The prevalence of pelvic floor disorders and their relationship to gender, age, parity and mode of delivery

Authors

  • Alastair H. MacLennan,

    Professor, Corresponding author
    1. Department of Obstetrics and Gynaecology, The University of Adelaide, Australia
      Correspondence: Professor A. MacLennan, University Department of Obstetrics and Gynaecology, 1st Floor Queen Victoria Building, Women's and Children's Hospital, 72 King William Road, North Adelaide, South Australia 5006, Australia.
    Search for more papers by this author
  • Anne W. Taylor,

    Epidemiologist
    1. Centre for Population Studies in Epidemiology, South Australian Department of Human Services, Adelaide, Australia
    Search for more papers by this author
  • David H. Wilson,

    Head of Unit
    1. Department of Obstetrics and Gynaecology, The University of Adelaide, Australia
    Search for more papers by this author
  • Don Wilson

    Professor
    1. Department of Obstetrics and Gynaecology, University of Otago, Dunedin, New Zealand
    Search for more papers by this author

Correspondence: Professor A. MacLennan, University Department of Obstetrics and Gynaecology, 1st Floor Queen Victoria Building, Women's and Children's Hospital, 72 King William Road, North Adelaide, South Australia 5006, Australia.

Abstract

Objectives To define the prevalence of pelvic floor disorders in a non-institutionalised community and to determine the relationship to gender, age, parity and mode of delivery.

Design A representative population survey using the 1998 South Australian Health Omnibus Survey.

Sample Random selection of 4400 households; 3010 interviews were conducted in the respondents'homes by trained female interviewers. This cross sectional survey included men and women aged 15–97 years.

Results The prevalence of all types of self-reported urinary incontinence in men was 4.4% and in women was 35.3% (P < 0.001). Urinary incontinence was more commonly reported in nulliparous women than men and increased after pregnancy according to parity and age. The highest prevalence (51.9%) was reported in women aged 70–74 years. The prevalence of flatus and faecal incontinence was 6.8% and 2.3% in men and 10.9% and 3.5% in women, respectively. Pregnancy (> 20 weeks), regardless of the mode of delivery, greatly increased the prevalence of major pelvic floor dysfunction, defined as any type of incontinence, symptoms of prolapse or previous pelvic floor surgery. Multivariate logistic regression showed that, compared with nulliparity, pelvic floor dysfunction was significantly associated with caesarean section (OR 2.5, 95% CI 1.5–4.3), spontaneous vaginal delivery (OR 3.4, 95% CI 2.4–4.9) and at least one instrumental delivery (OR 4.3, 95% CI 2.8–6.6). The difference between caesarean and instrumental delivery was significant (P < 0.03) but was not for caesarean and spontaneous delivery. Other associations with pelvic floor morbidity were age, body mass index, coughing, osteoporosis, arthritis and reduced quality of life scores. Symptoms of haemorrhoids also increased with age and parity and were reported in 19.9% of men and 30.2% of women.

Conclusion Pelvic floor disorders are very common and are strongly associated with female gender, ageing, pregnancy, parity and instrumental delivery. Caesarean delivery is not associated with a significant reduction in long term pelvic floor morbidity compared with spontaneous vaginal delivery.

INTRODUCTION

Pelvic floor disorders which result in urinary and anal (faecal or flatus) incontinence and uterovaginal prolapse undermine the quality of life for a large number of women of all ages throughout the world and have been linked to childbirth and ageing. Urinary incontinence affects between 17% and 45% of adult women1–4, and it is estimated that 0.4% to 17% of adult ambulatory women are faecally incontinent with an increasing prevalence with advancing age5. Similarly, uterovaginal prolapse is a common condition accounting for about 20% of major gynaecological surgery in developed countries6. These distressing and common problems also have significant cost implications for health services. Consequently, it is of great importance to identify possible aetiological factors in pelvic floor dysfunction with a view to its subsequent prevention or at least reduction of its impact.

