Pelvic floor muscle training works in athletes: significantly less leakage and measurably stronger contraction
The condition is trainable, not something to be worked around
Key Findings
Systematic review (PubMed, SportDiscus, Web of Science) of pelvic floor muscle training (PFMT) in female athletes or recreationally active women, for treatment or prevention of pelvic floor dysfunction; the stated aim concerned young, nulliparous female athletes. Random-effects meta-analyses could be performed on 6 of the included articles; pooled participant numbers are not given in the abstract.
| Outcome | Effect |
|---|---|
| Amount of urine leakage | significant reduction — SMD −1.13 (95% CI −1.84 to −0.41), p < 0.01 |
| Maximum voluntary contraction of the pelvic floor | significant increase — SMD 0.60 (95% CI 0.11 to 1.08), p = 0.02 |
| Vaginal resting pressure | no effect — SMD −0.05 (95% CI −1.27 to 1.17), p = 0.93 |
Authors' conclusion: PFMT increases pelvic floor muscle strength and reduces urine leakage in female athletes, and is a tool for both prevention and treatment.
The review's own background framing (cited, not its own result): athletes are almost 3 times more likely to experience urinary incontinence than sedentary women, with prevalence ranging from ~10.9% in low-impact sport such as cycling to ~80% in high-impact sport such as trampoline gymnastics.
Practical Application
Across 6 meta-analysed studies of female athletes and recreationally active women, pelvic floor muscle training reduced the amount of urine leakage (SMD −1.13, 95% CI −1.84 to −0.41) and increased maximal pelvic floor contraction (SMD 0.60, CI 0.11 to 1.08), with no effect on vaginal resting pressure. Leaking during effort is treatable — say so to the athlete.
Inference, not tested: the leakage effect was larger than the strength effect, so improvement is better judged on symptoms than on a strength number.
Impact level is a major risk modifier in the review's background figures (10.9% prevalence in cycling vs 80% in trampoline gymnastics). Not from this paper; the base's own practice: when an athlete moves toward higher-impact work — plyometrics, trail descents, track — that is a moment to ask (REF-727).
Not from this paper; the base's own practice: treat it like any other strength quality — it needs a programme and time, not a cue during a session.
⚠️ Refer out for the programme itself. Not from this paper; the base's own practice: a coach cannot verify a correct pelvic floor contraction, so route the athlete to a pelvic-health physiotherapist and build training around what they prescribe.
Limitations
Only 6 articles entered the meta-analyses, with wide confidence intervals — the leakage effect (−1.84 to −0.41) is convincingly non-zero but poorly bounded in size.
Inference, not tested: protocols (duration, supervision, dose, adherence) likely differ between trials; the abstract does not say which drives the effect.
The null result on vaginal resting pressure has a very wide interval (−1.27 to 1.17) — that is an uninformative comparison rather than evidence of no effect.
Mixed populations of athletes and recreationally active women; the abstract reports no long-term follow-up on whether gains persist when the programme stops.
No performance outcome. This is symptom and function evidence; whether treating incontinence improves training quality or performance is plausible but untested here.
How this entry was checked and changed
Why this level
regraded 2026-09-27 from A: meta-analyses of only 6 small, heterogeneous studies with wide confidence intervals
Content correction 2026-09-27
Level A -> A-: meta-analyses of only 6 studies with wide confidence intervals (rubric: SR/MA of mostly small trials).
Removed "a substantial fraction of women perform it incorrectly on verbal instruction alone" and "the trial-level evidence rests on supervised programmes" (not in the abstract); referral and programming advice labelled as the base's own practice.
Lead rewritten as the pooled result; the 10.9%/80% prevalence range marked as the review's background, not its result; symptom-vs-strength point labelled as inference; "mostly short interventions" removed.
Byline co-authors were wrong (Fernández-Ruiz, Roldán-Ruiz) -> Rodríguez-Longobardo, López-Torres, Guadalupe-Grau & Gómez-Ruano per the DOI record.