Bibliographie
Voici la version technique de la recherche derrière Exercise Breaks. Les 19 entrées ci-dessous — 17 articles évalués par les pairs, un éditorial d'accompagnement et une ligne directrice de pratique clinique — donnent le titre, les auteurs, la revue ou l'éditeur, le DOI ou l'URL, un court résumé du résultat, la nuance la plus importante, la source de financement quand elle a été vérifiée, et une étiquette de pertinence.
Étiquettes : Article principal (les prétentions de l'app reposent directement sur cette entrée), Connexe (contexte de soutien ou littérature parallèle), Contrainte (une entrée qui limite ce que l'app peut prétendre), Intervention (concerne la question de savoir si les rappels produisent un changement de comportement), Éditorial (commentaire sur un article primaire), ou Ligne directrice (une ligne directrice de pratique clinique). Le résumé en langage simple qui parcourt les mêmes preuves résultat par résultat est à /fr/research/.
Note sur la langue : les titres, résumés et nuances par article ci-dessous restent en anglais. Les articles cités sont publiés en anglais, et reproduire fidèlement la formulation des auteurs originaux compte plus que de traduire ; les abréger ou les traduire risquerait de modifier ce que les articles affirment. Le chrome de la page (titres, introduction, étiquettes, synthèse de financement) est en français.
Articles principaux
Trois articles ancrent Exercise Breaks plus directement que le reste de la bibliographie. L'écran Ressources dans l'app les nomme tous les trois ; cette page les traite avec un peu plus de détail.
Fang et al. 2026
Article principalMicro-exercise breaks every hour: a feasible strategy to improve metabolic health in sedentary office workers.
A 12-week randomized controlled trial in 86 sedentary office workers in Nanchang, China, testing the same protocol Exercise Breaks delivers — 3-minute micro-exercise breaks every hour during the 8-hour workday, 7 breaks per day, light-to-moderate bodyweight exercises. Significant improvements vs. control across fasting glucose, post-meal glucose, insulin sensitivity (~20% improvement in HOMA-IR), systolic and diastolic blood pressure, body composition, lipids, and self-reported energy and afternoon fatigue (the largest effect size in the study, d=0.98). 56% of participants who started with prediabetes had returned to normal blood sugar by week 12 vs. 13% in control. Adherence averaged 82% over 12 weeks. The closest thing in the literature to a direct test of the app's exact protocol.
Caveat: Single cultural context (three Nanchang institutions), n=86, predominantly overweight sample; the paper itself flags generalizability concerns to Western and normal-weight populations. Self-reported productivity / energy / fatigue outcomes were collected in an unblinded trial.
Funding: Verified — no specific grant from any public, commercial, or not-for-profit source. The trial was effectively unfunded. Authors declare no competing interests.
Duran et al. 2023
Article principalBreaking Up Prolonged Sitting to Improve Cardiometabolic Risk: Dose–Response Analysis of a Randomized Crossover Trial.
A randomized crossover trial in 11 middle-to-older-age adults (mean 57 y) at Columbia University. Each participant completed five 8-hour conditions in randomized order: a sedentary control plus four interval × duration combinations of light-intensity treadmill walking — 1 or 5 minutes every 30 or 60 minutes. The only paper in this literature that directly compares break doses head-to-head in the same design. Every break dose produced a 3–5 mm Hg drop in systolic blood pressure. Mood and fatigue improved at every dose except the minimum 1-minute / 60-minute arm. Significant glucose attenuation appeared at the 5-minute / 30-minute dose but not at hourly intervals.
Caveat: Very small sample (n=11; enrollment cut short by COVID), acute 8-hour lab sessions rather than chronic free-living behaviour. Predominantly healthy adults — effects in a hypertensive sample might differ.
Funding: Verified — Robert N. Butler Columbia Aging Center, an internal Columbia research center with no industry ties. Authors report no conflicts of interest.
Chastin et al. 2021
Article principalJoint association between accelerometry-measured daily combination of time spent in physical activity, sedentary behaviour and sleep and all-cause mortality: a pooled analysis of six prospective cohorts using compositional analysis.
