International Urogynecology Journal

, Volume 18, Issue 9, pp 1033–1042

The effect of mode of delivery, parity, and birth weight on risk of urinary incontinence

Authors

  • Thomas J. Connolly
    • Massachusetts General Hospital
    • New England Research Institutes
  • Sharon L. Tennstedt
    • New England Research Institutes
  • Carol L. Link
    • New England Research Institutes
  • John B. McKinlay
    • New England Research Institutes
Original Article

DOI: 10.1007/s00192-006-0286-4

Cite this article as:
Connolly, T.J., Litman, H.J., Tennstedt, S.L. et al. Int Urogynecol J (2007) 18: 1033. doi:10.1007/s00192-006-0286-4
  • 153 Views

Abstract

To examine the relationship between symptoms of urinary incontinence (UI) and mode of delivery, parity, and birth weight, data were obtained from a population-based random sample of 3,205 black, Hispanic, and white women age 30–79 in the Boston Area Community Health Survey. Measures include UI symptoms [≥3 (moderate/severe) Sandvik’s severity index]; reproductive history [live birth(s), no live births, never pregnant]; mode of delivery for live births (≥1 vaginal birth, cesarean delivery only); parity (1, 2, ≥3); and maximum birth weight of live births (<4,000 g, ≥4,000 g). Using logistic regression, women having ≥1 vaginal delivery had twice the odds of UI compared to women with no pregnancies (P = 0.002) or only cesarean deliveries (P = 0.032). There was no difference in odds of UI between cesarean delivery only and never pregnant, by parity or birth weight. Vaginal delivery contributes to risk of UI for black, Hispanic, and white women.

Keywords

Urinary incontinenceReproductive risk factors

Introduction

It is estimated that at least 10% of women experience symptoms of urinary incontinence (UI) to a significant degree [1] (Tennstedt et al. 2006, Prevalance and risk factors of urine leakage in a racially and ethnically diverse population of adults: the Boston Area Community Health [BACH] Survey. In review: unpublished data). Trauma to the pelvic floor during childbirth has been suggested as a major risk factor for symptoms of UI [2, 3]. Several obstetrical factors have been implicated in the risk of symptoms of UI. Vaginal delivery, particularly the initial delivery, has been associated with increased odds of UI [48]. Cesarean deliveries, on the other hand, have been reported as protective on the development of UI [914], with risk of UI being about half the risk associated with vaginal deliveries [3, 12, 13, 1517]. The relationship between parity and subsequent UI is unclear. Increasing parity is consistently associated with UI [1823], with reports of the effect diminishing with age [24, 25] in clinical populations but not in a population-based sample [26]. However, most studies have compared multiparous women with nulliparous women and have not simultaneously considered the effect of mode of delivery, with a recent exception. Lukacz et al. [27] have reported increased odds of stress UI for vaginally parous women when compared to nulliparous or cesarean deliveries, both labored and unlabored, in a large managed-care population. The evidence linking birth weight to subsequent UI is also inconsistent [4, 5]. Neurophysiologic studies have implicated birth weight with the risk of UI [28, 29], but multivariable analysis was not conducted, and hence, the results should be interpreted with caution.

Ability to draw clinically relevant conclusions from this body of research is hampered by design limitations, particularly the restricted and non-representative study populations and the short post-partum period of observation in many of these studies. Most of the large epidemiologic studies with population-based samples are limited to white women. To address the inconsistencies and limitations of these prior studies, we used a large population-based cohort of women with racial/ethnic diversity and across a broad age range to examine the relationship between UI and mode of delivery, parity, and birth weight, while controlling for multiple variables.

Materials and methods

The protocols and informed-consent procedures were approved by New England Research Institutes’ Institutional Review Board. All participants provided written, informed consent.

