Hysterectomy is not associated with increased risk of urinary incontinence-a northern Finland birth cohort 1966 study.

Heini Salo: Project development, data analysis, manuscript writing. Roosa Manninen: Manuscript writing. Anna Terho: Project development, manuscript writing. Johanna Laru: Project development, manuscript writing. Henri Sova: Project development, manuscript writing. Sari Koivurova: Project development, manuscript writing. Henna‐Riikka Rossi: Project development, manuscript writing.

This study was conducted according to STROBE guidelines for cohort studies and the ethics committee of Northern Ostrobothnia hospital district has approved the NFBC1966 study (EETTMK: 94/2011) on December 14, 2011.

Government funding for research. The Finnish Medical Association, Maija and Matti Vaskio Foundation. NFBC1966 received financial support from University of Oulu Grant no. 65354 and 24 000 692.

Background characteristics are presented in Table  1 . Women with hysterectomy were less often nulliparous, had more often ≥3 deliveries, and had lower level of education (Table  1 ). Preoperative characteristics, concomitant operations, indications for hysterectomy, and different surgical approaches are shown in Table  2 . Among hysterectomized women, LH was the most common surgical approach ( n  = 247, 53.6%), followed by VH ( n  = 107, 23.2%) and AH ( n  = 107, 23.2%) (Table  2 ). POP was the main indication for VH (10.9% for all hysterectomies, 40.6% for VH). The incidences of UI and POP diagnoses preceding hysterectomy were more frequent in the VH group compared to other approaches (preceding UI: 15.0% in VH, 6.9% in LH, and 2.8% in AH, p  = 0.003; preceding POP: 53.3% in VH, 6.9% in LH, and 2.8% in AH, p  < 0.001). Concomitant adnexal operations were most frequently performed in the LH group (Table  2 ). Characteristics of the study population. Data on parity and mode of deliveries were collected from the Finnish Medical Birth Register, and other variables (BMI, climacteric symptoms, smoking status, and education) from the 46‐year follow‐up questionnaire. Note : Data are reported as count numbers (percentages) or mean (standard deviation). p ‐value <0.05 was considered statistically significant. Differences in numbers vary in analyses as the result of missing data. Abbreviation: BMI, body mass index. Preoperative patient characteristics. BMI was collected from the 46‐year follow‐up questionnaire, and other variables from the Care Register for Health Care (CRHC). n  = 396 (85.9%) 26.77 (4.79) n  = 94 (87.9%) 26.61 (4.40) n  = 211 (85.4%) 26.67 (4.48) n  = 91 (85.0%) 27.16 (5.21) n  = 461 (100%) 45 (5.29) n  = 107 (100%) 45 (4.99) n  = 247 (100%) 45 (5.31) n  = 107 (100%) 44 (5.46) n  = 461 (100%) 36 (7.8%) n  = 107 (100%) 16 (15.0%) n  = 247 (100%) 17 (6.9%) n  = 107 (100%) 3 (2.8%) n  = 461 (100%) 77 (16.7%) n  = 107 (100%) 57 (53.3%) n  = 247 (100%) 17 (6.9%) n  = 107 (100%) 3 (2.8%) n  = 461 (100%) 7 (1.5%) n  = 107 (100%) 3 (2.8%) n  = 247 (100%) 3 (1.2%) n  = 107 (100%) 1 (0.9%) n  = 461 (100%) 2 (0.4%) n  = 107 (100%) 0 (0.0%) n  = 247 (100%) 1 (0.4%) n  = 107 (100%) 1 (0.9%) n  = 461 (100%) 108 (23.4%) n  = 107 (100%) 3 (2.8%) n  = 247 (100%) 87 (35.2%) n  = 107 (100%) 18 (16.8%) Note : Data are reported as count numbers (percentages) or mean (standard deviation). p ‐value <0.05 was considered statistically significant. Differences in numbers vary in analyses as the result of missing data. Abbreviations: BMI, body mass index; POP, pelvic organ prolapse; UI, urinary incontinence. After excluding women with preceding UI ( n  = 36, 7.8%), the final hysterectomy group size for