The prevalence of urinary incontinence 20 years after childbirth: a national cohort study in singleton primiparae after vaginal or caesarean delivery

Authors

  • M Gyhagen,

    1. Department of Obstetrics and Gynaecology, Sahlgrenska Academy at Gothenburg University, Gothenburg, Sweden
    Search for more papers by this author
  • M Bullarbo,

    1. Department of Obstetrics and Gynaecology, Sahlgrenska Academy at Gothenburg University, Gothenburg, Sweden
    2. Department of Obstetrics and Gynaecology, Södra Älvsborgs Hospital, Borås, Sweden
    Search for more papers by this author
  • TF Nielsen,

    1. Department of Obstetrics and Gynaecology, Sahlgrenska Academy at Gothenburg University, Gothenburg, Sweden
    2. Department of Obstetrics and Gynaecology, Södra Älvsborgs Hospital, Borås, Sweden
    Search for more papers by this author
  • I Milsom

    1. Department of Obstetrics and Gynaecology, Sahlgrenska Academy at Gothenburg University, Gothenburg, Sweden
    Search for more papers by this author

Professor I Milsom, Department of Obstetrics and Gynaecology, Sahlgrenska University Hospital, SE-416 85 Gothenburg, Sweden. Email ian.milsom@gu.se

Abstract

Objective  To investigate the prevalence and risk factors for urinary incontinence (UI) 20 years after one vaginal delivery or one caesarean section.

Design  Registry-based national cohort study.

Setting  Women who returned postal questionnaires (response rate 65.2%) in 2008.

Population  Singleton primiparae who delivered in the period 1985–1988 with no further births (n = 5236).

Methods  The Swedish Pregnancy, Obesity and Pelvic Floor (SWEPOP) study linked Medical Birth Register (MBR) data to a questionnaire about UI.

Main outcome measures  Prevalence of UI and UI for more than 10 years (UI > 10 years) were assessed 20 years after childbirth.

Results  The prevalence of UI (40.3 versus 28.8%; OR 1.67; 95% CI 1.45–1.92) and UI > 10 years (10.1 versus 3.9%; OR 2.75; 95% CI 2.02–3.75) was higher in women after vaginal delivery than after caesarean section. There was no difference in the prevalence of UI or UI > 10 years after an acute caesarean section or an elective caesarean section. We found an 8% increased risk of UI per current body mass index (BMI) unit, and age at delivery increased the UI risk by 3% annually.

Conclusions  Two decades after one birth, vaginal delivery was associated with a 67% increased risk of UI, and UI > 10 years increased by 275% compared with caesarean section. Our data indicate that it is necessary to perform eight or nine caesarean sections to avoid one case of UI. Weight control is an important prophylactic measure to reduce UI. Current BMI was the most important BMI-determinant for UI, which is important, as BMI is modifiable.

Introduction

In modern societies, women live the major part of their lives after giving birth to one or two children. Urinary incontinence (UI) is a common condition affecting adult women of all ages that may have a negative influence on their quality of life.1 Pregnancy, and in particular vaginal delivery, has been implicated in the etiology of UI.1,2 An increasing number of women request caesarean section for non-medical indications, and for some this demand appears to be motivated by a desire to prevent pelvic floor damage, including UI.

The etiology of UI is known to be multifactorial, but obesity and ageing, as well as obstetric trauma during childbirth, are known to be three of the most important risk factors.1,2 Although several studies have demonstrated an association between UI and vaginal delivery in the short- and medium long-term, the long-term effects of childbirth on the risk of UI remain controversial.3–6 The assessment of the influence of childbirth on urinary incontinence later in life has been hampered by the heterogeneity of study populations. Women of different ages and varying body weights have been included after a variable number of pregnancies, often with different modes of delivery. The aim of this study was therefore to compare the prevalence of UI 20 years after delivery in a cohort of women who had given birth to only one child after vaginal delivery or caesarean section.