Traditional predisposing factors are thought to be advancing age, childbearing, obesity and the menopause7. The relative importance of each of these factors has not yet been assessed, and in particular as regards childbearing there is no agreement whether it is pregnancy per se or parturition that predisposes to pelvic floor dysfunction8. The ideal study of a large cohort of women followed throughout their lives would be very costly and difficult, and would extend over more than three-quarters of a century. Second best, but still very useful and much cheaper, is a large cross sectional representative survey of a population where current symptoms of pelvic floor dysfunction at any age can be correlated with parity, mode of delivery and other variables that may influence long term urogenital and anal function.

The South Australian Health Omnibus Survey is a unique population survey9 which allows accurate collection each year of such health data. Investigators have used this survey previously to collect quality health data that have been published widely around the world (e.g. on the use and cost of alternative therapies10, hormone replacement therapy use and characteristics of users11,12, the prevalence of hysterectomy in Australia13 and postnatal depression14). The South Australian Health Omnibus Survey presents a special opportunity for inexpensive but accurate cross sectional population based epidemiological research. At a time when caesarean section rates are rising around the world and preferences are being expressed by women for elective abdominal delivery for fear of pelvic floor dysfunction in later life15, it is opportune to collect data to help clarify these issues.

This study, the first comprehensive population survey of pelvic floor dysfunction, was carried out to define the prevalence of the subjective symptoms of pelvic floor dysfunction in the community and to determine their relationship to gender, age, parity and mode of delivery.

METHOD

The data for these analyses were collected after ethical approval as part of the 1998 Autumn South Australian Health Omnibus Survey. The South Australian Health Omnibus Survey selected a large random sample of South Australian adults (≥ 15 years) from metropolitan Adelaide and from the country towns of South Australia (with a population > 1000 inhabitants) using a clustered, self-weighting, multi-stage, systematic area sample. The person in each household who last had a birthday was selected to be interviewed. No replacement interviews were permitted. From the 4400 households selected, 3010 persons were interviewed (response rate = 73.3%). The interviews were conducted in the respondents' homes. Hotels, hospitals and other institutions were not included in the sample. The data collected were weighted by sex, five-year age group and geographical area to the 1997 Australian Bureau of Statistics Estimated Residential Population16. The methodology of the South Australian Health Omnibus Survey is more fully explained in other papers17,18.

Questions related to pelvic floor disorders included whether men and women had, within the last year, experienced flatus incontinence (poor control of wind), experienced faecal incontinence (lost control of motions), lost urine when they coughed, laughed or sneezed (stress incontinence); or suddenly felt the urge to go to the toilet, but had accidentally wet themselves before reaching the toilet (urge incontinence). Severity of urinary incontinence was judged by use of aids (e.g. incontinence pads). They were also asked if they had symptoms from haemorrhoids, such as pain or bleeding. Female respondents were asked how many pregnancies they had had that had progressed beyond 20 weeks, and whether these had resulted in caesarean section, instrumental delivery (e.g. forceps or vacuum/ventouse suction delivery) or spontaneous vaginal deliveries (either vertex or breech). Women who had been delivered by caesarean section only were asked if any of these abdominal deliveries had been carried out during established labour. Women were asked if they had ever had a vaginal hysterectomy, a vaginal bladder repair, a vaginal rectal repair or any other bladder repair for incontinence, and if, more than three months after their last birth, they had experienced symptoms of prolapse (a feeling of something coming down in the vagina), excessive vaginal laxity during intercourse or introital dyspareunia (painful intercourse at the vaginal entrance). Finally they were asked whether they had required stitches after any of the vaginal deliveries and whether they were currently using, or had ever taken, hormone replacement therapy.

Demographic variables included in the analyses were gender, age, area of residence (metropolitan Adelaide, country South Australia), country of birth (Australia, UK/Ireland or other); highest educational attainment (trade/apprenticeship/diploma/bachelor degree or higher, or other), work status (full time/part time employed or other such as inactive), marital status (never married, married/defacto, separated/divorced or widowed) and gross annual household income (< $20,000, $20,001–$50,000, ≥$50,000 or not stated). The demographics of this population sample are shown in Table 1.

