Does functional training build athletes' endurance?
Across 13 RCTs in mostly martial-arts athletes aged 10-24.5, strength and power rose; field endurance did not differ clearly (g 0.798, −0.352 to 1.949). No VO2max; not hybrid.
Main caveat: Numbers that do not add up. Text total 530 with a mean of 35.33, while Table 3 sums to 478 (the abstract's figure) and 530/35.33 gives 15 trials, not 13 (our…
Wang et al. pooled randomised trials of high-intensity functional training (HIFT) in healthy athletes. The final 13 trials hold 478 athletes (the abstract and our sum of Table 3; the text says 530), most of them wrestlers, judoka, boxers, taekwondo and other martial-arts athletes, plus one volleyball and one aerobic-gymnastics trial.
In practice "HIFT" meant CrossFit in ten trials, CrossFit combined with rope skipping and other drills in one, a comparison of two CrossFit formats in one, and a programme called HIFT in one; at 2 to 4 sessions a week for 4 weeks to 10 months.
Upper-body strength (g = 0.414), lower-body strength (1.051), power (0.499), flexibility (3.167) and sport-specific performance (3.351) improved significantly; endurance (0.798) and agility (0.247) did not.
Our reading: useful as a controlled pool in athletes, with direction believable for strength and power, and the sizes of flexibility and sport-specific scores not believable. The non-significant endurance pool holds four small field-test studies, with a point estimate favouring HIFT and a very wide interval: it is an absence of evidence and not evidence of no effect.
It uses no VO2max.
Protocol
Search: PubMed, Scopus, Web of Science, EBSCOhost and Cochrane Library to 1 July 2023, plus Google Scholar and reference lists; English, peer-reviewed.
Eligibility (PICOS): healthy athletes of any sex or age; HIFT; two or more controlled groups; a physical-fitness or sport-specific outcome; randomised controlled trials.
Trials: n = 478 athletes in 13 trials of 20 to 60 each. 11 martial arts (432 athletes), 1 volleyball, 1 aerobic gymnastics. 8 male, 2 female, 3 not reporting sex. Ages 10 to 24.5.
Comparators (Table 3): usual or classical sport training, strength training, circular or interval training, or no additional training; one trial compares classic CrossFit with CrossFit AMRAP, with no non-HIFT arm, and still counts in the power pool.
Dose: 2 to 4 sessions a week; 4 to 12 weeks in most, 5, 6, 9 and 10 months in four. Sessions and adherence often not reported.
Quality: PEDro 5 or 6 in all (the authors call 5 or above high quality). None had allocation concealment, nor blinding of participants, therapists or assessors, and none intention-to-treat (Table 2).
Analysis: Comprehensive Meta-Analysis v3, random effects, Hedges' g, I2, Egger's test, leave-one-out. No subgroup analysis by sex, age or training variables.
Funding and interests: no specific funding; no competing interests.
Key Results
| Outcome | Trials / N | g [95% CI] | p | I2 | Egger p |
|---|---|---|---|---|---|
| Upper-body strength | 6 trials on strength; N = 209 | 0.414 [0.14 to 0.683] | 0.003 | 0.872% | 0.170 |
| Lower-body strength | N = 177 | 1.051 [0.126 to 1.977] | 0.026 | 0.00% | 0.160 |
| Power | 7; n = 226 | 0.499 [0.239 to 0.758] | < 0.001 | 0.00% | 0.084 |
| Endurance | 4; N = 122 | 0.798 [−0.352 to 1.949] | 0.174 | 20.245% | 0.274 |
| Agility | 3; N = 100 | 0.247 [−0.443 to 0.537, as printed] | 0.482 | 13.315% | 0.601 |
| Flexibility | 3; N = 104 | 3.167 [0.477 to 5.857] | 0.021 | 87.021% | 0.029 |
| Sport-specific performance | 7; N = 279 | 3.351 [1.780 to 4.922] | 0.000 | 11.554% | 0.005 |
What "endurance" was: the four trials' Table 3 outcomes are the Cooper test (12-min run), 30-s burpees, a 10 × 20 m shuttle and 1-min rope skipping, not VO2max; Table 3 lists no VO2max outcome in any trial (our reading; the pool's exact tests sit in a figure). So this pool does not measure the same thing as REF-905's "8% to 15%" VO2max.
What "sport-specific" was: competition score, rank position, attack strength, competition endurance, movement scores, unblinded in every trial and some of them judged (our reading). The abstract says "small to large"; on the paper's own scale 3.2-3.4 is very large.
