Exercise Intervention
Exercise and training interventions. A research review published by South Beach Longevity.
Exercise Intervention
Prescribed movement as prevention, treatment, and functional medicine — graded by outcome, not by enthusiasmPrescribed exercise is a medical intervention with a dose, a mode, an adherence problem, and a harm profile. It is not a mood. The record is strong for some diseases and some outcomes, surrogate-heavy for others, and thinner in unsupervised life than in supervised trials. This article grades that record. It is not a training plan.
Compiled by South Beach Longevity · 20 August 2026 Copyright 2026 Series SBL-41 / SP-EXERCISE-INTERVENTION · Register A scientific article Sources peer-reviewed human trials, consensus statements, meta-analyses, and labelled animal or in-vitro work · verified NCBI records Constraint This document describes published research. It is not medical advice. No human use, dose, route or schedule is recommended anywhere in this document.
How to read this document Every finding is labelled, in the sentence that reports it, by the kind of study that produced it. A supervised rehabilitation trial is not a free-living habit. A fall in HbA1c, blood pressure, liver fat, or VO₂ is not an event, a hospitalization, or a death. A guideline is a synthesis of a record, not a licence to treat every number in it as a prescription. Where two results conflict, both are given. Amounts and durations appear only as reported experimental or consensus parameters, always with the population attached. Nothing here is a recommendation. Findings are graded in place as established, strongly supported, emerging, plausible, or speculative. Two further labels mark careful absences rather than verdicts: not established, where the evidence is too thin to place a claim on the ladder at all — untested or insufficient, an absence of proof rather than disproof; and not supported, where the weight of evidence leans against a claim but stops short of a formal refutation. Evidence, guideline language, and individualized clinical decision-making are kept in separate sentences.
01 What prescribed exercise is, and four things it is not
Prescribed exercise is a structured, progressive, and time-limited change in bodily work, assigned for a named health, functional, or disease outcome. It is a branch of therapeutics that happens to use skeletal muscle. Pedersen and Saltin treated that fact as the field’s working claim: for a defined list of diagnoses, the human trial record is large enough that exercise belongs in the same conversation as drugs and procedures, not in a lifestyle appendix (Pedersen and Saltin, 2015). Booth and colleagues stated the complementary public-health claim: physical inactivity is itself a major cause of chronic disease, not merely the absence of a hobby (Booth, Gordon, and Chakravarthy, 2000; Booth, Chakravarthy, Gordon, and Spangenburg, 2002). Both papers are syntheses. They are strongly supported as maps of a large human literature. They are not a licence to treat every disease as if the effect size were the same.
Four things follow immediately and will be enforced for the rest of this document.
First, exercise is not physical activity, and physical activity is not sport. Physical activity is any bodily movement that raises energy expenditure. Exercise is planned, structured, and repetitive, with an objective. Sport adds rules, opponents, and a reason to hide an injury. The 2020 World Health Organization guidelines on physical activity and sedentary behaviour, and the 2018 Physical Activity Guidelines for Americans, are population maps of the first quantity (Bull et al., 2020; Piercy et al., 2018). Cardiac rehabilitation, pulmonary rehabilitation, LIFTMOR, HF-ACTION, and the LIFE mobility trial are tests of the second. Confusing the three is how a weekend walk inherits the evidence of a supervised interval programme, and how a marathon inherits the safety of a walk.
Second, exercise is not a polypill. Fiuza-Luces and colleagues used that metaphor because one behaviour moves several risk factors at once (Fiuza-Luces et al., 2013). The metaphor is pedagogically useful and scientifically dangerous. A behaviour that moves blood pressure, glucose, and mood is still a set of disease-specific effect sizes, not a substitute for an ACE inhibitor, metformin, or an antidepressant in the trials that tested those drugs. Naci and Ioannidis compared exercise and drug interventions across mortality outcomes in selected conditions and found that the comparison is possible, uneven, and not a slogan (Naci and Ioannidis, 2013). Naci and colleagues later compared exercise with antihypertensive drugs on systolic pressure in a network meta-analysis (Naci et al., 2019).
Third, exercise is not automatically safe because it is natural. Mittleman and colleagues showed that heavy exertion can trigger myocardial infarction, particularly in the habitually sedentary (Mittleman et al., 1993). Siscovick and colleagues had already shown that vigorous activity raises the short-term risk of cardiac arrest while lowering the habitual risk (Siscovick, Weiss, Fletcher, and Lasky, 1984). Albert and colleagues reported a similar triggering pattern for sudden death (Albert et al., 2000). Thompson, Franklin, and the American Heart Association statements on exercise-related events exist because the benefit and the trigger occupy the same behaviour (Thompson et al., 2007; Franklin et al., 2020). The ACSM preparticipation screening revision and PAR-Q+ exist because the older annual medical-clearance ritual was screening the wrong people (Riebe et al., 2015; Bredin, Gledhill, Jamnik, and Warburton, 2013).
Fourth, this document is not medical advice and not a training prescription. Reported minutes, intensities, set structures, and progression rules are experimental or consensus parameters attached to named populations. They are not instructions for any reader.
Sibling articles in this series take sports nutrition, strength training, and hypertrophy as performance and tissue problems. This title is the clinical-intervention cut of the same physiology. A time trial is not a hospitalization. A one-repetition maximum is not a fracture. Muscle protein synthesis is not sarcopenia.
02 Modes and the dose grammar
The American College of Sports Medicine’s quantity-and-quality position stand remains the field’s dose grammar: frequency, intensity, time, type, volume, and progression — FITT-VP — applied separately to cardiorespiratory, resistance, flexibility, and neuromotor training (Garber et al., 2011). Haskell, Nelson, and the ACSM/AHA public-health statements are the older, still-cited population translations of that grammar (Haskell et al., 2007; Nelson et al., 2007). Izquierdo and colleagues later restated the older-adult version as an international recommendation set (Izquierdo et al., 2021). Fragala and the NSCA restated the resistance-training version for older adults (Fragala et al., 2019). Chodzko-Zajko and colleagues remain the ACSM older-adult position stand (Chodzko-Zajko et al., 2009).
The modes compared in this article are not brands. They are different mechanical and metabolic questions.
Aerobic / endurance work raises heart rate and pulmonary ventilation for a sustained period. It is the mode behind most blood-pressure, VO₂, cardiac-rehabilitation, claudication, and pulmonary-rehabilitation trials.
Resistance / strength work loads skeletal muscle against an external resistance. It is the mode behind most sarcopenia, bone, and some glycaemia and depression trials.
Combined programmes do both. Sigal’s DARE trial and Church’s HART-D trial exist because the combination is not the sum of two press releases (Sigal et al., 2007; Church et al., 2010).
High-intensity interval training (HIIT) alternates brief hard efforts with recovery. Gibala, MacInnis, Weston, Milanović, Rognmo, Wisløff, Hannan, and Ellingsen (SMARTEX-HF) are the human record for that substitution, not Instagram (Gibala, Little, Macdonald, and Hawley, 2012; MacInnis and Gibala, 2017; Weston, Wisløff, and Coombes, 2014; Milanović, Sporiš, and Weston, 2015; Rognmo et al., 2004; Wisløff et al., 2007; Hannan et al., 2018; Ellingsen et al., 2017).
Walking is aerobic work at a speed most people already know. It is also the best-studied mode in peripheral artery disease and one of the few modes that survives translation out of the laboratory (Lane, Ellis, Watson, and Leng, 2014; McDermott et al., 2013; Treat-Jacobson et al., 2019).
Balance and neuromotor work is the mode behind most fall-prevention effects that survive a Cochrane filter (Sherrington et al., 2019; Gillespie et al., 2012). Flexibility is the mode most often included and least often shown to change a clinical endpoint.
Multicomponent programmes combine two or more of the above, usually in older adults. LIFE, Otago, and FINGER are different multicomponent questions, not one intervention wearing three names (Pahor et al., 2014; Campbell and Robertson, 2003; Ngandu et al., 2015).