Table 1.  Selected weighted sample demographics of the South Australian 1998 Health Omnibus Survey. Values are given as %.
 South Australia population
Gender 
 Male48.7
 Female51.3
Age 
 15–3436.1
 35–5435.6
 55–6410.8
 ≥6517.5
Income per annum 
 ≥$ 12,00010.9
 $12,001–$20,00014.4
 $20,000–$40,00024.1
 $40,001–$60,00018.5
 >$60,00017.4
Country of birth 
 Australia/Oceania75.3
 Overseas born24.7

Respondents were also asked whether they had used a range of health services in the last month. Co-morbidity variables included were medically confirmed diabetes, osteoporosis and arthritis. The reported presence of medically confirmed bronchitis, emphysema or current medically confirmed asthma was recoded into the presence or absence of respiratory problems. Respondents were also asked if, in the previous three months, they had coughed every day, most days, occasionally or never. Body mass index was derived from self-reported weight and height and recoded into four categories (underweight, normal weight, overweight and obese).

The SF-36 instrument was included in this survey. Scoring and interpretation of the SF-36 is already well documented in the Manual and Interpretation Guide19, and these methods were followed in scoring the questionnaires for this survey.

For the purposes of analysis in this study of the effect of pregnancy and mode of delivery on major types of pelvic floor dysfunction, the following four types of current or past pelvic floor disorders were grouped together as representative of pelvic floor dysfunction:

  • 1Urinary (urge or stress) incontinence.
  • 2Anal (faecal or flatus) incontinence.
  • 3Symptoms of prolapse.
  • 4Any type of bladder or vaginal repair (including vaginal hysterectomy).

Vaginal hysterectomy is included in this arbitrary definition, because in this population this route of hysterectomy has generally been performed as the route of choice when uterine prolapse is present.

Data were then analysed using SPSS Version 8.0 and Epi-Info 6.04 a; χ2 analyses were used to determine the relationships between the dependent and independent variables. The conventional P value of 0.05 was used for all tests of significance. Two multivariate analyses were fitted, one with urinary incontinence as the dependent variable, and one on women only with the four major forms of pelvic floor disorders (as previously defined) as the dependent variable.

To determine the inclusion of variables in logistic regression modelling, a P value of 0.25 was chosen as the critical value for statistical significance at the univariate level20. All independent variables that were statistically significant at the 0.25 level in each of the univariate analyses were entered into a logistic regression analysis to determine, first, the best joint predictors of women with pelvic floor dysfunction and, secondly, the best joint predictors of people with urinary incontinence.

Once a satisfactory model had been obtained, tests for interaction were performed on likely combinations of variables. Interaction terms were entered into the final model to determine whether a statistically significant improvement in the model was obtained. The presence of confounders was also assessed during the modelling process. To determine if multi-collinearity existed, principal component analysis was undertaken on all the independent variables in each of the final models. All relevant variables were entered into a factor analysis to determine the eigen values. The highest eigen value obtained from the factor analysis was divided by the lowest value. If the results of this division did not exceed 20, multi-collinearity was deemed not to be present21.

RESULTS

The total number of respondents was 3010 (48.7% men and 51.3% women). Among these there was a very high response rate (99.2%) to the pelvic floor function questions despite the personal nature of the questions. The mean ages and obstetric history of the women is shown in Table 2. The mean age of the men was 42.9 years. In the group delivered only by caesarean section, 36% of the women were not in established labour at the time of delivery.

Table 2.  Method of delivery. Values are given as n (%), unless otherwise indicated.
Method of deliveryNo. of womenMean ageSD
No births433 (28.0)29.916.9
Caesarean section(s) only100 (6.5)41.111.5
Only vaginal delivery660 (42.7)52.616.8
At least one instrumental delivery283 (18.3)51.315.7
Mixed caesarean and vaginal delivery58 (3.7)45.315.5
Refused to answer12 (0.8)54.514.4
TOTAL1546 (100.0)44.819.04

