OBJECTIVE: To describe the prevalence and severity of urinary incontinence in the 12-month postpartum period and to relate this incontinence to several potential risk factors including body mass index, smoking, oral contraceptives, breast-feeding, and pelvic floor muscle exercise.

METHODS: Participants were 523 women, aged 14 to 42 years, who had obstetrical deliveries. The women were interviewed in their rooms on postpartum day 2 or 3 and by telephone 6 weeks, 3 months, 6 months, and 12 months postpartum. Chart abstraction was conducted to obtain obstetrical data from the index delivery.

RESULTS: At 6 weeks postpartum, 11.36% of women reported some degree of urinary incontinence since the index delivery. Although the rate of incontinence did not change significantly over the postpartum year, frequency of accidents decreased over time.

In the generalized estimating equation, postpartum incontinence was significantly associated with seven variables: baseline report of smoking (odds ratio [OR] 2.934; P = .002), incontinence during pregnancy (OR 2.002; P = .007), length of breast-feeding (OR 1.169; P = .023), vaginal delivery (OR 2.360; P = .002), use of forceps (OR 1.870; P = .024), and two time-varying covariates: frequency of urination (OR 1.123; P =

< .001) and body mass index (OR 1.055; P = .005). Factors not associated with postpartum incontinence included age, race, education, episiotomy, number of vaginal deliveries, attendance at childbirth preparation classes, and performing pelvic floor muscle exercises during the postpartum period.

CONCLUSION: Postpartum incontinence is associated with several risk factors, some of which are potentially modifiable and others that can help target at-risk women for early intervention.

Urinary incontinence is a common occurrence during pregnancy, affecting 30% to 60% of women. Most often, it is a transient condition attributed to the enlarged uterus, fluctuating hormone levels, increased glomerular filtration rates, temporary changes in the urethrovesical angle, or other normal changes of pregnancy. Incontinence usually resolves within the first 3 months after delivery, but in some women, the problem persists after delivery or develops as a de novo problem. Unlike antenatal incontinence, postpartum incontinence is typically attributed to pathophysiological changes that occur as a result of delivery, such as bladder trauma, nerve or muscle injury, or damage to the urethra and its suspension.

Studies of postpartum incontinence report rates ranging from 6% (de novo incontinence) to 34.3%. Outcomes described less than 3 months postpartum may still be reflective of the acute injury sustained at delivery, but those with incontinence 3 months after a first delivery have an increased risk of long-term symptoms.

Previous research on the development of postpartum incontinence suggests a multifactorial etiology. Clinical and epidemiological studies generally agree that women who undergo vaginal delivery are at greater risk of incontinence than are women who are delivered by cesarean, presumably as a result of the detrimental impact on the pelvic floor. Further, it appears that the first vaginal delivery is when most women are likely to sustain this damage. Supporting these findings are several studies that have shown relationships between vaginal delivery and damage to pelvic floor innervation, as manifested by decreased pelvic floor muscle strength, changes in bladder neck position and mobility, and anal sphincter disruption.

Other factors that have been implicated in the etiology of incontinence are fetal factors, operative vaginal deliveries, prolonged second stage labor, and anesthesia. Data on these variables are mixed, with some reporting a clear relationship and others demonstrating no effect. Therefore, their etiological role is unclear and their impact may be transient. In addition, there is evidence for other risk factors, such as preconception incontinence and antenatal bladder neck mobility, indicating the possibility of inherent pelvic floor muscle or connective tissue weakness in the development of urinary incontinence.

The purpose of this study was to describe the prevalence and severity of self-reported urinary incontinence in the 12-month postpartum period and to identify the predictors of this incontinence. In addition to the commonly investigated obstetric factors, other potential predisposing factors were examined, including age, race, body mass index, parity, smoking, oral contraceptives, and breast-feeding.

Although childbirth preparation classes and pelvic floor muscle exercises are often recommended to help prevent incontinence and other pelvic floor problems in the postpartum period, little is known of their effects. Therefore, this study also examined whether women attended childbirth preparation classes and whether they practiced pelvic floor muscle exercises during pregnancy and at four time points throughout the 12-month postpartum period. Pelvic floor muscle exercise was included as a time varying covariate in a multiple-variable analysis to explore the possible protective effect on postpartum continence status.

MATERIALS AND METHODS
Participants were a convenience sample of 523 women who had obstetric deliveries between February 26, 1990, and March 1, 1991. Participants were contacted in their hospital rooms on postpartum day 2 or 3. The interviewer (JLL) rotated her visits among the three maternity wings of the hospital, approaching women who were alone in their rooms (ie, without hospital staff or visitors). The study was explained and patients who agreed to participate signed a consent form approved by the Magee Womens Hospital Institutional Review Board.

