REF-1398 · Beamish et al., 2025 · British Journal of Sports Medicine · Paper · PubMed · Free full text
Does pelvic-floor training after birth stop leaking?
In 7 trials it cut the odds of leaking by 37% (OR 0.63, 0.41-0.97; moderate certainty); no evidence it changed severity. Abdominal work narrowed a diastasis a little. No runners.
Main caveat: Heterogeneity. High on most pooled outcomes (I² 52-91%; anal incontinence 33%) and not explained by subgroups.
Beamish and colleagues reviewed whether exercise in the first year after childbirth prevents or treats pelvic-floor disorders and diastasis recti abdominis.
They searched eight databases to 12 January 2024 for studies of any design except case studies, with no language limit, in which exercise started in the first postpartum year (alone, or with a co-intervention such as biofeedback), against no exercise or a different exercise dose; outcomes were the risk and severity of urinary incontinence, anal incontinence, pelvic organ prolapse, sexual function and inter-rectus distance.
They included 65 studies (21,334 participants, 24 countries): 40 on pelvic-floor outcomes (23 RCTs, n = 3,298) and 28 on diastasis (14 RCTs, n = 626), plus 7 acute studies.
Pelvic-floor muscle training was the only exercise tested for pelvic-floor outcomes; it ran from once a week to three times a day, 5 to 60 min a session, from light to maximal contractions, started between the third trimester and 24 weeks after birth, and no programme explicitly used progressive overload.
Pelvic-floor training lowered the odds of urinary incontinence after the intervention (all 8 RCTs: OR 0.59, 0.39-0.89; training alone, 7 RCTs: OR 0.63, 0.41-0.97; moderate certainty) but showed no evidence of a difference in symptom severity (very low).
Anal incontinence odds were lower by a point estimate of 42% with a confidence interval that includes no effect (OR 0.58, 0.31-1.11; low). One RCT in women with an episiotomy or second-degree tear found 56% lower odds of prolapse (OR 0.44, 0.21-0.91); the two-trial pooled estimate showed no difference (OR 1.10, 0.15-8.01).
Abdominal muscle training reduced inter-rectus distance at rest by 0.52 cm more than control (5 RCTs; low) and during a head lift by 0.47 cm (2 RCTs; moderate in the text), from a starting gap of about 2.4 cm.
During single bouts, the gap narrowed with a head lift and widened with a transversus abdominis contraction, against the clinical belief that gentle transversus work closes a diastasis.
Our reading: the base had pelvic-floor training evidence only in mostly nulliparous athletes (REF-728); this adds trial evidence in the year after birth that such training lowers the chance of leaking, with modest certainty, nothing on how running or lifting affects the pelvic floor, and a small effect on a diastasis whose clinical meaning is unsettled.
Protocol
Design: systematic review and meta-analysis (PRISMA), protocol PROSPERO CRD42022359282 (registered 20 September 2022, updated 12 January 2024), part of the review series for the Canadian Society for Exercise Physiology 2025 Canadian Guideline for Physical Activity, Sedentary Behaviour and Sleep throughout the First Year Postpartum (REF-1394).
Search: MEDLINE, EMBASE, CINAHL, SPORTDiscus, EBM Reviews (Ovid), Scopus, Web of Science, ClinicalTrials.gov to 12 January 2024; reference lists and collaborator-nominated papers; no language or date limit.
Population: individuals in their first year after delivery (a study recruiting later was included only if the intervention started in the first year); interventions that began in pregnancy were included if most of the intervention was postpartum.
Intervention: any measured exercise or physical activity, acute or chronic, alone ("exercise-only") or with a co-intervention (biofeedback, electrical stimulation, abdominal binder, bracing, kinesiotaping); when a study had both, only the exercise-only arm was compared with control.
Outcomes: risk and severity of urinary incontinence (a critical guideline outcome), anal incontinence, prolapse and sexual function (important outcomes); inter-rectus distance (not a guideline outcome); measured from before to immediately after the intervention.
Included (n = 21,334 in 65 studies): pelvic-floor outcomes, 40 studies (n = 19,900): 23 RCTs (n = 3,298; 18 PFMT-only, 1 abdominal-training-only, 4 PFMT with co-interventions), 11 non-randomised (n = 1,456), 5 observational (n = 15,146); diastasis, 28 studies (n = 1,588): 14 RCTs (n = 626), 6 non-randomised (n = 530), 1 observational (n = 84); 7 acute studies.
Analysis: RevMan 5.4 random-effects models (Mantel-Haenszel for odds ratios from post-intervention events; inverse-variance mean or standardised mean difference of change scores, Hedges' g); a ≥25% reduction in odds called clinically significant; subgroups by timing of start (≤12 vs >12 weeks postpartum), delivery mode, exercise type, supervision; GRADE certainty assessed by one reviewer and checked by a second (amended from two independent raters for feasibility); risk of bias with JBI checklists.