The overall prevalence of pelvic floor disorders in men and women aged 15–97 years is shown in Table 3. Women experienced both urinary and anal incontinence more commonly than men. Nulliparous women experienced stress incontinence four times more often than men (10.9% and 2.5%, respectively; P < 0.01) and after the first pregnancy the prevalence rose to 37.4%. In both sexes the prevalence of each type of incontinence rose with age (Fig. 1). The overall prevalence of any form of urinary incontinence reported was 4.4% in men and 35.3% in women (OR 11.7; CI 8.9–15.4; P < 0.001). Female urinary incontinence was highest in the 70–74 five-year age group at 51.9% while the highest rate for men was 17.6% in the ≥ 75 year age group. Reports of pelvic floor surgery increased markedly after the age of 45 (Fig. 2). Overall, 46.2% of women were experiencing or had experienced pelvic floor dysfunction of at least one major type which was defined as stress or urge incontinence, flatus or faecal incontinence, current symptoms of vaginal prolapse or previous pelvic floor repair. At least two of these pelvic floor problems were reported by 21.9% of women, with 8.7% experiencing three or more types. These types of pelvic floor dysfunction increased greatly after the first pregnancy and further with subsequent pregnancies (Fig. 3).

Table 3.  Prevalence of pelvic floor disorders in the population aged . 15 years. Values are given as n (%), unless otherwise indicated.
 Male (n= 1464)Female (n= 1546)MaleOR (95% CI) FemaleP
Urinary incontinence     
 Stress only22 (1.5)322 (20.8)1.0017.2 10.9–27.5<0.01
 Urge only28 (1.9)45 (2.9)1.001.5 0.9–2.6)0.08
 Mixed15 (1.0)179 (11.6)1.0012.6 7.3–22.4<0.01
 TOTAL65 (4.4)546 (35.3)1.0011.7 8.9–15.4<0.01
Anal incontinence     
 Flatus100 (6.8)168 (10.9)1.001.7 1.3–2.2<0.01
 Faecal33 (2.3)54 (3.5)1.001.6 1.0–2.50.04
Previous pelvic repairs     
 Bladder (all)104 (9.4)
 Rectal34 (3.1)
 Uterine125 (11.2)
Symptoms of prolapse98 (8.8)
Difficulty defecation69 (4.7)162 (10.5)1.002.4 1.7–3.2<0.01
Haemorrhoids291 (19.9)467 (30.2)1.001.7 1.5–2.1<0.01
Vaginal laxity57 (5.2)
Painful intercourse43 (3.9)
Vaginal hysterectomy126 (8.1)
Figure 1.

Prevalence of urinary and anal incontinence by gender and age.

Figure 2.

Prevalence of women reporting privious pelvic floor surgery by age.

Figure 3.

Prevalence of women with pelvic floor dysfunction by parity.

Mode of delivery increased all the major types of pelvic floor dysfunction, in the order of the following modes of delivery: nulliparity (12.4%); caesarean section (43.0%); spontaneous vaginal delivery (58.0%); and any previous instrumental vaginal delivery (64.0%) (χ2 test for trend = 11.6; P < 0.01) (Table 4). The difference between the caesarean section only group and the vaginal birth group (spontaneous and instrumental combined) was 43.0% compared with 59.6% (χ2= 10.3; P < 0.01). There was no significant difference between women who had delivered only by elective caesarean section before labour, and those who had had a caesarean section during labour, as regards pelvic floor dysfunction (38.8%vs 45.3%, respectively; χ2= 0.4; P= 0.52), stress incontinence (35.1%vs 31.7%; χ2= 0.12; P= 0.73), urge incontinence (8.3%vs 10.9%; χ2= 0.17; P= 0.68) and flatus incontinence (8.1%vs 9.4%; χ2= 0.05; P= 0.83). When spontaneous vaginal deliveries with (n= 498) and without (n= 220) suturing following delivery were compared, there were no statistically significant differences in the prevalence of stress incontinence (39.1%vs 41.5%; χ2= 0.41; P= 0.52), urge incontinence (17.8%vs 19.5%; χ2= 0.31; P= 0.58), flatus incontinence (14.1%vs 12.0%; χ2= 0.58, P= 0.45) and faecal incontinence (5.5%vs 4.2%; χ2= 0.53; P= 0.47).

