A pelvic-floor programme done in pregnancy, from 28 weeks to delivery, lowered new stress incontinence at 6 weeks after birth from 13.9% to 8.7% (risk difference 5.17 percentage points, 95% CI 0.36-10.03) in a multicentre randomised trial of 764 first-time mothers without leaking at entry (382 per arm); the gap held at 3, 6 and 12 months, but participants were not blinded and part of the diagnosis was self-reported
Main caveat: Authors' own: participants (and physiotherapists) could not be blinded, which may bias the control data.
Gao and colleagues (Peking University People's Hospital and eight other centres) randomised 764 first-time pregnant women without stress urinary incontinence (SUI) or prolapse to PEFLOW or usual care.
PEFLOW combined daily pelvic-floor muscle training (slow 6-8 s holds and fast 1-s contractions, timed to breathing) with two weekly sessions of about 30 min of global postural exercise at four intensity levels, adjusted every 2 weeks by perceived exertion; it ran from 28 weeks' gestation to delivery and stopped there.
SUI was diagnosed by a positive cough or Valsalva stress test and/or a reported leak with coughing, sneezing or exercise. At 6 weeks after birth, the primary outcome, 8.7% of the exercise group and 13.9% of controls had SUI in the complete-case analysis, a 5.17-point difference; the two imputed analyses agreed (5.18 and 6.54 points).
Under worst-case imputation the exercise group also had less SUI at 3, 6 and 12 months after birth, but not at 37 weeks of pregnancy, and more women in the exercise group had good-to-strong pelvic-floor contractions (Modified Oxford Scale 4 or more) at 6 weeks, a difference that was no longer significant from 3 months. Incidence rose again in both groups by 12 months.
The authors report no programme-related adverse event and conclude that PEFLOW in pregnancy prevents postpartum SUI and should probably continue after birth.
Participants
1258 primiparous women recruited, 494 excluded, 764 randomised 1:1: exercise 382, control 382; 367 and 360 returned at 6 weeks after birth.
Eligibility: age 20-40, singleton pregnancy under 16 weeks; excluded: severe complications the programme might worsen, a history of SUI or pelvic organ prolapse, cervical insufficiency, recurrent miscarriage or induced labour.
Baseline (exercise vs control): median age 30 vs 29 years; median pre-pregnancy BMI 20.90 vs 21.00; 93% with a bachelor's degree or higher; 97% low occupational activity; MOS 4 or more in 50.8% vs 49.5%.
Training or sport participation: not reported.
Births (Table 4, exercise vs control): vaginal 59.7% vs 60.0%; episiotomy 40.6% vs 44.1% and laceration 65.2% vs 68.0% of vaginal births.
Protocol
Exercise group: physiotherapist-guided start, then home practice from 28 weeks to delivery in four blocks (28-30, 31-33, 34-36, 37-40 weeks) with an app and video. One pelvic-floor set = 5 moderate contractions held 6-8 s with 6-8 s rest, then 8 rounds of 5 maximal 1-s contractions with 10 s rest; done daily.Global postural exercise: 4 postures per level, about 30 min, twice a week, starting at level 2 and moved up or down on an RPE threshold of 15. Checks every 2 weeks at clinic visits; the protocol's weekly on-site supervised session became online video sessions during COVID-19.
Control group: usual care, obstetric examination every 2 weeks from 28 weeks; privately asked whether they did similar exercise.
Outcomes: primary, SUI incidence at 6 weeks after birth; secondary, SUI at 37 weeks' gestation and 3, 6 and 12 months after birth, and MOS by digital vaginal examination at the same points. SUI = positive stress test in the lithotomy position and/or reported leaking with cough, sneeze or exercise (following the ICIQ-UI SF).
Sample size: 293 per group for 90% power to detect 25.7% vs 15%, raised to at least 367 per group for 20% dropout.
Analysis: intention to treat by multiple imputation, worst-case imputation and complete cases for the primary outcome; worst-case imputation and generalised estimating equations for secondary outcomes; Wilson intervals with continuity correction.
Funding: two grants from China's National Key Technology R&D Program; funder had no role; no conflicts of interest reported.