Study design

The Boston Area Community Health (BACH) Survey is an epidemiologic survey of a range of urological and urogynecological conditions, conducted with a population-based random sample of black, Hispanic, and white men and women age 30–79. Screening was conducted at the household and then the individual level. Screenings were completed for 36% of the selected households; 30% of the households refused screening, and 34% of the households could not be contacted after at least 16 attempts to reach them by mail, telephone, or field visit. Individuals from selected census blocks were chosen according to eligibility rules to achieve our goal of approximately equal numbers of black, Hispanic, and white respondents in four age categories: 30–39, 40–49, 50–59, and 60–79 by gender. Sampling proceeded in five batches, each a random sub-sample of the overall BACH study. Eligibility rules varied by batch and were randomly assigned to selected households based on household demographics at the start of each batch. BACH eligibility criteria included: competent to sign informed consent and able to speak English or Spanish well enough to complete the survey. A total of 5,506 study participants were recruited over the period of April 2002 through June 2005. Participants included 3,205 women and 2,301 men; 1,770 blacks, 1,877 Hispanics, and 1,859 whites. Interviews were completed with 63.3% of the screener-identified eligible individuals from the selected households.

Measures

Data were obtained during a 2-h, in-person interview, conducted by a trained and certified phlebotomist/interviewer, generally in the subject’s home [30]. Detailed self-reported major symptoms of the urological and gynecological conditions of interest, as well as information on medical and reproductive history, major co-morbidities, prescription and over-the-counter medications, lifestyle factors (e.g. smoking, diet, and exercise), psychosocial factors, and medical care utilization were collected by interview. Wherever possible, the questions and scales employed on BACH were selected from published instruments with documented metric properties. Because the sample was racially/ethnically diverse by design, preference was given to those measures with available and validated Spanish versions to ensure cultural equivalence and sensitivity.

Urinary incontinence

Women were considered to have symptoms of UI using Sandvik’s severity index [31] based on frequency and amount of urine leakage. Women with an index score of 3 or more (moderate or severe UI) were considered to have symptoms of UI. Women with an index score less than 3 (mild UI) or no reported leakages of urine in the last 12 months were defined as not experiencing symptoms suggestive of UI. Frequency of urine leakage was coded as weekly (“one or more times per week” or “everyday”), monthly (“one or more times per month”), or occasionally (“less than once per month”). Amount of leakage was queried as “drops, small splashes, or more”. For women defined as having symptoms suggestive of UI, type of UI was determined based on responses to the following questions:

“During the last 7 days, how many times did you accidentally leak urine:
  1. (a)

    When you were performing some physical activity such as coughing, sneezing, lifting, or exercise?

     
  2. (b)

    When you had the strong feeling that you needed to empty your bladder but you couldn’t get to the toilet fast enough?

     
  3. (c)

    Without any particular physical activity or warning?”

     

Non-zero responses to (a) only were defined as stress-type leakage, to (b) only as urge-type leakage, to both (a) and (b) as mixed-type leakage, and to (c) only as other type of leakage. Zero-responses to all three questions were also classified as ‘other’ type of leakage.

Obstetrical factors

Based on self-report, reproductive history was defined as: (1) live birth(s), (2) pregnancies but no pregnancies resulting in live births, or (3) never pregnant. Women in the live births group had at least one live birth, but other pregnancies could have been stillbirths (130 pregnancies), miscarriages (718), abortions (634), or tubal/ectopic pregnancies (87). Women in the pregnancies resulting in no live births group included only pregnancies that were stillbirths (11 pregnancies), miscarriages (70), abortions (96), or tubal/ectopic pregnancies (12). The stillbirth, miscarriage, abortion, and tubal/ectopic pregnancy categories were not mutually exclusive, e.g., one woman could have reported a stillbirth and an abortion on separate occasions. In all analyses, the never pregnant group served as the reference group for comparison purposes. For women who reported live births, mode of delivery was categorized as either at least one vaginal birth or cesarean delivery only. The number of live births was categorized as one, two, three, or more. Maximum birth weight of all live births was considered as a dichotomous variable with two categories: <4,000 and ≥4,000 g.

