The long-term effectiveness of antenatal pelvic floor muscle training: eight-year follow up of a randomised controlled trial


Dr WI Agur, Urogynaecology Research Unit, Derriford Hospital, Plymouth, PL6 8DH, UK. Email


Objective  To determine the long-term effectiveness of antenatal pelvic floor muscle training (PFMT) on stress urinary incontinence (SUI).

Design  Eight-year follow up of a randomised controlled trial (RCT).

Setting  Acute NHS Teaching Trust.

Population  Participants in an RCT of antenatal PFMT 8 years previously.

Method  Participants were asked about the presence of SUI, impact on quality of life, frequency of performance of PFMT and details of subsequent deliveries.

Main outcome measure  The prevalence of SUI at 8 years.

Results  One hundred and sixty-four (71%) of the original 230 women responded. The significant improvement in postnatal SUI originally shown in the PFMT group compared with controls (19.2 versus 32.7%, P = 0.02) at 3 months was not evident 8 years later (35.4 versus 38.8%, P = 0.7). On direct questioning, 68.4% of the study group claimed that they still performed PFMT as taught during the study, with 38.0% of them performing this twice or more per week. There was no difference in outcome between those who performed PFMT twice or more per week compared with those performing PFMT less frequently. There were no differences in quality-of-life domains between the study and the control groups at 8 years.

Conclusion  The initially beneficial effect of supervised antenatal PFMT on SUI did not continue for a long term despite the majority claiming to still perform PFMT. These findings are in keeping with those of other studies and raise concerns about the long-term efficacy of PFMT. Strategies to improve compliance with PFMT are required.


Postnatal stress urinary incontinence (SUI) is a common problem affecting up to 34% of women,1 while 3–5% have faecal incontinence.2 The risk is higher, and incontinence is likely to persist in the long term after onset during pregnancy.3,4 Evidence suggests that vaginal delivery can contribute to damage to the pelvic floor through muscle and fascial injury as well as disrupting the nerve supply.5,6

Antenatal pelvic floor muscle training (PFMT) has been shown to reduce the incidence of postnatal SUI in the short term.7–9 However, in their 6-year follow-up study of postnatal PFMT, Glazener et al.10 showed a lack of long-term benefit with three-quarters of women still incontinent 6 years later.

The National Institute of Clinical Excellence (NICE)11 recommends PFMT for all women in a first pregnancy for prevention of SUI based upon data from two randomised controlled trials (RCTs).8,9 In the study of 268 women9 with antenatal bladder neck mobility (previously shown to be a risk factor for developing postnatal SUI12), it was found that fewer women in the intervention group had urinary incontinence compared with controls (19.2 versus 32.7%, P = 0.02) 3 months after delivery.9 We report in this study an 8-year follow-up study of these women to assess whether antenatal PFMT has a durable long-term effect and to determine the extent of women’s compliance with PFMT.


Participants were recruited in 1998–1999 during their 20-week visit to the antenatal clinic of an acute NHS Trust Teaching Hospital. All were primigravidae with bladder neck mobility12 on perineal ultrasound.13 Details of methodology, intervention and the outcome at 3 months following delivery have previously been described.9 Participants were randomised (by computer-generated pseudorandom numbers) to either monthly PFMT with supervision by a physiotherapist or a control group where they received only verbal advice and/or a leaflet on PFMT from their midwives. No record of the control group’s performance of PFMT was made, for example Oxford score or perineometry as in the study group.

Eight years later, the participants were contacted by letter inviting them to take part in this follow-up study. Letters were followed up by a telephone call to explain the study objectives and to determine whether or not they wished to take part. Following informed verbal consent, participants were asked to give details of their childbirth(s) over the past 8 years, the frequency of performing PFMT and the presence of SUI in the past 4–8 weeks. Duration of the phone call (from P.S.) averaged 10 minutes during which the symptom questionnaire used in the previous study9 and the King’s Health Questionnaire were completed. Inclusion criteria were participation in the initial study, being in good general health as well as willingness to complete the questionnaire by telephone.