Funding and interests: S-MR is funded by the Université du Québec à Trois-Rivières research chair in physical activity and maternal and neonatal health; MHD by a Christenson Professorship in Active Healthy Living; MJG by an Alberta Innovates—Health Solutions Summer Studentship. Competing interests: none declared. No device maker, federation or data platform is named.
Key Results
| Outcome (RCTs, after the intervention) | Studies, n | Effect (95% CI) | Certainty |
|---|---|---|---|
| Urinary incontinence, odds — PFMT alone vs control | 7 RCTs, 1,930 | OR 0.63 (0.41 to 0.97), I² 72% — 37% lower odds | moderate |
| Urinary incontinence, odds — all PFMT (incl. 1 with a co-intervention) | 8 RCTs, 2,005 | OR 0.59 (0.39 to 0.89), I² 71% — 41% lower odds | moderate (downgraded for indirectness) |
| Urinary incontinence, symptom severity (change) | 6 RCTs, 294 | SMD −0.04 (−0.49 to 0.42), I² 65% — no evidence of a difference | very low |
| Anal incontinence, odds | 3 RCTs, 694 | OR 0.58 (0.31 to 1.11), I² 33% — the interval includes no effect | low |
| Sexual function (FSFI change) | 2 RCTs, 118 | SMD 0.31 (−0.96 to 1.58), I² 91% — no evidence of a difference | very low |
| Pelvic organ prolapse, odds — PFMT alone (episiotomy or second-degree tear only) | 1 RCT, 123 | OR 0.44 (0.21 to 0.91) — 56% lower odds | moderate |
| Pelvic organ prolapse, odds — pooled (PFMT with a co-intervention + PFMT) | 2 RCTs, 197 | OR 1.10 (0.15 to 8.01), I² 84% — no evidence of a difference | low |
| Inter-rectus distance at rest — abdominal training vs control | 5 RCTs, 173 | MD −0.52 cm (−0.99 to −0.05), I² 91% | low |
| Inter-rectus distance at rest — abdominal training alone | 4 RCTs, 133 | MD −0.35 cm (−0.59 to −0.10), I² 52% | moderate |
| Inter-rectus distance during a head lift | 2 RCTs, 84 | MD −0.47 cm (−0.92 to −0.02), I² 63% | 'moderate' in the text; 'low' in the abstract |
Before the interventions: 62% (PFMT groups) and 63% (controls) had urinary incontinence (5 RCTs, n = 1,912; OR 1.18, 0.55 to 2.57); inter-rectus distance at rest was 2.39 ± 0.62 cm and 2.38 ± 0.77 cm, which the authors call comparable to a general adult mean of 2.2 cm.
Subgroups for urinary incontinence odds: no difference by when training started after birth (≤12 vs >12 weeks, p = 0.71), delivery mode (p = 0.74) or supervision (p = 0.38). For severity, one supervised group-based RCT reduced it more than individual home programmes (SMD −1.22, −2.21 to −0.23; p = 0.02 for the subgroup difference) — a single trial.
Acute bouts (7 studies): inter-rectus distance decreased during a head lift and increased during a transversus abdominis contraction, measured 2 cm above and below the umbilicus.
Not found: no RCT of general aerobic or resistance exercise for any pelvic-floor or diastasis outcome; no pelvic-floor study tested exercise without pelvic-floor training; non-randomised and observational data "partially support" the RCTs and showed no association of exercise with the odds of diastasis or prolapse severity (supplement, not read).
Overall certainty: high to very low across outcomes; downgraded mostly for imprecision (small samples) and indirectness (co-interventions such as education, biofeedback, electrical stimulation).
Mechanisms
Measured: after pelvic-floor training, fewer women reported urinary incontinence; after abdominal training, the gap between the rectus muscles was slightly smaller; in single bouts, a head lift narrowed and a transversus contraction widened it.
Authors' interpretation: pelvic-floor training is a relatively cheap, accessible first-line option; the trial programmes resembled an "intensive" programme by International Continence Society standards but never applied progressive overload, so future programmes should load the pelvic floor progressively through functional movements; early, symptom-guided return to loading during tissue healing might help, by analogy with tendon and ankle injuries (untested); exercise in the first 6-8 weeks may speed the natural narrowing of the inter-rectus distance (a hypothesis).
Authors' interpretation: a wide inter-rectus distance may not be a functional problem if the linea alba can stiffen under load; no clinically meaningful distance or diagnostic cut-off is agreed, so closing the gap is a questionable treatment target.
Inference: the lower odds of leaking with no change in severity fit a programme that moves some women from leaking to not leaking without the measures being sensitive to small changes in those who still leak; the review does not test this.
Practical Application
Does pelvic-floor training after birth stop leaking? In 7 trials it cut the odds of leaking by 37% (OR 0.63, 0.41-0.97; moderate certainty); no evidence it changed severity. Abdominal work narrowed a diastasis a little. No runners.
