Does easier functional training cost fitness?
In 30 functional athletes over 6 weeks, no clear difference vs threshold-heavy work at about 40% less heart-rate load; the "polarized" arm was really pyramidal. Not equivalence, not HYROX.
Main caveat: Null with n = 30 and 6 weeks. The sample-size calculation targets an interaction, not equivalence; no margin, no interval of the difference.
Held et al. asked whether the polarised intensity pattern used in endurance sport would work for high-intensity functional training (HIFT).
Thirty trained HIFT athletes (15 women) were allocated by minimisation to 6 weeks of either a "polarized" plan (POL, n = 16: three of four weekly sessions held below a heart rate matching the first lactate threshold, the fourth above the second) or their usual style (TRAD, n = 14), same exercises and the same training time, with live heart-rate feedback and coaches at every session.
Measured by heart rate, POL spent 78.4%, 13.0% and 8.6% of its time below, between and above the thresholds and TRAD 26.1%, 35.9% and 37.9%, so POL's heart-rate training load was much lower (785 vs 1,273 au). No group × time interaction reached significance for squat, deadlift, overhead press or high pull 1RM, VO2peak, lactate threshold, peak power or the benchmark Jackie.
Both groups got faster at Jackie and gained in deadlift, overhead press and high pull; the time effect was not significant for squat, VO2peak, lactate threshold or peak power. POL reported fewer negative wellbeing notes.
Our reading: no clear difference in a small, short trial, with an interesting load gap; not evidence that polarised functional training is as good or better, and the arm was not polarised by the authors' own account.
Protocol
Design: randomised controlled parallel-group trial by the authors' description; allocation "via minimization" with gender, age, height, body mass and peak power output as strata. Sample-size calculation: n = 28 for an effect of ηp2 = 0.06, power 0.80, with 30 enrolled. No blinding and no trial registration reported.
Who: n = 30 (POL n = 16, 8 women; TRAD n = 14, 7 women). Age 25.3 ± 4.6 vs 28.2 ± 4.6 y, experience 2.0 ± 1.8 vs 2.6 ± 1.7 y, prior training 6.6 ± 1.8 vs 6.4 ± 1.8 h/week. At least three HIFT sessions a week; no group differences at baseline by t-test on these variables.
Training: 6 weeks, 4 sessions a week, same exercises in both arms. POL: below a target heart rate at the first lactate threshold in three sessions, above the second lactate threshold in the fourth. TRAD: usual HIFT. Movement tempo, load or power adjusted to keep heart rate in range. Separate dedicated classes for each arm in two different gyms, coaches at every session. Detailed workouts are in supplemental files (not read here).
Load: heart-rate time in three zones (below LT1, LT1-LT2, above LT2), eTRIMP weights 1, 2, 3. Wellbeing from an online platform: negative notes for exhaustion, strain or injury.
Tests before and after: VO2peak, LT and peak power on a Wahoo Kickr bike ergometer (incremental then ramp); 1RM estimated from a set to failure at 95% of presumed 1RM by the Lombardi formula for squat, deadlift, overhead press and high pull; Jackie for time (1,000 m row, 50 thrusters at 20/15 kg, 30 pull-ups).
Analysis: 2 × 2 repeated-measures ANCOVA with the baseline value as covariate; Fisher's exact test for wellbeing.
Funding and interests: the authors state no financial support and no commercial or financial conflict of interest. Polar and Wahoo devices were used as equipment; no funding from them is reported.
Key Results
| Measure | TRAD pre → post | POL pre → post | Group × time p (ηp2) | Time p (ηp2) |
|---|---|---|---|---|
| Jackie (s) | 671 ± 120 → 599 ± 99 | 741 ± 171 → 668 ± 153 | 0.986 (0.00) | 0.002 (0.01) |
| High pull (% body mass) | 77.5 ± 18.4 → 86.9 ± 19.2 | 69.7 ± 24.8 → 73.6 ± 22.8 | 0.094 (0.09) | 0.001 (0.11) |
| Squat (% body mass) | 118.5 ± 29.1 → 121.6 ± 28.4 | 107.4 ± 35.0 → 113.9 ± 33.0 | 0.265 (0.03) | 0.172 (0.02) |
| Overhead press (% body mass) | 51.6 ± 8.3 → 55.0 ± 8.2 | 53.3 ± 17.4 → 56.1 ± 16.0 | 0.760 (0.00) | 0.020 (0.00) |
| Deadlift (% body mass) | 153.0 ± 37.5 → 163.0 ± 33.9 | 140.1 ± 33.2 → 148.0 ± 36.2 | 0.546 (0.02) | 0.001 (0.02) |
| Lactate threshold (W) | 176 ± 22 → 176 ± 30 | 197 ± 29 → 200 ± 32 | 0.681 (0.01) | 0.961 (0.01) |
| Peak power (W) | 305 ± 51 → 308 ± 51 | 299 ± 42 → 306 ± 46 | 0.485 (0.02) | 0.621 (0.02) |
| VO2peak (mL/min/kg) | 48 ± 5 → 47 ± 6 | 45 ± 9 → 45 ± 9 | 0.328 (0.02) | 0.100 (0.01) |
Intensity-distribution check: total time 602 ± 20 (POL) vs 601 ± 60 min (TRAD), p = 0.938. Below LT1 472 ± 57 vs 157 ± 29 min; LT1-LT2 78 ± 41 vs 216 ± 71 min; above LT2 52 ± 19 vs 228 ± 70 min. Distribution 78.4/13.0/8.6% vs 26.1/35.9/37.9%.The authors: POL "was better characterized by a pyramidal training distribution", TRAD by a threshold one; individually, four TRAD athletes were dominated by high intensity and two POL athletes were polarised. They note heart-rate time-in-zone under-reports high-intensity work.
