REF-1520 · Skaug et al., 2024 · British Journal of Sports Medicine · Paper · PubMed · Free full text

Will pelvic floor training help me leak less in functional fitness?

RCT, 47 women, 16 weeks at home: leak score −1.4 vs controls (CI −2.6 to −0.2), below the important-difference mark; 64% vs 8% felt better; none cured.

Main caveat: Effect is below the pre-stated important difference. Power was set for 2.5; the result was −1.4, with an upper reach beyond 1.58.
fairly strong evidenceMeasured on recreational to trained athletes

Skaug, Engh and Bø recruited women who habitually did CrossFit or functional fitness and reported stress urinary incontinence. Fifty-one were randomised, 26 to pelvic floor muscle training and 25 to continued usual training; four in the training group dropped out, none in controls.

The programme was 3 sets of 8–12 near-maximal contractions held for approximately 6–8 s, daily at home for 16 weeks, with weekly phone or SMS follow-up.

At 16 weeks the leakage score (ICIQ-UI-SF, 0–21) changed by −1.3 (95% CI −2.6 to −0.03) with training and 0.1 (−0.6 to 0.9) in controls, a between-group difference of −1.4 (−2.6 to −0.2); sensitivity imputations kept the result significant (−1.2 to −1.8). Asked directly, 64% vs 8% reported improved symptoms (relative risk 7.96, 2.03 to 31.19).

Pelvic floor strength, endurance and resting pressure showed no significant difference between groups, with wide intervals whose upper limits for strength and endurance pass the minimal detectable change, as the authors note; anal incontinence and prolapse symptoms did not differ either. Adherence averaged 70% (SD 23) and no adverse effects were reported.

The authors recommend the training as first-line. Our reading: the direction is credible and the cost is about 10 min a day, but the average change in the score is small, the score is self-reported by women who knew their group, and no woman was cured. It fits SYNTH-95's pelvic floor row and adds one trial in functional fitness.

It does not tell women whether their racing or lifting caused the leak.

Protocol

Design: assessor-blinded, parallel RCT with computer-generated block randomisation and sealed opaque envelopes; assessor and statistician blinded, participants and intervention physiotherapist not. Data collected May 2022 to April 2023 in Norway. n = 51 randomised (26 and 25); n = 47 analysed (22 and 25). Two others were excluded at baseline for being unable to perform a correct contraction.
Who: women ≥18, at least 6 months and 3 times a week of functional fitness, self-reported stress incontinence including leaking during exercise, ICIQ frequency and amount sum ≥3. Age 33.5 (SD 8.1); parous 49%; severity mild 49%, moderate 45%, severe 6%. Excluded: pregnancy, surgery for these conditions, childbirth within 12 months, injury in the last 6 months affecting exercise.
Training: taught by vaginal palpation; 3 sets of 8–12 maximum contractions of approximately 6–8 s daily, lying, kneeling, sitting or standing, progressed with position and 3–4 fast contractions; about 10 min a day for 16 weeks; weekly phone or SMS reminders; logged in the AthleteMonitoring app; SMS reminder if unlogged by 8PM. Unsupervised at home.
Control: kept functional fitness as usual, no pelvic floor training or education during the study; given the programme afterwards.
Primary outcome: change in total ICIQ-UI-SF score. Minimal important differences cited: 2.5 within group and 1.58 between treatments.
Secondary: global impression of improvement (7-point scale); vaginal manometry (resting pressure, strength, endurance); anal incontinence and prolapse symptoms; self-efficacy at 1 month.
Analysis: ANCOVA on week 16 with baseline as covariate; full analysis set of available participants; sensitivity with imputed scores for the four lost. Power: 24 per group (n=48), allowing 20% dropout, for a change of 2.5 at 80%, SD 2.8; aim 50.
Funding and interests: no specific grant; no competing interests declared. Author KB is named as founder of the original programme and taught it to the physiotherapists.

Key Results

OutcomeTraining (n=22)ControlBetween groups (95% CI)
ICIQ-UI-SF, change (primary)−1.3 (−2.6 to −0.03)0.1 (−0.6 to 0.9), n=25−1.4 (−2.6 to −0.2)
Reported improvement64%8%p<0.001, RR 7.96 (2.03 to 31.19)
Pelvic floor strength, cmH2O, change3.9 (SD 7.9)−1.0 (SD 8.3), n=243.8 (−0.8 to 8.4)
Pelvic floor endurance, cmH2O/sec, change28.0 (SD 82.4)−19.8 (SD 77.2), n=2439.28 (−1.5 to 80.1)
Resting pressure, cmH2O, change−2.1 (SD 6.6)−4.4 (SD 5.6), n=241.3 (−1.4 to 4.0)
Anal incontinence, prolapse symptomsno differenceno differencep 0.263-0.890
Sensitivity (n=26 vs 25, imputed ICIQ): −1.6 (−2.7 to −0.5) with −2 imputed; −1.3 (−2.4 to −0.2) with no change; −1.2 (−2.3 to −0.1) with +1.
Spread: SD of change in strength 7.9 (training) vs 8.3 (control), endurance 82.4 vs 77.2. The training arm's spread is about the same as the control's (lower for strength, slightly higher for endurance), so the authors' "various responses" is not proof of responders and non-responders.
Adherence: 70% (SD 23) of sessions; 59% adhered to more than 70%. No adverse effects. No cures.