Methods

A national survey of pelvic floor dysfunction, the Swedish Pregnancy, Obesity and Pelvic Floor (SWEPOP) study, was conducted in 2008. The population studied and their obstetric data were obtained from the Swedish Medical Birth Register (MBR). The MBR, which was started in 1973, is a national register that includes more than 98% of all births in Sweden. Data from all antenatal clinics and all obstetric units are sent to the MBR at the National Board of Health and Welfare. Obstetrical parameters from the delivery were obtained from the MBR. Caesarean sections performed before the onset of labour were denoted as elective caesarean sections, and caesarean sections performed during labour were denoted as acute caesarean sections. The weight and height of pregnant women had been measured by a midwife at the antenatal clinic, usually at 8–10 weeks of gestation, and was also obtained from the MBR. Maternal weight at delivery and the weight gain during pregnancy were recorded at the delivery unit, and were also obtained from the MBR. When individual data were initially examined it was noted that the maximum recorded body weight from the MBR was 99 kg. Because of a lack of data storage capacity in the 1980s, the MBR had decided to restrict the registration of ‘heavy women’ by recording up to two digits only. We therefore reviewed the patient records of the 300 women recorded as having a body weight of 99 kg to obtain the correct weights of these women.

The quality of this national database has been shown to be good and suitable for population studies of this type. A description of the MBR in English can be found at http://www.socialstyrelsen.se/register/halsodataregister/medicinskafodelseregistret/inenglish, and an evaluation of the MBR has been performed by Cnattingius et al.7 as well as by the National Board of Health and Welfare, and is available at http://www.socialstyrelsen.se/publikationer2002/2002-112-4. Inclusion criteria for participation in this study were singleton primiparae who delivered in the period 1985–1988 and who had no further births. Multifetal pregnancies were excluded. Ethical approval was obtained from the Regional and the National Ethic Review Boards (the Ethics Committee at Sahlgrenska Academy, Gothenburg University, and the National Board of Health and Welfare).

The results of this study have been reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement.

The questionnaire was sent to 9423 women who were asked to provide written, informed consent to participate and to complete a questionnaire (Figure 1). Women were excluded from the study, based on the answers in the questionnaire, if they affirmed multiparity (the misdiagnosis of ‘parity’ is predominantly related to immigration, as the first birth in Sweden is sometimes assumed to be the first birth ever), or a multifetal or current pregnancy (Figure 1). The results regarding anal incontinence and genital prolapse will be reported separately.

Figure 1.

 Flow chart of the women who gave birth to one child in the period 1985–1988, as identified from the Swedish Medical Birth Register (MBR).

The 31-item questionnaire included questions about current height and weight, urinary or anal incontinence and genital prolapse, menstrual status, hysterectomy, the menopause, and hormone treatment, etc. UI was defined according to the International Continence Society and by the question ‘Do you have involuntary loss of urine?’.8 Participants reporting UI were grouped according to the duration of UI (UI < 5 years, 5–10 years, or >10 years). The severity of UI (frequency and volume of leakage) was assessed using the Sandvik score.9 After three mailing cycles during a 4-month period the questionnaire was returned by 6148 women (65.2%).

Maternal body mass index (BMI) was categorized as normal (<25), overweight (≥25–29.9), and obese (≥30) according to the WHO classification,10 and was calculated for each woman according to weight and height measurements in early pregnancy at 8–10 weeks of gestation (early pregnancy BMI), at delivery (delivery BMI), and 20 years after delivery (current BMI).

Characteristics of the sample population and the non-responders

The proportion of missing data varied between 0 (age) and 15.9% for hysterectomy in the population cohort. There was little difference in the proportions of missing data between groups, e.g. the proportion of missing data for hysterectomy, which had the greatest proportion of missing data, was 15.5% (620/3995) in the vaginal delivery group and 17% (205/1204) in the caesarean section group. The non-responders were 1.6 years younger (49.6 ± 5.9 versus 51.2 ± 5.9 years; < 0.001), and were more often overweight or obese (37 versus 27%; < 0.001 and had an infant birthweight <4000 g (43 versus 48%; < 0.003) compared with the responders.

Statistical analysis

Statistical analysis was performed with sas 9.1 (SAS Institute Inc., Cary, NC, USA). For cohort characteristics the chi-square test was used to compare categorical variables and the Students t-test was used for continuous variables. A P-value of <0.05 was considered statistically significant. Adjusted frequencies and odds ratios for UI were calculated using a covariance analysis model to obtain effect measures. A logistic regression model was used to assess risk factors for UI while controlling for potential confounding factors. Potential risk factors used in the analysis were mode of delivery, maternal age at delivery, maternal BMI (at delivery and current), hysterectomy, hormone replacement therapy, gestational age, infant birthweight and head circumference. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated from the model. The prevalence data permitted the calculation of the number of caesarean sections needed to avoid one case of UI using the number needed to treat principle (NNT). The NNT was calculated as the inverse of the absolute risk reduction, where risk reduction was the difference of adjusted UI prevalence between vaginal delivery and caesarean section.