WHO 2020 states a population dose: 150–300 minutes a week of moderate-intensity aerobic activity, or 75–150 minutes of vigorous-intensity activity, or an equivalent combination, plus muscle-strengthening on two or more days, with additional balance work for older adults, and a separate warning that sedentary time is not cancelled by a single bout (Bull et al., 2020). Piercy and the 2018 U.S. guidelines say substantially the same thing to a U.S. audience (Piercy et al., 2018). Those numbers are public-health consensus parameters. They are strongly supported as a description of the observational and trial synthesis those committees judged. They are not a rehabilitation prescription, not a bone-loading prescription, and not a depression protocol.
03 Clinical endpoints versus surrogates
This is the load-bearing distinction of the article.
A clinical outcome is an event or a function a patient can name without a laboratory: death, myocardial infarction, hospitalization for heart failure, major mobility disability, a fall with fracture, a claudication distance that changes a life, a depression score that crosses a remission threshold, a quality-of-life instrument that was the trial’s pre-specified endpoint.
A surrogate is a measurement asked to stand in for that event: VO₂peak, HbA1c, systolic pressure, intrahepatic triglyceride, BMD T-score, hippocampal volume, C-reactive protein, endothelial function, a clock, a myokine.
Kodama and colleagues showed that cardiorespiratory fitness is a strong observational predictor of mortality (Kodama et al., 2009). Myers and colleagues showed the same in a referred veterans cohort (Myers et al., 2002). Blair and colleagues showed it earlier in the Aerobics Center (Blair et al., 1989). Ross and colleagues argued that fitness belongs in clinical assessment (Ross et al., 2016). Those papers make fitness established as a risk marker. They do not make every training-induced rise in VO₂ a proof that events will fall. HF-ACTION improved fitness and health status and did not, on its primary analysis, reduce the composite of all-cause mortality or hospitalization (O’Connor et al., 2009; Flynn et al., 2009). Look AHEAD improved weight, fitness, and many risk factors and did not reduce the primary cardiovascular composite (Look AHEAD Research Group, 2013). CALERIE-adjacent fitness papers in the sibling nutrition titles make the same point from the other direction: a better mass-relative VO₂ can coexist with a worse absolute capacity.
The grade rule used below is mechanical. An intervention that changes a surrogate and an event is established or strongly supported for the event. An intervention that changes only the surrogate is graded for the surrogate and labelled as such. An intervention that changes a pathway cartoon is plausible or speculative until a human outcome appears.
04 Adherence, selection, and the healthy-participant problem
The first adversarial question is not optional colour. It is the main threat to every claim in Parts Two to Four.
People who enrol in exercise trials are not a random sample of the disease. They can attend, they can walk, they can tolerate a run-in, and they have already said yes to effort. People who remain in the trial are a further subset. The benefit estimated in that subset is real for that subset. It is not automatically the benefit of writing “exercise” on a discharge summary.
HF-ACTION is the clearest cardiac example: a large, multicentre, randomized trial of exercise training in heart failure in which median exercise time and adherence were lower than the protocol asked for, and the primary endpoint missed on the unadjusted analysis (O’Connor et al., 2009). Look AHEAD is the clearest metabolic example: an intensive lifestyle intervention that produced large early weight and fitness changes, then a slow fade, then a neutral cardiovascular composite (Look AHEAD Research Group, 2013; Wing et al., 2013). The LIFE trial is the clearest geriatric example: a structured activity programme that reduced major mobility disability in at-risk older adults who were, by design, already able to walk 400 metres (Pahor et al., 2014).
Ekelund, Lee, Arem, Kyu, Wen, and Sattelmair describe observational dose–response curves that are strongly supported as associations and not established as causal proofs of the same magnitude in the unselected sick (Lee et al., 2012; Wen et al., 2011; Arem et al., 2015; Kyu et al., 2016; Sattelmair et al., 2011; Ekelund et al., 2016). The healthy-participant problem is not an argument that exercise is useless. It is an argument that the slogan and the trial are different objects.
Supervision is part of the same problem. Pulmonary rehabilitation, cardiac rehabilitation, and supervised treadmill walking for claudication are interventions that include a building, a timetable, and another human being. Home-based walking can work; McDermott and colleagues tested that claim instead of assuming it (McDermott et al., 2013). Unsupervised advice is what most of the world actually receives. Umpierre and colleagues separated “exercise advice” from structured training in type 2 diabetes and found that the structured programmes moved HbA1c (Umpierre et al., 2011). That distinction will be reused whenever a guideline sentence and a trial sentence threaten to collapse.
05 Guideline architecture
Guidelines are not evidence. They are committees reading evidence and writing sentences that a health system can implement.
WHO 2020 and the 2018 U.S. Physical Activity Guidelines are population documents (Bull et al., 2020; Piercy et al., 2018). ACSM position stands (Garber; Pescatello; Colberg; Donnelly; Chodzko-Zajko; Schmitz/Campbell) are professional-society documents (Garber et al., 2011; Pescatello et al., 2004; Colberg et al., 2010; Donnelly et al., 2009; Chodzko-Zajko et al., 2009; Campbell et al., 2019). ADA Standards address physical activity inside diabetes care (Colberg et al., 2016). ACC/AHA documents address hypertension, peripheral artery disease, and exercise testing (Whelton et al., 2018; Gerhard-Herman et al., 2017; Fletcher et al., 2013). ESC documents address sports cardiology and rehabilitation (Pelliccia et al., 2021; Piepoli et al., 2014). ATS/ERS addresses pulmonary rehabilitation (Spruit et al., 2013). OARSI addresses osteoarthritis (Bannuru et al., 2019). ASCO and ACSM/ACS address cancer (Ligibel et al., 2022; Rock et al., 2022; Campbell et al., 2019). EWGSOP2 and ICFSR address sarcopenia and frailty (Cruz-Jentoft et al., 2019; Dent et al., 2019). ACP addresses low-back pain (Qaseem et al., 2017). KDIGO addresses CKD lifestyle. Cochrane reviews sit underneath many of those sentences and are cited as reviews, not as trials.
Table B compares those documents on dose, mode, supervision, and the outcome they actually claim. A guideline that recommends exercise for a disease is not thereby a proof that exercise changes that disease’s hardest endpoint.
06 Cardiovascular disease and cardiac rehabilitation
Powell, Thompson, Caspersen, and Kendrick stated the modern observational claim in 1987: physical activity is inversely associated with coronary-heart-disease incidence (Powell et al., 1987). Lee, Sattelmair, and later dose–response syntheses kept that association and tried to price it in minutes (Lee et al., 2012; Sattelmair et al., 2011). That is established as an association. It is not, by itself, a rehabilitation trial.
Exercise-based cardiac rehabilitation after myocardial infarction or coronary revascularization is the intervention trial. Cochrane reviews by Heran, Anderson, and then Dibben found that exercise-based rehabilitation reduces cardiovascular mortality and hospital admission in people with coronary heart disease, with less certainty for total mortality in the more recent evidence, and with the usual caveats about older trials, mixed programmes, and incomplete blinding (Heran et al., 2011; Anderson et al., 2016; Dibben et al., 2021). That package is strongly supported for the outcomes those reviews pre-specified. It is not a proof that a gym membership is cardiac rehabilitation. Cardiac rehabilitation, as tested, is a multicomponent programme: supervised exercise plus education, risk-factor management, and a health system. Kabboul and others have tried to unpick components; the exercise core is necessary and not always sufficient.
HIIT in coronary disease is a mode question inside that programme. Rognmo and colleagues reported larger VO₂ gains with high-intensity aerobic intervals than with moderate continuous training in a small coronary sample (Rognmo et al., 2004). Hannan’s systematic review treated HIIT as effective for fitness in coronary disease (Hannan et al., 2018). Fitness is the surrogate. Events remain the Cochrane rehabilitation outcome, not the HIIT-versus-moderate contrast.