The prevalence of SUI at 8 years was assessed using the symptom questionnaire used in the original study.9 A severity question was used to ascertain whether leakage occurred once a week, twice or more per week or daily. Participants with no leakage in the previous 4 weeks were regarded as ‘continent’.

The original trial sample size had been calculated as 128 test subjects and 128 controls (based on the results of the previous study12) to obtain a 5% two-sided significance level and an 80% power to detect the difference in postpartum urinary incontinence among participants in the two groups, allowing for a 25% attrition rate.9

Two members of the research team (P.S. and W.I.A.) entered the 8-year follow-up data onto database software. Statistical analysis studied the relationship between subsequent childbirth(s), PFMT performance and urinary symptoms comparing all with the results of the original study (M.W.).

Nominal categorical data were tested using exact forms of the chi-square test. For 2×2 tables with a control (layer) variable, the Cochrane–Mantel–Haenszel test was used. Mann–Whitney nonparametric tests were used to compare groups for measurements on an ordered, categorical scale. P < 0.05 were considered significant. Logistic regression model was used for multivariate analysis as SUI at 8 years was common (greater than one-third), and adjusted relative risks (RR) and their confidence intervals were estimated using log-binominal generalised linear models.14


A total of 164 (71%) of the 230 participants with available postnatal data from the original study responded 8 years later (Figure 1). No obvious difference was found between participants who responded at the original study and those who responded 8 years later regarding age (27.6 versus 28.1 years), mode of index delivery (caesarean section rates: 14.5 versus 15.2%), body mass index (24.5 versus 24.5) or the presence of urinary incontinence (25.7 versus 26.2%).

Figure 1.

Study profile. BMN, bladder neck mobility.

The number of nonresponders was higher from the original treatment group compared with the original control group (34.2 versus 22.7%, P = 0.004). Nonresponders were younger than responders, but there were no other significant differences (Table 1).

Table 1.  Comparison between responders and nonresponders regarding demographic data and SUI at 3 months
 RespondersNonrespondersP value
Mean age in years (SD)28.0 (5.6)25.8 (5.8)0.002
Mean body mass index (SD)24.5 (4.2)23.7 (4.0)0.13
Caesarean section rate at index delivery (%)14.914.90.9
Postpartum SUI at 3 months (%)

The mean age at the time of the original study was 27.6 years (range: 16–47 years). Eight years later, 114 (69.5%) participants had at least one further delivery, while parity was unchanged (i.e. para 1) in the remaining 50 (30.5%). More women in the control group went on to have further deliveries compared with the treatment group (80.0 versus 58.2%, P = 0.004).

The significant improvements in postnatal urinary incontinence seen in the PFMT group compared with controls at 3 months in the original trial (19.2 versus 32.7%, P = 0.02) were not evident 8 years later (35.4 versus 38.8%, P = 0.75). The severity of urinary incontinence showed little difference between the PFMT and the control groups (Table 2). There was no difference between the groups on any of the eight domains of the King’s Health Questionnaire.

Table 2.  SUI severity 8 years following the index delivery
SUI severityIntervention (n = 79), n (%)Control (n = 85), n (%)Significance
  1. NS, not significant.

Absent51 (64.6)52 (61.2)NS
Mild (≤1 per week)10 (12.7)12 (14.1)
Moderate (2–6 per week)12 (15.2)15 (17.6)
Severe (≥7 per week)6 (7.6)6 (7.1)
Total28 (35.5)33 (38.8)

There was a small number of spontaneous remissions in the control group: 12 women who had SUI at 3 months were now dry at 8 years (14.1%). The majority of those who were continent at 8 years were also dry at 3 months: in the control group, 56 women were dry at 3 months and 40 of them were dry at 8 years (71%) and in the study group, 65 were dry at 3 months and 47 of them (72%) were dry at 8 years.