Results

Basic characteristics of the women grouped according to mode of delivery are shown in Tables 1 and 2. The mean follow-up time after delivery was 21.5 years. (SD 1.5 years) in the vaginal delivery group and 21.8 years (SD 1.1 years) in the caesarean section group. Women delivered by caesarean section were older [current age 53.7 years (SD  6.3  years) compared with an age of 50.4  years (SD  5.6  years) in the vaginal delivery group; < 0.001], and gave birth to an infant with a lower birthweight (< 0.001) at a lower gestational week (< 0.001), compared with women who were delivered vaginally (Table 1). The proportion of women aged ≥35 years at delivery was higher (< 0.001) in the caesarean section group, whereas the proportion of infants with a birthweight ≥3500 g was lower (< 0.001) in the caesarean section group compared with the vaginal delivery group.

Table 1. The basic characteristics of the women grouped according to mode of delivery
 Vaginal delivery mean (range) (%)
n = 3995
Caesarean section (elective and acute) mean (range) (%)
n = 1204
Difference
95% CI
P
  1. A Student’s t-test was used for statistical comparison between groups.

Age at delivery29.0 (15–46)31.9 (15–45)2.9 (2.6 to 3.3)<0.001
Age ≥ 35 years17.4%35.6%18.2 (15.3 to 21.2)<0.001
BMI in early pregnancy23.0 (15.0–45.6)23.0 (15.2–41.7)0.0 (−0.3 to 0.3)=0.94
BMI at delivery28.3 (17.2–50.3)28.3 (18.3–47.2)0.0 (−0.3 to 0.3)=0.95
BMI at delivery ≥2579.0%78.1%0.9 (−4 to 2)=0.56
BMI current26.1 (14.5–63.0)26.3 (16.1–53.8)0.2 (−0.1 to 0.5)=0.16
BMI current ≥2551.3%51.2%0.1 (−3 to 3)=0.98
Infant birthweight (g)3585 (850–5680)   3294 (820–5615)291 (245 to 336)<0.001
Infant birthweight ≥3500 g55.8%38.2%17.6 (14.4 to 20.7)<0.001
Gestational age (weeks)39.7 (24–45)38.5 (27–43)1.2 (1.1 to 1.4)<0.001
Hysterectomy7.9%9.9%2.0 (0.00 to 0.04)=0.05
Estrogen therapy7.1%10.0%2.9 (0.8 to 4.9)<0.01
Table 2. Cohort characteristics for the women who underwent elective or acute caesarean section
 Elective caesarean section mean (range) (%)
n = 766
Acute caesarean section mean (range) (%)
n = 438
Difference
95% CI
P
  1. A Student’s t-test was used for statistical comparison between groups.

Age at delivery32.5 (15–45)30.9 (18–45)−1.64 (−2.32 to −0.95)<0.001
Age ≥35 years40.7%26.7%−14.0 (−19.4 to −8.6)<0.001
BMI early pregnancy22.4 (15.2–41.7)23.9 (15.4–41.7)1.52 (1.00 to 2.04)<0.001
BMI at delivery27.6 (18.3–47.2)29.3 (18.9–46.5)1.75 (1.20 to 2.30)<0.001
BMI at delivery ≥2574.7%83.6%8.9 (3.8 to 14.0)<0.001
BMI current25.8 (18.9–50.8)27.0 (18.1–53.8)1.16 (0.55 to 1.77)<0.001
BMI current ≥2547.1%58.2%11.1 (5.2 to 17.0)<0.001
Infant birthweight (g)  3157 (820–5615) 3534 (950–5310)377 (286 to 468)<0.001
Infant birthweight ≥3500 g27.2%57.5%30.3 (24.7 to 36.0)<0.001
Gestational age (weeks)38.0 (27–43)39.4 (27–43)1.44 (1.15 to 1.73)<0.001
Hysterectomy10.1%9.5%−0.01 (−0.04 to 0.03)=0.75
Estrogen therapy11.5%7.4%−0.04 (−0.08 to 0.00)<0.05