Table 4.  Prevalence of pelvic floor dysfunction by method of delivery for females. Values are given as n (%).
 No births (n= 433)Caesarean section (n= 100)Spontaneous vaginal delivery (n= 718)Instrumentalvaginal delivery (n= 283)
Stress47 (10.9)33 (33.0)293 (40.8)123 (43.5)
Urge19 (4.4)10 (10.0)136 (19.1)56 (19.8)
Flatus23 (5.3)9 (9.0)91 (12.7)45 (15.9)
Faecal7 (1.6)4 (4.0)33 (4.6)11 (3.9)
Bladder repairs2 (0.5)1 (1.0)67 (9.2)38 (11.0)
Rectal repairs2 (2.0)15 (2.1)17 (6.0)
Prolapse symptoms64 (8.9)34 (12.0)
Vaginal hysterectomy1 (0.2)7 (7.0)77 (10.7)38 (13.4)
Any one of the above conditions74 (12.4)43 (43.0)417 (58.0)181 (64.0)

When both types of vaginal delivery (spontaneous and instrumental) were combined and compared with the caesarean section only group, only urge incontinence was associated with a significantly higher prevalence following vaginal delivery (OR 2.2, CI 1.1–4.5, P= 0.02). Stress incontinence (OR 1.5; CI 0.9–2.3; P= 0.09), flatus incontinence (OR 1.8; CI 0.8–3.8; P= 0.12) and faecal incontinence (OR 1.5; CI 0.5–4.9; P= 0.51) were not significantly associated with the route of delivery.

Univariate analysis showed that increasing age, parity, marital status, work status, education level, low income, increasing weight, mode of delivery, poor quality of life, respiratory problems, coughing, high use of health services, visit to psychiatrist, diabetes, osteoporosis and arthritis were all significantly associated with pelvic floor dysfunction.

The results of the multivariate logistic regression are shown in Table 5. Increasing parity after the first child was not significantly associated with the further risk of pelvic floor dysfunction. However, older age, increasing weight and obesity, method of delivery, coughing, osteoporosis and arthritis all contributed significantly to the logistic regression model describing women with pelvic floor dysfunction (Model χ2= 355.9, df = 12, P < 0.01). No interactions were found. No multi-collinearity existed. Multivariate analysis excluding nulliparous women showed no significant difference between caesarean section and spontaneous vaginal delivery (OR 1.4; 95% CI 0.8–2.2) but a significant association between caesarean section and instrumental delivery (OR 1.8; 95% CI 1.1–2.9).

Table 5.  Multivariate associations between pelvic floor dysfunction and variables of interest for females. BMI = body mass index.
VariableOR(95% CI)P
Age (years)   
 15–341.0  
 35–542.1(1.5–3.0)< 0.001
 ≥553.1(2.1–4.5)< 0.001
Method of delivery   
 No births1.0  
 Caesarean only2.5(1.5–4.3)< 0.001
 Vaginal only3.4(2.4–4.9)< 0.001
 At least one forceps4.3(2.8–6.6)< 0.001
 Both vaginal and caesarean4.7(2.3–9.3)< 0.001
BMI   
 Underweight1.0  
 Normal weight1.4(0.9–2.2)0.123
 Overweight2.0(1.3–3.2)0.003
 Obese2.6(1.6–4.3)< 0.001
Coughing   
 Never/ occasionally1.0  
 Most/every day1.6(1.1–2.3)0.023
Osteoporosis   
 No1.0  
 Yes1.8(1.0–3.2)0.049
Arthritis   
 No1.0  
 Yes1.8(1.3–2.5)< 0.001

Univariate analysis performed with urinary incontinence as the dependent variable found that female gender, increasing age, country of birth, marital status, education level, work status, household income, body mass index, quality of life, respiratory problems, coughing, high use of health services, diabetes, osteoporosis and arthritis were all significantly associated with incontinence.

Multivariate logistic regression was performed with all types of urinary incontinenc as the dependent variable. Female gender (OR 13.8, 10.3–18.7), older age (OR 3.3, 2.3–4.8), marital status (OR 2.5, 1.7–3.5), obesity (OR 2.6, 1.1–4.1), coughing and poor quality of life (OR 1.6, 1.2–2.2) all proved to contribute significantly (Model χ2= 717.2, df = 12, P < 0.001). No interactions were found. No multi-collinearity existed.

In this study 13.8% of women and 3.1% of men reporting urinary incontinence wore some kind of extra protection or incontinence aid and were defined as having severe incontinence. A person with severe incontinence was statistically significantly more likely to be female, older, widowed, to have arthritis and to have a household income of less than $20,000 per year. Severe incontinence was not related to increasing parity (after the first pregnancy) or method of delivery when compared with other women with incontinence.

