Urinary incontinence includes stress incontinence, urge incontinence and mixed incontinence. The term ‘stress incontinence’ can be used either to describe a symptom or as a medical diagnosis. When describing a symptom it refers to the involuntary leakage of urine, usually on exertion or on sneezing or coughing (Abrams et al 2002). As a diagnosis it refers to involuntary loss of urine when the intravesical pressure exceeds that of the urethra, with no simultaneous detrusor contraction. The International Continence Society refer to the latter as ‘genuine stress incontinence’, but this diagnosis should only be made following urodynamic investigation (Cardozo & Khullar 1995). The definition of stress incontinence used in this chapter, unless otherwise specified, refers to a symptom rather than a diagnosis. Urge incontinence is the complaint of involuntary leakage accompanied by or immediately preceded by urgency (Abrams et al 2002). Some women have mixed incontinence, which is a combination of the two. Most studies of incontinence in postpartum women are of stress incontinence, which is the most common childbirth-related type, but some do not specify type and a few separately distinguish stress and urge incontinence.

The prevalence of incontinence in the general female population varies according to the definition used and the age groups included, but it is very common (Yarnell et al., 1981, Burgio et al., 1991, O’Brien et al., 1991 and Hannestad et al., 2000). Many observational studies have now examined the prevalence of urinary incontinence in postpartum women, again with variations in prevalence according to timing and methods of ascertainment. Some studies have assessed women over time and have found many symptoms to be persistent.

In a cross-sectional study of 1505 women in New Zealand, contacted by postal questionnaire at 3 months postpartum, 23.9% reported stress incontinence and a further 10.4% reported urge incontinence (Wilson et al 1996). A UK postal questionnaire survey of 11,701 women contacted 1–9 years after birth found that 20.6% reported stress incontinence that had started for the first time within 3 months of the delivery and lasted beyond 6 weeks (MacArthur et al 1991). Three-quarters of the symptomatic women reported symptoms lasting for at least 1 year.

A large multicentre longitudinal study recruited 7879 women at 3 months postpartum and 4214 were followed up at 6 years (MacArthur et al 2006). Of the women contacted at 6 years, 24% had urinary incontinence at both times (most had stress or mixed incontinence); of those who had been symptomatic when first contacted at 3 months, 73% still had symptoms.

Prevalence of urinary incontinence at 6 months postpartum in a Canadian cross-sectional study of 2492 primiparous women, with ascertainment using a postal questionnaire continence severity score index, was 29.6% (Hatem et al 2005). Type of incontinence was also ascertained, showing that 43% of symptomatic women had stress incontinence, 6% had urge, 13% had mixed, and for the remaining 38% type was unspecified. Farrell et al (2001), also in a Canadian study, recruited 690 primiparae and contacted them up to 6 months postpartum, when prevalence of urinary incontinence (type not specified) was 26%.

A national cohort study in Sweden (Schytt et al 2004), of 2390 women followed from early pregnancy using postal questionnaires, found prevalence of stress incontinence to be 22% at 12 months postpartum. Burgio et al (2003) in a US cohort study, following a convenience sample of 523 women just after birth and at 6 weeks, 3, 6 and 12 months, found prevalence of urinary incontinence (type not specified) ascertained by telephone interview at 12 months after birth to be 13.3%. Prevalence at the various contact points was very similar.

In an Australian study of 1336 women sent postal questionnaires at 6–7 months postpartum, prevalence of stress incontinence was 11% (Brown & Lumley 1998). This was lower than in most other studies but the wording of the questionnaire was whether their stress incontinence had been ‘a problem’ for them, and there is evidence to suggest that postpartum women do not always consider stress incontinence to be a problem (see below). Another Australian study contacted 1295 women immediately postpartum and at 8, 16 and 24 weeks and, using the same wording as the above study, found prevalence of stress incontinence to be 19% at 8 weeks and 11% by 24 weeks (Thompson et al 2002).

Sleep & Grant (1987a), in following up the population of a trial of perineal management regimes (see guideline on Perineal Pain and Dyspareunia), found a similar prevalence of involuntary loss of urine at the 3-year follow-up (23%) as at the 3-month follow-up (Sleep et al 1984). In a European longitudinal questionnaire-based survey of health problems at 5 and 12 months after birth, 1.7% of women in Italy and 7.6% in France reported urinary incontinence (type not specified) at 5 months. At the 12-month follow-up, this symptom was reported by 5.0% and 14% of women respectively. The prevalences were lower than in other studies, especially in Italy, but in both countries showed an increase over time (Saurel-Cubizolles et al 2000).