Covariates

Covariates were selected based on prior evidence and include age, self-reported race/ethnicity (black, Hispanic, white as reference group), body weight/obesity [body mass index (BMI), measured height in meters and weight in kilograms, and waist-to-hip ratio (WHR)], physical activity as measured by the Physical Activity Scale for the Elderly (PASE), alcoholic drinks including beer, wine, and hard liquor consumed per day, menopausal status, self-reported history of urinary tract infections (UTIs), hysterectomy, and type of hysterectomy. Surgery for UI symptoms was of interest but was precluded from analysis because of small numbers. Age and WHR were considered as continuous variables; the remaining variables were treated as categorical measures.

Analysis

Before multivariate modeling, characteristics of the BACH study population by moderate/severe UI status were tested using t tests for continuous measures and chi-square tests for categorical measures. Odds ratios (ORs) [with 95% confidence intervals (CIs)] were calculated to describe the bivariate relationships between moderate/severe UI and these characteristics.

This study aims to separately investigate the relationships between moderate/severe UI and three reproductive factors: (1) mode of delivery, (2) parity, and (3) maximum birth weight. For instance, when investigating mode of delivery, our interest lies in quantifying the impact that mode of delivery has on moderate/severe UI status without accounting for the other reproductive factors. Similarly, we investigated the relationships between parity and birth weight with UI. Therefore, for each reproductive factor, we first consider a model to describe the relationship between UI and the reproductive factor, adjusting for the design parameters, age, and race/ethnic group. Then, for each reproductive factor, we consider a model to describe the relationship between UI and the reproductive factor adjusting for additional covariates to ensure that the relationship remains in the presence of other intervening variables. Each of these models is described below.
  1. 1.

    Mode of delivery

    Mode of delivery was defined as a composite measure with four categories: one or more vaginal deliveries, only cesarean deliveries, no live births, and never pregnant. To predict the odds of moderate/severe UI based on mode of delivery controlling for age and race/ethnicity, a logistic regression model was fit (Table 2). ORs with 95% CIs were presented for mode of delivery, age, and race/ethnicity; CIs that cover one denote that the group was not statistically significantly different from the reference group. The Wald F test P value gives the P value for testing the overall difference for each factor; e.g., the P = 0.009 in Table 2 indicates that there is at least one significant difference between the at least one vaginal, only cesarean deliveries, no live births, and never pregnant groups. The individual t test P values quantify where the specific two-way differences occur; e.g., the P = 0.032 in Table 2 denotes a statistically significant difference in the odds of moderate/severe UI between the at least one vaginal group and the only cesarean delivery group.

    To ensure that the relationship between mode of delivery and moderate/severe UI did not change, controlling for additional factors other than age and race/ethnicity, a backward elimination technique was used to determine which additional variables needed to be controlled for. It was found that BMI and prior UTIs were the only significant covariates after multivariate adjustment.

    Additional analyses were also performed to decide if the relationship between UI and mode of delivery remained in other situations. To investigate whether the relationship between UI and mode of delivery changed by race/ethnicity, an interaction term was included. To ascertain whether the relationship between UI and mode of delivery remained constant across age groups, a logistic regression model adjusting for race/ethnicity was fit for each age group.

     
  2. 2.

    Parity

    Parity was defined as a composite measure with five categories: one live birth, two live births, 3+ live births, no live births, and never pregnant. To predict the effect of parity on the odds of moderate/severe UI, a logistic regression model adjusting for age and race/ethnicity was fit (Table 3). Because the aim was to assess the effect of parity on UI for those having a live birth and this effect on UI may vary by type of birth, a covariate to indicate cesarean delivery was included in the model. As with mode of delivery, a multiple logistic regression model adjusting additionally for BMI and prior UTIs was also fit. Models stratifying by age were also fit to determine whether relationships remained similar by age group.

     
  3. 3.