Pelvic-floor training belongs in the first year after birth, before and alongside impact (A-, transfer): training lowered the odds of urinary incontinence, and the effect did not differ by whether it began before or after 12 weeks postpartum (p = 0.71) or by delivery mode (p = 0.74), tests of subgroup difference; how to train it (dose, progression) is not settled by these trials, so the programme belongs to a pelvic-health physiotherapist.Not from this paper; the base's own practice: leaking, heaviness or a bulge sends the runner to a pelvic-health physiotherapist and back a step in the return plan (SYNTH-36 postpartum line; REF-1397 screen).
A diastasis alone is not shown to be a reason to avoid core or running work (our inference from this review and [REF-1397]; no trial tested running with a diastasis): abdominal training reduced the gap by about 0.35-0.52 cm on a starting gap near 2.4 cm (low to moderate certainty), and the authors question whether the gap is a meaningful target at all; the Delphi in REF-1397 did not agree on inter-recti distance as a return-to-running milestone.A transversus contraction widened the gap acutely, so the belief that gentle transversus work closes a diastasis is not supported.
Prolapse: the 56% lower odds come from one trial in women after an episiotomy or second-degree tear; the pooled estimate showed no difference. Do not promise that pelvic-floor training prevents prolapse.
What it does not show: anything about running, cycling, lifting or other general exercise and the pelvic floor after birth (no such trial was found); athlete outcomes; long-term effects beyond the end of the interventions.
Limitations
Heterogeneity. High on most pooled outcomes (I² 52-91%; anal incontinence 33%) and not explained by subgroups; populations ranged from all postpartum women to only incontinent women, only those with an anal sphincter injury, or women with incontinence and back pain; programmes and "standard care" differed widely.
Few trials for several outcomes. Prolapse benefit from one trial in a specific population; head-lift distance from two; sexual function from two; anal incontinence from three with a wide interval.
Certainty grading. One reviewer graded certainty and a second checked it (a protocol amendment). The head-lift result is 'moderate' in the results text and grouped as 'low' in the abstract.
No athletes and no running. No trial measured leakage during running or sport, or tested aerobic or resistance exercise.
Inter-rectus distance. No agreed diagnostic cut-off or clinically meaningful change; measurement sites and tools varied; the baseline gaps were within normal adult values, so the trials say little about large diastases (the authors ask for studies above 5 cm).
Supplement not read. Non-randomised and observational results and the GRADE tables are in the online supplement.
Funding and interests: university research chair, professorship and studentship funding; competing interests none declared. Four authors (Beamish, Davenport, Ali, Ruchat) are also authors of the guideline it informs (REF-1394).
How this entry was checked and changed
Why this level
systematic review and meta-analysis of RCTs, held below A because heterogeneity is high on most pooled outcomes (I² 72% for leakage odds, 91% for inter-rectus distance at rest; 33% for anal incontinence), most outcomes rest on few, small trials (prolapse benefit: one RCT; head-lift distance: two), certainty ranges from moderate to very low, and publication bias could not be assessed (fewer than 10 studies per analysis)
Full text (checked 2026-10-28)
65 studies (n=21 334 participants) from 24 countries were included (full text)
Pelvic floor muscle training reduced the odds of urinary incontinence by 37% (seven randomised controlled trials (RCTs), n=1930; OR 0.63, 95% CI 0.41 to 0.97, I^2 72%) and pelvic organ prolapse by 56% (one RCT, n=123; OR 0.44, 95% CI 0.21 to 0.91) (full text)
Post-intervention pooled estimate for all PFMT RCTs: eight RCTs, n=2005; OR 0.59, 95% CI 0.39 to 0.89, I^2 71%; certainty 'moderate', downgraded for indirectness (full text)
No difference in change in urinary incontinence symptom severity (six RCTs, n=294; SMD −0.04, 95% CI −0.49 to 0.42, I^2 65%), 'very low' certainty (full text)
Pelvic organ prolapse, two RCTs (n=197): OR 1.10, 95% CI 0.15 to 8.01, I^2 84%; the single PFMT trial included only individuals who had an episiotomy or second-degree perineal tear (full text)
Inter-rectus distance at rest: MD –0.52, 95% CI −0.99 to −0.05, I^2 91% (five RCTs, n=173), 'low' certainty; abdominal muscle training-only, MD –0.35, 95% CI −0.59 to −0.10, I^2 52% (four RCTs, n=133) (full text)
Head lift task: two RCTs, n=84; MD −0.47, 95% CI −0.92 to −0.02, I^2 63%; certainty rated 'moderate' (full text)
Anal incontinence: three RCTs, n=694; OR 0.58, 95% CI 0.31 to 1.11, I^2 33%, 'low' certainty (full text)
Subgroup analyses for urinary incontinence: timing of intervention initiation in postpartum, p=0.71; mode of delivery, p=0.74; level of supervision, p=0.38 (full text)
RCTs that implemented general exercise interventions (eg, whole-body aerobic and/or resistance exercise) were not found (full text)
No intervention explicitly implemented the training principle of progressive overload (full text)
Funding: S-MR, Université du Québec à Trois-Rivières research chair; MHD, Christenson Professorship in Active Healthy Living; MJG, Alberta Innovates—Health Solutions Summer Studentship; competing interests: None declared (full text)