Association of epidural analgesia during labor and early postpartum urinary incontinence among women delivered vaginally: a propensity score matched retrospective cohort study

This single-center retrospective cohort study was conducted at International Peace Maternity and Child Health Hospital (IPMCH) from December 2020 to February 2022. At 42–60 days postpartum, maternity appointment to the institutional Pelvic Floor Rehabilitation Center for pelvic floor related examinations. The physician will consult each woman if she suffers from symptoms related to pelvic floor disorders through a standardized questionnaire. Results were recorded into an electronic health record and matched to the maternal baseline data in the electronic medical record by a unique hospitalization number.

Only primiparous women who delivered vaginally were included to avoid confounding by number of deliveries as well as mode of delivery. The exclusion criteria were as follows: (1) Preterm births; (2) Twin births; (3) Postpartum visits beyond 42 to 60 days; (4) Without baseline data (e.g., height, weight, and labor summaries).

The protocol was approved by the Ethics Committee of IPMCH (GKLW-2023-024-01), and the requirement for individual consent was waived. The study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline.

Since the cost of epidural anesthesia was covered by medical insurance in Shanghai, the choice of epidural anesthesia was determined by the individual's wishes, not by socioeconomic status. Prior to the procedure, the obstetrician and anesthesiologist jointly assessed the basic condition of the parturient women to exclude contraindications. The timing of anesthesia was chosen when the cervix was dilated 3–4 cm, and the epidural catheter was implanted in the L2-L3 intervertebral space. The rate of infusion was adjusted as necessary to maintain the labor analgesic effect without causing motor blockage. Following that, professional nursing personnel will attend to the women and notify the obstetrician if labor progress is sluggish or the fetal heartbeat is weak, and the obstetrician will determine whether to continue the labor or intervene artificially.

The primary outcome was UI, which was defined by International Urogynecological Association (IUGA) and International Continence Society (ICS) as any involuntary urine leakage [22]. Women were further assessed by a physician using the International Consultation on Incontinence Questionnaire Short Form (ICIQ-UI SF) [23] if they self-reported symptoms of urine leaking after delivery. The type of UI was determined primarily by Sect. 6 of the questionnaire. SUI was diagnosed in participants who chose "leaks when you cough or sneeze"/"leaks while you are physically active/exercising" from the list of options. UUI was diagnosed in participants who chose "leaks before you can get to the toilet"/"leaks when you are asleep"/"leaks when you have finished urinating and are dressed " from the list of options. While both symptoms were present in the participants, MUI was diagnosed. This questionnaire is now available in Chinese, and its test validity and accuracy have been well validated [24].

Baseline characteristic variables included maternal age, prenatal BMI (calculated from early pregnancy weight and height), pregnancy weight gain ratio (pregnancy weight gain divided by weight at the early pregnancy), abortion history, complications (hypertensive disorders; gestational diabetes/pre-pregnancy diabetes; others, defined as anemia, impaired liver and kidney function, and abnormal thyroid function), gestation week, induction of labor (oxytocin, prostaglandin, and cervix balloon mechanical induction of labor), infant weight, and infant head circumference.

Delivery characteristic variables included first stage of labor (time from regular uterine contractions to cervix fully dilated), second stage of labor (time from cervix full dilated to complete delivery of fetus), presence of perineal lacerations, use of episiotomy, and use of instrumental birth.

The frequency (percentage) of categorical variables and the median (interquartile range) or mean (standard deviation) of continuous variables were used to report descriptive statistics. Propensity-matched scoring was applied to achieve the balance of baseline data in the exposed and control groups (i.e., minimal confounding). A multivariate logistic regression model was used to calculate the propensity score, with epidural anesthesia as the dependent variable. Age, prenatal BMI, pregnancy weight gain ratio, abortion history, complications, gestation week, infant weight, and infant head circumference were the covariates factors. The caliper width was set at 0.02 and the matching procedure was completed with a 1:1 ratio and no replacement (greedy matching method). The absolute standardized mean difference (SMD) was used to estimate the balance of baseline data between the two groups before and after matching. SMD value lower than 0.1 was considered a good balance. Distribution of propensity scores in the sFigure 1 of supplementary material.

In the matched cohort, odds ratio (OR) and their 95% confidence interval (CI) of outcomes were estimated for women with and without epidural analgesia use. Previous studies have reported stage of labor, perineal lacerations, episiotomy, and instrumental birth as potential risk factors for the development of postpartum UI. Considering that these factors could not be classified as baseline characteristics, we included them in a binary logistic regression model to obtain an adjusted ORs in the post-matched cohort.