The frequency of performing PFMT was recorded from direct questioning and graded as ‘frequent’ (2–6 per week) or ‘infrequent’ (≤1 per week). In the study group at 8 years, 68.4% reported that they still performed PFMT and 38.0% of them claimed to be undertaking this twice or more per week. Although participants were not asked about the nature of their daily activities or using the pretimed contraction (the ‘knack’15), there was no difference in incontinence ‘rates’ between those performing PFMT twice or more per week and those performing PFMT less frequently.

Table 3 and Figure 2 show the relationship between subsequent deliveries and the presence of SUI in the two groups. In the PFMT group, 46 women had further pregnancies compared with 68 women in the control group. Only one-third of those who had further deliveries had SUI compared with almost half of those whose parity was unchanged. There was a trend to less SUI in the PFMT group at 8 years but only in those who had further pregnancies compared with those who did not (26 versus 37%, not significant).

Figure 2.

Effect of subsequent deliveries on prevalence of SUI 8 years following participation in the antenatal PFMT trial.

Table 3.  Relationship between reporting SUI at 8 years and further deliveries
 Intervention (n = 79)Control (n = 85)Total (%)Significance
  1. NS, not significant.

Women without subsequent birth(s)
Number331750 (30.5) 
SUI present (%)16 (48)8 (47)24 (48)NS
Women with subsequent birth(s)
Number4668114 (69.5) 
SUI present (%)12 (26)25 (37)37 (32.5)NS

When assessing mode of delivery and 8-year outcomes, of those who had a caesarean section (CS) in the first pregnancy (either in labour or elective; n=25), 3 (25%) had SUI at 8 years in the control group compared with 2 (15%) in the study group. Those who did not have CS in first pregnancy had a slightly higher (but nonsignificant) incidence of SUI at 8 years (41% of women in the control group compared with 39% in PFMT group) (Figure 3).

Figure 3.

Effect of mode of index delivery 8 years ago on current prevalence of SUI in participants.

When instrumental deliveries are assessed (for the index or subsequent pregnancies), of 16 women in the control group, 25% had SUI at 8 years compared with 31% (13 women) in the PFMT group (nonsignificant). This might suggest that PFMT is not protective following instrumental delivery. For noninstrumental deliveries, the figures are 38% for the control group and 34% for the PFMT group. Figure 3 details the effect of the mode of index delivery on the current prevalence of SUI.

Univariate analysis of the factors that may affect the risk of SUI at 8 years after their first birth revealed four to be significant or of significance: age, mode of index delivery, whether there were further deliveries and the presence of SUI at 3-month postpartum to the index delivery.

When adjusted for age alone, caesarean section showed a significant protective effect (RR = 0.42, P = 0.03) compared with vaginal delivery, as did having further deliveries (RR = 0.65, P = 0.025 compared with no further deliveries). No interactions were found between these two factors. The strongest association was with SUI at 3-month postpartum (RR = 1.98, P < 0.001). The multivariate analysis failed when we attempted to adjust this factor for age. Excluding age, the adjusted RR when the other three factors are modelled together are shown in Table 4. Only SUI at 3-month postpartum was a significant risk factor in the multivariate analysis.

Table 4.  Adjusted RR for SUI 8 years in a multivariate analysis
FactorAdjusted RR* (95% CI)P value
  • *

    Adjusted for all other variables in the model.