Haemorrhoid symptoms were more commonly reported by women than men (30.2%vs 19.9%; χ2= 42.6; P < 0.01) and in both sexes symptoms suggestive of haemorrhoids increased with age (Fig. 4). A highly significant increase (from 8.6% to 37.9%, χ2= 82.8, P < 0.001) in the prevalence of symptoms of haemorrhoids occurs in women after their first pregnancy which did not significantly increase with subsequent pregnancies (parity 2: 38.4%; parity ≥ 3: 40.0%). Of women who had never given birth, 8.5% reported haemorrhoids, 36.3% of whom had had a caesarean section, 38.3% a vaginal delivery and 41.5% instrument deliveries.

Figure 4.

Prevalence of symptoms of haemorrhoids by gender and age.

Overall, 4.0% of women reported introital dyspareunia. This symptom was mostly reported by women who had delivered vaginally rather than abdominally. Those who had had instrumental delivery experienced discomfort more commonly than those who delivered spontaneously (5.7%vs 3.3%; χ2= 2.8; P= 0.09). Women who had never given birth reported dyspareunia most often, and the prevalence of this condition decreased with increasing parity (χ2 for trend = 8.1; P < 0.005). Overall, 5.2% of women reported vaginal laxity. It was reported more commonly by younger women (15–44 years) than older women (≥ 45 years) (8.0% and 2.9; χ2= 14.8; P < 0.01). Use of hormone replacement therapy at any time was reported by 42.6% of all women aged > 50 years and was not associated with the reporting of any type of incontinence.

Quality of life data were calculated for the summary physical and mental health dimensions of the short-form (SF-36) questionnaire. For women with any type of incontinence or symptoms of prolapse, both the physical and mental health summary scores were statistically significantly lower than the population mean of 50 and for those without these conditions (Table 6). Women with pelvic repairs and symptoms of haemorrhoids were statistically significantly lower on the physical health summary score but not on the mental health score.

Table 6.  SF-36 quality-of-life summary scale scores for people reporting types of pelvic floor dysfunction. Values are given as mean (SD), unless otherwise indicated.
 nPhysical health summary scale scoreMental health summary scale score
  1. *Statistically significantly different (P > 0.05) from population without that disorder.

Stress incontinence53845.9 (10.8)*48.6 (11.0)*
Urge incontinence26742.8 (11.2)*48.1 (11.4)*
Flatus incontinence26843.7 (12.0)*46.9 (12.3)*
Faecal incontinence8740.1 (13.1)*43.9 (11.2)*
Any previous pelvic repair19042.8 (11.3)*49.9 (10.7)
Current symptoms of prolapse9844.4 (11.3)*48.6 (10.6)*
Haemorrhoids75946.8 (10.7)*49.5 (10.4)

DISCUSSION

This study, which is the first comprehensive cross sectional population survey of pelvic floor dysfunction, has shown that, outside institutional care, 46.2% of women and 11.1% of men currently acknowledge some form of major pelvic floor dysfunction. Lifetime risk of such problems is even higher and is of epidemic proportions in later life. This study confirms that there is a major sex difference in these problems, greatly increasing in parous women following the first pregnancy. However, even in women who have never had a pregnancy beyond 20 weeks of gestation there is a much higher prevalence of pelvic floor dysfunction than in men (e.g. stress incontinence occurring in 10.9% compared with 1.5%). This study did not collect data on nulliparous women who may have had pregnancies which did not reach 20 weeks of gestation. Thus, any association between early pregnancy and later pelvic floor function cannot be assessed in this study.

The data suggest that it is pregnancy rather than parturition that contributes most to pelvic floor dysfunction in later life. Caesarean section, as the only mode of delivery, is not associated with a significant reduction in most types of pelvic floor morbidity compared with spontaneous vaginal delivery. However, there is a consistent trend for instrumental vaginal delivery to increase further the risk of nearly all types of pelvic floor morbidity. This has been described previously in many studies, and both structural and neurological damage during labour and delivery have been incriminated22,23. It is likely that most of these instrumental deliveries were forceps deliveries, as until recently ventouse deliveries have been uncommon in South Australia.

