I train, then sit all day: does it matter?
Unknown. In 84 general adults thigh activity faded across a long sit, a little less after harder activity. No health outcome, no athletes selected: it cannot say if training offsets anything.
Main caveat: No health outcome. Muscle activity only; the title's metabolic health is inference from other studies.
Brakenridge and colleagues asked what thigh muscles do during ordinary sitting. Adults from Finland and China wore EMG-sensing shorts with an accelerometer through waking hours, with no behavioural instructions, and the analysis took each accelerometer-defined sitting bout and asked how often EMG rose above a 3 µV baseline.
Activity fell with bout duration on a power-law curve, was lower in people whose usual bouts were 25 minutes or longer, and was somewhat higher in bouts that followed longer or more intense activity. The authors offer a speculation that the health associations of prolonged sitting could be due to this fall in muscle activity, and suggest breaking up sitting.
Our reading: it is a useful first free-living description of a suspected mechanism, measured in muscle rather than by hip movement. It tests neither the mechanism nor any remedy. "Implications for metabolic health" in the title is the authors' extrapolation from other work: this study measured no metabolic outcome.
Protocol
Who: n = 84 after removing 26 for poor synchronisation and 2 for artefacts (112 before, our arithmetic). EMG24, n = 55: 20-76 y, no major illness, Jyväskylä. BREAK, n = 14: women, overweight or obese (BMI ≥ 25) with no metabolic disorder, 20.8 y, Hangzhou. OPTIMUS, n = 15: BMI ≥ 25 with non-insulin-dependent type 2 diabetes, 63.1 y, Mikkeli. Overall 54 women (64.3%), 43.0 (18.7) y, BMI 25.8 (4.9).
Measures: textile EMG shorts (Myontec, Suunto), quadriceps and hamstrings in EMG24, plus gluteals in the other two; normalised to maximal voluntary contraction; calves not measured. Sitting from a hip-worn accelerometer with a machine-learning classifier (EMG24) or a thigh-worn Fibion (BREAK, OPTIMUS). Median 2.0 synchronised days at about 12 h a day, 30.3 sitting bouts a day.
Outcome: EMG activity, the share of time above a 3 µV baseline during sitting. First and last minute of each bout removed. No glucose, lipid, blood pressure, fitness or performance outcome. BMI, waist and body fat describe the sample and BMI is a covariate.
Predictors: time within the bout; usual bout duration (a person-level measure, split at 25 min); duration and intensity of the bout of activity directly before the sit, intensity graded against each person's laboratory standing and walking EMG.
Analysis: non-linear power-law and exponential fits; linear and mixed models with random intercepts; adjustment for age, sex and BMI is printed only for the preceding-activity models (Figure 2). Estimated marginal means with intervals; the text prints no contrasts or P values for these results.
Funding and interests: EMG24: Academy of Finland 128643, Ministry of Education 42/627/2010, Urheiluopistosäätiö. BREAK: Fundamental Research Funds for the Central Universities S20250035, National Natural Science Foundation of China 82003485 and 82111530074. OPTIMUS: Academy of Finland 332731. The authors declare no competing interests.
Key Results
| Sitting EMG activity, min per hour of sitting (95% CI) | Lower group | Higher group |
|---|---|---|
| Usual bout ≥25 vs <25 min, overall | 6.55 (5.25 to 7.85) | 10.42 (9.06 to 11.77) |
| — EMG24 | 7.74 (6.45 to 9.03) | 9.94 (8.69 to 11.20) |
| — BREAK | 6.44 (−0.10 to 12.98) | 11.71 (5.40 to 18.02) |
| — OPTIMUS | 3.12 (1.99 to 4.25) | 7.53 (5.07 to 9.99) |
| Preceding bout <3 vs >10 min, overall | 7.76 (6.53 to 9.09) | 9.87 (8.49 to 11.37) |
| Preceding standing vs above walking, overall | 6.69 (5.43 to 8.08) | 9.84 (8.00 to 11.86) |
| Preceding standing vs above walking, EMG24 | 8.48 (6.94 to 10.18) | 13.49 (10.86 to 16.41) |
| Preceding standing vs above walking, BREAK | 4.13 (1.70 to 7.64) | 6.96 (3.56 to 11.49) |
The curve: overall fit, EMG activity % = 25.72 × t^−0.461 (t in minutes), power law preferred to exponential (BIC 709 vs 826). Bouts over 90 minutes were under 1% of bouts.
The 25 minutes does not follow from the printed fit: the paper says this equation reaches 10%, 5% and 1% at 8.9, 12.2 and 25.7 min. Put in those times, it gives about 9.4%, 8.1% and 5.8%; it reaches 10%, 5% and 1% at about 8, 35 and 1,150 min (our arithmetic). Sample-specific functions are in a supplement we did not read, and the mismatch is unexplained.