Heterogeneity: I2 low to none on six pools and high (87.021%) on flexibility. Low I2 sits oddly with intervals this wide for these N (lower-body strength 0.126 to 1.977 on N = 177, endurance on N = 122), so the low values do not reassure (our inference).
The correction (PLoS ONE 2024): a notice of republication. On 25 January 2024 the article was republished because Xinzhi Wang had been displayed as the corresponding author instead of Kim Geok Soh. The notice changes no data or conclusion. The version read here is the Europe PMC one, carrying the correction link.
Mechanisms
Measured: none. The review pools performance tests; it holds no physiological measures of adaptation.
Authors' interpretation: HIFT recruits more muscle, stimulates neuroendocrine responses and improves lactate tolerance and cardiorespiratory function. The neuroendocrine claim cites Glassman's beginner's guide in the CrossFit Journal, and none of it was tested in the pool.
Inference: since many comparators are usual sport practice, adding or swapping in general strength and conditioning for combat athletes who do little of it can explain large gains in strength, power and flexibility, not something unique to the CrossFit format.
Practical Application
Does functional training build athletes' endurance? Across 13 RCTs in mostly martial-arts athletes aged 10-24.5, strength and power rose; field endurance did not differ clearly (g 0.798, −0.352 to 1.949). No VO2max; not hybrid.
Good for: the honest note that in this controlled pool of athletes (an earlier CrossFit pool of four short studies, not limited to athletes: REF-1492), CrossFit-type work raised strength and power, while its endurance benefit was not established.
Not good for: "HIFT does not build the engine" (four small studies, wide interval, no VO2max) or "HIFT builds it by 8% to 15%" (REF-905, a scoping figure in other populations). Quote both with who they measured. The sport-specific and flexibility sizes (above 3) should not be quoted at all.
The authors' dose (at least two sessions a week for at least four weeks) is their reading, with no dose analysis behind it.
HYROX (transfer): not from this paper; the base's own practice: the engine is built with running (SYNTH-71); this review gives no reason to expect mixed sessions alone to replace it, and no proof they cannot.
Limitations
Numbers that do not add up. Text total 530 with a mean of 35.33, while Table 3 sums to 478 (the abstract's figure) and 530/35.33 gives 15 trials, not 13 (our arithmetic). The flow text says 36 articles reached full-text evaluation, then lists exclusions adding to 587 (336, 192, 38, 21); 602 fits the arithmetic (both our sums).The agility interval −0.443 to 0.537 is not centred on 0.247. Endurance is 0.798 in Results, 0.831 in the Discussion. The endurance sample is called "100 female martial arts athletes" while Table 3 shows the Ambroży and Avetisyan trials as male.
Discussion against result. The Discussion calls HIFT "an effective alternative to traditional endurance training" although the pool was not significant; the Conclusion states it had no significant impact.
Bias. Significant Egger tests on flexibility and sport-specific performance. No blinding, concealment or intention-to-treat in any trial; the PEDro label "high" means a score of 5.
Clustering. Mostly combat sports; three trials share a first author (Osipov 2017, 2019, 2022).
Young and mixed. Children of about 11 pooled with adults of 24; durations 4 weeks to 10 months; no subgroups; comparators differ.
Selection. English only; PROSPERO number as printed (shorter than the usual format; not looked up).
How this entry was checked and changed
Why this level
A- by design (a meta-analysis of small trials), held one step lower: implausibly large pooled effects with significant Egger tests on exactly those two outcomes, no trial with allocation concealment or any blinding, and reporting errors in the sample size, the flow chart, one confidence interval and the endurance estimate. Read from the full text (CC BY)
Full text (checked 2026-10-30)
The total number of subjects collected in the 13 studies was 530, ranging from 20 to 60, with an average sample size of 35.33; martial arts athletes 432 (full text)
HIFT has a moderate but no significantly effect on endurance (ES = 0.798; 95% CI = -0.352-1.949; p = 0.174; Egger test p = 0.274; N = 122), I2 = 20.245% (full text)
Sport-specific performance ES = 3.351; 95% CI = 1.780-4.922; Egger's test p = 0.005; N = 279; I2 = 11.554% (full text)
Flexibility ES = 3.167; 95% CI = 0.477-5.857; p = 0.021; Egger test p = 0.029; I2 = 87.021% (full text)
Agility ES = 0.247; 95% CI = -0.443-0.537; p = 0.482 (full text)
PEDro scale scores ranged from 5 to 6; none met the criteria for allocation concealment or blinding of participants, therapists and assessors (full text)
Notice of Republication: republished on January 25, 2024, to correct an error in the author list (full text)