Two sensitivity analyses were undertaken. First, considering that age, prenatal BMI, pregnancy weight gain ratio, infant weight and infant head circumference related to postpartum UI, and that epidural anesthesia prevalence vary according to the rate of induction of labor, we stratified participants according to age (< 35 years, ≥ 35 years), prenatal BMI (< 25 kg/m2, ≥ 25 kg/m2), pregnancy weight gain ratio (< 20%, ≥ 20%), infant weight (< 4000 g, ≥ 4000 g), infant head circumference (< 34 cm, ≥ 34 cm), and induction of labor (no, yes). Second, considering possible confounding causality, women with UI during pregnancy may already have abnormal pelvic floor function and are more likely to report symptoms of UI after delivery. We analyzed whether the association would change if only individuals who had UI at postpartum were selected.

Statistical and graphing software were done with R version 4.1.3. All statistics were two-sided tests, and P < 0.05 was considered statistically significant.

A total of 13 627 women were delivered in the IPMCH from December 2020 to February 2022, and 8 437 women were excluded as follows: 7 920 were delivered via cesarean section or/and parity ≥ 2; 220 were preterm delivery; 3 were twin births; 214 had a visits beyond 42 to 60 days postpartum; and 80 had missing baseline data. The final 5 190 participants were enrolled in the study, of which 3 709 had epidural anesthesia at delivery. The propensity score-matched cohort included 2 894 primiparous women, 1 447 each in the epidural and non-epidural groups (Fig. 1).

The former part of Table 1 describes the baseline characteristics of participants. Before propensity matching, women who underwent epidural analgesia had a higher gestation week (39 [38, 40] vs. 39 [38, 40], P < 0.001, SMD = 0.205), a higher infant weight (3317.00 [343.39] vs. 3222.13 [349.62], P < 0.001, SMD = 0.274), and a higher incidence of induced labor (1614 [43.5] vs. 393 [26.5], P < 0.001, SMD = 0.362). After propensity matching, a total of 1 447 individual pairs of study participants were matched and successfully entered the final cohort, with well-balanced baseline data (SMD less than 0.1).

The latter part of Table 1 describes the delivery characteristics of the participants. In the propensity-matched cohort, women with epidural anesthesia during labor had a longer first (450 [330, 600] vs. 240 [180, 360], P < 0.001) and second (45 [30, 68] vs. 32 [20, 55], P < 0.001) stages of labor. The rate of perineal laceration was higher in the epidural group compared to the non-epidural group (2258 [73.6] vs. 2119 [69.1], P = 0.005), but no statistically significant differences were found in the incidence of instrument birth (194 [13.4] vs. 194 [13.4], P = 1.000) and episiotomy (365 [25.2] vs. 412 [28.5], P = 0.054) between the two groups.

In the propensity-matched cohort, the incidence of UI was higher in the epidural analgesia group than in the no epidural analgesia group (440 [30.4] vs. 326 [22.5]; adjusted OR 1.50, 95% CI 1.24–1.81). Further studies of each subtype of UI found similar results in SUI (307 [21.2] vs. 229 [15.8]; adjusted OR, 1.38 [95% CI, 1.12–1.71]). The difference of the incidence of UUI (80 [5.5] vs. 58 [4.0]; adjusted OR, 1.45 [95% CI, 0.99–2.15]) and MUI (53 [3.7] vs. 39 [2.7]; adjusted OR, 1.52 [95% CI, 0.95–2.45]) between the two groups was not statistically significant (Table 2).

For each subgroup, those who received epidural analgesia were matched with those who did not before the sensitivity analysis. The propensity score-matched group had well-balanced covariates with no statistically significant differences. We observed significantly interactions between age (P for interaction < 0.001) and infant weight (P for interaction = 0.006) on the relationship between epidural analgesia and early postpartum SUI. The association between the epidural analgesia and early postpartum SUI was more pronounced among older women (> 35 y) and women with macrosomia (infant weight ≥ 4000 g), compared with their counterparts (Fig. 2).

We reanalyzed our data after excluding participants who had UI during pregnancy. The levels of ORs for UI and different subtypes of UI between the two groups were comparable to the overall participants (Fig. 3).

This study had a large sample size, but it still had several limitations. First, the current study is a retrospective study, which has an inherent selection bias and some important variables (e.g., the intensity of postpartum physical activity, and breastfeeding or not in the postpartum period) cannot be collected. Second, urodynamic parameters such as bladder urine residue, urinary flow rate, and other indicators were not acquired. Due to the considerable medical costs involved, these were not used as standard postpartum examinations. Despite the presence of professionals on hand to guide patients, the use of questionnaire scales to determine the diagnosis of urinary incontinence and its type is still subjective, especially since a lot of times there is an under-reporting of the condition due to inherent shame by patients. Third, this study was limited to the early postpartum period, and further investigation is needed regarding changes in the incidence of UI over time and the association between early and persistent UI. Finally, almost all of the participants in this study were residents of Shanghai, China, an economically more developed region of China. Thus, our results may not be generalizable to all Chinese people. Further studies need to be conducted to replicate our findings in a nationally representative sample.