On postpartum day 2 or 3, an interviewer obtained informed consent and administered a structured questionnaire that used a standard interview format. The information acquired included demographics, obstetric and gynecologic history, urinary and fecal incontinence history, current medication use, history of urinary tract infections, and questions about bladder habits, smoking habits, presence of a chronic cough, height, and attendance at prenatal and child birth preparation classes.

When obtaining the continence history, the interviewer asked, “Have you ever experienced any difficulty controlling urination? Have you ever had any accidental loss of urine, even a small amount?” Because many women use different language to describe incontinence, a negative response was followed by rephrasing the question, “Have you ever wet yourself?” Those who acknowledged any incontinence were asked several questions about onset, course, when it occurred, under what circumstances, and the frequency and volume of urine loss.

Follow-up telephone interviews were conducted by trained interviewers at 6 weeks, 3 months, 6 months, and 12 months postpartum. These telephone interviews used standard questionnaires to collect information about the period of time since the last contact. At these time points, participants were queried about their continence status, including questions about frequency, volume, and circumstances of urine loss, as well as how disturbing the incontinence was and how much it restricted their activities. Other questions addressed frequency of urination (daytime and nighttime), use of oral contraception, breast-feeding, employment status, smoking, coughing, urinary tract infections, current weight, and feelings of depression.

At the 6-week follow-up, participants were asked a series of questions about their knowledge and use of pelvic floor muscle exercises, including whether they had ever learned them, how they were taught, and whether they did them during and after their pregnancy. At later contacts, they were asked if they were currently doing their exercises, how often, and for how long. At 12 months, they were asked if they were pregnant again, and if so, for how long.

Chart reviews were conducted to extract information on the patient’s labor and delivery course: anesthesia, length of second-stage labor, type of delivery (vaginal versus cesarean delivery), episiotomy, use of forceps, infant’s head circumference, and vaginal and anal sphincter lacerations.

Descriptive statistics were used to characterize the participants and their continence status at 6 weeks, 3 months, 6 months, and 12 months postpartum. Data for women who had delivered again or were pregnant again at the 12-month follow-up were excluded from the analysis. The primary objective of the analysis was to determine the risk factors associated with urinary incontinence over four time points within the postpartum year. The longitudinal nature of the data involves correlation among the outcomes at 6 weeks, 3 months, 6 months, and 12 months postpartum. Because of this correlation among the outcomes, logistic regression could not be used to examine for associations between urinary incontinence and predictor variables. To account for this correlation, generalized estimating equations were used.36

Similar to a repeated-measures analysis of variance for continuous outcomes, the dependent variable for the generalized estimating equation (GEE) analysis is an individual’s urinary incontinence status observed repeatedly at four separate time points. The longitudinal measures are analyzed collectively by accounting for the correlations among longitudinal outcomes for an individual. Furthermore, analyzing the data longitudinally allows for interpretation of the risk factors after adjusting for the effect of time. The benefits of the longitudinal analysis over four cross-sectional analyses (separate analyses for 0.5, 3, 6, and 12 months) are that it protects against inflated type I error rates, and it accounts for correlation among outcomes. Furthermore, the interpretation of the odds ratio (OR) is the ratio of the odds of being incontinent for individuals with the risk factor relative to the odds of being incontinent for individuals without the risk factor over the first year postpartum.

To identify potential predictors, GEE models predicting urinary incontinence longitudinally were built using the time variable and the predictor under investigation. If the P value of the predictor was less than .15, the predictor was retained for investigation in a multivariable model. Potential predictors included demographic variables, obstetric variables, fetal factors, history of smoking and coughing, parity, nocturia, fecal incontinence, childbirth preparation classes, and breast-feeding. Time-varying covariates included frequency of urination, oral contraceptives, current smoking, body mass index, and pelvic floor muscle exercises. Potential interaction between time since delivery and the risk factor were carefully examined. However, none of these interactions proved to be significant.

The maximum model under investigation included all variables identified by GEE analyses assessing one risk factor at a time. Backward elimination was used to remove variables that did not contribute significantly in the prediction of urinary incontinence in the presence of the other predictor variables. The backward elimination procedure stopped when all predictors achieved an alpha level of .05 significance.

