REF-1437 · Kwok et al., 2022 · International Journal of Environmental Research and Public Health · Paper · PubMed · Free full text

Does deep-water running keep my running fitness?

A review of 11 small trials found it about as good as land training for VO2max (effect sizes 0.06-0.10 in two runner trials, 3-6 weeks). Low quality; no injured runners.

Main caveat: No pooling; results are single-study effect sizes.
moderate evidenceMeasured on non-athletes to highly trained athletes

Kwok and colleagues (Hong Kong Polytechnic University, St Vincent's Hospital Melbourne) systematically reviewed the effects of deep-water running on cardiorespiratory fitness, physical function and quality of life in healthy and clinical populations.

Six databases were searched to February 2022 (PROSPERO CRD42020154988); interventional studies in adults with at least half the intervention as deep-water running qualified, and standardised mean differences were calculated where means and SDs were reported.

Eleven trials (7 randomised) with 287 participants were included: four in physically active or trained adults, three in inactive or obese women, two in healthy or community-dwelling older adults and one in low back pain. Programmes ran 30 to 70 min, 2 to 5 times a week, for 3 to 18 weeks, almost all with a flotation device.

Against land training, deep-water running had similar effects on VO2max — SMD 0.06-0.10 in the two athletic trials, −0.39 to 0.28 in the untrained healthy trials, −0.83 in back-pain patients. Against no exercise, one trial found a large effect (SMD −1.28) and one did not.

For running outcomes, time to fatigue (SMD 0.82, n = 12) and 2-mile time (SMD −0.41 to −0.27 against running and cycling) did not differ. No meta-analysis was done because 7 trials scored as poor quality and the studies were heterogeneous.

The authors conclude that deep-water running appears comparable to land-based training for improving cardiorespiratory fitness, and that small, low-quality studies limit further conclusions.

Our reading: the only synthesis of deep-water running in the base, useful mainly because it lets a guide cite a review instead of two 1990s trials — but for athletes it adds no data beyond REF-922 and REF-1428 plus one 3-week trial, gives no dose, and contains nothing on heart rate in water (REF-1433) or on injured runners.

Protocol

Design: systematic review guided by PRISMA; registered on PROSPERO (CRD42020154988).
Search: SPORTDiscus, MEDLINE, CINAHL, AMED, Embase, Cochrane Library to February 2022, plus reference and citation tracking; two reviewers screened and extracted independently, a third resolved disagreement.
Inclusion: randomised or non-randomised interventional studies in adults; deep-water running at least 50% of the intervention (or with passive treatments); outcomes of cardiorespiratory fitness, physical function related to walking or running, or quality of life. Excluded: other aquatic exercise (shallow-water running, underwater treadmill), under-18s, non-experimental designs.
Sample: n = 11 trials, 287 participants, 10 to 60 per study; 7 randomised controlled trials, 2 longitudinal, 2 quasi-experimental (Table 2).
Athletic trials: McKenzie 1991, 12 competitive runners (6 water, 6 land), 3 weeks, 30-min sessions 5 times a week, no flotation device; Wilber 1996, 16 trained male distance runners (8 water, 8 treadmill), 6 weeks, 30-60 min 5 times a week, flotation device; Eyestone 1993, 32 men (10 water running, 11 cycling, 11 road running), 6 weeks, 20-30 min, flotation device in a diving pool.
Quality: selected Downs and Black items (reporting, external validity, bias, confounding, power); 7 trials scored poor (≤14).
Analysis: SMD with 95% CI per study (RevMan); no pooling.
Funding and interests: no external funding; the authors declare no conflict of interest.