    Maximum birth weight

    A composite measure including birth weight and reproductive history (no live births and never pregnant) was constructed. Next, to predict the odds of moderate/severe UI based on maximum birth weight controlling for age and race/ethnicity, a logistic regression model was fit (Table 4). Because the aim was to assess the effect of maximum birth weight in women who had live births and because type of delivery affects the relationship with UI, an indicator variable for cesarean delivery was included in the model. As with the mode of delivery and parity analyses, a multiple logistic regression model was then fit adjusting for age, race/ethnicity, BMI and prior UTIs. Additional models stratifying by age were also considered to decide if the relationship between birth weight and UI remained constant across age.

     

P values <0.05 were deemed statistically significant; this does not account for multiple comparisons. A multiple imputation technique [32, 33] was used to impute missing values occurring for some of the UI and reproductive history measures and other covariates. Because of design requirements, the BACH subjects had unequal probabilities of selection into the study. To account for this unequal probability of selection, all analyses were weighted and conducted in Statistical Analysis Software (SAS) [34] and Survey Data Analysis (SUDAAN) [35]. All reported percentages were weighted.

Results

Of the 3,205 women included in this analysis, 1,070 women classified themselves as black, 1,111 as Hispanic, and 1,024 as white. Of the 3,205 women, 390 were defined as having symptoms of moderate/severe UI (13% using survey weights). Of the remaining women, 515 women were classified as having symptoms of mild UI, and 2,300 women reported no leakage. The distribution of symptoms of UI type was as follows: 69 (25%) with stress UI only, 60 (13%) with urge UI only, 223 (55%) with mixed UI, and 38 (7%) categorized as other type. Because of the small numbers of women who experienced stress UI only and urge UI only, analyses by symptoms of UI type were not presented.

Regarding reproductive history, the majority (n = 2,621) reported at least one live birth, 164 women had been pregnant at least once but had no live births, and 420 women reported that they had never been pregnant. Of the 2,621 women reporting at least one live birth, the vast majority of women (86.9%) reported at least one vaginal delivery; 329 women (13.1%) reported cesarean deliveries exclusively. Most (76.8%) reported a maximum birth weight <4,000 g and 590 women (23.2%) reported maximum birth weight ≥4,000 g. Numbers of live births were reported as one (22.3%), two (33.4%), or three or more (44.3%).

As reported in Table 1, women reporting symptoms of UI were more likely to be white; have BMI 30 or more; have a high WHR; score lower on the PASE (less active); be pre-, post-, or surgical menopause; have had prior UTIs; or have had a hysterectomy or had surgery for UI. Bivariate analyses show that almost all of the covariates of interest, with the exception of alcohol use and type if hysterectomy, are related to symptoms of moderate/severe UI.
Table 1

Study characteristics comparing women with and without symptoms of urinary incontinence (UI)a; odds ratios (95% confidence intervals) from separate logistic regression models for the bivariate relationships of each covariate with UI

Variable

Total

Without symptoms of UI

With symptoms of UI

P value*

Odds ratio (95% CI)

Age

49.2 ± 13.3

48.5 ± 13.1

53.4 ± 13.4

<0.001

1.03 (1.01, 1.04)

Race/ethnic group

   

0.046

 

 Black

29.9

30.1

28.8

 

0.85 (0.58, 1.25)

 Hispanic

13.3

14.0

8.4

 

0.54 (0.31, 0.92)

 White

56.8

55.9

62.8

 

1.00 (reference)

Body mass index

   

0.003

 

 <25 kg/m2

33.3

34.3

26.1

 

1.00 (reference)

 25–29 kg/m2

28.6

29.5

22.2

 

0.98 (0.60, 1.60)

 30+ kg/m2

38.1

36.2

51.7

 

1.86 (1.18, 2.93)

Waist-to-hip ratio

0.83 ± 0.08

0.82 ± 0.08

0.84 ± 0.09

0.006

15.96 (2.20, 115.96)

Physical activity (PASE score)

   

0.002

 

 <100

27.8

26.7

36.1

 

1.00 (reference)