The prevalence of UI (Table 3) was 67% higher (OR 1.67; 95% CI 1.45–1.92) after a vaginal delivery (40.3%) compared with women who had been delivered by caesarean section (28.8%). From the prevalence data available on UI it was possible to calculate using the NNT principle that it is necessary to perform eight or nine caesarean sections to avoid one case of UI. Furthermore, the prevalence and risk increase of UI persisting for more than 10 years almost tripled for vaginal delivery compared with caesarean section. The prevalence of UI for >10 years after vaginal delivery was 10.1% compared with 3.9% after caesarean section (OR 2.75; 95% CI 2.02–3.75). There were, however, no significant differences in the prevalence of UI (27.1 versus 24.4%; OR 1.15; 95% CI 0.88–1.51) or UI for more than 10 years (6.5 versus 5.1%; OR 1.30; 95% CI 0.79–2.14) between women delivered by acute caesarean section or by elective caesarean section, respectively.

Table 3. Crude and adjusted prevalence and odds ratios of urinary incontinence, and urinary incontinence persisting for more than 10 years, in relation to mode of delivery
  Caesarean section (%)Vaginal delivery (%)Crude OR (95% CI)Caesarean section (%)Vaginal delivery (%)Adjusted* OR (95% CI)
  1. *Adjusted for BMI current, BMI at delivery, maternal age, gestational weeks, infant birthweight, and head circumference.

Urinary incontinence30.040.21.56 (1.36–1.80)28.840.31.67 (1.45–1.92)
Urinary incontinence for >10 years4.610.02.30 (1.73–3.08)3.910.12.75 (2.02–3.75)

The prevalence of UI was higher after vaginal delivery compared with caesarean section for each current BMI class (BMI <25, 25–29.9, and ≥30), with differences ranging from 11 to 14% (Table 5). Again, using the NNT principle, we calculated the number of caesarean sections that would need to be performed to prevent one case of UI for the different BMI groups (nine for a BMI <25; seven for a BMI of 25–29.9, and eight for a BMI ≥30). The combined effect of BMI and mode of delivery was substantial, for example the adjusted frequency of UI after caesarean section with a current BMI <25 was 24.7%, whereas the frequency more than doubled to 54.8% after a vaginal delivery with a current BMI ≥30 (Table 5). When using ‘normal BMI’ as a reference, the risk of UI increased significantly for both overweight and obese women after both modes of delivery. The increase in risk of UI in obese women more than doubled in comparison with women of normal BMI after vaginal delivery, and more than tripled after caesarean section (Table 6). In the logistic regression analyses we found an 8% (range 6–10%) increased risk of UI per BMI unit increase, and the increased rate of UI was apparent for both modes of delivery (Table 4).

Table 4. The results of logistic regression analysis of possible risk factors for urinary incontinence (odds ratio; 95% CI
Risk/confounding factorsOR (95% CI)
  1. *A total of 3488 women contributed to the model.

Vaginal delivery1.71 (1.41–2.08)
BMI current1.08 (1.06–1.10)
BMI at delivery0.98 (0.96–1.00)
Maternal age at delivery1.03 (1.02–1.04)
Gestational weeks1.01 (0.96–1.06)
Infant birthweight (kg)1.00 (0.98–1.02)
Infant head circumference ≥36 cm1.06 (0.84–1.34)
Hysterectomy1.21 (0.90–1.63)
Hormone replacement therapy yes versus no1.34 (0.98–1.82)

Because of an interaction term between infant birth weight and mode of delivery (< 0.01) we made separate analyses of infant birth weight for the caesarean section group and vaginal delivery group. The prevalence of UI following vaginal delivery was higher than after caesarean section in all infant birthweight groups, except for weights <3000 g. For women who delivered vaginally rates of incontinence increased with increasing infant birthweight, but this was not observed after caesarean section (Table 5). Logistic regression analysis in the total cohort failed to demonstrate a significant increase of UI risk for infant birthweights ≥4500 g (Table 6).