Wasfy and Baggish reviewed the other end of the dose: whether very high recreational endurance volumes eventually harm the heart (Wasfy and Baggish, 2016). That literature is emerging as a U-shaped or plateau discussion in recreational athletes. It is not a reason to withhold walking from a post-infarct patient, and it is not a reason to treat a marathon as therapy.
07 Hypertension
Cornelissen and Smart’s meta-analysis of exercise training and blood pressure remains the quantitative centre of the human record: endurance, resistance, and isometric work all lower blood pressure, with the largest mean effects in the hypertensive strata and with the usual heterogeneity of small trials (Cornelissen and Smart, 2013). Whelton, Chin, Xin, and He’s earlier meta-analysis of aerobic exercise had already made the direction of the aerobic effect difficult to deny (Whelton et al., 2002). Pescatello and the ACSM hypertension position stand translated that record into FITT language (Pescatello et al., 2004; Pescatello et al., 2015). Carpio-Rivera and colleagues separated the acute post-exercise hypotensive effect from the training effect (Carpio-Rivera, Moncada-Jiménez, Salazar-Rojas, and Solera-Herrera, 2016). Fagard’s narrative is the short version: exercise is good for blood pressure, and the size of the good is not the size of a drug at the top of a titration schedule (Fagard, 2006).
The 2017 ACC/AHA hypertension guideline includes physical activity in nonpharmacological therapy (Whelton et al., 2018). That is a guideline sentence. The trial underneath a lifestyle-plus-DASH claim is Blumenthal and colleagues’ randomized work combining the DASH diet with exercise and weight reduction, which lowered blood pressure more than DASH counselling alone in unmedicated overweight adults with high-normal or high pressure (Blumenthal et al., 2010). That is strongly supported for clinic and ambulatory pressure in that sample. It is not a hard-outcome hypertension trial. SPRINT was a drug-target trial and is not spent here as exercise evidence.
Established: structured aerobic training lowers blood pressure in adults with elevated pressure. Strongly supported: resistance and isometric modes also lower pressure, with a thinner trial base. Not established: exercise as monotherapy for outcome-level prevention of stroke or heart failure in severe hypertension.
08 Obesity
Donnelly and the ACSM position stand on physical activity and weight loss stated the awkward quantitative fact: the exercise dose that reliably produces clinically meaningful weight loss is larger than the dose that improves fitness, and diet remains the larger lever for scale weight (Donnelly et al., 2009). Swift, Johannsen, Lavie, Earnest, and Church restated the same architecture: exercise contributes to weight loss, matters more for prevention of regain and for fitness, and is easy to over-claim from small trials (Swift et al., 2014). Ross and colleagues showed that exercise can reduce visceral fat and improve metabolic risk with little or no scale-weight change (Ross et al., 2000; Ross et al., 2004). That last finding is strongly supported and is the reason “the scale did not move” is not a null metabolic result.
Look AHEAD randomized adults with type 2 diabetes and overweight or obesity to an intensive lifestyle intervention — diet plus exercise — versus diabetes support and education. Weight, fitness, waist, and many risk factors improved. The primary composite of cardiovascular death, non-fatal myocardial infarction, non-fatal stroke, or hospitalized angina did not (Look AHEAD Research Group, 2013). Subsequent Look AHEAD papers described the durability and the fade of the weight effect (Wing et al., 2013; Gregg et al., 2016). Look AHEAD is established as a test of intensive lifestyle against a cardiovascular composite in type 2 diabetes. It is not a proof that weight loss is useless, and it is not a proof that exercise is useless. It is a proof that a surrogate package can move while the pre-specified event composite does not.
STRRIDE, led by Kraus and Slentz, remains the cleanest dose-and-mode experiment in overweight adults: different amounts and intensities of aerobic training, with and without resistance, change fitness, lipoproteins, and visceral fat in a dose-responsive way (Kraus et al., 2002; Slentz et al., 2004). Those are metabolic surrogates. They are strongly supported as such.
09 Type 2 diabetes
The prevention record and the treatment record are different experiments.
Tuomilehto and colleagues, in the Finnish Diabetes Prevention Study, and Knowler and the U.S. Diabetes Prevention Program Research Group, showed that a lifestyle programme including dietary change and physical activity reduces incident type 2 diabetes in high-risk adults, and that the lifestyle arm out-performed the placebo and, in the U.S. trial, compared favourably with metformin on the diabetes-incidence endpoint (Tuomilehto et al., 2001; Knowler et al., 2002). The DPP 10-year follow-up showed that some of that prevention persists, with attenuation (Diabetes Prevention Program Research Group, 2009). That is established for diabetes incidence in those high-risk samples. It is a multicomponent lifestyle result, not an isolated exercise result.
For people who already have type 2 diabetes, Boulé, Haddad, Kenny, Wells, and Sigal’s meta-analysis showed that exercise training reduces HbA1c (Boulé et al., 2001). Snowling and Hopkins compared modes (Snowling and Hopkins, 2006). Umpierre and colleagues showed that structured training, not mere advice, is what moves HbA1c, and that more than 150 minutes a week was associated with larger reductions in that synthesis (Umpierre et al., 2011). Sigal’s DARE trial randomized aerobic, resistance, or both, and found that combined training reduced HbA1c more than either mode alone (Sigal et al., 2007). Church’s HART-D trial, in a different sample, found that combination training reduced HbA1c relative to controls while aerobic or resistance alone did not meet the same threshold (Church et al., 2010). Colberg and the ADA/ACSM joint position, and later ADA Standards language, treat both aerobic and resistance work as part of diabetes care (Colberg et al., 2010; Colberg et al., 2016).
Established: structured exercise lowers HbA1c in type 2 diabetes; combined aerobic and resistance work is the best-supported mode package. Strongly supported: lifestyle programmes that include exercise prevent diabetes in high-risk adults. Not established: exercise as a substitute for glucose-lowering drugs on microvascular or macrovascular hard outcomes. Look AHEAD is the warning label on that last sentence.
10 Metabolic syndrome
Metabolic syndrome is a cluster, not a disease. Alberti and colleagues’ harmonized definition is the naming paper (Alberti et al., 2009). Exercise trials in this space usually report the components — waist, triglycerides, HDL-cholesterol, blood pressure, fasting glucose — or a composite score. STRRIDE is again the cleanest dose experiment (Kraus et al., 2002; Slentz et al., 2004). Pedersen and Saltin include the cluster in their “exercise as medicine” inventory (Pedersen and Saltin, 2015).
The grade is strongly supported for component improvement under structured aerobic or combined training. It is not established for a unique syndrome-level event that is not already counted under diabetes, hypertension, or coronary disease. Treating the cluster as a fifth disease with its own mortality trial is a category error.
11 MASLD and NAFLD
The name moved. Rinella and the multi-society consensus retired “NAFLD” in favour of metabolic dysfunction–associated steatotic liver disease (MASLD) without thereby creating a new exercise literature (Rinella et al., 2023). The trials were done under the old name.
Keating, Hackett, George, and Johnson reviewed exercise and NAFLD and found that exercise reduces liver fat even when weight loss is modest (Keating et al., 2012). Hashida and colleagues compared aerobic and resistance modes (Hashida et al., 2017). Hallsworth and colleagues showed that resistance exercise can reduce liver fat (Hallsworth et al., 2011). Houghton and colleagues reported that exercise reduces liver lipids and improves some metabolic measures in NAFLD (Houghton et al., 2017). Keating later examined aerobic-dose questions (Keating et al., 2015). Orci and colleagues’ systematic review sits on the same surrogate: intrahepatic fat, enzymes, insulin sensitivity (Orci et al., 2016).
Strongly supported: structured exercise reduces liver fat in NAFLD/MASLD. Emerging: mode differences once total energy expenditure is matched. Not established: exercise as a therapy that changes cirrhosis, hepatocellular carcinoma, or liver-related death. Those events were not the endpoints of the exercise trials. Any sentence that spends a proton-density fat-fraction change as if it were a transplant avoided is a category error.
