We conducted a retrospective cohort study of women who gave birth between July 2018 and May 2019 at the Kameda Medical Center, Chiba Prefecture, Japan. An average of 600 deliveries is managed at our medical center annually. This study received the institutional ethical review board approval (approval ID, 19–052).
All applicants were informed and enrolled by mail, which included the informed consent paper. Hence, informed consent was received from all participants online before initiation of the study.
PFD was evaluated using the PFDI-20 questionnaire. The PFDI-20 consists of 20 items of condition-specific measures of pelvic symptoms. It has three subscales: Pelvic Organ Prolapse Distress Inventory with six items (POPDI-6), Colorectal-Anal Distress Inventory with eight items (CRADI-8), and Urinary Distress Inventory with six items (UDI-6).
Each question relates to the presence of an individual symptom. If the symptom is present, the responder scores it on a 4-point scale to indicate the extent of botheration; score 1 is not at all, and 4 is quite a bit.
POPDI-6, CRADI-8, and UDI-6 are each scored similarly. First, subscale scores for each scale are obtained by taking the mean value of all items answered within each subscale. These subscale scores are determined by multiplying each scale’s mean score by 25, which gives each subscale score a range of 0–100. Therefore, the total scale score is calculated by summing the subscale scores (a range of 0–300). A high score indicates more symptoms and higher levels of inconvenience by the symptoms. The questionnaire was based on the validated Japanese language version of the Pelvic Floor Distress Inventory-20 (PFDI-20).10 The questionnaire was put into a web form without any modification from the original form.
Participants were sent a questionnaire along with an explanatory document about PFDI-20, 6–15 months after delivery. The inclusion criteria for the study were as follows: non-preterm delivery, single pregnancies, and fluency in Japanese. The exclusion criteria for the study were as follows: women who either did not respond or responded incompletely to the questionnaire. In this study, no sample size calculation was performed a priori, and all available pregnant women were included during the study period.
We divided all women into the following groups: C-section, epidural (vaginal delivery with epidural anesthesia), and non-epidural (without epidural anesthesia).
The PFDI-20 questionnaire was answered at 6–15 months postpartum, reflecting the state during that period. A “yes” on the questionnaire represented a positive result, indicating the occurrence of PFD. Study IDs were assigned to each woman, and their medical information was obtained from the medical records.
We obtained age, parity, maternal body weight before the pregnancy, mode of delivery, and the result of delivery mode. Continuous variables with normal distribution were presented as mean, median values. We compared the maternal, fetal, and neonatal characteristics and PFDI-20 (POPDI, CRADI, and UDI) in each group. Continuous variables were compared using the t-test and categorical variables using the Chi-square test. A linear regression model for PFDI-20 (POPDI, CRADI, and UDI) was constructed, adjusting confounding factors (age, parity, and pre-pregnancy body weight). All confounding factors were referred to from the findings of previous study. The causal directed acyclic graphs (DAGs) was used to identify the potential interplay among variables.11 The supplementary figure shows our DAGs (Supplemental Data 1).
It is said that maternal age and pre-pregnancy body weight influence mode of delivery due to maternal obstetrical complications.12 13 Moreover, previous delivery mode affects the choice of delivery mode in future pregnancies; for example, the rate of vaginal birth after cesarean seemed to be reduced by 10–30% approximately in Japan.14 Additionally, these factors influence the PFD;15 therefore, we consider these factors as confounding factors. Statistical significance was defined as P < 0.05, and all statistical analysis was conducted by R version 3.6.3 22/07/2011 23:54:00 (R Foundation for Statistical Computing, Vienna, Austria).
We assumed that any missing data was missing completely at random; hence, we conducted all analyses using the available data.