 100–249

53.6

53.8

52.6

 

0.72 (0.50, 1.03)

 250+

18.6

19.6

11.4

 

0.43 (0.25, 0.72)

Alcoholic drinks per day

   

0.943

 

 None

41.6

41.4

43.3

 

1.00 (reference)

 <1

43.3

43.6

41.2

 

0.90 (0.63, 1.30)

 1–2

12.9

12.9

12.9

 

0.95 (0.51, 1.76)

 3+

2.2

2.2

2.7

 

1.18 (0.34, 4.05)

Menopause status

   

0.003

 

 Pre

24.9

26.5

13.5

 

0.40 (0.14, 1.10)

 Peri

22.2

21.9

24.3

 

0.87 (0.34, 2.21)

 Post

22.2

21.7

25.8

 

0.93 (0.39, 2.20)

 Surgical

15.7

14.7

22.6

 

1.20 (0.50, 2.85)

 Hormones

12.1

12.3

10.3

 

0.66 (0.24, 1.82)

 Undetermined

2.9

2.8

3.6

 

1.00 (reference)

Prior urinary tract infections

   

<0.001

 

 Yes

45.0

42.9

59.6

 

1.96 (1.36, 2.83)

 No

55.0

57.1

40.4

 

1.00 (reference)

Hysterectomy

   

0.003

 

 Yes

15.4

14.4

22.6

 

1.74 (1.26, 2.40)

 No

84.6

85.6

77.4

 

1.00 (reference)

Type of hysterectomy

   

0.859

 

 Vaginal

25.9

25.7

26.9

 

1.05 (0.51, 2.13)

 Abdominal

74.1

74.3

73.1

 

1.00 (reference)

Surgery for incontinence

   

0.002

 

 Yes

2.4

1.5

8.8

 

6.25 (2.99, 13.07)

 No

97.6

98.5

91.2

 

1.00 (reference)

aPercentages are given for categorical variables, mean, and standard deviation given for continuous variables.

*P value from a t test to test the relationship between age and symptoms of urinary incontinence; remaining P values are from chi-square tests to test the relationship between study characteristic and symptoms of urinary incontinence

The odds of moderate/severe UI differed by the composite measure of reproductive history and type of delivery (i.e., at least one vaginal delivery, only cesarean deliveries, no live births, and never pregnant; P = 0.009), adjusting for age and race/ethnicity (Table 2). Women having at least one vaginal delivery had significantly higher odds of moderate/severe UI compared both to women who had never been pregnant (P = 0.002) and to women who had only cesarean deliveries (P = 0.032). There was no difference in the odds of moderate/severe UI between women who had only cesarean deliveries and women who had never been pregnant (P = 0.703). After also adjusting for BMI and prior UTIs, the significant relationship between vaginal delivery and moderate/severe UI remained (P = 0.013; data not shown). This relationship did not differ by race/ethnicity (data not shown). Stratifying by age group (not shown), the strongest effect of mode of delivery on risk of UI occurred for the youngest age group (30–39 years). Women 30–39 years having at least one vaginal delivery had significantly increased odds of moderate/severe UI compared either to women who had never been pregnant (OR 4.51, CI 1.18, 17.20, P = 0.027) or to women having cesarean deliveries only (OR 10.67, CI 2.17, 52.49, P = 0.004). However, there was no difference in the odds of moderate/severe UI between women 30–39 who had cesarean deliveries and those who had never been pregnant (OR 0.42, CI 0.06, 2.81, P = 0.373).
Table 2

Results of a logistic regression model to predict the odds of urinary incontinence based on mode of delivery, adjusting for age and race/ethnicity using data from the 3,205 women in the Boston Area Community Health (BACH) Survey

Covariate

Odds ratio with 95% confidence interval

P value for Wald F testa

P value for individual t test

Reproductive history and mode of delivery

   

 At least one vaginal delivery

2.10 (1.30, 3.39)