12 Heart failure
HF-ACTION is the largest randomized trial of exercise training in chronic heart failure with reduced ejection fraction: 2,331 patients, supervised then home-based aerobic training versus usual care. The primary composite of all-cause mortality or hospitalization was not significantly reduced on the unadjusted analysis; adjustment for highly prognostic baseline characteristics produced a nominally significant reduction; safety did not signal a new hazard; VO₂ and health status improved (O’Connor et al., 2009; Flynn et al., 2009). That is established as the trial. It is strongly supported for fitness and health status. It is not established as an unadjusted mortality benefit.
Cochrane and related reviews (Taylor; Long; Smart / ExTraMATCH) sit around that trial and older, smaller ones, generally finding reductions in hospital admission and improvements in exercise capacity, with more cautious language on mortality (Taylor et al., 2014; Long et al., 2019). Hambrecht and Giannuzzi reported antiremodeling and endothelial effects in smaller mechanistic trials (Hambrecht et al., 2000; Giannuzzi et al., 2003). Those are strongly supported as physiology. They are not HF-ACTION.
Wisløff and colleagues’ small aerobic-interval trial in heart failure reported a large VO₂ advantage for intervals over moderate continuous training (Wisløff et al., 2007). SMARTEX-HF, the later multicentre test of that idea, did not reproduce a clear superiority of high-intensity interval training over moderate continuous training on reverse remodeling; training intensity in the interval arm was lower than prescribed, which is itself an adherence finding (Ellingsen et al., 2017). HIIT in HFrEF is therefore emerging as a fitness tool and not established as a superior clinical strategy.
HFpEF is a different disease with a thinner exercise record. Kitzman, Edelmann (Ex-DHF), Pandey, and Mandic reported that training improves peak VO₂ and some quality-of-life measures in older adults with heart failure and preserved ejection fraction, without a mortality trial of HF-ACTION’s size (Kitzman et al., 2010; Edelmann et al., 2011; Pandey et al., 2015). Grade: strongly supported for fitness and symptoms; insufficient for hard events.
13 COPD
Pulmonary rehabilitation is the COPD exercise intervention that has been tested as a programme, not as a slogan. McCarthy, Puhan, and the Cochrane reviews found that rehabilitation improves dyspnoea, fatigue, exercise capacity, and health-related quality of life in people with COPD, with effects large enough to matter to a patient and with the usual caveats about programme heterogeneity (McCarthy et al., 2015; Puhan et al., 2016). Spruit and the ATS/ERS statement is the professional-society translation (Spruit et al., 2013).
Established: pulmonary rehabilitation improves symptoms, capacity, and quality of life in COPD. Strongly supported: the exercise core is necessary; education alone is not the same intervention. Not established: rehabilitation as a substitute for bronchodilators, oxygen, or smoking cessation on mortality. Hospitalization effects exist in some syntheses and are more fragile than the symptom effects.
14 Peripheral vascular disease
Supervised exercise therapy for intermittent claudication is one of the cleanest indications in this article. Lane, Ellis, Watson, and Leng’s Cochrane review found that exercise improves walking performance in claudication (Lane et al., 2014). Gardner and colleagues’ work on the dose and structure of walking — including the use of claudication pain as a training signal — is the laboratory backbone (Gardner, Katzel, Sorkin, and Goldberg, 2001). McDermott and colleagues compared treadmill exercise, resistance training, and, separately, home-based walking, and they argued that six-minute-walk distance is a more relevant outcome than a treadmill test the patient will never take again (McDermott et al., 2009; McDermott et al., 2013). Treat-Jacobson and the AHA scientific statement, and Gerhard-Herman and the 2016 AHA/ACC PAD guideline, place supervised exercise as first-line therapy for claudication (Treat-Jacobson et al., 2019; Gerhard-Herman et al., 2017).
Established: supervised walking programmes improve pain-free and maximal walking distance in intermittent claudication. Strongly supported: home-based walking can work when it is actually a programme, not a pamphlet. Not established: exercise as a replacement for revascularization when limb threat, not claudication, is the problem.
15 Chronic kidney disease
Heiwe and Jacobson’s Cochrane review of exercise training in adults with chronic kidney disease found improvements in physical fitness, walking capacity, and some cardiovascular and health-related quality-of-life measures, on a base of small, heterogeneous trials (Heiwe and Jacobson, 2011). Johansen’s narrative and Howden, Coombes, and Isbel’s experimental work sit in the same thin but consistent record: people with CKD can train; VO₂ and function move; hard renal endpoints are almost never the primary outcome (Johansen, 2007; Howden et al., 2012). Greenwood and colleagues have tested whether training can be implemented in nephrology services (Greenwood et al., 2015). KDIGO lifestyle language includes physical activity; that is a guideline sentence on a modest trial base.
Strongly supported: exercise training improves fitness and function in CKD, including some dialysis samples. Emerging: implementation inside dialysis units and transplant pathways. Insufficient: exercise as a therapy that slows GFR decline or reduces kidney-failure events. Those sentences, if written, would be speculative.
16 Osteoporosis
Howe and colleagues’ Cochrane review of exercise for preventing and treating osteoporosis in postmenopausal women found that exercise can improve bone mineral density by a small amount, with different modes loading different sites (Howe et al., 2011). Kelley’s meta-analyses of walking and of lumbar-spine BMD are the older quantitative backbone and are limited by the small absolute changes BMD trials can detect (Kelley, 1998; Kelley, Kelley, and Tran, 2001). Martyn-St James and Carroll’s walking meta-analysis is the caution: walking alone is a weak osteogenic stimulus (Martyn-St James and Carroll, 2008).
LIFTMOR is the trial that took the osteogenic-stimulus problem seriously. Watson, Weeks, Horan, and Beck randomized postmenopausal women with low bone mass to high-intensity resistance and impact training versus a low-intensity control and reported improvements in lumbar and femoral BMD and in function, with a safety discussion that is part of the result (Watson et al., 2018). Giangregorio and the Too Fit To Fracture recommendations are the professional-society attempt to write that stimulus into practice without pretending that walking groups are LIFTMOR (Giangregorio et al., 2014).
Strongly supported: site-specific loading and progressive resistance can improve BMD modestly and can improve function. Established as a limitation: BMD is a surrogate for fracture. Not established: community walking programmes as fracture-prevention therapy. Fracture-endpoint exercise trials of adequate size remain scarce. Pharmacologic fracture trials are not spent here as if they were exercise trials.
17 Osteoarthritis
Fransen, McConnell, Harmer, Van der Esch, Simic, and Bennell’s Cochrane review of land-based exercise for knee osteoarthritis found that exercise reduces pain and improves function, with effects that are clinically relevant and that fade after the programme stops (Fransen et al., 2015). Bennell’s narrative is the clinical translation (Bennell, Hinman, and Wrigley, 2014). Bannuru and the 2019 OARSI guideline place exercise — land-based, mind-body, and, in some strata, aquatic — as core treatment (Bannuru et al., 2019). Messier’s IDEA trial tested intensive diet, exercise, or both in overweight adults with knee osteoarthritis and found that the combination reduced pain and, on some measures, joint load more than exercise alone (Messier et al., 2013).
Established: structured exercise reduces pain and improves function in knee osteoarthritis. Strongly supported: hip OA follows a similar but thinner record. Not established: exercise as a disease-modifying therapy that changes radiographic progression the way a patient hopes a “cartilage drug” would. The outcome that holds is the one the patient can feel.
18 Chronic low-back pain
Hayden, van Tulder, Malmivaara, and Koes’s reviews found that exercise therapy reduces pain and improves function in chronic non-specific low-back pain relative to no treatment or usual care, with no single named mode winning a league table (Hayden et al., 2005; Hayden et al., 2021). van Middelkoop and colleagues reached a similar conclusion (van Middelkoop et al., 2010). Qaseem, Wilt, McLean, and Forciea’s ACP guideline, building on Chou’s evidence review, recommends exercise as a first-line nonpharmacologic treatment for chronic low-back pain (Qaseem et al., 2017; Chou et al., 2017). Oliveira and colleagues’ later guideline review is a map of how many societies said some version of the same sentence (Oliveira et al., 2018).
