Journal Pre-proof The influence of pregnancy, parity, and mode of delivery on urinary incontinence and prolapse surgery – a national register study

The long-term risk of urogenital reconstructive surgery in women with cesarean deliveries 56 only was on par with that in nulliparous women. BACKGROUND: The long-term effects of vaginal delivery, parity, and pregnancy on the pelvic floor remain uncertain and controversial issues. In comparison with studies using self- reported symptoms, surgical register data may offer a more valid mean for evaluating the 85 relative influence of these risk factors. This study used data from three high-quality nationwide registers, the Swedish National Quality Register of Gynecological Surgery, the Swedish Medical Birth Register, and the Total population register, to evaluate the contribution of vaginal and cesarean delivery, parity, and factors not related to childbirth on the long-term risk of reconstructive urogenital surgery. deviation; IQR interquartile range; CI confidence interval; BMI body mass incontinence; pelvic prolapse; a Mann-Whitney U test was used for between groups of continuous variables and Fisher´s


Introduction 129
Pelvic organ prolapse (POP) and stress urinary incontinence (SUI) are common female pelvic 130 floor disorders (PFDs), which have a significant impact on well-being and quality of life and 131 cause both personal suffering and costs to society. [1][2][3] Approximately every fourth woman 132 reports one or more PFD. 4 Estimates for the lifetime risk of pelvic floor surgery is currently 133 up to one in five women, 5-8 but the actual need may be even higher. 9 At present, the 134 preponderance of evidence from epidemiologic and imaging studies suggests that vaginal 135 delivery (VD) is the leading cause of SUI and POP, 10-13, including the more surgical 136 demanding forms of SUI and POP. 14 Whether cesarean deliveries can avoid PFDs later in life 137 is controversial; two studies have indicated that the protective effect of cesarean delivery on 138 PFDs diminishes over time and even disappears after multiple deliveries. 15,16 139 Most prior epidemiologic studies on POP and UI are based on patient reporting, which relies 140 on women's subjective perception and willingness to report. Another way to assess the impact 141 of pregnancy and childbirth is to study the epidemiology of surgical treatment of PFDs as 142 reconstructive surgery presupposes an objectively confirmed condition and bothersome 143 symptoms affecting quality of life. Nowadays, women in many countries have access to 144 urogenital surgical treatment thanks to state-sponsored public health care systems that make it 145 available at an affordable cost. Those who qualify for surgical treatment may offer a suitable 146 measure to evaluate risk factors for PFDs. 147 There are still crucial and unresolved issues regarding the relative impact of pregnancy, 148 vaginal delivery, parity, and factors unrelated to childbirth on the long-term risk of PFDs. It is 149 of particular interest and yet unknown whether the hormonally triggered changes during 150 pregnancy are temporary and reversible during involution or if they persist, becoming 151 aggravated with time. The aim of this study was, therefore, to use data from three high-152 quality national registers, the Swedish National Quality Register of Gynecological Surgery 153 (GynOp) 17 performed annually under local, regional, or general anesthesia. All women planned for 168 surgery received information about GynOp and the possibility of declining participation or 169 opt-out at any time. Data were registered consecutively, including a preoperative evaluation 170 (postal-or web-based questionnaires), hospital records from admission, surgery, and 171 discharge, and a questionnaire 1-year postoperatively. 6% of all women declined participation 172 or returned a blank or unusable form (2019). 17 GynOp was started in 1997 and was intended All women ≥45 years with SUI or POP surgery in 2010-2017 were eligible for the study 180 (n=59,415). The three study cohorts were designed to measure the risk of surgery after 181 vaginal delivery, after pregnancy separate from vaginal delivery, and the sum effect of factors 182 unrelated to childbirth (Supplemental Figure S1). Accordingly, the total study cohort was 183 stratified into 1. nulliparous women, 2. women with first and all subsequent deliveries by 184 cesarean delivery, and 3. women with ≥1 VDs, regardless of additional cesarean deliveries 185 ( Figure 1A, Supplemental Table S1). The number of births (0, 1, 2, 3, ≥4) was registered for 186 all women ( Figure 1B). Information about parity in GynOp was cross-checked with the 187 Baseline characteristics were almost similar between the parous groups. Nulliparous women 222 were older than the parous groups, and in addition, among those with SUI surgery, also more 223 often hysterectomized, taking estrogen, and reported preoperative urge UI more often (Table  224 1A+B  Figure 3). 244 The effect of parity on the AR of surgery differed markedly between the VD and the CD 245 group. With nullipara as a reference, the cumulative AR of surgery increased consistently in 246 women with ≥1VDs for both POP and SUI surgery (Table 3; Figure 3). The AR of POP 247 surgery was 2.3 per 1000 women (95% CI 2.2-2.5) in the nulliparous group and increased 17-248 fold to 38.7 per 1000 women (95% CI 37.8-39.8) with four or more VDs. The increase in the 249 AR of SUI surgery was consistent but less pronounced in women with VDs from AR 2.7 per 250 1000 women (95%CI 2.6-2.9) in nullipara to 14.5 per 1000 women (95%CI, 13.9-15.1) after 251 four or more births (5-fold) ( Table 3). There were significant trends in the AR of surgery 252 according to parity in the cesarean delivery group (a positive trend for SUI and a negative 253 trend for POP surgery, both P<.0001) (Table 3; Figure 3), but the effect was small and on par with that in the nulliparous women. The first vaginal birth carried the highest risk increase for 255 both POP (AR difference 11.5 per 1000 women [95%CI 11.1-11.9, P<0.0001]) and SUI 256 surgery (AR difference 6.2 per 1000 women [95%CI 5.8-6.5, P<0.0001]), higher than for all 257 subsequent births (Table 3;  This pattern was more pronounced for SUI surgery (2 nd birth added 10% of that in the first 261 birth) than for POP surgery (2 nd birth added 38% of that in the first birth) ( Table 3). 262