A pooled compositional analysis of six prospective accelerometry cohorts from the UK, US, and Sweden (n>130,000 adults, 4-year follow-up). Found that 3 minutes of moderate-to-vigorous activity — or 12 minutes of light activity — per hour of sitting was associated with roughly 30% lower odds of early death. The "3 minutes per hour of sitting" framing is exactly how Exercise Breaks structures its protocol, and this is the largest pooled accelerometry analysis to express its finding in those units. Multiple alternative activity-sleep-sitting "cocktails" yielded similar 30% reductions, suggesting the population signal is robust to mix.
Caveat: Observational, not causal. Sicker people sit more and die sooner, which confounds analyses of this kind; the population-scale signal can't be assigned to the breaks themselves.
Funding: Verified — 17 grants from the US National Institutes of Health (NHLBI, NIDDK, NCI, NIA, NINDS), the British Heart Foundation, and the UK Medical Research Council. All public-sector. Competing interests: none declared. (One co-author is affiliated with Accelting, his own consultancy on accelerometer-analysis methods — an academic-methodology consultancy, not industry pharma.)
Le reste de la bibliographie
Les 16 entrées restantes — 14 articles, un éditorial d'accompagnement et une ligne directrice de pratique clinique — dans l'ordre du document source. Chacune est étiquetée selon sa pertinence par rapport aux prétentions de l'app.
Jenkins et al. 2019
ConnexeDo stair climbing exercise "snacks" improve cardiorespiratory fitness?
The original randomized trial that put "exercise snacks" into the literature. n=24 sedentary young adults performed three vigorous stair-climbing bouts per day, 3 days per week for 6 weeks, with 1–4 hours of recovery between bouts. Peak VO₂ was modestly but significantly higher post-intervention (P=0.003).
Caveat: The protocol is vigorous-intensity stair climbing in roughly 20-second efforts — different from Exercise Breaks' 3-minute light-to-moderate breaks. Cite for the term and concept, not as direct support for the app's defaults.
Funding: Not separately verified for this entry.
Islam, Gibala, Little 2022
ConnexeExercise snacks: a novel strategy to improve cardiometabolic health.
A narrative review that formalizes "exercise snacks" as isolated bouts of one minute or less of vigorous activity performed several times per day, separated by recovery. Synthesizes the cardiorespiratory and cardiometabolic evidence across the cluster. The best single paper to cite when explaining how the literature defines "exercise snacks" — and where Exercise Breaks departs from that definition (3 minutes, light-to-moderate, not ≤1 minute vigorous).
Caveat: Narrative review, not a quantitative meta-analysis.
Funding: Not separately verified for this entry.
Little et al. 2019
ConnexeSprint exercise snacks: a novel approach to increase aerobic fitness.
A follow-up paper from the same group as Jenkins 2019, testing sprint-based "exercise snacks" for aerobic fitness. Useful as evidence that the exercise-snacks literature is more than one paper, though it doesn't add new support for the app's specific 3-minute light-to-moderate protocol.
Caveat: Same group, same cluster as Jenkins 2019 and Islam 2022; sprint intensity sits well above the app's light-to-moderate regime, so effect sizes don't transfer.
Funding: Not separately verified for this entry. (Source bibliography flags the DOI as "look up by title" — not pulled in this pass.)
Stamatakis et al. 2022
ConnexeAssociation of wearable device-measured vigorous intermittent lifestyle physical activity with mortality.
UK Biobank cohort of 25,241 self-reported non-exercisers measured by wrist accelerometry over a median 6.9-year follow-up (852 deaths). A median of 3 vigorous intermittent lifestyle physical activity (VILPA) bouts per day of 1–2 minutes each was associated with a 38–40% reduction in all-cause and cancer mortality and a 48–49% reduction in cardiovascular mortality vs. no VILPA. Near-linear dose-response.
Caveat: Observational. The activity intensity is vigorous, which Exercise Breaks does not require — these effect sizes don't transfer to a light-to-moderate regime.
Funding: Not separately verified for this entry. (Stamatakis discloses equity in "Complement 1," a VILPA-related company, in his subsequent papers — see Koemel 2025 below.)
Nature Medicine editorial 2022
ÉditorialVigorous intermittent lifestyle physical activity improves mortality risk.
An editorial accompanying Stamatakis 2022 that summarizes the headline finding for a lay audience: "as little as 3–4 minutes of VILPA per day is associated with substantially reduced mortality risk." Useful as a clean lay-summary citation; not a substitute for citing the underlying paper.