Established: exercise therapy helps chronic non-specific low-back pain on pain and function. Strongly supported: the effect is modest, the programmes are heterogeneous, and the search for the one correct exercise is a marketing project. Not established: exercise as imaging-driven therapy for a named structural lesion that was not the inclusion criterion of the trials.
19 Frailty, sarcopenia, and fall prevention
Fried’s phenotype paper is the frailty naming study, not an exercise trial (Fried et al., 2001). Fiatarone and colleagues’ 1994 New England Journal of Medicine trial in 100 frail nursing-home residents (mean age 87.1 years) is the existence proof that very old, frail people can gain strength and gait speed with high-intensity resistance training (Fiatarone et al., 1994). Liu and Latham’s Cochrane review of progressive resistance training in older adults found improvements in physical function (Liu and Latham, 2009). Cruz-Jentoft and EWGSOP2 define sarcopenia as a muscle-disease construct with muscle strength as the front door (Cruz-Jentoft et al., 2019). Dent and ICFSR write the frailty and sarcopenia practice guidelines (Dent et al., 2018; Dent et al., 2019). Bauer and ESPEN address the protein half of the same problem; that half is owned by sibling nutrition titles and is not re-litigated here.
The LIFE trial is the mobility-disability trial: 1,635 sedentary adults aged 70–89 years with physical limitations, able to walk 400 metres, randomized to a structured physical-activity programme versus health education. Structured activity reduced major mobility disability (Pahor et al., 2014). That is established for that endpoint in that sample.
Fall prevention is a separate endpoint. Gillespie, Sherrington, Hopewell, and the Cochrane reviews found that exercise programmes — especially those that challenge balance and include multiple components — reduce the rate of falls in older people living in the community (Gillespie et al., 2012; Sherrington et al., 2019; Hopewell et al., 2018). Sherrington’s updated systematic review is the dose-and-mode statement: programmes that challenge balance and are delivered at a sufficient dose reduce falls; walking-only programmes are a weaker instrument (Sherrington et al., 2017). The Otago Exercise Programme is a named home-based package with its own trial record (Campbell and Robertson, 2003).
Established: progressive resistance improves strength and function in older and frail adults; structured activity can prevent major mobility disability; targeted exercise reduces community falls. Strongly supported: the effective fall programme is not “walk more.” Not established: exercise as a treatment that reverses the Fried phenotype the way an antibiotic reverses bacteraemia.
20 Cancer survivorship
Schmitz and the ACSM cancer roundtable, and Campbell and the 2019 ACSM exercise guidelines for cancer survivors, are the professional-society map: during and after treatment, aerobic and resistance exercise improve fitness, fatigue, quality of life, and several treatment-related impairments, with disease-site and treatment-phase caveats (Schmitz et al., 2010; Campbell et al., 2019). Ligibel and the 2022 ASCO guideline address exercise, diet, and weight management during cancer treatment (Ligibel et al., 2022). Rock and the American Cancer Society nutrition and physical activity guideline address the survivor population (Rock et al., 2022). Cormie and COSA wrote the Australian “exercise as an adjunct to cancer treatment” statement (Cormie et al., 2018). Courneya’s START and related randomized trials are the breast-cancer exercise backbone (Courneya et al., 2007). Buffart and colleagues’ individual-patient meta-analysis examined moderators of quality-of-life effects (Buffart et al., 2017).
Strongly supported: exercise during and after cancer treatment improves fatigue, fitness, and quality of life for several common cancers, most robustly studied in breast cancer. Emerging: site-specific programmes for prostate, colorectal, and haematologic cancers; prehabilitation. Not established: exercise as a cancer-survival drug. Observational associations between activity and cancer incidence or mortality (Moore et al., 2016; Kyu et al., 2016) are strongly supported as associations and are not randomized survival trials of a prescribed programme.
21 Depression and anxiety
Cooney, Dwan, Greig, Lawlor, Rimer, Waugh, McMurdo, and Mead’s Cochrane review found that exercise improves depressive symptoms, with the effect shrinking when only the most rigorous trials are kept (Cooney et al., 2013). Schuch, Vancampfort, Richards, Rosenbaum, Ward, and Stubbs’s later meta-analysis reported larger effects and argued that earlier syntheses were conservative (Schuch et al., 2016). Kvam, Kleppe, Nordhus, and Hovland reached a similar “exercise is a treatment” conclusion (Kvam et al., 2016). Rebar and colleagues’ meta-meta-analysis treated physical activity as associated with less depression and less anxiety across a stack of meta-analyses (Rebar et al., 2015). Stubbs and colleagues isolated anxiety (Stubbs et al., 2017). Gordon and colleagues isolated resistance training and depressive symptoms (Gordon et al., 2018).
Blumenthal’s SMILE programme is the trial clinicians actually argue about: structured aerobic exercise versus sertraline versus placebo in major depression, with later SMILE-adjacent papers on exercise versus pharmacotherapy (Blumenthal et al., 1999; Blumenthal et al., 2007). Singh and colleagues tested high- versus low-intensity weight training in older adults with depression (Singh, Clements, and Fiatarone, 1997; Singh et al., 2005). Those are randomized treatment trials. They are strongly supported for symptom reduction in defined samples. They are not a licence to withdraw an antidepressant without a clinician, and they are not a proof that a walk cancels a melancholic syndrome.
Strongly supported: structured exercise reduces depressive symptoms; resistance training participates in that effect. Emerging: anxiety as a primary endpoint. Not established: exercise as uniformly equivalent to first-line pharmacotherapy or psychotherapy across severity strata. The Cochrane caution about study quality is part of the result, not a footnote.
22 Cognition
Lautenschlager and colleagues randomized older adults with subjective memory impairment to a 24-week home-based physical-activity programme and reported a modest cognitive benefit on the Alzheimer Disease Assessment Scale (Lautenschlager et al., 2008). Erickson and colleagues showed that aerobic training increased hippocampal volume in older adults — a surrogate with a famous figure (Erickson et al., 2011). Northey, Cherbuin, Pumpa, Smee, and Rattray’s meta-analysis found that exercise interventions improve cognitive function in adults older than 50 (Northey et al., 2018). Sofi, Hamer, and Norton’s observational syntheses associate higher activity with lower risk of cognitive decline or dementia (Sofi et al., 2011; Hamer and Chida, 2009; Norton, Matthews, Barnes, Yaffe, and Brayne, 2014). Livingston and the Lancet standing commission include physical inactivity among modifiable dementia-risk factors (Livingston et al., 2020). FINGER tested a multidomain intervention — diet, exercise, cognitive training, vascular-risk management — and improved a composite cognitive score in at-risk older adults (Ngandu et al., 2015). EXERT is the later U.S. test of exercise in mild cognitive impairment; its existence is the field admitting that FINGER was not a single-domain exercise trial (Baker et al., 2022).
Strongly supported: exercise interventions can improve some cognitive-test scores in older adults; inactivity is associated with later dementia. Emerging: hippocampal and network surrogates. Not established: prescribed exercise as a single-domain therapy that prevents Alzheimer disease. FINGER is a multidomain result. Spending it as a walking prescription is a category error.
23 Dose–response
Arem and colleagues, using six pooled cohorts, described a non-linear association between leisure-time physical activity and mortality, with most of the associated benefit present by the range the guidelines name and with additional activity adding smaller increments (Arem et al., 2015). Sattelmair’s coronary-disease dose–response, Kyu’s multi-disease meta-analysis, Wen’s Taiwanese cohort “15 minutes a day” paper, and Saint-Maurice’s life-course leisure-time analysis are the same shape in different clothes (Sattelmair et al., 2011; Kyu et al., 2016; Wen et al., 2011; Saint-Maurice et al., 2019). O’Donovan’s “weekend warrior” analysis asks whether compressing the dose into one or two days retains the association (O’Donovan et al., 2017). Moore and colleagues associated leisure-time activity with risk of 26 cancers (Moore et al., 2016). Ekelund and colleagues asked whether activity attenuates the sitting association (Ekelund et al., 2016).