The prevalence of symptoms of pelvic floor dysfunction may not correlate with objective tests for these morbidities. In our study 13.8% of women and 3.1% of men reporting urinary incontinence wore some kind of extra protection or incontinence aids and were defined as having severe incontinence. Multivariate analysis of only those women who had severe incontinence showed a similar association with gender, first pregnancy and age and no association with mode of delivery. A history of vaginal hysterectomy was intentionally included in the compendium of conditions associated with pelvic floor dysfunction, as in South Australia this route of hysterectomy is generally chosen when a significant degree of uterine prolapse is present. Exclusion of this surgically corrected group would have reduced the true incidence of genital prolapse, and major differences in the prevalence of this operation were seen with parity and mode of delivery (Table 4). No significant differences were seen in the analysis when vaginal hysterectomy was excluded from the definition of pelvic floor dysfunction.

There is increasing public awareness that pelvic floor dysfunction and future quality of life in women are associated with pregnancy and childbirth. In some countries there has been a recent trend for women to request elective caesarean section in their first and subsequent pregnancies in the hope of avoiding future urinary and anal incontinence, vaginal prolapse and laxity, and haemorrhoids. Indeed a recent regional survey24 of female obstetricians in the UK found that 31% would choose an elective caesarean section in an uncomplicated pregnancy mostly for those reasons. Our cross sectional observational study does not suggest that caesarean section (before or during labour) avoids or greatly reduces many of these problems. The association of urinary incontinence with abdominal delivery was slightly less than that following all types of vaginal birth in this study, but there was no reduction in anal incontinence or haemorrhoids associated with caesarean birth as the only route of delivery. Only symptoms of prolapse, vaginal laxity and bladder repairs were significantly reduced in association with abdominal delivery. These long term cross sectional correlations are at variance with Wilson et al.8 whose short term longitudinal study showed a 60% reduction in urinary incontinence three months after delivery following one caesarean section as compared with a vaginal delivery. However, the small numbers of women having three or more caesarean sections in the study by Wilson et al.8 also had a similar prevalence of urinary incontinence (35%) as those women delivered vaginally. These findings were explained by possible vesical denervation following several caesarean sections and also increasing age, factors which could also apply to the present study. It is also important to emphasise that the prevalence of urinary incontinence in women who have given birth vaginally is probably under-estimated in the present study because of successful treatment (i.e. there was an increase in bladder repairs reported by this group) (Table 4). Thus the reduction in urinary incontinence in the caesarean section group could have been relatively greater than the raw data suggest.

A possible cause of the long term effect of pregnancy on pelvic floor function is the influence of the hormones of pregnancy on connective tissues. The hormone relaxin is secreted by the ovaries, placenta and decidua during human pregnancy with levels of relaxin highest in the second trimester25. This hormone has a putative role in connective tissue remodelling in the human and has been associated with pelvic girdle relaxation syndrome and other structural effects that reduce the tensile strength of connective tissue26. Considerable remodelling of connective tissue occurs in the uterine body, cervix, pelvic joints and perineal tissues in late pregnancy and parturition. On the basis of evidence from animal and in vitro experiments, it is postulated that in the human relaxin activates the collagenolytic system by increasing collagenase and collagen peptidase. It also activates fibroblasts to produce new collagen but achieves a net dilution of the connective tissue, reducing the viscosity of the ground substance by increasing its water content. Thus, the tensile strength of the target connective tissues is weakened5,27. Other contributing factors could be the increasing and prolonged pressure on the pelvic floor muscles and ligaments during pregnancy, and hereditary connective tissue disorders which might explain the association seen in this study between pelvic floor dysfunction and osteoporosis and arthritis.

