Exercise Changes Six Metabolic Systems. Weight Isn't One of Them.

Before writing the prescription, decide what you are treating. Hepatic steatosis, insulin resistance, blood pressure, apnea severity, depression, bone density, and all-cause mortality each have their own evidence-based dose. They are not interchangeable, and a single generic recommendation will under-treat most of them.

Practical move: pick the one or two targets that matter most for this patient, write those doses, and let the rest come along for the ride.

The insulin-sensitizing effect of exercise is largely a prolonged acute effect that lasts roughly 24 to 72 hours after a session, and it reverts to baseline after about 15 days of stopping. That physiology has a direct prescribing consequence: the patient who trains four short sessions a week is insulin-sensitized nearly all the time, while the patient who does one long weekend session is sensitized for two days out of seven.

Write: movement on most days, ideally never more than 48 hours between sessions.

Add for postprandial control: 10 minutes of walking after each main meal lowered postprandial glucose 12% overall and 22% after the evening meal in adults with type 2 diabetes, outperforming the same total time walked once daily.

Reference: Reynolds AN et al. Diabetologia. 2016;59(12):2572-2578. Bajpeyi S et al. J Appl Physiol. 2009;106(4):1079-1085.

The 2023 meta-analysis of randomized trials found that exercise made patients 3.5 times more likely to achieve a 30% relative reduction in MRI-measured liver fat, that the effect required at least 750 MET-minutes per week, and that treatment response was independent of clinically significant (greater than 5%) weight loss.

750 MET-minutes is roughly 150 minutes of brisk walking, or 75 minutes of vigorous activity. Both aerobic and resistance modalities work: cycling cut hepatic triglyceride 21% in four weeks and resistance training cut liver lipid 13% in eight weeks, both at unchanged body weight.

Pitfall: telling a MASLD patient to "lose 10% of your body weight" when they are not going to, instead of writing a dose they can actually hit that works at stable weight.

Reference: Stine JG et al. Am J Gastroenterol. 2023;118(7):1204-1213.

The 2023 network meta-analysis of 270 randomized trials ranked isometric training first for resting blood pressure reduction, ahead of aerobic, dynamic resistance, combined, and HIIT.

Isometric: -8.24/-4.00 mmHg. Combined: -6.04/-2.54. Dynamic resistance: -4.55/-3.04. Aerobic: -4.49/-2.53. HIIT: -4.08/-2.50.

Isometric wall squats ranked best for systolic; isometric running best for diastolic. A typical protocol is four bouts of two minutes with two minutes of rest, three times a week.

Why this matters: an 8 mmHg systolic reduction is in the range of a single antihypertensive agent, requires no equipment, and takes about 15 minutes.

Reference: Edwards JJ et al. Br J Sports Med. 2023;57(20):1317-1326.

This is the protocol that reduced AHI from 32.2 to 24.6 over 12 weeks without significant weight loss, while the stretching control group got worse. Pooled across trials the AHI reduction is about 6.3 events per hour, with no significant change in BMI.

It will not replace CPAP in severe disease. It does shift severity categories in some patients, improves daytime sleepiness by about 3 points on the Epworth, increases slow-wave sleep, and raises VO2peak by roughly 4 mL/kg/min, which no CPAP machine does.

Say it out loud to the patient: "This works on your apnea even if you do not lose a pound." Most patients with OSA have been told the opposite their whole lives.

Reference: Kline CE et al. Sleep. 2011;34(12):1631-1640. Iftikhar IH et al. Lung. 2014;192(1):175-184.

In the BMJ network meta-analysis, walking or jogging produced g -0.63, yoga -0.55, and strength training -0.49, against CBT at -0.55 and SSRIs at -0.26. Higher intensity produced larger effects. Supervised and group formats improved adherence.

A defensible starting prescription: 30 to 45 minutes of moderate-to-vigorous activity, three to five times weekly, for at least 12 weeks, supervised or in a group if at all possible.

Honest framing: the 2026 Cochrane update found the pooled effect was not statistically significant when restricted to the six highest-quality trials. Exercise is a reasonable first-line or adjunctive treatment; it is not a reason to withhold medication or therapy from someone who needs them.

Reference: Noetel M et al. BMJ. 2024;384:e075847. Clegg AJ et al. Cochrane Database Syst Rev. 2026;(1):CD004366.

This is the most surprising number in the whole literature and the easiest sell in clinic. Muscle-strengthening activity shows a J-shaped dose-response for all-cause mortality, cardiovascular disease, and cancer, with maximum benefit at roughly 30 to 60 minutes per week. Beyond that the curve flattens and may bend back up. For diabetes risk the curve is L-shaped, with benefit accruing up to about 60 minutes per week.

Two 20-minute sessions a week is a therapeutic dose. That is a very different conversation than "you need to join a gym."

For bone: twice-weekly high-intensity resistance and impact training raised lumbar spine BMD 2.9% in postmenopausal women with low bone mass, with one minor adverse event across the trial.

For falls: add balance and functional training. It cut fall rate 24% as a single modality, the highest-certainty finding in the Cochrane review.