0.009

0.032*, 0.343**, 0.002***

 Only cesarean deliveries

1.13 (0.59, 2.17)

 

0.540**, 0.703***

 No live births

1.42 (0.62, 3.28)

 

0.407***

 Never pregnant

1.00 (Reference)

  

Age

1.02 (1.01, 1.04)

0.007

 

Race/ethnic group

   

 Black

0.77 (0.51, 1.14)

0.046

 

 Hispanic

0.51 (0.28, 0.90)

  

 White

1.00 (Reference)

  

aThe Wald F test quantifies whether the odds of moderate/severe UI varies by the four groups (at least one vaginal delivery, only cesarean deliveries, no live births, or never pregnant).

*Compared with the only cesarean deliveries group

**Compared with the no live births group

***Compared with the never pregnant group

The effect of mode of delivery on risk of UI was not statistically significant in older age groups (40–49, 50–59, and 60–79 years). Specifically, women 40–49 years having at least one vaginal delivery did not have significantly increased odds of moderate/severe UI compared to women who had never been pregnant (OR 1.73, CI 0.57, 5.29, P = 0.332); for women 50–59 years OR 1.64, CI 0.75, 3.62, P = 0.217 and for women 60–79 years OR 1.47, CI 0.71, 3.06, P = 0.302. In addition, there were no differences in the odds of moderate/severe UI between women 40–49, 50–59, and 60–79 years who had cesarean deliveries and those who had never been pregnant (OR 1.05, CI 0.30, 3.65, P = 0.936 for 40–49 age group; OR 1.49, CI 0.45, 4.87, P = 0.510 for 50–59 age group; OR 0.70, CI 0.14, 3.52, P = 0.664 for 60–79 age group).

There was an overall difference between women who have had one child, two children, three or more children, no live births, and no pregnancies (P = 0.030), adjusting for age, race/ethnicity, and cesarean delivery (Table 3). Specifically, compared with women who had never been pregnant, the odds of moderate/severe UI were higher for women who had had one (P = 0.016), two (P = 0.002), or three or more (P = 0.031) children. However, there were no detectable differences in the odds of moderate/severe UI by number of children; that is, odds of moderate/severe UI did not appear to increase with parity. The results were similar when adjusting for BMI and prior UTIs and also when stratifying by age.
Table 3

Results of a logistic regression model to predict the odds of urinary incontinence based on parity, adjusting for age, race/ethnicity, and cesarean deliveries, only using data from the 3,205 women in the Boston Area Community Health (BACH) Survey

Covariate

Odds ratio with 95% confidence interval

P value for Wald F testa

P value for individual t test

Reproductive history and number of children

   

 One

2.14 (1.15, 3.97)

0.030

0.651*, 0.522**, 0.394***, 0.016****

 Two

2.43 (1.40, 4.22)

 

0.195**, 0.192***, 0.002****

 Three or more

1.79 (1.06, 3.03)

 

0.601***, 0.031****

 No live births

1.42 (0.62, 3.28)

 

0.407****

 Never pregnant

1.00 (Reference)

  

Age

1.02 (1.00, 1.04)

0.003

 

Race/ethnic group

   

 Black

0.78 (0.53, 1.16)

0.071

 

 Hispanic

0.54 (0.30, 0.95)

  

 White

1.00 (Reference)

  

Cesarean delivery only

0.52 (0.30, 0.93)

0.027

 

aThe Wald F test quantifies whether the odds of moderate/severe UI varies by the five groups (one, two, three or more, no live births, or never pregnant).