analyses was 425 (LH n  = 230, 54.1%; AH n  = 104, 24.5%, and VH n  = 91, 21.4%) (Figure  1 ). Mean follow‐up time after hysterectomy was 9 (±5.3) years. After the exclusion, there was no significant difference between the incidence of de novo UI in the hysterectomy group and the prevalence of UI in reference group ( n  = 24, 5.6% vs. n  = 166, 4.7%, p  = 0.416). Furthermore, regarding different UI subtypes, the incidence of SUI or other UI was comparable between the study groups (Table  3 ). The rate of UI operations did not differ between the hysterectomy and the reference groups ( n  = 14, 3.3% vs. n  = 87, 2.5%; p  = 0.323) (Table  3 ). The mean time from hysterectomy to incontinence diagnosis was 4.6 (±3.6) years, and to incontinence operation 6.0 (±3.9) years. The incidence of de novo urinary incontinence (UI) and UI operation for women with hysterectomy and the prevalence of UI and UI operation in women without hysterectomy. Note : Data reported as count numbers (%). p ‐value <0.05 was considered statistically significant. Abbreviations: SUI, stress urinary incontinence; UI, Urinary incontinence. When examining different hysterectomy approaches separately, the incidence of de novo UI diagnosis was more frequent in the VH group compared to other surgical approaches and prevalence of UI in reference group; VH 12.1% vs LH 3.9% ( p  = 0.006), AH 3.8% ( p  = 0.031), and reference group 4.7% ( p  = 0.001) (Figure  2 ). The incidence (%) of UI in women who have not undergone hysterectomy and de novo UI in different surgical approaches. p ‐values according to Independent‐Samples t ‐test. AH, abdominal hysterectomy; LH, laparoscopic hysterectomy; UI, urinary incontinence; VH, vaginal hysterectomy. In univariate regression analyses, hysterectomy did not associate with an increased risk of any subtype of UI (overall UI: OR 1.20, 95% CI 0.77–1.86; SUI: OR 1.51, 95% CI 0.89–2.55; and other UI: OR 0.80, 95% CI 0.36–1.74). However, the results of univariate analyses for de novo UI varied depending on the surgical approach. Compared to the reference group, the risk of UI, especially SUI, was significantly increased in the VH group (overall UI: OR 2.76, 95% CI 1.44–5.28 and SUI: OR 3.02, 95% CI 1.36–6.70), but not in the LH and AH groups (Table  4 ). Odds ratios (ORs) and their 95% confidence intervals (CIs) for overall urinary incontinence (UI), stress urinary incontinence (SUI), and other UI in women with or without hysterectomy. Note : Model 1: parity, BMI, smoking. Model 2: pelvic organ prolapse. Model 3: vaginal delivery. Model 4: education level. Model 5: all covariates. Statistically significant values ( p  < 0.05) are presented in bold. In multivariate analyses, the results were similar to univariate analysis after adjusting for parity, BMI, and smoking (model 1), vaginal delivery (model 3), and the level of education (model 4) (Table  4 ). After adjusting for preceding POP diagnosis (model 2), hysterectomy was associated with a significantly decreased risk for overall UI and UI other than SUI. Regarding different surgical approaches, when adjusting for preceding POP (model 2), the risk for any type of UI was not increased in any hysterectomy group. Instead, in the LH group, the risk of overall UI was decreased after adjusting for preceding POP. The hazard curves estimating the cumulative risk of UI in women after hysterectomy in different surgical approaches are shown in Figure  3 . Cumulative hazard curves for UI in women after hysterectomy in different surgical approaches. AH, abdominal hysterectomy; LH, laparoscopic hysterectomy; UI, urinary incontinence; VH, vaginal hysterectomy.