These are strongly supported observational dose maps. They are not HF-ACTION. They are not LIFE. A person who cannot walk 400 metres is not in Arem’s leisure-time cohorts in the same way. Table D keeps the guideline dose, the trial doses, and the observational doses in separate columns for that reason.
Wasfy and Baggish’s high-dose cardiac discussion is the right-hand tail of the same curve, not a cancellation of the left (Wasfy and Baggish, 2016).
24 Do modes produce meaningfully different outcomes?
Yes, when the outcome is mode-specific. No, when the outcome is a generic “health” noun.
Resistance training is not optional for sarcopenia, bone-loading, and some glycaemia and depression questions (Fiatarone et al., 1994; Watson et al., 2018; Sigal et al., 2007; Gordon et al., 2018). Aerobic training is not optional for claudication, pulmonary rehabilitation, and most blood-pressure and VO₂ questions (Lane et al., 2014; McCarthy et al., 2015; Cornelissen and Smart, 2013). Combined training is the better glycaemia package in DARE and HART-D (Sigal et al., 2007; Church et al., 2010). Balance-challenging work is the fall-prevention package (Sherrington et al., 2019). Flexibility is the mode that most often appears in tables and least often carries a clinical endpoint.
HIIT versus moderate continuous training is the comparison that consumes the most ink per unit of clinical difference. Weston, Wisløff, and Coombes found larger VO₂ gains with HIIT in lifestyle-induced cardiometabolic disease (Weston et al., 2014). Milanović, Sporiš, and Weston found HIIT and endurance training both improve VO₂max, with a possible HIIT edge (Milanović et al., 2015). Gibala and MacInnis explained the physiology (Gibala et al., 2012; MacInnis and Gibala, 2017). SMARTEX-HF is the reminder that a VO₂ edge in a small, tightly coached sample is not automatically a remodeling or event edge in a multicentre heart-failure trial (Ellingsen et al., 2017). Hannan is the coronary-disease fitness synthesis (Hannan et al., 2018).
Strongly supported: mode matching to the outcome matters. Emerging: HIIT as a time-efficient fitness substitute in selected clinical samples. Not established: HIIT as a generally superior clinical-event strategy. A safety clause: interval work raises the intensity that Mittleman, Siscovick, and Albert already showed can trigger events in the untrained (Mittleman et al., 1993; Siscovick et al., 1984; Albert et al., 2000).
25 Supervision, progression, and implementation
The efficacy–effectiveness gap is not a philosophical problem. It is a delivery problem.
Supervised cardiac and pulmonary rehabilitation, supervised claudication walking, and the LIFE activity centres are interventions that include transport, a timetable, a staff member, and a culture of showing up. Home-based programmes can approach those effects when they keep the programme — coaching, progression, measurement — and not when they keep only the pamphlet (McDermott et al., 2013; Campbell and Robertson, 2003). Umpierre’s advice-versus-training split in diabetes is the metabolic version of the same sentence (Umpierre et al., 2011).
Progression is part of FITT-VP (Garber et al., 2011). Trials that do not progress load, duration, or intensity are not testing the intervention they name; they are testing a plateau. LIFTMOR progressed. Fiatarone progressed. Many “exercise advice” arms do not.
Implementation is the reason most eligible patients never see cardiac or pulmonary rehabilitation, the reason supervised claudication therapy is guideline-endorsed and still unevenly available, and the reason a article that stopped at p < 0.05 would be a brochure. This document records that gap. It does not fill it with a startup.
26 Safety, adverse events, and contraindications
Harms are under-reported relative to benefits in the exercise-trial literature. That is an established feature of the record, not a rumour. Small trials mention “no serious events” without a denominator that could have found one. Large trials (HF-ACTION, LIFE, Look AHEAD) are the better safety documents because they had a chance to see rare events.
What the better documents say is not that exercise is harmless. It is that, in screened adults, the event rate during sessions is low, the net population association is protective, and the short-term trigger risk is real in the untrained and the recently silent-ischaemic (Thompson et al., 2007; Franklin et al., 2020; Mittleman et al., 1993; Siscovick et al., 1984; Albert et al., 2000). Riebe’s ACSM screening revision and PAR-Q+ moved the field from “get a doctor’s note” toward a risk-stratified algorithm (Riebe et al., 2015; Bredin et al., 2013; Warburton, Gledhill, Jamnik, Bredin, and McKenzie, 2011). Pelliccia and the ESC sports-cardiology document address the athlete and the congenital or inherited-disease edge (Pelliccia et al., 2021). Maron’s sudden-death work addresses the competitive-athlete edge (Maron et al., 2009). Those edges are not the LIFE sample.
Contraindication language in this article is descriptive of the cited statements, not a decision algorithm. Unstable coronary syndromes, decompensated heart failure, uncontrolled arrhythmia, severe symptomatic aortic stenosis, acute pulmonary embolism, acute myocarditis, and some ophthalmologic or musculoskeletal crises appear in the screening documents as reasons not to start an exercise test or a vigorous programme that day (Fletcher et al., 2013; Riebe et al., 2015). Cancer-specific, dialysis-specific, and post-fracture cautions live in the specialty statements (Campbell et al., 2019; Johansen, 2007; Giangregorio et al., 2014). Individual clearance is a clinical act. It is not performed here.
Musculoskeletal adverse events — delayed-onset soreness, tendinopathy, a fall during a balance session, a vertebral symptom after an ill-chosen load — are the common harms. They are under-counted when trials only report “serious” cardiac events. LIFTMOR’s safety discussion is the model for a high-load trial that does not hide the question (Watson et al., 2018).
27 How large are the effects relative to standard therapy?
Naci and Ioannidis asked this directly for mortality in selected conditions and found that exercise and some drug interventions can be compared, that the comparison is sensitive to which trials enter, and that exercise is not uniformly weaker or stronger (Naci and Ioannidis, 2013; Naci et al., 2019). That is emerging-to-strongly-supported as a comparative-effectiveness claim and not a reason to stop an indicated drug.
Disease by disease, the honest relative-size sentences are these.
In claudication, supervised exercise is first-line therapy in the AHA/ACC guideline alongside, not instead of, risk-factor drugs and, when needed, revascularization (Gerhard-Herman et al., 2017). In COPD, pulmonary rehabilitation is additive to inhalers, not a substitute (Spruit et al., 2013). In hypertension, the mean exercise effect on systolic pressure is smaller than the effect of a well-chosen first drug in a typical titration, and the two are not competitors in the ACC/AHA document (Cornelissen and Smart, 2013; Whelton et al., 2018). In type 2 diabetes, DPP lifestyle out-performed placebo and compared well with metformin on incidence; Look AHEAD lifestyle did not beat education on a cardiovascular composite; glucose-lowering drugs with outcome trials are not cancelled by either result (Knowler et al., 2002; Look AHEAD Research Group, 2013). In HFrEF, HF-ACTION did not replace ACE inhibitors, beta-blockers, mineralocorticoid antagonists, or later SGLT2 inhibitors (O’Connor et al., 2009). In depression, SMILE-class trials make exercise a treatment option, not an automatic equivalent of combined pharmacotherapy and psychotherapy in severe illness (Blumenthal et al., 2007; Cooney et al., 2013). In osteoporosis, exercise is not a bisphosphonate on a fracture Kaplan–Meier curve (Howe et al., 2011; Watson et al., 2018). In cancer, ASCO and ACSM place exercise beside treatment, not over it (Ligibel et al., 2022; Campbell et al., 2019).
The relative-size question is therefore not “exercise or medicine.” It is “for which outcome is exercise the therapy, the adjunct, or the slogan.”