Various obstetric/midwifery practices have been suggested to minimise trauma of the genital tract at childbirth. Liberal episiotomy rates do not appear to reduce trauma, early postnatal complications28 or urinary incontinence29,30. Data from this survey show no differences in the prevalence of long term pelvic floor morbidity whether or not vaginal or perineal suturing was required after spontaneous vaginal delivery. However, instrumental delivery is associated with higher rates of pelvic floor dysfunction, particularly urinary incontinence and bladder repairs. Some countries have reported recent reductions in instrumental deliveries, with a corresponding rise in caesarean section rates; and it may be that a reduction in ‘difficult’ instrumental deliveries will in turn reduce future long term pelvic floor morbidity. It is not possible in this study to distinguish difficult from less difficult instrumental deliveries. A randomised trial comparing elective caesarean section with vaginal delivery, with lifelong follow up of pelvic floor function, may never be possible practically; so descriptive studies such as this may be the main source of data for some time to come. Currently different policies as to maternal position in labour, style of pushing and perineal massage have not been studied in regard to long term pelvic floor dysfunction. However, pelvic floor muscle exercises appear to be of benefit in the treatment of urinary stress incontinence31 and should be encouraged after delivery. Other policies that may reduce the prevalence of pelvic floor morbidity are the avoidance of excess weight and of factors that exacerbate coughing (e.g. smoking), and possibly the treatment of oestrogen deficiency.

Although the prevalence of pelvic floor dysfunction is very high in this population study, these figures are likely to be an underestimate, particularly in the very elderly, as nursing homes and hospitals are excluded from this health omnibus survey. Urinary and anal incontinence are very common in elderly women, especially those in institutional care32,33. This survey confirms an escalation in incontinence rates, symptoms of prolapse, and associated pelvic floor surgery with increasing age.

As seen in other studies9,34, multivariate analysis of the data from this study showed an association between incontinence and increasing body mass index and coughing with a significant reduction in quality of life with all types of incontinence. Multivariate analysis also found persistent associations between osteoporosis and arthritis, and the major types of pelvic floor disorder described in this study. Reasons for these associations are not clear, but the link may be a genetic predisposition to connective tissue and structural disorders and/or oestrogen deficiency.

Introital dyspareunia was reported most commonly by those who had never given birth and after the first birth; vaginal trauma and suturing are more commonly associated with the first vaginal birth, and perineal discomfort is common for many months postnatally30. Paradoxically, younger women complained more of vaginal laxity than older women.

Symptoms of haemorrhoids had a slightly different pattern of prevalence than for other types of pelvic floor morbidity. They were common in men (20%) but were still more common in women (30%). Their prevalence rapidly increases after one pregnancy from 9% to 38%, and are not significantly affected by the mode of delivery. Increasing age was strongly associated with an increase in symptoms of haemorrhoids.

All types of urinary and anal incontinence in this study were associated with a reduction in both mental and physical quality of life scores, particularly in the case of faecal incontinence. The approximate prevalence of 2% and 4% of faecal incontinence in ambulant men and women, respectively, is in keeping with the literature35. All types of incontinence can greatly interfere with a person's ability to enjoy life, go on excursions and function outside the home. Furthermore incontinence incurs considerable cost in sanitary aids, clothes and institutional care. The results of this cross sectional population study are in keeping with a recent postal questionnaire study of 7879 female respondents where three months after delivery 26% and 4% complained of flatus and faecal incontinence, respectively25. In that study forceps delivery was significantly associated with faecal incontinence (OR 1.94) and caesarean section reduced early faecal incontinence by nearly a half (OR 0.58) compared with spontaneous vaginal delivery.

This population study confirms that urinary and anal incontinence and other symptoms of genital prolapse are a major problem in society and are strongly associated with female gender, ageing, first pregnancy, increasing parity and reduced quality of life. The majority of elderly women report at least one type of major pelvic floor morbidity. Pregnancy, more than birth, appears to contribute to long term pelvic floor dysfunction. Caesarean section is not associated with a significant reduction in the risk of pelvic floor morbidity compared with spontaneous vaginal delivery, and thus elective caesarean section is apparently not an effective way to reduce the prevalence of most subsequent pelvic floor disorders, except when instrumental vaginal delivery can thereby be avoided. As a lifelong randomised intervention study of mode of delivery is unlikely ever to be performed, data from such cross sectional population studies may have to be the basis of advice to women who are concerned that the route of delivery may influence future pelvic floor dysfunction and their quality of life.

Acknowledgements

This study was supported by research grants from the University of Otago and the University of Adelaide.

Ancillary