Our population‐based cohort study showed that hysterectomy is not an independent factor for increasing the postoperative incidence of UI or any UI subtype. Surprisingly, after eliminating the effect of preceding POP, hysterectomy was even associated with a decreased risk of UI. Furthermore, we found that hysterectomy does not increase the risk of subsequent UI operations. Regarding different surgical approaches, only VH was associated with an increased risk of de novo UI. However, also this association was explained by preceding POP. Our results are in accordance with previous literature. A recent study of 260 women hysterectomized by abdominal approach found that self‐reported urge UI as well as other bladder and urinary symptoms decreased in 3 years postoperatively. 4 A prospective Danish follow‐up study of 108 women showed that hysterectomy is not significantly associated with the risk of de novo UI, and the surgical approach (abdominal/vaginal) did not affect the risk of de novo UI. 15 Respectively, another Danish cohort study showed no significant difference in UI incidence after hysterectomy compared to diverse control groups (laparoscopic cholecystectomy and transcervical endometrial resection) during a 10‐year follow‐up. 16 In a Swedish cohort study in which women with preoperative UI symptoms were excluded, no difference in de novo UI symptoms after hysterectomy was observed compared to reference group during a 2‐year follow‐up. 21 A substantial proportion of our study population suffered from UI symptoms prior to hysterectomy, especially in the VH group. This is in line with a previous study of 16 182 hysterectomies with a 1‐year postoperative follow‐up, which showed that almost a third of women undergoing hysterectomy have UI preoperatively. In their study, 13.3% reported remission of UI symptoms after hysterectomy, and the remission rate was increased in women who had POP as an indication of hysterectomy. 17 In our study, we could not assess UI remission rate, unfortunately, but the results of Bohlin et al. upport our interpretation about the significance of POP for the risk of de novo UI after hysterectomy. However, also conflicting results have been published. In a prospective cohort study comparing hysterectomy with endometrial ablation in the treatment of dysfunctional uterine bleeding, hysterectomy was found to be an independent risk factor for de novo UI. 11 The discrepancies between the results of previous literature may be due to the differences in study designs, the heterogeneity of the study populations, inconsistency in accounting for confounding factors, and the variability in the surgical approaches as well as the follow‐up times. Hysterectomy has been shown to be associated with an increased risk of subsequent UI surgery. A Swedish population‐based cohort study showed that hysterectomy more than doubled the risk of lifetime UI surgery compared to women without hysterectomy, especially in the first five‐year period postoperatively. 9 Another large Swedish cohort study with 118 601 hysterectomized women found hysterectomy to be associated with an increased risk of subsequent SUI surgery and POP. 13 Both of these studies found that preoperative POP was associated with a higher risk for subsequent surgeries after hysterectomy. 9 , 13 In a recent Danish register‐based study of 83 370 hysterectomies, the risk of SUI surgery was almost threefold in women who had undergone hysterectomy compared to reference women. 12 However, these studies did not consider all incontinence‐related confounding factors, such as BMI, although parity and/or vaginal deliveries were included in the adjustments. In our study, no increased risk of subsequent UI surgery after hysterectomy was found, in accordance with another Finnish register‐based study of more than 5000 hysterectomized women. 22 They also showed the surgical approach of hysterectomy to be important when considering the risk of postoperative UI; the risk for SUI operation was over twice higher after VH compared to AH. 14 This is in line with our results, where VH was associated with an increased risk of de novo UI, with the risk being highest for SUI. However, the increased risk disappeared after adjusting for preceding POP. We found no increased risk of de novo UI after LH or AH, again in line with Tulokas et al. Thus, we suggest that the previously observed association between hysterectomy and increased risk of UI may be explained by underlying factors, mainly pre‐operative POP, rather than the hysterectomy or the surgical approach independently. The NFBC1966 as a general population‐based cohort forms an excellent dataset to evaluate the effect of hysterectomy on the risk of de novo UI at a population level. Our unselected birth cohort enabled us to form a controlled study design, including participants with no significant differences regarding BMI, vaginal delivery, menopausal status, or the ability to reach access to public high‐quality healthcare. Furthermore, we were able to combine our cohort data with high‐quality Finnish national registers. The Finnish CRHC has previously been confirmed to be valid for different medical conditions, 20 enabling the adjustment for preceding POP. The Finnish Medical Birth Register provided reliable data on parity status and mode of deliveries. We were able to exclude women with pre‐hysterectomy UI to assess specifically de novo UI. Furthermore, also hysterectomies performed due to malignant indications were included in the cohort, although few in numbers. The extensive 46‐year questionnaire survey enabled us to consider a wide range of potential covariates such as education and smoking. Finally, we were able to provide results on the effect of hysterectomy on different subtypes of UI, as well as the effect of different hysterectomy approaches on the risk of de novo UI. As for limitations, this study represents women with hospital‐based diagnoses only, as we identified women with UI using ICD‐10 codes from the CRHC. Thus, our study probably underestimated the incidence of UI. However, the same underestimation occurred in both study groups, so this limitation should not affect the results significantly. It should also be considered that the number of cases, especially for UI other than SUI, as well as UI operations, remained low. As the UI cases were identified using ICD‐10 codes, we were unable to separate urge and mixed UI subtypes. In addition, excluding women with preceding UI in the hysterectomy group may have caused a slight bias, while in the reference group, lifetime UI prevalence was included. It is well known that the incidence of UI increases with age. Unfortunately, the results of our study do not represent elderly population, as our data were collected until the year 2020, when the cohort particpants were 54 years old. Our vastly considered confounding factors were collected from the 46‐year questionnaire data; thus, they are not necessarily representative of the time points of the hysterectomy or the UI diagnosis. Moreover, as LEF13 code is used for vaginal hysterectomy for prolapse with and without colporrhaphy, we were not able to determine the proportion of concomitant prolapse procedures in our data. Due to the study design and the heterogeneity of indications for hysterectomy and different surgical approaches, we could only assess the independent association, not causality, between hysterectomy and de novo UI. Finally, the regional birth cohort with homogenous ethnic and cultural backgrounds may diminish the global generalizability of these results.