Caveat: Editorial commentary, not original research — defer to the underlying Stamatakis 2022 paper for any analytical claim.
Funding: Editorial — no separate funding statement; any disclosure attaches to the underlying paper.
Koemel et al. 2025
ConnexeVigorous intermittent lifestyle physical activity (VILPA) and mortality risk among US adults: a wearables-based national cohort study.
An independent US replication of Stamatakis 2022 in a nationally representative NHANES 2011–14 sample of 3,293 non-exercisers measured by wrist accelerometry. A median of ~5 VILPA bouts per day was associated with 44% lower all-cause mortality risk (HR 0.56, 95% CI 0.39–0.82). The authors note in the abstract that the effect attenuates when participants with prevalent CVD or cancer are excluded, suggesting some reverse causation.
Caveat: Observational. The authors' own caveat about reverse causation should travel with any quote of this paper.
Funding: Not separately verified for this entry. Co-author Stamatakis discloses equity in "Complement 1," a company whose products relate to VILPA; the disclosure is appropriately reported in the paper.
Ahmadi et al. 2024
ConnexeDevice-measured vigorous intermittent lifestyle physical activity (VILPA) and major adverse cardiovascular events: evidence of sex differences.
A VILPA analysis with sex-disaggregated estimates for major adverse cardiovascular events. Useful as supporting context for cardiology-audience outreach.
Caveat: Citation incomplete in the source bibliography (full author list and journal not pulled); not verified in detail. Same vigorous-intensity caveat as the rest of the VILPA literature.
Funding: Not separately verified for this entry.
Diaz et al. 2017
ConnexePatterns of Sedentary Behavior and Mortality in U.S. Middle-Aged and Older Adults: A National Cohort Study.
REGARDS cohort, n=7,985 adults aged ≥45, hip accelerometry, median 4-year follow-up (340 deaths). Both total sedentary time and mean sedentary bout length were independently associated with all-cause mortality in a dose-dependent fashion — highest risk in those with both high total sitting (>12.5 h/day) and long uninterrupted bouts (≥30 min). Foundational evidence for the premise that how sitting accumulates matters, not just how much.
Caveat: Observational, not causal — must be noted in any copy that cites the bout-length finding.
Funding: Not separately verified for this entry.
Dunstan et al. 2012
ConnexeBreaking up prolonged sitting reduces postprandial glucose and insulin responses.
A randomized three-period crossover trial in 19 overweight/obese adults aged 45–65. Compared uninterrupted sitting against 2-minute light-intensity walks every 20 minutes and 2-minute moderate-intensity walks every 20 minutes. Both activity-break conditions reduced 5-hour postprandial glucose iAUC vs. uninterrupted sitting (P<0.01). The most-cited primary RCT for the "break up sitting" protocol.
Caveat: Acute laboratory effect, small sample, specific population. The interval tested was 20 minutes, not 60.
Funding: Not separately verified for this entry.
Peddie et al. 2013
ConnexeBreaking prolonged sitting reduces postprandial glycemia in healthy, normal-weight adults: a randomized crossover trial.
A randomized crossover trial in 70 healthy normal-weight adults across three 9-hour conditions: prolonged sitting; one 30-minute continuous walk then sitting; or 1 minute 40 seconds of walking every 30 minutes. The pattern of frequent short breaks reduced insulin iAUC more than either prolonged sitting or a single 30-minute walk (P<0.001), with the same pattern for glucose. The headline finding — frequent short breaks beat one longer block — is the single most direct support for the philosophy behind Exercise Breaks.
Caveat: Acute, healthy young population, 30-minute interval (not 60).
Funding: Not separately verified for this entry.
Gao et al. 2024
ContrainteImpact of Prolonged Sitting Interruption on Blood Glucose, Insulin and Triacylglycerol in Adults: A Systematic Review and Meta-Analysis.
A network meta-analysis of randomized trials comparing different sitting-interruption frequencies. The "every 30 minutes" interruption had the highest probability of being the best frequency for glycemic outcomes (SUCRA 81.8% for glucose, 77.5% for insulin); every 20 minutes also significant. Longer intervals were not the top-ranked interventions. The paper that most directly challenges the app's 60-minute default: it does not say 60-minute is useless, but it says 30-minute has the strongest acute glycemic evidence — which is why the app does not claim "research-backed 60-minute interval" for acute glucose specifically.
