduled sessions on average (range 72–100%). This high adherence rate is typical of supervised, short-term trials and does not reflect what would happen in real-world, unsupervised settings.

**Progression protocol:** The resistance training was progressive. Every two weeks, participants were retested on their 1RM for each exercise, and the training weight was adjusted upward to maintain 75–85% of their current maximum. This ensured that the stimulus remained challenging throughout the 16 weeks.

**Statistical approach:** The primary analysis was intention-to-treat — meaning all participants were analyzed in the group they were originally assigned to, regardless of whether they completed the intervention. This is the conservative approach and preserves the benefits of randomization. Between-group differences were analyzed using analysis of covariance (ANCOVA) with baseline values as covariates, which adjusts for any baseline differences between groups. Results are reported as mean ± standard error (SE). P-values <0.05 were considered statistically significant.

**What this design can and cannot prove:**

**Can prove:** That supervised, high-intensity resistance training causes improvements in glycemic control, body composition, and blood pressure compared to no exercise in this specific population over 16 weeks. The randomization and control group allow causal inference.

**Cannot prove:** That resistance training is better than other forms of exercise (e.g., aerobic training, combined training) — there was no aerobic exercise comparison group. Cannot prove long-term sustainability or real-world effectiveness — the supervised setting is artificial. Cannot prove that the effects generalize to other populations (younger adults, non-Latino ethnicities, insulin users, people with complications). Cannot determine the optimal dose (frequency, intensity, volume) — only one dose was tested.

**Methodological weaknesses:**

No aerobic exercise comparison group, so we cannot say resistance training is superior or inferior to the more commonly prescribed aerobic exercise for diabetes.

No sham exercise control (e.g., stretching or light walking), so the control group received no attention or social contact from trainers. The exercise group got 48 sessions of one-on-one attention, which could produce psychosocial benefits independent of the exercise itself.

Small sample size (n=62 total, n=31 per group) limits statistical power for subgroup analyses and increases the risk of baseline imbalances despite randomization.

Single ethnic group (Latino) limits generalizability.

Medication changes were made by participants' own physicians, not by a standardized protocol, introducing variability.

Muscle glycogen was measured only in a subset of participants (n=20 per group) due to the invasiveness of muscle biopsies.

**Primary outcome — glycemic control:**

HbA1c decreased in the resistance training group from 8.7 ± 0.3% to 7.6 ± 0.2% (mean change -1.1 percentage points). The control group showed no change (8.6 ± 0.3% to 8.7 ± 0.3%). The between-group difference was statistically significant (P = 0.004).

Fasting plasma glucose decreased in the resistance training group from 10.6 ± 0.6 mmol/L to 9.3 ± 0.5 mmol/L (mean change -1.3 mmol/L). The control group showed no significant change (10.4 ± 0.6 to 10.6 ± 0.6 mmol/L). Between-group difference P = 0.05.

Fasting insulin did not change significantly in either group.

**Medication changes:**

72% of participants in the resistance training group reduced their diabetes medication dosage (either oral agents or dose reduction) over the 16 weeks.

In the control group, 42% of participants increased their diabetes medication dosage.

This difference was statistically significant (P = 0.01).

**Muscle glycogen:**

Muscle glycogen stores increased in the resistance training group from 60.3 ± 3.9 to 79.1 ± 5.0 mmol glucose/kg muscle (a 31% increase).

Muscle glycogen stores decreased in the control group from 61.4 ± 7.7 to 47.2 ± 6.7 mmol glucose/kg muscle (a 23% decrease).

Between-group difference P = 0.05.

**Body composition:**

Lean body mass increased in the resistance training group by +1.2 ± 0.2 kg and decreased in the control group by -0.1 ± 0.1 kg (between-group P = 0.01).

Total fat mass decreased in the resistance training group by -0.9 ± 0.2 kg and increased in the control group by +0.7 ± 0.2 kg (between-group P = 0.01).

Trunk fat mass decreased in the resistance training group by -0.7 ± 0.1 kg and increased in the control group by +0.8 ± 0.1 kg (between-group P = 0.01).

Body weight did not change significantly in either group (resistance training: -0.5 ± 0.3 kg; control: +0.6 ± 0.3 kg; P = 0.10).

**Blood pressure:**

Systolic blood pressure decreased in the resistance training group by -9.7 ± 1.6 mmHg and increased in the control group by +7.7 ± 1.9 mmHg (between-group P = 0.01).