Contrary to the hypothesis, in our population‐based cohort study, hysterectomy performed by any surgical approach does not appear to have an independent association with the risk of any de novo UI subtype. Instead, POP appears to have a more significant effect on the anatomical and physiological changes in pelvic floor function than the surgical trauma caused by the operation itself. Such information is of critical importance in gynecologists' decision‐making and when counseling women on the associated risks related to hysterectomy.

Hysterectomy is one of the most common surgical procedures in gynecology. In the US, approximately 600 000 women undergo a hysterectomy each year, and the most common indications are uterine fibroids, dysfunctional bleeding, chronic pelvic pain, pelvic organ prolapse (POP), endometriosis, pelvic mass, and endometrial cancer. 1 , 2 In Finland, 5000 hysterectomies were performed in 2022, and laparoscopy (LH) was the most common surgical approach (59%), followed by vaginal (VH, 25%) and abdominal (AH) hysterectomy (17%) (Finnish Institute for Health and Welfare). Hysterectomy has been suggested to disrupt the anatomy and the supporting structures of the pelvic floor, causing structural and functional changes such as alterations in bladder and bowel functions, as well as sexual dysfunction and dyspareunia. 3 , 4 , 5 There is a clinical and patient‐oriented concern about whether hysterectomy predisposes to de novo urinary incontinence (UI). UI is a common, often underdiagnosed and undertreated condition, diminishing quality of life markedly. The prevalence ranges from 37.5% in women aged 30–50 years up to 77% in the elderly. 6 Stress urinary incontinence (SUI) is the most common subtype (45.9%), followed by urgency urinary incontinence (31.1%) and mixed urinary incontinence (18.1%). 7 Existing data on the association between hysterectomy and the risk of de novo UI are partially conflicting and have mainly focused on de novo SUI. Majority of previous studies have shown hysterectomy to be associated with an increased risk of UI, especially SUI. 8 , 9 , 10 , 11 Furthermore, two Nordic studies showed an increased risk of subsequent SUI surgery after hysterectomy, 12 especially if performed vaginally. 13 The predominance of VH among patients needing post‐hysterectomy SUI surgery was also seen in a Finnish study, 14 suggesting that the surgical approach of hysterectomy might be associated with the incidence of de novo SUI. On the contrary, some studies have reported no risk of de novo UI after hysterectomy 15 , 16 or even urinary symptom relief after hysterectomy. 17 However, most of the previous literature on this subject is based on non‐controlled hospital‐derived data, with only three studies reporting population‐based controlled settings. 9 , 12 , 13 UI is associated with the same risk factors that predispose to hysterectomy, with pathophysiology overlapping other pelvic floor dysfunctions—thus a substantial proportion of women already have UI symptoms prior to hysterectomy. Due to the unclear effect of hysterectomy on the risk of UI, we aimed to assess the independent association between any total hysterectomy and the risk of de novo UI. Based on previous literature, the hypothesis of our study was that hysterectomy increases the risk of subsequent incontinence.