Caveat: MDPI's Applied Sciences has mixed peer-review reputation; spot-check methodology before leaning hard on this single paper. The 30-minute-is-best signal is consistent across multiple primary studies regardless.
Funding: Not separately verified for this entry.
Dempsey et al. 2020
ConnexeInterrupting prolonged sitting with repeated chair stands or short walks reduces postprandial insulinemia in healthy adults.
An acute trial finding that the type of break (chair stands ≈ walking) matters less than the fact of the interruption itself. Both produced comparable reductions in postprandial insulinemia.
Caveat: Acute laboratory effect, small sample, healthy adults; the equivalence between break types holds within light-to-moderate intensity but not necessarily across intensity regimes.
Funding: Not separately verified for this entry.
American Diabetes Association — Standards of Care
Ligne directriceStandards of Medical Care in Diabetes — sedentary-break recommendations.
The ADA's Standards of Medical Care in Diabetes recommend interrupting bouts of sitting with brief activity every 30 minutes, with a special note for type-2 diabetes benefit. Not a primary source, but "the ADA recommends X" is a useful credibility anchor for lay audiences who won't read a primary RCT — primary-source papers should sit behind it.
Caveat: Clinical guideline, not original research; the recommended interval is 30 minutes (closer to Gao 2024's strongest acute glycemic evidence than to Exercise Breaks' 60-minute default).
Funding: Clinical practice guideline produced through the ADA's institutional process, not a per-paper grant.
Ekelund et al. 2016
ContrainteDoes physical activity attenuate, or even eliminate, the detrimental association of sitting time with mortality? A harmonised meta-analysis of data from more than 1 million men and women.
A harmonised meta-analysis of 16 prospective cohort studies, n=1,005,791, with 84,609 deaths over follow-ups of 2 to 18+ years. High levels of moderate-intensity physical activity (~60–75 min/day) appear to eliminate the increased mortality risk associated with high sitting time. TV-viewing time retained some residual risk even at high activity. The most-cited paper on whether exercise compensates for sitting — and the reason Exercise Breaks does not claim "you can't compensate for sitting with exercise."
Caveat: The 60–75 min/day threshold is far above the WHO 150 min/week guideline (~22 min/day); most desk workers don't hit it. Also addresses all-cause mortality only — not the acute physiological outcomes during the workday that Fang and Duran cover.
Funding: Verified — UK Medical Research Council programme grant MC_UU_12015/3 partly funded Ekelund. No relevant conflicts of interest.
Ekelund et al. 2020
ContrainteJoint associations of accelerometer-measured physical activity and sedentary time with all-cause mortality: a harmonised meta-analysis in more than 44,000 middle-aged and older individuals.
A harmonised meta-analysis of nine prospective cohorts (n=44,370, 3,451 deaths) using accelerometer-measured activity rather than self-report. About 30–40 minutes per day of moderate-to-vigorous physical activity attenuated the association between sedentary time and mortality risk. The stronger version of the 2016 finding, with better measurement methodology — confirms that sufficient daily MVPA does compensate for sitting at the all-cause mortality endpoint.
Caveat: 30–40 min/day is closer to the WHO 150 min/week guideline but most office workers still don't hit it. Same scope limit as the 2016 paper — addresses mortality, not acute workday physiology.
Funding: Not separately verified at the funding-statement level. Same lead author and institutional pattern as Ekelund 2016, no reason to expect a different disclosure profile.
Leppe-Zamora et al. 2025
InterventionThe effect of computer prompt in breaks of sedentary behaviour among office workers: a systematic review and meta-analysis.
A systematic review and meta-analysis of 18 randomized trials (1,164 office workers) testing computer-prompt software vs. control. Pooled reduction in workday sitting: −12.46 min/day (95% CI −18.12 to −6.80). Pooled increase in steps: +1,029.99 steps/day (95% CI 815.97 to 1,244). Secondary cardiometabolic and work-performance outcomes trended in favour of prompts but did not reach statistical significance. The only meta-analysis that isolates computer-prompt interventions specifically — direct support for the basic premise that being reminded to move produces objective behaviour change.
Caveat: Effect size is modest (~12 min/workday). Honest framing: meaningful and real, not transformative on its own. GRADE certainty was low to moderate.