Reference: Momma H et al. Br J Sports Med. 2022;56(13):755-763. Watson SL et al. J Bone Miner Res. 2018;33(2):211-220. Sherrington C et al. Cochrane Database Syst Rev. 2019;1:CD012424.

A systematic review of 17 trials in adults with overweight and obesity found that adherence was predicted by supervision and group format, not by exercise type, session length, or baseline body weight. The active ingredient in a successful exercise prescription is often another human being.

Practical: refer to a supervised program, a group class, or a qualified exercise professional (usreps.org) rather than handing over a printed sheet. The dose-response curve is steepest going from zero to something, so protecting the first eight weeks matters more than optimizing the protocol.

Reference: Rodrigues MSA et al. J Obes. 2026;2026:4164477.

The single highest-yield number on this list, and the one almost no primary care practice records. Each 1-MET gain associates with roughly 13% lower all-cause mortality and 15% fewer cardiovascular events.

How to get it without a CPET: a non-exercise estimated MET equation from age, sex, BMI, resting heart rate and self-reported activity; a submaximal step or treadmill protocol; or the 6-minute walk distance in deconditioned patients. Perfect precision is not the point. A documented direction of travel is.

Pitfall: 6-minute walk distance is a submaximal functional test, not VO2max. Do not present the two as the same measurement.

Visceral fat falls with training even when total weight does not, and visceral fat is the depot that drives the metabolic risk. In the trial of men who exercised without losing weight, visceral fat fell significantly (P=0.001) while the scale did not move at all.

Measure at the top of the iliac crest, same technique, same person, every visit. A 2 cm reduction at stable weight is a real result and it is invisible on a scale.

Grip strength predicted all-cause mortality better than systolic blood pressure did in the PURE cohort (HR 1.37 vs. 1.15 per standard deviation), and each 5 kg decrement carried HR 1.16.

A hand dynamometer costs less than a blood pressure cuff. The 30-second sit-to-stand needs a chair and a timer. Both change measurably within eight weeks of resistance training, which makes them the fastest source of visible progress for a patient whose weight is flat.

HbA1c is the slow signal. Fasting insulin and HOMA-IR move sooner and often move when weight does not, given the roughly 74% rise in insulin sensitivity documented at unchanged weight, BMI and waist.

The highest-impact teaching tool available: a CGM and a post-meal walk. The patient watches their own postprandial excursion flatten in real time. Nothing in a chart is as persuasive as a patient seeing their own curve change.

Remember the half-life. Check glycemic measures within a day or two of a training session, not after a two-week gap, or you will be measuring detraining.

Given that exercise reduces hepatic fat independent of weight loss, MASLD is one of the clearest places to show a patient an objective win at stable weight. ALT and FIB-4 are cheap and repeatable; MRI-PDFF is the research standard and occasionally worth it.

Recheck at 12 weeks if the patient is hitting 750 MET-minutes per week. Sooner than that and you are measuring noise.

Expect somewhere between 4 and 8 mmHg systolic depending on modality, with isometric training at the top of the range. That is a clinically meaningful change and it shows up within weeks.

Use home readings averaged over a week rather than a single office measurement, so a good result is not lost in white-coat noise.

Epworth Sleepiness Scale, PHQ-9, and GAD-7 all cost nothing and all move with training. Pooled apnea trials showed a 3.3-point Epworth improvement, and the depression effect sizes for walking, yoga and strength training exceed those reported for SSRIs in the same network analysis.

Patients report these changes long before any lab does. Capturing them turns a subjective "I feel better" into a documented outcome, which is what keeps someone training through month three.

This is the part of the conversation that gets lost in the GLP-1 era, and it belongs in every visit.

Fitness. In the S-LITE trial, VO2peak normalized to fat-free mass rose 6.1 mL/kg FFM/min with exercise versus placebo, and 5.3 with exercise plus liraglutide. Liraglutide alone was not significant (2.3, 95% CI -0.4 to 4.9). The drug took weight off. It did not make anyone fitter.

Endogenous GLP-1. One year of exercise after weight loss raised late-phase postprandial GLP-1 secretion 37% (95% CI 20 to 57), while GLP-1 receptor agonist treatment produced no change in the body's own secretion.

Lean mass. Lean tissue accounted for roughly 25% of weight lost with tirzepatide in SURMOUNT-1 and a higher fraction in the semaglutide substudies. Strength appears better preserved than muscle size, but resistance training remains the only intervention that reliably adds fat-free mass during a deficit.

The framing that holds up: not exercise versus medication. Exercise alongside it, doing the part the molecule does not do. The World Obesity Federation's 2026 position statement makes the same point from the other direction: physical activity is a core component of obesity care and not a silver bullet, and the barriers are environmental and structural rather than a failure of individual will.

Reference: Jensen SBK et al. Sports Med. 2026;56(7):1785-1800. Holt J et al. Obesity (Silver Spring). 2026;34(1):51-57. Lundgren JR et al. N Engl J Med. 2021;384(18):1719-1730. Gowers Z et al. Obes Rev. 2026:e70103.