REF-1571 · dos Santos et al., 2022 · JAMA Internal Medicine · Paper · PubMed

Can I fit my week into 1-2 days?

Against 3+ sessions at the same total, death HR 1.08 (0.97-1.20): a difference not shown, sameness not proven. General adults, self-report; cohort, not proof.

Main caveat: Abstract only. Group sizes, covariates, sensitivity analyses and the handling of early deaths were not read.
moderate evidenceMeasured on non-athletes and recreational people

The authors asked whether the recommended weekly activity carries the same mortality association when concentrated into fewer days. Active people split into weekend warriors and regularly active; the inactive were the reference. Weekend warriors and regularly active people were each compared with the inactive, then with each other.

The authors conclude that people meeting the guideline may experience the same benefit either way. Our reading: that wording goes further than the data, because the interval still allows a meaningfully higher hazard and no equivalence test appears in the abstract.

Protocol

Exposure, by self-report: inactive (<150 min/week of moderate to vigorous activity); active (≥150 min/week of moderate or ≥75 min/week of vigorous). Active split into weekend warrior (1-2 sessions/week) and regularly active (≥3 sessions/week). The abstract describes no repeat measurement.
Who: mean age 41.4 (SD 15.2) years; 192 432 women (50.8% as printed); 209 432 non-Hispanic White (67.8% as printed).
Outcomes: all-cause, cardiovascular and cancer mortality; 3.6 million person-years; 21 898 deaths, 4130 from cardiovascular disease and 6034 from cancer.
Adjustment: for total activity in the pattern contrast; other covariates are not in the abstract.
Funding and interests: not in what was read.

Key Results

ComparisonOutcomeHR (95% CI)
Weekend warrior vs inactiveall-cause0.92 (0.83-1.02)
Regularly active vs inactiveall-cause0.85 (0.83-0.88)
Weekend warrior vs regularly active, same totalall-cause1.08 (0.97-1.20)
Weekend warrior vs regularly active, same totalcardiovascular1.14 (0.85-1.53)
Weekend warrior vs regularly active, same totalcancer1.07 (0.87-1.31)
The authors' conclusion is firmer than this row. The authors say weekend warriors had lower mortality than the inactive, yet that interval includes 1.00 for all-cause death. The abstract says cause-specific findings against the inactive were similar but prints no figures.
Reporting. 192 432 of 350 978 is 54.8% and 209 432 is 59.7% (our arithmetic), against 50.8% and 67.8% as printed. These may be survey-weighted figures; the abstract does not say.

Mechanisms

Measured: nothing physiological; only self-reported activity and death records.
Authors' interpretation: the total amount, more than how it is spread, may be what matters.
Inference: people already ill are likelier to be inactive at survey, and healthier people are likelier to be active, so the contrast with the inactive may overstate any effect of activity. The inactive group also includes anyone under 150 min/week, not only the sedentary, which would pull the contrast the other way.

Practical Application

Can I fit my week into 1-2 days? Against 3+ sessions at the same total, death HR 1.08 (0.97-1.20): a difference not shown, sameness not proven. General adults, self-report; cohort, not proof.
Good for: someone training for health whose week only allows one or two sessions: this cohort did not show the pattern to be worse for mortality (HR 1.08, 0.97-1.20), though the interval allows a modest excess.
Not good for: saying the patterns are equal. The all-cause interval reaches a 20% higher hazard and the cardiovascular one 53% (our arithmetic, from the upper limits).
Not good for: performance or fitness. Session frequency against VO2max has been randomised elsewhere, in other samples (REF-1329, REF-1292).
Not from this paper; the base's own practice: in runners, a single run much longer than the past month's longest is its own load question (SYNTH-72).

Limitations

Abstract only. Group sizes, covariates, sensitivity analyses and the handling of early deaths were not read. Two published letters and the authors' reply were not read either.
Self-report of frequency, duration and intensity, with no repeat measure described.
Reverse causation and healthy-user bias cannot be excluded by this design.
Non-significant is not equivalent. See Key Results.
Young mean age (41.4) and a median 10.4 years of follow-up leave deaths concentrated in the older part of the sample (inference).
No fitness, injury or performance outcome.
How this entry was checked and changed
Why this level

(from the abstract) — large observational cohort, the rubric's ceiling for this design however many people it holds. Group sizes and the adjustment set were not read, and no full text is held