28 Standing constraint
This document describes published research on prescribed exercise as a health and disease intervention. It is not medical advice. It is not a training plan. It does not diagnose, clear, prescribe, or dose for any person. No human use, dose, route, or schedule is recommended anywhere in this document. Individual decisions about exercise in the presence of disease belong to a clinician and the person in front of them, using information this article does not have.
Disease-by-intervention matrix
| Condition | Best-supported mode | Outcome that holds | Outcome that does not | Grade |
|---|---|---|---|---|
| Coronary disease / CR | Aerobic, usually in a multicomponent programme | CV mortality, admission (Cochrane CR) | A gym membership as CR | Strongly supported |
| Hypertension | Aerobic; resistance and isometric also | Blood pressure | Stroke/HF events as exercise monotherapy | Established (BP) |
| Obesity | Aerobic ± resistance; diet still the scale lever | Fitness, visceral fat, some weight | Look AHEAD CV composite | Strongly supported / established miss |
| Type 2 diabetes | Combined aerobic + resistance | HbA1c; incident diabetes (lifestyle) | Microvascular hard outcomes as exercise-only | Established (HbA1c, prevention) |
| Metabolic syndrome | Aerobic or combined | Component scores | A unique syndrome mortality trial | Strongly supported |
| MASLD/NAFLD | Aerobic or resistance | Liver fat | Cirrhosis, HCC, liver death | Strongly supported (surrogate) |
| HFrEF | Aerobic (supervised then home) | Fitness, health status; adjusted admission/death | Unadjusted HF-ACTION primary | Strongly supported / mixed |
| HFpEF | Aerobic | VO₂, symptoms | Mortality | Strongly supported (surrogate) |
| COPD | Pulmonary rehabilitation (aerobic + strength) | Dyspnoea, QoL, capacity | Mortality replacement for inhalers | Established |
| PAD / claudication | Supervised walking | Walking distance | Limb-threat salvage | Established |
| CKD | Aerobic ± resistance | Fitness, function | GFR slope, kidney failure | Strongly supported (function) |
| Osteoporosis | High-load resistance / impact | BMD, function | Fracture endpoint at drug-trial scale | Strongly supported (BMD) |
| Osteoarthritis | Land-based strengthening / aerobic | Pain, function | Radiographic disease modification | Established |
| Chronic LBP | Exercise therapy, no single winner | Pain, function | Imaging-guided structural cure | Established |
| Frailty / sarcopenia | Progressive resistance ± multicomponent | Strength, function | Phenotype “cure” | Established / strongly supported |
| Mobility disability | Structured activity (LIFE) | Major mobility disability | Generalizability to non-walkers | Established |
| Falls | Balance-challenging, sufficient dose | Fall rate | Walking-only as the programme | Established |
| Cancer survivorship | Aerobic + resistance | Fatigue, fitness, QoL | Survival as a prescribed-programme effect | Strongly supported |
| Depression / anxiety | Aerobic and/or resistance | Symptom scores | Uniform drug equivalence | Strongly supported |
| Cognition | Aerobic ± multicomponent | Some test scores | Alzheimer prevention as exercise-only | Emerging / not established |
Guideline comparison
| Document | Population | Dose / mode named | Outcome claimed | What it is not |
|---|---|---|---|---|
| WHO 2020 (Bull et al.) | All ages, population | 150–300 min moderate or 75–150 vigorous + strength ≥2 d; balance in older adults | Population health, NCDs, function | A rehabilitation prescription |
| US PAG 2018 (Piercy et al.) | US population | Same architecture as WHO | Population health | A disease-specific protocol |
| ACSM FITT-VP (Garber 2011) | Apparently healthy adults | Mode-specific FITT-VP | Fitness, health | A cardiac-rehab manual |
| ACSM / AHA older adults (Nelson; Chodzko-Zajko; Izquierdo; Fragala) | Older adults | Aerobic + strength + balance | Function, health | A falls-only protocol |
| ACC/AHA HTN 2017 (Whelton) | Adults with high BP | Activity as nonpharmacologic therapy | BP, risk | A drug-replacement trial |
| ADA / ACSM diabetes (Colberg) | Type 2 diabetes | Aerobic + resistance | Glycaemia, health | A CV-outcome trial |
| ACC/AHA PAD 2016 (Gerhard-Herman) | Claudication | Supervised exercise first-line | Walking, symptoms | A pamphlet |
| ATS/ERS PR (Spruit) | COPD | Rehabilitation programme | Symptoms, capacity, QoL | An inhaler |
| ESC sports cardiology (Pelliccia) | Athletes / known disease | Eligibility and risk | Safety of sport | A population walk |
| OARSI 2019 (Bannuru) | Knee / hip OA | Core exercise | Pain, function | A cartilage drug |
| ACP LBP (Qaseem / Chou) | Low-back pain | Exercise among nonpharmacologic options | Pain, function | A surgical algorithm |
| ACSM / ASCO / ACS cancer (Campbell; Ligibel; Rock) | Cancer treatment / survivorship | Aerobic + resistance, site caveats | Fatigue, function, QoL | A survival drug |
| EWGSOP2 / ICFSR (Cruz-Jentoft; Dent) | Sarcopenia / frailty | Resistance + multicomponent | Strength, function | A myostatin drug |
| KDIGO lifestyle | CKD | Activity as lifestyle | Fitness, risk | A GFR trial |
RCT and meta-analysis matrix
| Study or synthesis | Design | n / scope (as reported) | Intervention | Primary or headline outcome | Result class |
|---|---|---|---|---|---|
| HF-ACTION (O’Connor 2009) | RCT | 2331 HFrEF | Supervised then home aerobic | All-cause death or hospitalization | Missed unadjusted; improved VO₂ / status |
| Look AHEAD (2013) | RCT | Adults with T2D + overweight | Intensive lifestyle | CV composite | Neutral |
| DPP (Knowler 2002) | RCT | High-risk adults | Lifestyle vs metformin vs placebo | Incident diabetes | Lifestyle reduced incidence |
| Finnish DPS (Tuomilehto 2001) | RCT | High-risk adults | Lifestyle | Incident diabetes | Reduced incidence |
| DARE (Sigal 2007) | RCT | T2D | Aerobic, resistance, both | HbA1c | Combined best |
| HART-D (Church 2010) | RCT | T2D | Aerobic, resistance, both | HbA1c | Combined reduced HbA1c |
| STRRIDE (Kraus / Slentz) | RCT | Overweight adults | Aerobic dose / intensity | Fitness, lipids, fat | Dose-responsive surrogates |
| LIFE (Pahor 2014) | RCT | 1635 older adults | Structured activity | Major mobility disability | Reduced |
| SMILE / Blumenthal | RCT | Major depression | Exercise vs sertraline vs placebo | Depression scores | Exercise active |
| SMARTEX-HF (Ellingsen 2017) | RCT | HFrEF | HIIT vs moderate | Reverse remodeling | No clear HIIT superiority |
| LIFTMOR (Watson 2018) | RCT | Postmenopausal low bone mass | High-intensity resistance + impact | BMD, function | Improved vs low-intensity |
| IDEA (Messier 2013) | RCT | Knee OA + overweight | Diet, exercise, both | Pain, joint load | Combination strongest |
| FINGER (Ngandu 2015) | RCT | At-risk older adults | Multidomain | Cognitive composite | Improved |
| Lautenschlager 2008 | RCT | Subjective memory impairment | Home activity | ADAS-Cog | Modest benefit |
| Fiatarone 1994 | RCT | Institutionalized very old | High-intensity resistance | Strength, gait | Improved |
| Cochrane CR (Anderson / Dibben) | MA | CHD rehabilitation | Exercise-based CR | CV death, admission | Benefit |
| Cochrane HF (Taylor / Long) | MA | Heart failure | Exercise-based rehab | Admission, capacity | Benefit; cautious mortality |
| Cochrane COPD (McCarthy / Puhan) | MA | COPD | Pulmonary rehab | QoL, capacity, symptoms | Benefit |
| Cochrane claudication (Lane) | MA | PAD | Exercise | Walking | Benefit |
| Cochrane CKD (Heiwe) | MA | CKD | Exercise training | Fitness, function | Benefit; small trials |
| Cochrane bone (Howe) | MA | Postmenopausal women | Exercise | BMD | Small BMD gain |
| Cochrane knee OA (Fransen) | MA | Knee OA | Land-based exercise | Pain, function | Benefit |