RESULTS
Participants ranged in age from 14 to 42 years (mean, 28.6 years; median, 29.0). The majority (82.4%) were white, 16.1% were black, and 1.2% were Hispanic or Asian. Fifty-eight percent had completed high school or technical/vocational school, and 36% had completed college or an advanced degree. Parity ranged from 1 to 6 births (mean, 1.9; median, 2.0), including the current delivery, and 41.5% of patients were primiparous.

Continence status before and during pregnancy has been reported previously and is summarized here. Before the index pregnancy, 132 women (25.2%) had experienced some degree of incontinence. Most of these (80.0%) said that the incontinence was associated with a previous pregnancy and/or childbirth (20.8%), and in 86.1%, it was with their first child. Of the 523 women, 311 (59.5%) reported involuntary loss of urine during the index pregnancy.

Of the original 523 women interviewed, 493 (94.3%) were reinterviewed 6 weeks postpartum, 483 (92.4%) at 3 months, 460 (88.0%) at 6 months, and 426 (81.5%) at 12 months. One woman had delivered another child, and 42 women were pregnant again at the time of the 12-month follow-up. Data from the time points affected by these pregnancies were excluded from subsequent analysis.

The prevalence and characteristics of incontinence in the 12-month postpartum period are presented in Table 1.The prevalence of incontinence remained approximately the same across the 12-month period. Examination of the dropouts suggests that the small rise in the rate of incontinence at the 12-month interview was the result of selective attrition. Although the rate of incontinence did not change, the frequency of accidents decreased from 15.9 per week at the 6-week interview to 7.9 at 3 months, 8.6 at 6 months, and 4.8 at 12 months. Similarly, there was a small reduction over time in the percentage of respondents who reported that incontinence was disturbing or restricted their activities.

DISCUSSION
The longitudinal pattern of incontinence in this study was similar to that of most previous studies of incontinence during parturition, which demonstrate a high rate of incontinence during pregnancy, followed by resolution and lower rates in the postpartum period. After a 59.6% rate of incontinence during pregnancy, the proportion of women reporting incontinence dropped to 11.4% at 6 weeks after delivery and remained essentially stable throughout the 12-month postpartum period.

Although the proportion of women reporting urinary incontinence did not change significantly over the course of the next year, the frequency of accidents dropped from a mean 15.9 per week to about half this amount at the 3- and 6-month follow-up and to about 5 per week at the 12-month follow-up. Similarly, the volume of accidents appeared to decease somewhat over time, as did the restriction of activities. Thus, urinary incontinence continued to affect about 10% of women, although its severity gradually dissipated over the 12-month period.

The results of this study indicate that undergoing vaginal delivery more than doubled a woman’s risk of experiencing postpartum incontinence. These findings contribute to the growing body of evidence that implicates vaginal delivery as a factor in the development of urinary incontinence and other pelvic floor disorders. The only other fetal or obstetric factor that was associated with incontinence was the use of forceps, which nearly doubled the risk of incontinence in this study. This association has been reported previously, whereas others have found no significant relationship. Postpartum urinary incontinence was not related to other obstetric or fetal factors including anesthesia, length of second stage labor, laceration, head circumference, or birth weight of the heaviest neonate.

The strongest predictor of postpartum incontinence was self-reported frequency of urination. Interpretation of this finding is compromised by the inability to determine the direction of the effect. It is possible that women develop the habit of frequent voiding during pregnancy and that this habit persists in the postpartum period, contributing to reduced bladder capacity and urge incontinence. It is equally possible that frequent urination is a concomitant rather than a risk factor for incontinence, or a behavior adopted in response to incontinence to avoid leakage episodes.

Smoking was another significant predictor of incontinence in the postpartum period. Previous literature provides some evidence for a link between smoking and incontinence, although results have been mixed. It has been suggested that smoking may increase the risk of incontinence by causing chronic coughing. An interesting association found between incontinence and breast-feeding may demonstrate a transient effect attributable to hormonal changes in the postpartum period.

Obesity is an established risk factor for incontinence in the epidemiological literature. In this study, the data on body mass index suggest that being overweight is a factor in the postpartum period as well. It is hypothesized that excess weight places extra pressure on the pelvic floor, compromising the outlet.

It has been stated that incontinence rarely develops in the puerperium. In fact, much of postpartum incontinence actually begins during pregnancy and persists after delivery. Although a less common event, there were 26 cases in this study in which women were continent during pregnancy and developed de novo incontinence after delivery. Indeed, incontinence during the index pregnancy was a significant predictor of postpartum incontinence, which is consistent with previous studies. Furthermore, incontinence during pregnancy is not only related to postpartum incontinence, but has been reported as a risk factor for incontinence at 5 years past delivery. Also, preconception incontinence predicts postpartum incontinence. In light of the fact that antepartum incontinence and postpartum incontinence are thought to have different mechanisms, these findings highlight the possibility of intrinsic factors in the etiology of both.