Key Results

ComparisonOutcomeStudiesSMD
Deep-water vs land training, athleticVO2maxMcKenzie 1991, Wilber 19960.06 to 0.10 (similar)
Deep-water vs land training, untrained healthyVO2maxDavidson 2000, Eyestone 1993−0.39 to 0.28
Deep-water vs land training, low back painVO2maxKanitz 2019−0.83
Deep-water vs no exerciseVO2maxBroman 2006; Michaud 1995−1.28 (significant); −0.83 (not significant)
Deep-water vs land runningtreadmill time to fatigueMcKenzie 19910.82 (similar)
Deep-water vs running and cycling2-mile run timeEyestone 1993−0.41 to −0.27 (similar)
Deep-water vs land trainingquality of lifeKanitz 2019; Assis 20060.03 to 0.05
Heart rate: three studies measured peak heart rate but reported no means and SDs, so no effect size could be calculated.
Quality of the three athletic trials (Table 1): none concealed allocation or blinded assessment, none reported a power calculation; Wilber 1996 was not randomised (quasi-experimental), and Eyestone 1993 did not account for losses to follow-up.
A classification that does not add up: the Results list Eyestone 1993 among the four studies in "physically active or trained adults", the VO2max section among the "untrained healthy population"; Table 2 gives its entry criterion as a 1.5-mile run in under 10'45.

Mechanisms

Measured (per study): similar VO2max change with deep-water and land training in athletic and untrained groups; larger gains against no exercise in sedentary older people.
Authors' interpretation: buoyancy removes vertical ground-reaction forces and lowers joint loading, so high-intensity work is possible with less orthopaedic stress; immersion restricts chest expansion and raises the work of breathing; larger effects in sedentary older people reflect a low starting fitness.
Inference: the comparison with land training in runners tests holding, not building, fitness over 3-6 weeks; with 6-11 people per arm a moderate difference could be missed.

Practical Application

Does deep-water running keep my running fitness? A review of 11 small trials found it about as good as land training for VO2max (effect sizes 0.06-0.10 in two runner trials, 3-6 weeks). Low quality; no injured runners.
How to dose it: this review gives no dose (the authors say so).The tested athletic programmes ran 3-5 times a week for 20-60 min (McKenzie and Wilber 5 times a week for 30-60 min; Eyestone 20-30 min, 3-5 times a week), Wilber's with 90-100% and 70-75% VO2max sessions (REF-922); the base's own practice (not from this review) is to set intensity by effort and structure with heart-rate targets about 10 bpm below land (REF-1433, our rounding); recreational runners steering by feel undershoot (REF-1438).
What it does not show: anything in injured runners, beyond 6 weeks in athletes, in women runners, on running economy beyond REF-922, or on readiness of bone and tendon to run again (PHYS-37, PHYS-38).

Limitations

No pooling; results are single-study effect sizes.
Low quality: 7 of 11 trials scored poor on the Downs and Black items; most lacked concealed allocation and blinded assessment.
Mixed populations; only three small trials in runners, all men, 3-6 weeks.
Classification slip (Eyestone 1993 counted as trained in one place, untrained in another) and a design count that lists all 11 references for the 7 randomised trials; Table 2 resolves it (7 R, 2 L, 2 Q).
No heart-rate synthesis and no dose recommendation.
Supplementary tables (search strategy, eligibility, quality items) not read.
Funding and interests: none received; no conflicts declared.
How this entry was checked and changed
Why this level

a registered systematic review (the rubric's A- line is an SR/MA with serious heterogeneity or mostly small trials) held at B+: nothing was pooled, 7 of the 11 trials scored poor quality (Downs and Black ≤14), and the athlete question rests on three small trials in healthy runners (McKenzie 1991, n = 12; Wilber 1996, n = 16, in the base as REF-922; Eyestone 1993, n = 32, REF-1428)

Full text (checked 2026-10-29)
A total of 11 clinical trials (7 randomized controlled trials) with a total of 287 participants were included (full text)
Among athletic populations, VO2 max ranged from SMD 0.06–0.10 [26,27]; for the untrained healthy population SMD −0.39–0.28 [19,28] (full text)
Meta-analyses were not completed due to Downs and Black poor-quality score of (≤14) [10,19,20,25,26,27,28] (full text)
Eyestone et al. (1993) found a similar effect of DWR to both land-running and cycling for two-mile run time (SMD = −0.41 to −0.27) (full text)
The duration of DWR interventions varied from 30 to 70 min, 2 to 5 sessions per week, lasting between 3 to 18 weeks (full text)
Three studies evaluated the changes in HR peak [10,25,27], but did not report means and SD; therefore, SMDs could not be calculated (full text)
Wilber et al (1996): 16 (16/0), aerobically trained male distance runners; Mckenzie et al. (1991): 12 (12/0), competitive runners (full text)
This research received no external funding; the authors declare no conflict of interest (full text)