"5 extra minutes" per hour: the abstract says more frequent and intense interruptions result in this. No derivation is printed. The nearest figure is one contrast in one cohort, EMG24 by intensity, 13.49 minus 8.48 (our arithmetic; BREAK's usual-bout gap, 11.71 minus 6.44, is also near 5, but it compares people, not interruptions).Overall the contrasts are about 3.9 (usual bout), 3.1 (intensity) and 2.1 (duration) min per hour. Frequency of interruption was not itself a predictor, and the design cannot show "result in".
Carry-over overlap: the overall preceding-activity intervals overlap, and no test of the difference is printed, while the Discussion says preceding activity significantly influences the bout. It appeared to matter least in OPTIMUS, by the authors' own account.
Projection, not a result: the authors estimate that breaks every 25 minutes across 10 h of sitting could add about 1.5 h of muscle activity. No calculation is shown; the person-level gap gives about 39 min (3.87 × 10, our arithmetic) and comes from comparing people, not from changing anyone's sitting.
Mechanisms
Measured: thigh EMG presence during sitting by time in the bout, by person's habit, by what came before. Nothing downstream of it.
Authors' interpretation: falling muscle activity is a potential mechanism for the health associations of prolonged sitting, and more active interruptions may carry over; they call it a speculation and ask for further study.
Inference: lower activity in habitual long sitters may reflect who they are as much as what the sitting does: between-person contrast, with no adjustment printed for it, and the diabetes cohort lowest of all. Calf muscles were not measured. EMG drift, sweat and skin temperature are acknowledged sources of error, and over hours they could change a low-amplitude signal as a bout lengthens.
Practical Application
I train, then sit all day: does it matter? Unknown. In 84 general adults thigh activity faded across a long sit, a little less after harder activity. No health outcome, no athletes selected: it cannot say if training offsets anything.
What it adds for an athlete: only that a quiet thigh is what a long sit looks like. Muscle silence after hard training is not shown to be bad, and rest after a session is not what this study tested (inference).
Do not quote: "sitting over 25 minutes shuts muscles off" (the printed curve does not support the cut-off, and activity stays above zero), or "breaks add 5 minutes an hour" (no derivation printed; the overall contrasts are about 2 to 4, observational).
If asked what to do: Not from this paper, and no guide in the base sets it; plain practice, no evidence read: standing or walking a little in a long desk day is a cheap habit worth offering, not a training tool and not a recovery method, and it does not replace sleep, fuelling or the session.
Limitations
No health outcome. Muscle activity only; the title's metabolic health is inference from other studies.
Observational and cross-sectional in effect. About two days per person in most; no change in sitting was imposed. The authors' experimental support is a separate trial of theirs in overweight and obese men (not read).
Sample. 84 people, 28 removed first, three very different cohorts (one all women and overweight, one with diabetes, median days 1.5 to 8.0). None recruited as athletes, no measure of training.
Measurement. Different accelerometers and placements across cohorts, a hip-worn device in the largest; EMG summarised per second in one cohort and per minute in the others; 10 and 25 Hz devices. Authors acknowledge sensor shift, adiposity, sweat and temperature could bias EMG.
Reporting. The abstract gives 46% female and mean age 43.7; Table 1 gives 64.3% and 43.0 (its counts sum to 54 of 84). The fit vs its stated thresholds and the "5 minutes" as above. Reference numbers in the text do not match the printed list (for example, "ref 29" for the 10% muscle-activity study points to a fidgeting cohort), so cited support could not be traced by number. Supplements not read; data and code on request.
Funding and interests as printed: public research funds and one foundation (Urheiluopistosäätiö); no competing interests declared. Shorts came from Myontec and Suunto; no support from them is stated.
How this entry was checked and changed
Why this level
cross-sectional measurement study under the rubric's 100 for B+ (the n is 84 people; the 7684 are bouts within them), held below B because the outcome is a marker with no health outcome and the abstract's 25-minute floor cannot be rebuilt from the printed equation
Full text (checked 2026-11-02)
84 participants (EMG24 n = 55, BREAK n = 14, OPTIMUS n = 15) after removal for poor synchronisation (n = 26) and artefacts (n = 2); 7684 sitting bouts (full text)
Overall power-law fit 25.72 × min^−0.461, stated to reach 10%, 5% and 1% at 8.9, 12.2 and 25.7 minutes; power law BIC = 709, exponential BIC = 826 (full text)
Longer usual sitting bout (≥25 min) 6.55 mins/hour [5.25, 7.85] vs shorter (<25 min) 10.42 mins/hour [9.06, 11.77] (full text)
Preceding activity <3 min 7.76 mins/hour [6.53, 9.09] vs >10 min 9.87 [8.49, 11.37]; standing 6.69 [5.43, 8.08] vs above walking 9.84 [8.00, 11.86] (full text)
EMG activity defined as the proportion of periods above signal baseline (3 μV) during accelerometer-estimated sitting bouts (full text)
Table 1: females 54 (64.3%), age 43.0 (18.7); the abstract states 46% female and mean age 43.7 (18.7) (full text)
The authors declare no competing interests; the OPTIMUS study was funded by Academy of Finland #332731 (full text)