Further deliveries
One or more0.72 (0.51–1.01)0.058
Mode of index delivery
Normal delivery1 
Assisted0.65 (0.38–1.14)0.131
Caesarean section0.49 (0.23–1.06)0.069
SUI at 3-month postpartum
Present1.98 (1.39–2.82)<0.001


This is the first long-term follow-up study of the effectiveness of antenatal PFMT in prevention of SUI; a good response rate was achieved at 8 years (71% of those who completed the original trial and 61% of those originally randomised). As there were more nonresponders in the treatment group, this has the potential to bias the results. It is also possible that some nonresponders might have improved with PFMT, and there is no evidence that they had undergone continence surgery. While antenatal PFMT has been shown to reduce the incidence of postnatal SUI in the short term,7–9 the data suggest that the improvement was not maintained 8 years later. It should be noted, however, that the attrition rate of the current study (39% of those originally randomised) is higher than that used to calculate the sample size of the original study (25%), which inevitably reduces the power of this current 8-year follow-up study. Nevertheless, the results are in keeping with those of a postnatal PFMT study, which showed that three-quarters of the women still had urinary incontinence 6 years following the index delivery.10 It has also been shown that the initially marked difference in urinary symptoms shown between women groups 6 months after starting an intensive PFMT programme is no longer maintained 15 years later.16 The previous evidence suggests that the difference in the number of participants performing PFMT (at 1 year) had disappeared 6 years later10 and that only 28% of women still perform PFMT in the long term.16 Our data support these findings.

Possible reasons for loss of benefit might be poor compliance in the study group (only 38.0% claimed to be training twice or more per week) and that women in the control group were doing regular PFMT, which in some women might have been effective; the so-called ‘Hawthorne effect’ of an RCT (also described as ‘avis effect’ by Bøet al.17). It is possible that women in the control group might have been given verbal advice on PFMT in a subsequent pregnancy, but this is unlikely to have been supervised training with a physiotherapist or a specialist nurse. It is not possible to confirm whether the study group who had been individually taught PFMT 8 years earlier (or the control group) were still performing PFMT correctly at 8 years. It seems reasonable to assume, however, that the study group should still be aware of the correct technique compared with controls.

As there is no information on how many participants in the control group received verbal advice and/or the relevant leaflet 8 years ago, we cannot make a firm conclusion that supervised PFMT confers no extra benefit over simply giving verbal advice/information leaflet. The loss of the favourable outcome of PFMT 8 years later is, in our view, because of lack of compliance rather than an effect of a certain method of administering the training. The aim of the current study was not to compare the outcome of the two methods, and the results do not appear to confirm or deny the superiority of either method over the other after 8 years.

The mode of delivery showed no significant difference in the prevalence of SUI between participants who had instrumental deliveries and those who delivered spontaneously or by caesarean section. This is in keeping with the findings of previous studies18,19 where the benefit from caesarean section was found to be only transient. In one study, caesarean delivery seemed to protect against the development of postnatal SUI, but 3 months after delivery, the statistically significant influence of the mode of delivery had disappeared.18 Although the association between SUI and childbirth variables (including instrumental deliveries) was found to be strong in bivariate analysis in another study, the effects of these variables were insignificant in multivariate analysis.19 The main determinant of the presence of SUI at 8 years seems to be the presence of SUI at 3 months, and this is consistent with the findings of previous studies.10,20

This study presents results of the longest follow up of a cohort of participants who undertook antenatal PFMT with a satisfactory 71% response rate and attempts to look at compliance and the effects of subsequent pregnancies.

It lends further support to the view that antenatal bladder neck mobility is a risk factor for postnatal and longer term SUI.12 The prevalence of SUI seen in the control group (38.8%; Table 2) is higher than the published prevalence rates for primiparous women.21

Despite the disappointing long-term results using antenatal PFMT and those of postnatal PFMT,10 nonetheless, the former has been recommended by the 3rd International Consultation on Incontinence (ICI)22 and NICE11 as a form of prevention of postnatal SUI in primigravidae. Further research into motivation techniques is required to improve compliance with PFMT (e.g. a national training programme starting in primary care).


The initially beneficial treatment effect of supervised antenatal PFMT on SUI did not continue long term. These findings are in keeping with those of other authors and raise questions about the long-term efficacy of PFMT in prevention of SUI. Strategies to improve compliance with PFMT are required.

Details of ethics approval

The procedures of the study received ethics approval from the regional ethics committee.


Local NHS Trust.