Key Results
SUI incidence (exercise vs control; risk difference in percentage points, 95% CI):
| Time point and analysis | Exercise | Control | Difference | P |
|---|---|---|---|---|
| 6 weeks after birth, complete case (primary) | 32/367 (8.7%) | 50/360 (13.9%) | 5.17 (0.36-10.03) | .03 |
| 6 weeks, multiple imputation | 34/382 (9.0%) | 54/382 (14.2%) | 5.18 (0.66-9.71) | .03 |
| 6 weeks, worst-case imputation | 47/382 (12.3%) | 72/382 (18.8%) | 6.54 (1.21-11.86) | .01 |
| 37 weeks' gestation, worst case | 84/382 (22.0%) | 98/382 (25.7%) | 3.66 (printed −2.56 to −9.86) | .23 |
| 3 months after birth, worst case | 42/382 (11.0%) | 68/382 (17.8%) | 6.81 (1.64-11.97) | .007 |
| 6 months after birth, worst case | 50/382 (13.1%) | 72/382 (18.8%) | 5.76 (0.37-11.13) | .03 |
| 12 months after birth, worst case | 74/382 (19.4%) | 114/382 (29.8%) | 10.47 (4.18-16.66) | <.001 |
Pelvic-floor strength, MOS 4 or more (worst case):
| Time point | Exercise | Control | Difference (95% CI) | P |
|---|---|---|---|---|
| 37 weeks' gestation | 50.8% | 49.7% | 1.05 (−6.20 to 8.28) | .78 |
| 6 weeks after birth | 17.8% | 7.9% | 9.95 (5.05-14.87) | <.001 |
| 3 months | 25.7% | 22.3% | 3.4 (−2.84 to 9.61) | .27 |
| 6 months | 33.0% | 27.2% | 5.76 (−0.93 to 12.38) | .08 |
| 12 months | 33.0% | 27.5% | 5.5 (−1.20 to 12.13) | .10 |
Within the exercise group (not randomised): women whose exercise level reached 80% or more of the required duration or intensity had less SUI than those at 50% to under 80% at 37 weeks (7.8% vs 16.8%), 6 months (2.7% vs 11.9%) and 12 months (6.9% vs 16.9%); differences at 6 weeks (8.6% vs 11.0%) and 3 months (2.7% vs 7.6%) were not significant.
Safety (exercise vs control, no tests reported): premature rupture of membranes 17.6% vs 23.5%; hypertensive disorders of pregnancy 5.9% vs 4.1%; fetal distress 0% vs 0.8%; postpartum haemorrhage 0.3% vs 3.8%; no programme-related adverse event.
Practical Application
Pelvic-floor training done through late pregnancy cut new leaking after birth in first-time mothers who did not leak before: about 9 in 100 leaked at 6 weeks against about 14 in 100 without it (5.17 percentage points fewer, 95% CI 0.36-10.03), with gaps of 5.76-10.47 points at 3, 6 and 12 months under worst-case counting and none at 37 weeks of pregnancy. This is prevention started in pregnancy; it says nothing directly about treating leaking that is already there after birth.
How it bears on [SYNTH-95]: it supports the guide's "pelvic floor from pregnancy on", which the guide grounds in the Canadian guideline's full text (REF-568) and in this randomised trial.It qualifies, and does not contest, the rule not to promise that training stops leaking: REF-1398 and REF-1709 pooled training after birth, measured as odds or risk ratios, whereas this trial started in pregnancy and reports risk differences, so the numbers are not comparable; and even here about 1 in 11 trained women leaked at 6 weeks (complete cases) and about 1 in 5 at 12 months (worst-case counting).
More training went with less leaking, in a non-randomised comparison. Women who reached at least 80% of the required duration or intensity leaked least, a comparison within the exercise group that is not randomised; leaking rose again in both groups by 12 months after the programme had stopped at delivery. The authors' suggestion to continue it after birth was not tested here.
Inference, not tested here: for a pregnant athlete, a daily pelvic-floor set of slow holds and fast contractions alongside normal training is a low-cost addition; whether it prevents leaking with running or jumping after birth was not measured separately.
Not treatment advice: whether to start pelvic-floor training in pregnancy, and how, is for the obstetric provider and a pelvic-health physiotherapist to decide.
Limitations
Authors' own: participants (and physiotherapists) could not be blinded, which may bias the control data; postpartum continuation of the programme was not part of the protocol.
The diagnosis was a stress test and/or a reported leak, so part of it rests on unblinded participants' answers; how many diagnoses came from each route is not reported.
The primary 6-week difference has a lower limit of 0.36 points; the sample size assumed 25.7% incidence in controls, while 13.9% was seen.
The text says all secondary SUI differences were significant, but the 37-week difference was not (P = .23), and its interval is printed with a sign error; the 37-week pelvic-floor counts (194 and 190 with MOS 4 or more) are identical to the baseline counts in Table 1, and the Methods list an assessment at 28 weeks where the rest of the paper uses 37 weeks.
Secondary outcomes used worst-case imputation only; the adherence comparison is not randomised; safety differences were not tested.
On-site supervised sessions were replaced by online sessions during COVID-19; how well control women avoided similar exercise is described only by private interviews.
The registration date is not given in the article; the protocol paper is cited from 2021, after recruitment began in August 2020 (the registry entry was not read).
First-time mothers in Chinese hospitals, highly educated and mostly in sedentary jobs; no athletes.
Publicly funded (China's National Key Technology R&D Program); no industry funding or conflicts reported.
How this entry was checked and changed
Why this level
a registered trial with 382 per arm, read in full, held below A because the full text does not give the registration date (so pre-registration is not confirmed), participants were unblinded while the diagnosis could rest on their own answer to "when does urine leak?", and in-person supervision was replaced by video calls during COVID-19