*Compared with the two children group

**Compared with the three or more children group

***Compared with the no live births group

****Compared with the never pregnant group

There was an overall difference in the odds of moderate/severe UI between the <4,000 g group, the ≥4,000 g group, the no live birth group, and those never pregnant (P = 0.023), adjusting for age, race/ethnicity and cesarean delivery (Table 4). Compared with women who had never been pregnant, the odds of moderate/severe UI were about two times higher both for women with maximum birth weight <4,000 g group (P = 0.002) and ≥4,000 g (P = 0.028). However, there was no detectable difference in the odds of moderate/severe UI between the two maximum birth weights (P = 0.710). These results are unchanged when additionally adjusting for BMI and prior UTIs or when stratifying by age.
Table 4

Results of a logistic regression model to predict the odds of urinary incontinence based on maximum birth weight, adjusting for age, race/ethnicity, and cesarean deliveries only, using data from the 3,205 women in the Boston Area Community Health (BACH) Survey

Covariate

Odds ratio with 95% confidence interval

P value for Wald F testa

P value for individual t test

Reproductive history and maximum birth weight

   

 <4,000 g

2.15 (1.32, 3.50)

0.023

0.710*, 0.323**, 0.002***

 >4,000 g

1.97 (1.07, 3.60)

 

0.466**, 0.028***

 No live births

1.42 (0.62, 3.28)

 

0.406***

 Never pregnant

1.00 (Reference)

  

Age

1.02 (1.00, 1.04)

0.007

 

Race/ethnic group

   

 Black

0.76 (0.51, 1.13)

0.044

 

 Hispanic

0.51 (0.29, 0.90)

  

 White

1.00 (Reference)

  

Cesarean delivery only

0.54 (0.31, 0.95)

0.032

 

aThe Wald F test quantifies whether the odds of moderate/severe UI varies by the four groups (<4,000 G, 4,000+ G, no live births or never pregnant).

*Compared with the <4,000 g group

**Compared with the no live births group

***Compared with the never pregnant group

Discussion

The BACH survey included a sample of more than 3,000 black, Hispanic, and white women aged 30–79 years. In this population of women, we found that having at least one vaginal delivery was the consistent obstetrical risk factor of moderate/severe UI. The definition of incontinence used in this study differs from the International Continence Society definition of any involuntary leakage of urine [36]. We focused on moderate/severe UI because we consider incontinence of this severity to be clinically meaningful to women, likely to have an impact on quality of life and to influence care seeking. Our definition of mode of delivery is consistent with that of McKinnie et al. [17]; that is, the group with at least one vaginal delivery included women with cesarean deliveries as well. However, in contrast to their finding of no protective effect of cesarean delivery on risk of UI, the finding from this study is consistent with that from other large epidemiological studies [1, 6, 7, 37] as well as studies of clinical populations [3, 10, 12, 13]. However, unlike those studies that have reported higher risk of UI for women with cesarean deliveries only when compared to nulliparous women [17, 3840], we found no difference in the odds of UI between these two groups of women. This could be explained by our focus on moderate/severe UI rather than any incontinence as in several of these studies. Lukacz et al. [27] also reported no difference in stress UI between women with cesarean delivery only and nulliparous women, although severity of symptoms was not specified. The clinical importance of this finding is supported by theorized pathophysiologic changes as well as observed injury to the levator ani or nerves during vaginal delivery [6, 12, 41] as contributors to incontinence risk. These theorized changes and injuries are further supported by results of neurophysiologic and imaging studies [28, 42].

The effect of mode of delivery on odds of UI is significant only for women in the 30–39 age group and not in older age groups. This finding is supportive of the link between a pathophysiologic injury occurring with vaginal delivery and subsequent UI that possibly resolves with time. However, we also suspect that, with aging, other factors contribute to the risk of UI that might diminish the impact of delivery mode on incontinence risk.

Importantly, and not addressed by most studies to date, this association of vaginal delivery and subsequent risk of UI was observed for black, Hispanic, and white women. Although white women have consistently reported the highest rates of UI [43, 44], and as was also found in the BACH sample (Tennstedt et al. 2006, Prevalance and risk factors of urine leakage in a racially and ethnically diverse population of adults: the Boston Area Community Health [BACH] Survey. In review: unpublished data), the risk of UI was affected similarly for all three race/ethnic groups by mode of delivery.