Table 5. Crude and adjusted prevalence and odds ratio of urinary incontinence in relation to mode of delivery, stratified for each risk factor
Prevalence of urinary incontinence
  Caesarean section (%)Vaginal delivery (%)Crude OR (95% CI)Caesarean section (%)Vaginal delivery (%)Adjusted* OR (95% CI)
  1. *Adjusted for BMI current, BMI at delivery, maternal age, gestational weeks, infant weight and head circumference.

Infant birthweight (g)
<300033.037.61.23 (0.91–1.64)35.537.41.08 (0.81–1.49)
3000–349928.638.61.57 (1.22–2.00)27.339.11.71 (1.33–2.19)
3500–399930.940.51.53 (1.14–2.05)27.641.31.85 (1.36–2.50)
4000–449927.841.51.85 (1.28–2.67)25.341.32.08 (1.42–3.03)
≥450023.150.43.39 (1.63–7.04)21.248.83.56 (1.68–7.53)
Infant head circumference (cm)
<3629.839.91.56 (1.34–1.82)29.040.11.64 (1.40–1.91)
≥3626.943.12.06 (1.41–3.01)24.048.82.46 (1.66–3.63)
BMI early pregnancy
<2530.937.51.34 (1.12–1.61)29.737.81.44 (1.20–1.74)
25–29.929.244.51.94 (1.37–2.76)27.644.92.14 (1.50–3.05)
≥3045.551.91.29 (0.67–2.50)47.751.91.18 (0.61–2.28)
BMI at delivery
<2529.432.71.17 (0.84–1.63)28.932.91.20 (0.86–1.68)
25–29.930.839.31.45 (1.17–1.80)30.239.51.51 (1.21–1.87)
≥3030.546.31.96 (1.49–2.59)28.446.92.93 (1.68–2.95)
Current BMI
<2524.534.11.59 (1.29–1.97)24.735.61.68 (1.36–2.08)
25–29.929.341.71.73 (1.34–2.23)28.142.41.88 (1.46–2.43)
≥3045.352.81.35 (1.01–1.81)41.654.81.71 (1.27–2.29)
Maternal age at delivery (years)
<2322.835.61.87 (1.18–2.96)22.836.01.90 (1.20–3.02)
23–2927.939.01.65 (1.26–2.17)26.038.31.77 (1.34–2.33)
30–3433.842.91.47 (1.14–1.89)33.243.21.53 (1.19–1.97)
≥3530.443.01.72 (1.33–2.23)32.342.71.56 (1.21–2.01)
Table 6. Adjusted additional risks of urinary incontinence in relation to stratified risk factors, grouped according to mode of delivery*
  Caesarean section (95% CI)Vaginal delivery (95% CI)
  1. *Adjusted for maternal age, gestational age in weeks, BMI at term, BMI current, infant birthweight and infant head circumference.

BMI current
<25Ref.Ref.
25–29.91.50 (1.11–2.03)1.32 (1.14–1.53)
≥303.27 (2.34–4.59)2.50 (2.10–2.98)
Age at term
<23 yearsRef.Ref.
23–29 years1.29 (0.73–2.26)1.22 (1.02–1.46)
30–34 years1.84 (1.06–3.18)1.49 (1.23–1.80)
≥35 years1.66 (1.00–2.74)1.49 (1.21–1.84)
Infant birthweight
<4500 gRef.Ref.
≥ 4500 g0.66 (0.33–1.29)1.23 (0.87–1.76)
Infant head circumference
<36 cmRef.Ref.
≥ 36 cm0.86 (0.59–1.25)1.07 (0.88–1.29)

The multivariable analysis (Table 4) did not demonstrate any significant increased risk of UI associated with infant head circumference. However, there was an increased risk of UI after vaginal delivery compared with caesarean section regardless of fetal head circumference. The risk increase associated with vaginal delivery in comparison with caesarean section was stronger for fetal head circumferences ≥36 cm than for head circumferences <36 cm: ORs 2.46 (95% CI 1.66–3.63) versus 1.64 (95% CI 1.40–1.91) (Table 5). Nor were there any differences in UI prevalence in the women grouped according to fetal head circumference (Table 6) after both modes of delivery.