Diastolic blood pressure decreased in the resistance training group by -4.0 ± 1.2 mmHg and increased in the control group by +3.1 ± 1.4 mmHg (between-group P = 0.05).

**Strength:**

Upper body strength (chest press 1RM) increased by 33% in the resistance training group (from 28.5 ± 1.8 to 37.9 ± 2.1 kg) with no change in controls.

Lower body strength (leg press 1RM) increased by 42% in the resistance training group (from 97.3 ± 5.6 to 138.2 ± 6.8 kg) with no change in controls.

All strength gains were statistically significant (P < 0.001).

**Lipid profile:**

Total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides did not change significantly in either group.

The 1.1 percentage point reduction in HbA1c is clinically meaningful. For context, metformin (the most common first-line diabetes medication) typically reduces HbA1c by 1.0–1.5 percentage points. So the effect of 16 weeks of resistance training was roughly equivalent to adding a diabetes medication — and indeed, 72% of exercisers were able to reduce their actual medication.

The 31% increase in muscle glycogen means that after training, participants' muscles could store about 19 more grams of glucose per kilogram of muscle tissue. Since muscle is the primary site of glucose disposal after meals, this increased storage capacity directly improves blood sugar control. The control group lost 23% of their muscle glycogen over the same 16 weeks — likely due to age-related decline and continued sedentary behavior — meaning the net effect of resistance training versus doing nothing was a 54% difference in muscle glycogen stores.

The blood pressure reduction of -9.7 mmHg systolic is comparable to what you might expect from a low-dose blood pressure medication. The fact that control participants' blood pressure rose by +7.7 mmHg means the net difference between groups was 17.4 mmHg — a very large effect.

The body composition changes were modest in absolute terms (1.2 kg lean mass gained, 0.9 kg fat lost) but meaningful because they occurred without significant weight loss. This is a classic "body recomposition" effect — participants got stronger and leaner without changing their weight much, which is particularly valuable for older adults who should avoid weight loss due to risk of sarcopenia (age-related muscle loss).

**Acknowledged by authors:**

Small sample size limits generalizability and statistical power for subgroup analyses.

Single ethnic group (Latino) — results may not apply to other populations.

No aerobic exercise comparison group, so relative efficacy versus other exercise modalities is unknown.

Medication changes were made by participants' personal physicians, not by a standardized protocol, introducing potential confounding.

Muscle biopsy data were available only for a subset of participants.

**Critical reader observations:**

No sham control: The exercise group received 48 sessions of one-on-one attention from a trainer, while the control group received no equivalent contact. This means the social support, accountability, and attention could have contributed to some outcomes (e.g., medication adherence, dietary changes) independent of the exercise itself.

No dietary control: Participants were instructed to maintain their usual diet, but dietary intake was not measured or controlled. If exercisers spontaneously improved their diet (common when people start exercising), some of the glycemic improvement could be due to diet rather than exercise.

Short duration: 16 weeks is long enough to see metabolic changes but not long enough to assess whether these effects persist, whether medication reductions are safe long-term, or whether the exercise program is sustainable without supervision.

High supervision level: The 89% adherence rate was achieved with one-on-one supervision, free access to a gym, and regular encouragement. Real-world adherence to resistance training in older adults is typically much lower.

No intention-to-treat analysis for all outcomes: While the primary analysis was intention-to-treat, the muscle biopsy data came from a subset of completers, which could introduce selection bias.

Industry funding: The study was supported by the National Institutes of Health (NIH) and the U.S. Department of Agriculture (USDA), so no obvious industry bias. However, the exercise equipment was provided by a manufacturer (Cybex), though this is unlikely to have influenced results.

No assessment of adverse events: The paper does not report whether any participants experienced injuries, muscle soreness, or other adverse effects from the resistance training. This is a notable omission for an exercise study in older adults.

Baseline HbA1c was relatively high (8.7%), meaning participants had poorly controlled diabetes. It is unclear whether similar effects would be seen in people with better-controlled diabetes (HbA1c <7.5%) where there is less room for improvement.

For someone running their own n=1 experiment:

**What to test:**

High-intensity progressive resistance training (weightlifting) targeting all major muscle groups. Use compound exercises: leg press (or squats), chest press, seated row, knee extension, knee flexion, and some form of overhead press. Aim for 3 sets of 8–10 repetitions per exercise, with the weight heavy enough that the last 2–3 reps of each set are very difficult to complete (rating of perceived exertion 8–9 out of 10).

Frequency: 3 non-consecutive days per week (e.g., Monday, Wednesday, Friday). Each session should take 45–60 minutes including warm-up