Several studies have examined severity or effects on lifestyle of postpartum stress incontinence, with varying findings. Bick & MacArthur (1995), in a questionnaire survey of 1278 women at 6–7 months postpartum, assessed symptom severity using a 100 mm visual analogue scale (VAS), median score being 28 (100 most severe). Although most symptomatic women did not rate stress incontinence as severe, 47% reported wearing pads at some time to protect against their involuntary leakage. Many of the women seemed to view stress incontinence as an expected consequence of childbirth (Bick & MacArthur 1995). In the study by Wilson et al (1996), 129 (25%) of the 516 women with urinary incontinence reported using pads. In the study by Schytt et al (2004) only 2% of the sample had stress incontinence which they considered to be a problem for them. Burgio et al (2003) found that only 6% of women who were symptomatic of incontinence at 12 months felt this restricted their activity at all, although 63% felt that it did disturb them somewhat and 12% felt it disturbed them extremely. In the Canadian study by Hatem et al (2005), two quality of life measures were used and both were found to be significantly worse in the women with urinary incontinence than those without. Dolan et al (2004) measured quality of life using the King’s Health Questionnaire and found that among 370 primiparae followed to 3 months postpartum, for 71.1% symptoms had an impact on their life, although for 87.5% of these women this impact was only minor. In a 6-year longitudinal study (MacArthur et al 2006) 12% of symptomatic women had daily or more symptoms and a further 21% had symptoms weekly or more, 23% sometimes and 11% always used a pad to protect against leakage. On a VAS score to assess overall extent of the problem, median score was only 25 but 47% of symptomatic women reported an effect on hygiene, 16% on home life, 35% on social life and 13% on sex life. Mean anxiety and depression scores were also significantly lower in women with incontinence.

Although childbirth is generally considered to be a major cause of urinary incontinence, and general population studies show it to be more frequent in parous than nulliparous women (Thomas et al., 1980, Yarnell et al., 1981, Yarnell et al., 1982, Jolleys, 1988, Assassa et al., 2000, Hannestad et al., 2000 and Rortveit et al., 2003), the precise causative role of the various factors is still not fully understood.

Urinary incontinence is common during pregnancy and a few cohort studies have followed women and found pregnancy incontinence to be an important risk factor for postpartum incontinence. Schytt et al (2004) found that 21.7% of primiparae and 28.6% of multiparae had involuntary loss of urine in the last trimester of pregnancy; this was found to be a strong predictor of postpartum stress incontinence in both vaginal and caesarean section deliveries. Burgio et al (2003) found that 59.5% of the 523 women in their study experienced involuntary loss of urine at some time during the pregnancy and that this was a strong predictor of postpartum incontinence. Viktrup et al (2006) followed up 241 women 12 years after their first pregnancy. They found that onset during pregnancy as well as onset within the first 3 months was predictive of symptoms at 12 years.

Obstetric factors have been examined in both epidemiological and urodynamic studies. Caesarean section has consistently been found to be associated with less subsequent incontinence (MacArthur et al., 1991, Wilson et al., 1996, Assassa et al., 2000, Hannestad et al., 2000, Burgio et al., 2003, Schytt et al et al., 2004, Glazener et al., 2006 and Viktrup et al., 2006). This is consistent with findings of urodynamic investigations of pelvic floor damage (Snooks et al 1986). Even though there is a reduced risk after caesarean section, longitudinal studies have found the prevalence, even in women who only ever delivered operatively, was still about 15% (Hannestad et al., 2000 and MacArthur et al., 2006).

Several urodynamic investigations have found pelvic floor damage to be more common after a longer second-stage labour and the delivery of a bigger baby, but findings relating to forceps delivery have been inconsistent (Snooks et al., 1984, Snooks et al., 1986, Allen et al., 1990 and Meyer et al., 1998). In the surveys by MacArthur et al., 1991 and Brown and Lumley, 1998, stress incontinence was found to be more common after forceps delivery and longer second-stage labour. However, the two are closely interrelated and the effect of forceps disappeared (MacArthur et al 1991), or became only marginally significant (Brown & Lumley 1998), after taking duration of second-stage labour into account. Several other large studies have found no association with forceps or instrumental deliveries on multivariate analyses (Farrell et al., 2001, MacArthur et al., 2001, Schytt et al et al., 2004 and Glazener et al., 2006), although some smaller studies have found a weak positive association (Viktrup, 2002, Burgio et al., 2003 and Dolan et al., 2004). Thompson et al (2002) found an association between stress incontinence and instrumental delivery only at 8 weeks but not at 16 and 24 weeks postpartum.

Older maternal age has been found in numerous studies to be a risk factor for urinary incontinence (MacArthur et al., 1991, Fritel et al., 2004, Schytt et al et al., 2004 and Glazener et al., 2006) as well as multiparity (Assassa et al., 2000, MacArthur et al., 1991, MacArthur et al., 2006, Schytt et al et al., 2004 and Wilson et al., 1996). High body mass index (BMI) or obesity (Wilson et al., 1996, Burgio et al., 2003, Schytt et al et al., 2004 and Viktrup et al., 2006) as well as heavier infant birthweight (MacArthur et al., 1991 and Hojberg et al., 1999) have also been identified as risk factors in some studies.