Load: eTRIMP 785 ± 71 vs 1,273 ± 126 au over 6 weeks (p ≤ 0.001, SMD = 4.55), "about 40% less".
Time in the plan: 602 min over 6 weeks is about 100 min a week and 25 min a session (our arithmetic), well under the 6.4-6.6 h/week prior training; the paper monitors "endurance/HIFT-related parts", so the figure may leave out other work. What else they did is not stated.
Wellbeing: poor wellbeing, overexertion or injury notes less likely in POL (Fisher p = 0.019; odds ratio 0.12, 95% CI 0.01-0.82). Self-reported, one wide interval; event counts are in a figure.
No between-group intervals: the paper gives p and ηp2 for the interaction, no confidence interval of the difference, so how large a loss is ruled out cannot be read from it.
Mechanisms
Measured: 40% lower heart-rate load with the same session count and clock time, no significant interaction on any outcome, and no significant time effect on VO2peak, lactate threshold, peak power or squat.
Authors' interpretation: high-intensity work drives adaptation but a large share risks inadequate recovery; a polarised/pyramidal plan might give similar gains with less recovery cost. Stress hormones and recovery were not measured here.
Inference: the eTRIMP gap is a heart-rate score. Loaded lifts and gymnastics raise heart rate poorly, and the authors cite work finding session RPE better than TRIMP for HIFT, so "about 40% less load" may not hold for the muscles. Our inference from that limit.
Inference: the aerobic markers did not move in either arm, so this trial cannot show that an easier mix keeps aerobic gains in functional athletes; it shows no difference where neither arm gained. The ride test may also be insensitive to adaptation in rowing and whole-body work.
Practical Application
Does easier functional training cost fitness? In 30 functional athletes over 6 weeks, no clear difference vs threshold-heavy work at about 40% less heart-rate load; the "polarized" arm was really pyramidal. Not equivalence, not HYROX.
Good for: answering an athlete who fears every mixed session must be hard. Over 6 weeks, trained functional athletes who kept most of their heart-rate time low did not clearly lose strength or benchmark speed (B+, a null, n = 30).
Not good for: saying polarised is as good or better, or for a long block. High pull moved in TRAD's favour (77.5 to 86.9 vs 69.7 to 73.6% of body mass, interaction p = 0.094, ηp2 = 0.09, moderate on the paper's scale), so a real cost in some outcomes is not excluded.
HYROX (transfer): not from this paper; the base's own practice: run 3-4 times a week, most of it easy, and treat SkiErg and rower as intervals near threshold, not maximal pulls (SYNTH-71). This trial adds only that lowering the intensity share showed no clear short-term price.
Limitations
Null with n = 30 and 6 weeks. The sample-size calculation targets an interaction, not equivalence; no margin, no interval of the difference. The title's "does not affect" and the conclusion's "not affected" say more than the design can.
Allocation and setting. Minimisation is reported as random assignment; each arm apparently trained in its own gym ("two separate HIFT gyms"; the wording is not explicit), so gym and coach effects may travel with the group. No blinding of testers reported.
Baseline gaps. POL was slower at Jackie (741 vs 671 s) and had a higher lactate threshold (197 vs 176 W); ANCOVA adjusts for this, in a wide-spread sample the authors call heterogeneous.
The label. By heart rate, POL was pyramidal and TRAD threshold, with individual patterns varying inside the arms. No arm tested a true polarised distribution, and the zones are heart-rate time, not session goals.
Test specificity. Aerobic tests on a bike, no running or rowing test; one benchmark workout; 1RMs estimated from sets to failure, not lifted.
Reporting. The abstract gives time effects of 0.01 ≤ ηp2, while Table 2 prints 0.00 for overhead press (p = 0.020). No adverse-event counts in the text.
Interests. None declared; no funding.
How this entry was checked and changed
Why this level
parallel RCT with n = 30 (the rubric's 20-39 band), read from the full text (CC BY). It sits at the low edge: a null, 6 weeks, no blinding reported, apparently each arm in its own gym, and the session descriptions are in a supplement not read here
Full text (checked 2026-10-30)
POL and TRAD revealed a training intensity distribution (LiT, ThT, and HiT) of 78.4%, 13.0%, and 8.6% and 26.1%, 35.9%, and 37.9% (full text)
The polarized training group was better characterized by a pyramidal training distribution, and the high-intensity training group by a threshold training distribution (full text)
POL (785 ± 71 au) vs TRAD (1,273 ± 126 au) eTRIMP; total training volume 602 ± 20 vs 601 ± 60 min (p = 0.938) (full text)
No group × time effects (p ≥ 0.094; ηp2 ≤ 0.09); Jackie 671 ± 120 to 599 ± 99 s in TRAD and 741 ± 171 to 668 ± 153 s in POL (full text)
VO2peak time effect p = 0.100, lactate threshold time effect p = 0.961, squat time effect p = 0.172 (full text)
Wellbeing: Fisher's exact test p-value = 0.019, odds ratio 0.12, 95% confidence interval 0.01-0.82 (full text)