In contrast to clinical studies that have reported increased risk of UI with increasing parity [3, 12, 13, 2325, 39, 45], we found that symptoms of UI did not increase with the number of vaginal deliveries or the number of births. This is consistent with another large cohort study that also focused on risk of moderate/severe UI [16]. Finally, consistent with the lack of parity effect on UI, we found no significant effect of increasing birth weight on risk of moderate/severe UI.

Unfortunately, there were some questions that could not be addressed with data from the BACH Survey. Other studies have reported that increased risk of UI is associated with other obstetric factors such as UI during pregnancy [7, 8, 46] or with older maternal age at first delivery [13]. These factors were not available in the dataset for analysis. However, as previously stated, the BACH results suggest that the effect of type of delivery diminishes with age as also reported in another cohort study [16].

Data were not collected on the reason(s) for cesarean deliveries. Therefore, we were not able to address the inconsistent findings in other studies regarding a difference in risk of UI in the case of elective cesarean delivery versus cesarean delivery for obstructed labor. It is possible that structural and functional damage to the pelvis resulting from a difficult or prolonged labor makes the risk of incontinence for cesarean deliveries in these circumstances similar to that of vaginal delivery. Finally, investigation of the obstetrical risk factors for types of UI would have provided useful information. However, even in this large sample of women, the number of participants with pure stress UI symptoms and pure urge UI symptoms was insufficient for this analysis. The race/ethnic diversity of the sample likely contributes, in particular, to the lower rate of pure stress UI symptoms, as this type of UI is found most frequently in white women [4750].

The BACH survey offers numerous advantages for these analyses, particularly the large community-based, not clinical, sample of women from three race/ethnic groups and across a broad age range. However, certain design limitations of the study must also be acknowledged. First, like any study of this type, participants are subject to recall bias with questions regarding their obstetrical history. Validation of obstetrical factors and history with birth registry information or medical records was beyond the scope of the study. Because some studies [5153] have suggested that women have difficulty recalling specific events during their labor, we limited obstetrical information to mode of delivery, parity, and birth weight. Second, the BACH Survey is a study of symptoms; therefore, we are not reporting on clinically ascertained cases of UI. However, symptoms are important, as they influence quality of life and drive health care utilization.

Finally, the completion rate of this epidemiologic survey could be considered modest. However, the rate of 63% (percent of screener-identified eligible individuals who were interviewed) is respectable in light of multiple challenges to field studies, including the following: historical decline in survey response rates in general; difficulty obtaining telephone numbers, given increased use of cell phones replacing conventional landlines; the introduction of “do not call” legislation and increased use of caller ID services; suspicion of older and inner-city populations, especially minorities, toward research; more specific to the BACH Survey, a burdensome and possibly inconvenient protocol that includes questions perceived as personal or sensitive; and urologic health being a “hard sell”, especially among young adults with no apparent disease. In addition, the BACH cooperation rate of 63% is similar to that of the 2004 Behavioral Risk Factor Surveillance System (BRFSS) rate for Massachusetts (60%) [54]. When the BACH sample data are compared to samples from three national surveys [the National Health and Nutrition Examination Survey (NHANES), the National Health Interview Survey (NHIS) and the BRFSS] on sociodemographic and health-related variables including level of education, marital status, employment status, history of diabetes, hypertension, asthma, smoking status, and general health status, most of the BACH estimates are comparable. This supports the generalizability of these findings to the broader population of women in the US.

In conclusion, this large-scale epidemiological study identified vaginal delivery as the one obstetrical factor consistently contributing to lifetime risk of UI. Of further importance, this risk was observed overall for a sample consisting of black, Hispanic, and white women. An increased risk of UI with vaginal delivery should not be used to justify elective cesarean deliveries. The results of this study underscore the importance of clinical studies to identify underlying mechanisms that are amenable to clinical intervention and can inform the development of clinical guidelines for care.

Acknowledgment

Funding for the BACH Survey was provided by NIDDK (NIH) DK 56842.

Copyright information

© International Urogynecology Journal 2007