The finding that antepartum incontinence is a predictor of postpartum incontinence also opens an opportunity to identify women who experience incontinence during pregnancy, to counsel them about their increased risk and offer them preventative strategies.

Approximately three-fourths of these new mothers had learned to do pelvic floor muscle exercises, and about half reported doing them during pregnancy, although fewer persisted into the postpartum period. In the analyses, pelvic floor exercises during pregnancy or the postpartum period did not reduce the occurrence of urinary incontinence. In the subanalysis of multiparous women, pelvic floor muscle exercise was actually associated with an increased risk of urinary incontinence in the postpartum period. Although this study was prospective, it was not randomized, and it is not possible to determine the direction of this effect. After this study was conducted, randomized trials were reported in which women randomized to pelvic floor muscle exercise during pregnancy had lower rates of incontinence in late pregnancy and the postpartum period. Other intervention studies have documented the value of postpartum pelvic floor muscle exercises for reducing incontinence up to 12 months later. It is most likely that the association between incontinence and pelvic floor muscle exercises in the present study occurred because incontinent women adopt exercises as a management strategy to control their incontinence.

In conclusion, this study and others have identified a number of risk factors for incontinence in the postpartum period. In addition to this, there is evidence that incontinence in the postpartum period is highly predictive of incontinence beyond the postpartum period. This suggests that the factors influencing postpartum incontinence, especially if they are intrinsic factors, have implications for a woman’s lifetime continence status. Although pelvic floor muscle exercises and other conservative measures are often recommended to pregnant women, they are usually not taught in a manner intensive enough to have significant benefit. Given the evidence of their beneficial effects, it would seem reasonable to systematically offer these conservative interventions to women identified as being at risk, and implement them fully so that they may have maximal benefit for women who need them most.

Supported in part by Research Career Development Award K04 AG00431 from the National Institute on Aging.

doi:10.1016/j.obstetgynecol.2003.09.013

Received May 23, 2003. Received in revised form August 22, 2003. Accepted September 4, 2003.

REFERENCES

_{1. Francis WJA. The onset of stress incontinence. J Obstet Gynaecol Br Commonw 1960;67:899–903.}

2. Iosif S. Stress incontinence during pregnancy and in puerperium. Int J Gynaecol Obstet 1981a;19:13–20.

3. Sleep J, Grant A. Pelvic floor exercises in postnatal care. Midwifery 1987;3:158–64.

4. Viktrup L, Lose G, Rolff M, Barfoed K. The symptom of stress incontinence caused by pregnancy or delivery in primiparas. Obstet Gynecol 1992;79:945–9.

5. Dimpfl TH, Hesse U, Schussler B. Incidence and cause of postpartum urinary stress incontinence. Eur J Obstet Gynecol Reprod Biol 1992;43:29–33.

6. Viktrup L, Lose G, Rolff M, Barfoed K. The frequency of urinary symptoms during pregnancy and puerperium in the primipara. Int Urogynecol J 1993;4:27–30.

7. Burgio KL, Locher JL, Zyczynski H, Hardin JM, Singh K. Urinary incontinence during pregnancy in a racially mixed sample: Characteristics and predisposing factors. Int Urogynecol J 1996;7:69–73.

8. Mason L, Glenn S, Walton I, Appleton C. The prevalence of stress incontinence during pregnancy and following delivery. Midwifery 1998;120–9.

9. Meyer S, Schreyer A, DeGrandi P, Hohlfeld P. The effects of birth on urinary continence mechanisms and other pelvic-floor characteristics. Obstet Gynecol 1998;92: 613–8.

10. Chaliha C, Kalia V, Stanton SL, Monga A, Sultan AH. Antenatal prediction of postpartum urinary and fecal incontinence. Obstet Gynecol 1999;94:689–94.

Kathryn L. Burgio, PhD, Halina Zyczynski, MD, Julie L. Locher, PhD, Holly E. Richter, PhD, MD, David T. Redden, PhD and Kate Clark Wright

From the Birmingham/Atlanta Geriatric Research, Education, and Clinical Center, Department of Veterans Affairs Medical Center, and School of Medicine, Center for Aging, and School of Public Health, University of Alabama at Birmingham, Birmingham, Alabama; and University of Pittsburgh, School of Medicine, Pittsburgh, Pennsylvania.