Principal findings 266
Vaginal delivery was the sole dominant risk factor for subsequent urogenital surgery and was 267 overrepresented, whereas nulliparous women and those delivered exclusively by cesarean 268 delivery were underrepresented. After one or more pregnancies, in women delivered by 269 cesarean delivery, the AR of POP and SUI surgery was negligible and on par with that in 270 nulliparous women. An increasing number of vaginal births was associated with a cumulative 271 increased AR of surgery for POP and SUI. The first vaginal birth brought the largest, and the 272 second vaginal birth the smallest additive AR for POP and SUI surgery. 273

Results in context 274
The crucial contribution to this study was the inclusion of a large group of nulliparous women 275 who had undergone urogenital surgery and the possibility of identifying a corresponding 276 reference group of women at risk in the total female population aged ≥45 years. Known parity 277 and mode of delivery made it possible to calculate the relative and absolute risk of surgery for 278 all cohorts. In addition, a nulliparous cohort allowed an assessment of the combined effect of 279 factors unrelated to childbirth for evaluating the long-term effect of pregnancy, separate from 280 vaginal delivery, and to measure the additive AR of surgery at the first and all subsequent 281 births. 282 When exploring the epidemiology of PFDs, accessing a significantly large group of nulliparas 283 has been a significant problem. In a cross-sectional survey from Norway, only one out of 252 284 nulliparas was identified as having self-reported POP surgery among 1123 surgical 285 procedures. 24  with the increasing number of vaginal births indicates a dose-response relationship. However, 309 a monotonic biological gradient does not necessarily prove but suggests causality behind the 310 statistical associations. 31 A recent review stated that it remains controversial whether an 311 elective cesarean delivery is protective for the long-term effects of VD. 32 Our results 312 regarding the long-term effects of one or more pregnancies unequivocally showed that ≥1 313 cesarean delivery compared to ≥ 1 vaginal delivery was preventive for the cumulative excess 314 risk of POP and SUI surgery, which increased with the number of vaginal births. We could 315 not find any substantial difference in risk for prolapse and incontinence surgery between truly 316 nulliparous women and those having all their deliveries by acute or elective cesarean 317 deliveries. 318 A remaining unresolved issue concerns the risk of surgery for POP and SUI according to the 319 order of vaginal births. In this study, the first vaginal birth was associated with the largest risk 320 increase of the AR for both POP and SUI surgery, and there were subsequent risk increases 321 with subsequent deliveries. The second vaginal birth was associated with the lowest risk 322 increase of AR for both POP and SUI surgery, followed by a subsequent increase of the AR 323 describing a biphasic process. This does not harmonize with the linear increase found in 324 studies that did not use nulliparous women as reference for SUI, 28 POP, 29 and UI, and POP. 14 325 However, some previous population-and register-based studies presented prevalence data 326 demonstrating a strong effect of the first birth on the prevalence of UI, 33 SUI, 34, and POP, 35 as 327 well as an overall biphasic pattern with increasing vaginal parity, all of which used nulliparous women as reference. Introducing nulliparous women as baseline references for 329 these comparisons is pivotal and may explain why the results differ between studies. 330

Clinical implications 331
The long-term effect of one or more cesarean deliveries on the risk of reconstructive 332 urogenital surgery, representing the effect of one or more pregnancies, was similar to that in 333 nulliparous women who are unaffected by childbirth. This is crucial information for health 334 care economic calculations and women's autonomous decision regarding their preferred mode 335 of delivery. 336

Research implications 337
This study contributes to the Life Span Study approach to the epidemiology and health care

Strengths and limitations 347
The strengths of this study include the large cohorts based on prospective data from national 348 registers considered to be of high quality. It is a major strength to include a reference group of 349 nulliparous women to evaluate the long-term effect of pregnancy in the CD group. There was 350 no risk of recall bias concerning obstetric history as information on parity and mode of 351 delivery was cross-checked with national registers. Further, the Swedish public health care 352 system is state-sponsored, and medical assessment and surgery are available to all citizens at 353 an affordable cost, reducing the risk of bias due to socio-economic factors. 354 It may be considered a limitation that some obstetric information that may contribute to the    Note: SUI, stress urinary incontinence; POP, pelvic organ prolapse; y, years. a There were 4 12 women with uncertain parity, thus, the total number of women with known parity was 19,076. 13 b There were 11 women with uncertain parity, thus, the total number of women with known 14 parity was 38,717. c Women with both incontinence and prolapse procedures were included in 15 each treatment category. d Set according to the rate of nulliparous women aged ≥ 45 in 2017 16 in the TPR. e Based on women born in 1960 in the MBR. The rate between women with ≥1 17 CD and those with ≥1 VD was 7.7% versus 92.3%, and given the rate for nulliparous women, 18 the final distribution was 13.8%, 79.6%, and 6.6%. From these percentages, the number of 19 women in all cohorts according to mode of delivery and parity (0, 1, 2, 3, ≥4) was calculated 20 from the total number of women in 2017 (n=2,309,765) down to the singular.  births is fixed in each parity group, but the location may differ (e.g., ≥1 CD followed by 1 VD with ≥1subsequent CD = VD parity 1 with additional CD). 38,717 (11 missing), 19,080 (4 missing).