The management of urinary incontinence mainly involves the use of pelvic floor muscle exercises (PFME), either in isolation or combined with some form of biofeedback, although there has been one small trial on the use of weighted vaginal cones. Since the last edition of this book there have been several additional RCTs and systematic reviews investigating the effects of behavioural techniques on urinary incontinence in childbearing women. Some of these have evaluated antenatal interventions aimed at preventing postpartum incontinence, so are not strictly relevant in a chapter on postnatal management. In the most relevant systematic review, ‘Physical therapies for prevention of urinary and faecal incontinence in adults’, most of the included trials are in childbearing women (Hay-Smith et al 2002). However, this review includes trials with antenatal and postnatal interventions. Another systematic review of pelvic floor exercises, both during and after pregnancy, did not include trials which assessed treatment rather than prevention of urinary incontinence (Harvey 2003). So rather than describe the overall findings of these reviews, we describe here the individual trials in which the intervention to reduce incontinence was delivered postnatally.

Sleep & Grant (1987b) undertook an RCT including 1800 postpartum women of any parity who had had a vaginal delivery. Whilst still in hospital, all women received group-based instruction on pelvic floor exercises from the obstetric physiotherapist, which was standard care. Women in the intervention group also received individual instruction daily whilst in hospital from a midwife co-ordinator. At 3 months postpartum the prevalence of urinary incontinence was similar in both groups (odds ratio (OR) 1.00, 95% confidence interval (CI) 0.83–1.2).

An RCT in Australia included 720 postnatal women who were considered to be at higher risk of urinary incontinence (had instrumental delivery or baby 4000 g+) (Chiarelli & Cockburn 2002). Control was usual care and the intervention was two sessions of individual physiotherapist instruction, one before hospital discharge and the second at 8 weeks postpartum. Significantly fewer women in the intervention group reported urinary incontinence at 12 weeks postpartum relative to controls (OR 0.81, 95% CI 0.66–0.99).

The effects of PFME and bladder training in the treatment of women who were symptomatic of postnatal incontinence at 3 months postpartum (thus not included in the systematic review of prevention) were examined in an RCT in New Zealand (Wilson & Herbison 1998). This trial found significantly less urinary incontinence at 12 months postpartum in the intervention group, but because just over half the women withdrew prior to 12-month assessment relative to only 22% of controls, substantial bias was possible. The intervention was tested again in a much larger multicentre RCT with good follow-up and similar in the treatment and standard care controls (Glazener et al 2001). Symptomatic women (n = 747) were randomly allocated, after stratifying for mode of delivery, frequency of symptoms and parity (para 4 vs 3 or less), to the control (standard postnatal advice) or the intervention group. The intervention was intensive home-based instructions on PFME and bladder retraining from a specially trained midwife or health visitor at 4, 7 and 9 months postpartum. Assessment at 12 months found that significantly fewer of the intervention group were still symptomatic (59.1% vs 69%, diff 9.1%, 95% CI 1.0–19.7%) and significantly more were performing PFME. When these women were followed up at 6 years postpartum, however, there was no difference between groups either in incontinence symptoms or in corresponding PFME practice.

A small non-randomised trial recruited 107 primiparae 2 months after vaginal delivery, some of whom were symptomatic, and randomised them to either 12 one-to-one sessions of pelvic floor re-education with a physiotherapist or standard care. No difference was found between groups in the proportion with urinary incontinence at 10-month postpartum follow-up, but among those who had been symptomatic at recruitment, a significantly higher proportion were ‘cured’ in the intervention group (Meyer et al 2001).

A non-randomised study of 81 matched pairs comparing intensive postpartum PFME weekly therapy sessions with standard care (Morkved & Bo 2000) found that those who had this therapy programme had significantly less incontinence at 12 months postpartum.

In a systematic review of the effects on urinary incontinence of weighted vaginal cones (Herbison et al 2002), the only trial of childbearing women (Wilson & Herbison 1998) showed no benefit, but this was relatively small and there was substantial and unbalanced drop-out so more evidence on this is needed.

Evidence from RCTs on whether postnatal PFME are effective in managing urinary incontinence therefore is not entirely conclusive, although all but the earliest trial, which had the least intensive therapy, showed positive effects in reducing incontinence up to 12 months postpartum. It is very likely that PMFE will be more effective if performed correctly, and adequate advice and training are necessary to do this (Wilson et al., 1987, Bo et al., 1988, Bump et al., 1991 and Morkved and Bo, 1996).