Funding: Authors at Universidad del Desarrollo (Chile), Pontificia Universidad Católica de Chile, and Universidad de los Andes (Colombia). All academic; no competing interests declared. The funding statement itself was not extracted in the verification pass.
Qui a financé cette recherche
Une question raisonnable à se poser de n'importe quel produit appuyé par la recherche est de savoir qui a payé pour la recherche. Le profil de financement et de conflits d'intérêts à travers les 19 entrées ci-dessus :
Vérifié sans conflit — le trio d'articles principaux plus les articles « Contrainte » et l'article « Intervention ». Fang 2026 n'a pas été financé (aucune subvention d'aucune source publique, commerciale ou sans but lucratif) et les auteurs ne déclarent aucun conflit d'intérêts. Duran 2023 a été financé par le Robert N. Butler Columbia Aging Center, un centre de recherche interne de Columbia sans lien avec l'industrie ; aucun conflit déclaré. Chastin 2021 a été financé par 17 subventions des US National Institutes of Health, de la British Heart Foundation et de l'UK Medical Research Council — pur financement du secteur public ; aucun conflit d'intérêts déclaré. Ekelund 2016 a été financé en partie par l'UK Medical Research Council ; aucun conflit d'intérêts pertinent. Ekelund 2020 n'a pas été vérifié séparément au niveau de la déclaration de financement, mais suit le même auteur principal et la même configuration institutionnelle. Les auteurs de Leppe-Zamora 2025 dans des universités chiliennes et colombiennes ne déclarent aucun conflit d'intérêts ; la déclaration de financement elle-même n'a pas été extraite dans ce passage.
Divulgation signalée — Stamatakis 2022 et Koemel 2025. Stamatakis déclare une participation au capital de « Complement 1 », une entreprise dont les produits sont liés à l'activité physique intermittente vigoureuse. La divulgation est rapportée de façon appropriée dans les articles. Exercise Breaks ne cite pas les réductions de mortalité du VILPA comme appuyant ses propres prétentions, car l'app utilise une intensité légère à modérée plutôt que le régime vigoureux que la littérature VILPA décrit — donc la divulgation ne façonne pas matériellement le texte de l'app.
Non vérifié séparément — les neuf articles restants : Jenkins 2019, Islam et al. 2022, Little 2019, Ahmadi 2024, Diaz 2017, Dunstan 2012, Peddie 2013, Gao 2024 et Dempsey 2020. Aucun de ces articles n'appuie directement les prétentions de l'app. Si l'un d'eux passait plus tard à un rôle d'appui direct, la déclaration de financement devrait être vérifiée avant que cela ne se produise.
Cas particuliers. L'éditorial de Nature Medicine accompagnant Stamatakis 2022 n'a pas de déclaration de financement distincte — les éditoriaux sont des commentaires sur les articles principaux du numéro, et toute divulgation est rattachée à l'article sous-jacent. Les Standards of Care de l'American Diabetes Association est une ligne directrice de pratique clinique produite par le processus de gouvernance institutionnel de l'ADA plutôt que par une subvention par article ; ses divulgations vivent dans la structure de l'ADA, pas dans cette bibliographie.
Aucun des articles principaux ne rapporte de financement industriel. Aucun de leurs auteurs ne déclare de conflits d'intérêts financiers pertinents pour les interventions de pauses sédentaires.
Il vaut la peine de le dire clairement : l'absence de conflits d'intérêts financiers n'est pas la même chose que l'absence de biais. Les chercheurs qui ont bâti des carrières dans ce domaine ont des intérêts professionnels et de réputation à ce que la littérature continue de constater que les pauses d'exercice aident, de la même façon que les nutritionnistes qui ont bâti des carrières autour d'une catégorie d'aliments ont des intérêts à ce que cette catégorie continue de paraître bonne. La protection structurelle contre ce genre de biais est la réplication par des groupes indépendants dans différents pays — ce que l'ensemble des articles principaux a effectivement : Fang en Chine, Duran aux États-Unis, Chastin groupant des cohortes US/Royaume-Uni/Suède, Ekelund menant un consortium international basé en Norvège, Leppe-Zamora groupant Chili/Colombie. La réplication trans-pays et trans-institutionnelle de la même direction d'effet est ce qui rend cette littérature digne de confiance au-delà des articles individuels.