The prevalence of UI was 10% higher in women aged ≥35 years at delivery compared with women aged <23 years who had undergone caesarean section, and was 7% higher in women of the same ages who had a vaginal delivery (Table 5). In the logistic regression analysis a higher maternal age was associated with an increased risk of UI (OR 1.03; 95% CI 1.02–1.04), which corresponds to an annual risk increase of 3% per year.

Discussion

The risk of developing UI was found to be 67–71% higher after vaginal delivery than after caesarean section, and the prevalence of UI persisting for more than 10 years almost tripled after vaginal delivery compared with caesarean section. We found no difference in the prevalence of UI, or UI persisting for more than 10 years, between women delivered by acute or elective caesarean section, indicating that it is the later stages of delivery, when the fetus passes through the pelvic floor, that leads to the increased risk of UI. Maternal weight was also an important risk factor, and in the logistic regression analyses we found an 8% increased risk of UI per BMI unit increase, and this influence of BMI on UI was apparent for both modes of delivery. Current BMI was the most important BMI-determinant for UI, and this finding is important, as BMI is modifiable. For women who delivered vaginally rates of incontinence increased with increasing infant birthweight, but this was not observed after caesarean section. The prevalence of UI increased with maternal age and there was an annual increase in UI prevalence of 3% per year.

The main strengths of this study are the use of a large, national, population-based cohort of singleton primiparae and the high response rate. There are advantages of studying singleton primiparae, as the first delivery is considered to exert the greatest increase in risk for UI, even if subsequent deliveries contribute to a further increase in the risk of UI.1,11 Including multiparous women would disrupt the obstetric homogeneity, and as most risk factors also covariate with time/age this would also confound the effect measures of the analysis. The inclusion of singleton primiparae regardless of maternal health status, and maternal and fetal complications, is considered a strength as it allows for a greater generalisation of results, and therefore gives a better basis for consultation about elective caesarean section on request. The weight and height data during pregnancy were objectively measured at the antenatal clinics, and a validated questionnaire was used.9,12

Some limitations of the present investigation must also be considered. First, women with incontinence may be more predisposed to participate in such a study, and therefore the rate of UI might be overestimated. Secondly, the symptoms of UI were self-reported. However, several studies have shown that self-reported symptoms are consistent and valid when assessing current UI and changes in incontinence severity over time, which applies to our study.11,13 This study also lacks information on whether UI was present before or during pregnancy, or whether it started after delivery. However there is little evidence to suggest any difference in UI prevalence before the first pregnancy or during pregnancy in women grouped according to mode of delivery. It was not possible to assess the importance of the length of the second stage of delivery, as this is unfortunately not documented in the MBR. Obstetric techniques and parameters have varied over time (fewer episiotomies, increasing number of vacuum extractions and severe lacerations, older mothers, higher BMI, and heavier children), which may also influence the clinical interpretation of our results. It may also in some respects seem unrepresentative to study the consequences of giving birth to only one child. However, UN data show that total fertility rates (TFR) are rapidly declining globally, and the predicted TFR in the middle of this century is predicted to be <2.0 children/woman, and in many developed countries the TFR is already between 1.0 and 1.5.14 Analyses of the non-responder group suggest a small selection bias on our results acting in both directions (younger women and smaller children leading to an overestimation of results; overweight/obesity leading to an underestimation of results).

In this study, caesarean section was often used as a reference for comparison with vaginal delivery to quantify the effect of vaginal birth on UI. The baseline outcome after caesarean section can then be interpreted as representing the risk of pregnancy itself, and the risk of vaginal delivery is represented by the risk of pregnancy plus the risk of vaginal delivery, and hence the difference between vaginal delivery and caesarean section is therefore a measure (in terms of UI prevalence and risk) of vaginal birth trauma. Even if the nulliparous pelvis represents the best available clinical model of normal function, the prevalence of UI in nulliparous women of childbearing age has been reported to be 10–15%.2,15 Urinary leakage preceding pregnancy in nulliparous women has been shown to be a strong precursor for the increased prevalence of UI 4–12 years postpartum.1,15 Pregnancy in itself, independent of labour and delivery practice, has been reported to be a risk factor for postpartum UI,4,16 especially if the incontinence started during the first trimester.17 Several studies have demonstrated that postpartum UI is a risk factor for UI after varying terms of follow-up.3–5