| Cochrane LBP (Hayden) | MA | Chronic LBP | Exercise therapy | Pain, function | Benefit |
| Cochrane depression (Cooney) | MA | Depression | Exercise | Symptoms | Benefit; quality-sensitive |
| Cochrane falls (Gillespie / Sherrington) | MA | Older adults | Exercise / multifactorial | Falls | Benefit if balance challenged |
| Cornelissen 2013 | MA | BP | Endurance / resistance / isometric | Blood pressure | Reduction |
| Umpierre 2011 | MA | T2D | Structured vs advice | HbA1c | Structured moves HbA1c |
| Schuch 2016 | MA | Depression | Exercise | Symptoms | Larger effect than some earlier MAs |
| Northey 2018 | MA | Adults >50 | Exercise | Cognition | Improved test scores |
| Weston 2014 / Milanović 2015 | MA | Clinical / mixed | HIIT vs endurance | VO₂ | HIIT fitness edge |
Dose–response table
| Source class | Reported dose parameter | Outcome | What the number is |
|---|---|---|---|
| WHO 2020 / US PAG 2018 | 150–300 min·wk⁻¹ moderate or 75–150 vigorous + strength ≥2 d | Population NCD / function | Consensus parameter, not a trial arm |
| Umpierre 2011 | Structured training; >150 min·wk⁻¹ associated with larger HbA1c drop | HbA1c | Meta-analytic stratum |
| ACSM weight (Donnelly 2009) | Higher aerobic volumes for scale weight than for fitness | Weight | Position-stand synthesis |
| Arem 2015 | Leisure-time MET-hours; most associated mortality benefit near guideline range | All-cause mortality | Pooled observational |
| Sattelmair 2011 | Activity dose vs CHD risk | Incident CHD | Observational dose–response |
| Wen 2011 | ~15 min·d⁻¹ associated with lower mortality vs inactive | Mortality | Observational cohort |
| LIFE 2014 | Centre-based activity, progressed walking and strength | Major mobility disability | Randomized programme dose |
| HF-ACTION 2009 | Protocol 36 supervised sessions then home, 30 min | Death / hospitalization | Randomized; adherence < protocol |
| LIFTMOR 2018 | Twice-weekly high-intensity resistance + impact, 8 months | BMD | Randomized high-load dose |
| Sherrington 2017/2019 | Balance challenge + sufficient dose | Falls | Review dose statement |
| HIIT syntheses | Brief hard intervals, various 4×4 and low-volume protocols | VO₂ | Surrogate dose |
Adherence and safety matrix
| Setting | Adherence fact | Safety fact | Translation limit |
|---|---|---|---|
| HF-ACTION | Below-protocol exercise minutes | No new training hazard signal | Efficacy ≠ completed dose |
| Look AHEAD | Large early uptake; fade over years | Lifestyle safety acceptable | Surrogates moved; CV composite did not |
| LIFE | Centre-based programme sustained enough to move disability | Falls / events monitored in old, limited adults | Enrollees could already walk 400 m |
| SMARTEX-HF | Interval arm under-shot prescribed intensity | HIIT not shown superior | Supervision does not guarantee intensity |
| Claudication supervised | Attendance is the intervention | Ischaemic pain is used as a signal | Home pamphlet ≠ supervised ET |
| Pulmonary rehab | Programme completion drives the estimate | Exertional desaturation / comorbidity | Effectiveness falls when the building is far |
| LIFTMOR | Supervised high load, twice weekly | Safety discussion is part of the paper | Community classes are not this protocol |
| Depression RCTs | Completers analyses inflate effects (Cochrane caution) | Medical screening varies | Healthy, motivated depressives ≠ all MDD |
| Observational PA cohorts | Survivors who still exercise | Residual confounding, reverse causation | Association ≠ prescription |
| Preparticipation (Riebe; PAR-Q+) | Screening exists because events exist | Trigger risk highest in sedentary + vigorous | Clearance is clinical, not documentary |
Evidence handling
Study type is named in the sentence that uses the result. Randomized trials outrank observational cohorts for causal claims about programmes. Cochrane and society syntheses outrank single small trials for direction, and remain secondary to the landmark trial when the landmark trial is the thing being discussed. Guidelines are labelled as guidelines. Animal and in-vitro work, if it appears, is labelled as such and does not carry a clinical grade above plausible.
Conflicting results are both given. HF-ACTION’s unadjusted miss and adjusted signal are both given. Look AHEAD’s risk-factor success and event-composite miss are both given. SMARTEX-HF’s failure to reproduce Wisløff’s HIIT superiority is given next to Wisløff. Cochrane depression’s quality-sensitive shrink is given next to Schuch’s larger estimate. The primate-style lesson from the sibling nutrition titles applies: when two good studies disagree, the disagreement is the finding.
Amounts are experimental or consensus parameters. They are not prescriptions.
Project 06 was queried read-only (D:\PROJECT_06_KB\project06_catalog.sqlite, identity PROJECT_06, PROJECT_05_IMPORT_STATUS=NOT_IMPORTED). Hits informed discovery. They did not become citations. Project 07 news was scanned as discovery and is not evidence. Peptide-article HTML files were catalogued (125 files). HOUSESTYLE.md and HOUSE_STYLE.md through A48 were read. Full prose read of every peptide article is not claimed.
Adversarial resolution
Six adversarial lenses were applied to the reviewed record. Identifiers that do not appear in VERIFIED_CITATION_STORE.jsonl are marked UNVERIFIED in the ledger and are not used as load-bearing proof.
1. Adherence / healthy-participant selection. Yes, benefits are exaggerated when completers, run-in survivors, and already-mobile adults are treated as “patients with the disease.” HF-ACTION, Look AHEAD, LIFE, and the Cochrane depression quality filter are the exhibits. The exaggeration is a bias, not a proof of nullity.
2. Surrogates in place of clinical outcomes. Yes, routinely: VO₂, HbA1c, blood pressure, liver fat, BMD, hippocampal volume. The article grades those as surrogates. It spends events only where events were measured (Cochrane CR, Look AHEAD, HF-ACTION, LIFE, DPP incidence, falls).
3. Effect size versus standard therapy. Exercise is first-line for claudication walking distance, pulmonary-rehab symptoms, chronic LBP and OA pain/function, and fall risk. It is adjunctive for HFrEF, coronary disease, diabetes, hypertension, cancer care, and depression. It is not a replacement for outcome-proven drugs in HFrEF, fracture, or severe hypertension. Naci’s comparisons are the explicit relative-size literature and are sensitive to trial mix.
4. Supervised efficacy versus real-world adherence. The gap is established. Umpierre’s advice-versus-training split, McDermott’s home-versus-supervised PAD work, rehabilitation attendance, and SMARTEX-HF’s missed interval intensity are the exhibits. Implementation is part of the intervention.
5. Mode differences. Meaningful when the outcome is mode-specific (resistance for muscle and bone; walking for claudication; balance for falls; combined for HbA1c). Overstated when the outcome is a generic wellness noun. HIIT’s most reliable advantage is VO₂, not events.
6. Under-reported harms. Yes. Small trials are underpowered for rare cardiac events; musculoskeletal harms are often invisible; screening documents exist because the trigger risk is real. Large-trial safety is more trustworthy than “no events” in an n = 24 interval study.
Verdict on the owner’s question: prescribed exercise is a real, disease-specific intervention whose strongest effects are on function, symptoms, selected risk factors, diabetes incidence, fall rate, claudication distance, and rehabilitation-era coronary outcomes — and whose weakest habit is the conversion of those effects into a universal, unsupervised, mode-indifferent, event-level medicine.
References
Continue reading