Mechanisms

Measured: symptoms improved with no significant between-group change in strength, endurance or resting pressure.
Authors' interpretation: the improvement may have come from morphology (bladder neck and bowel position, narrowing of the levator hiatus) or automatic muscle response to pressure, unmeasured here. Less supervision and only 4 months may explain the small strength change, against 6-month supervised trials.
Authors' interpretation: lower baseline severity than the group used to set the 2.5 MID, and half of the sample being nulliparous, may have limited the measurable change. A MID of 1.33 has been proposed for moderate severity.
Inference: a self-reported score and global impression in unblinded women, against no-attention controls, are open to expectation. That the global result (64% vs 8%) is far larger than the score change fits that possibility, as well as the authors' reading that the score under-detects benefit.

Practical Application

Will pelvic floor training help me leak less in functional fitness? RCT, 47 women, 16 weeks at home: leak score −1.4 vs controls (CI −2.6 to −0.2), below the important-difference mark; 64% vs 8% felt better; none cured.
When she asks: leaking during jumps, running or lifting is common in this group, and this is a low-risk 10 minutes a day with a realistic expectation: some improvement, not a cure, by 16 weeks. The authors recommend it as first-line and suggest supervision and longer training (more than 6 months) may do more; that suggestion is not tested here. Correct contraction was taught by a physiotherapist by palpation before starting, and two women of 53 could not do it.
Do not change her training on this paper. Controls kept training as usual and training load was not recorded; it does not show that fewer jumps or a lighter barbell helps. Not from this paper; the base's own practice: keep training, refer to a pelvic-health physiotherapist (SYNTH-95).

Limitations

Effect is below the pre-stated important difference. Power was set for 2.5; the result was −1.4, with an upper reach beyond 1.58. The authors say so and argue that severity changes the threshold.
Self-report, unblinded participants, no-attention control. The assessor's blinding protects the manometry, not the primary outcome.
Unequal dropout. Four of 26 lost in the training group, none of 25 in control; analysis of available cases, with imputations that hold up. The target of 48 already allowed for 20% dropout, so the 47 analysed do not fall short of it; the limit is that it was set for a 2.5-point effect.
Training not tracked. Functional fitness load and exercise types were not monitored, tampons and other devices were not asked about, and the questionnaire has no question on specific exercises such as rope jumping (authors' limitations).
Selected sample. Recruited on social media with symptoms; 85% with a degree; Norwegian; one trial. Only 16 weeks and no follow-up afterwards.
Numbers that do not add up. Table 4 prints "15 (64)" for improvement while the rows above give 14 (0+3+11) and 14/22 is 64% (our arithmetic). The −3 imputation row prints −1.8 with CI −2.3 to −0.6, which is lopsided around its estimate, unlike the others.
Not HYROX, and no performance outcome.
How this entry was checked and changed
Why this level

RCT at or above 40 (51 randomised, 47 analysed), concealed allocation and blinded assessor; the sample met its power target only for a 2.5-point effect, the primary outcome is self-report in unblinded participants, and the observed effect is below the minimal important difference

Full text (checked 2026-10-30)
Fifty-three women recruited; two excluded; 26 participants randomised to the PFMT group and 25 to the control group; four dropped out of the PFMT group and none in the control group (full text)
Mean change in ICIQ-UI-SF −1.3 (95% CI: −2.6 to −0.03) in the PFMT group and 0.1 (95% CI: −0.6 to 0.9) in control; mean difference −1.4 (95% CI: −2.6 to −0.2) (full text)
A change of 2.5 has been identified as the MID and 1.58 as between-treatment MID (full text)
64% versus 8% reported improved symptoms (p<0.001, RR: 7.96, 95% CI 2.03 to 31.19) (full text)
PFM strength between-group 3.8 (−0.8 to 8.4) cmH2O; endurance 39.28 (−1.5 to 80.1) cmH2O/sec (full text)
None of the participants reported a complete cure for SUI (full text)
The programme took about 10 min per day; mean 70% (SD: 23) of sessions completed; no adverse effects were reported (full text)
Funding: no specific grant declared; competing interests: none declared (full text)