There is still no general agreement whether or not the long-term maternal effects of the two delivery modes differ with regard to prevalence of UI. A prospective multicentre study did not show a significant difference of risk for bothersome UI between women delivered by one or more vaginal deliveries and women delivered by one or more caesarean sections.18 Also, the omnibus survey of MacLennan et al.19 could not demonstrate an increased risk for any type of UI after vaginal delivery, when compared with caesarean section. In these studies, however, the caesarean section groups were relatively small and heterogeneous with respect to parity. On the other hand, the EPINCONT study demonstrated a 1.7-fold increased age-adjusted risk of UI after one or more vaginal deliveries, compared with one or more caesarean sections. The age-standardized prevalence rate of UI was 15.9% for the caesarean section group and 21.0% for the vaginal delivery group.2 The study population was younger and the follow-up time shorter in the EPINCONT-study compared with the cohorts and the follow-up time of our study. Other later studies have also indicated an increased risk of UI following vaginal delivery compared with caesarean section.20–22

Several studies have reported that a higher BMI is a risk factor for UI,1 and cross-sectional studies have confirmed this association in middle-aged women.23,24 We found an increased risk of UI of ≈8% per unit BMI. Our findings correspond with those of others who showed a risk increase varying between 2 and 10% per unit increase of BMI.25,26 The overall assessment of the relationship between BMI and prevalence of UI in this study indicated that there was a dose–response relationship between BMI and UI, whereas the effect of mode of delivery (i.e. vaginal delivery or caesarean section) appeared to be constant, regardless of maternal BMI status.

The results of this study indicated that current BMI was the most important determinant for UI, and this finding is important, as BMI is modifiable. The resolution of UI has been demonstrated after weight loss.27 Intervention by non-surgical means or laparoscopic gastric bypass surgery indicates that there is a dose–response association between the prevalence of UI and the magnitude of weight reduction.28 The strong association between the prevalence of UI and current BMI is encouraging, as it means that it is never too late to achieve an improvement of UI through weight reduction and weight control.

The negative effect of vaginal delivery on urinary continence is consistent with results of several clinical studies that have demonstrated poor urethral support and increased urethral mobility after vaginal delivery, leading to UI.29,30 Impaired urethral function could also be shown after vaginal delivery,29 but this was not observed after caesarean section.31

In conclusion, the risk of developing UI and experiencing UI persisting for more than 10 years was higher 20 years after a vaginal delivery compared with a caesarean section. The prevalence did not differ between women delivered by acute or elective caesarean section, indicating the importance of the later stages of delivery when the fetus passes through the pelvic floor for the occurrence of UI in later life. Weight control was also shown to be an important preventive measure to reduce UI. Our data also provide a quantification of the importance of mode of delivery and body weight for the risk of future UI. The results of this study indicate that one has to perform eight or nine caesarean sections to avoid one case of UI. However, there may be other advantages regarding the possible protective effect of caesarean section on future pelvic floor function, such as a reduced prevalence of vaginal prolapse, which could be included in the decision of whether or not caesarean section is advantageous. Vaginal delivery and BMI have been shown to be important risk factors for UI, but operative delivery by caesarean section also involves a degree of risk for morbidity and mortality over and above that of vaginal delivery.32

Disclosure of interests

We declare that we have no conflict of interests.

Contribution to authorship

All authors were involved in the conception and design of the study, the acquisition of data, and in the interpretation of the results, as well as the writing of the article. All authors approved the final version of the submitted article. MG and IM take full responsibility for the integrity of the data and the accuracy of the data analysis.

Details of ethics approval

Ethical approval for the SWEPOP study was obtained from the Regional and National Ethic Review Boards (the Ethics Committee at Sahlgrenska Academy, Gothenburg University, ref. no. 381-07, 13 August 2007 and the National Board of Health and Welfare, ref. no. 34-9148/2007, 26 October 2007).

Funding

The study was supported by a National LUA/ALF grant no. 11315 and the Region of Västra Götaland, and by grants from The Göteborg Medical Society and the Hjalmar Svenssons Fund. The funding sources had no role in the study design, data analysis, data interpretation, or writing of the report. MG and IM had full access to all study data and had final responsibility for the decision to submit for publication.

Acknowledgements

We thank Ms Marianne Sahlén and Ms Anja Andersson for help with data registration, and Björn Areskoug MSc for expertise in statistical programming.

Ancillary