The authors declare no conflicts of interest.

Our study was a prospective population‐based cohort study. The study cohort consists of the women included in The Northern Finland Birth Cohort 1966 (NFBC1966). 18 NFBC1966 is a general population‐based cohort consisting of 96.3% of all expected births during 1966 in northern Finland (12 055 mothers, 12 058 live‐born children, 5889 females) (University of Oulu: Northern Finland Birth Cohort 1966. University of Oulu). 19 Enrollment for this database began at the 24th gestational week and thereafter data (postal questionnaire and/or clinical measurements) have been collected at ages 1, 14, 31, and 46 years. The original purpose of this cohort was to study risk factors involved in pre‐term birth and intrauterine growth retardation, and the consequences of these early adverse events on subsequent morbidity and mortality. We combined cohort data to register data from the Care Register for Health Care (CRHC) maintained by the Finnish Institute for Health and Welfare, between 1968 and 2020. 20 The CRHC contains all diagnoses and surgical operations for all hospital visits in Finland using International Classification of Diseases (ICD‐10) codes and Classification of Surgical Procedure (NCSP) codes. Furthermore, the Finnish Medical Birth Register data were combined with cohort data. Women who had undergone hysterectomy ( n  = 461) were identified in the CRHC using the NCSP codes, and the operations were classified according to surgery approach into VH (LCD10, LCD40, LEF13, and LEF14), LH (LCD01, LCD04, LCD11, LCD31, and LCD97), and AH (LCD00, LCD30, and LCD96) subgroups. Women without the aforementioned codes were considered the reference group ( n  = 3495). UI diagnoses were collected from the CRHC using ICD‐10 codes (N39.3 for SUI, and N39.4, N39.8, and N39.9 for other UI), and UI operations with NCSP codes (LEG10, LEG12 and LEG13, and LEG96). In addition, urethral bulking agent injection treatments (KDV20 and KDV22) and urological procedures (KDG01, KDG20, and KDG96) were also included as UI operations. All register data were collected until the end of 2020, and for each operation and diagnosis, only the code reported for the first time was included for each woman. For the hysterectomy group, women with preoperative UI diagnoses were excluded from the outcome analyses, and all UI diagnoses without an existing pre‐hysterectomy diagnosis were considered de novo incontinence as we aimed to specifically investigate the effect of hysterectomy on surgery‐induced UI. The flowchart of study groups is shown in Figure  1 . Flowchart of the study population, n (%). AH, abdominal hysterectomy; CRHC, Care Register for Health Care; LH, laparoscopic hysterectomy; NCSP, National Classification of Surgical Procedure; UI, urinary incontinence; VH, vaginal hysterectomy. The age at the time of hysterectomy was calculated, and the main diagnosis determined at the operation visit was considered as an indication for hysterectomy: uterine fibroid (D25*), POP (N81*), dysfunctional bleeding (N92* and N93.8), dysmenorrhea/endometriosis (N94* and N80*), malignancy (C53*, C54*, C56*, D06*), or other. Parity status and mode of deliveries were obtained from the Finnish Medical Birth Register. Parity was categorized as nulliparous, one or two deliveries, and three or more deliveries, and mode of deliveries was dichotomized into women with no vaginal deliveries and women with one or more vaginal deliveries. Variables including body mass index (BMI), menopausal status, smoking status, and education were collected from the 46‐year follow‐up questionnaire mailed to all NFBC1966 participants living in Finland in 2012. Education was categorized into basic (basic or vocational school), secondary (college degree), and tertiary (polytechnic or university degree). Menopausal status was defined by asking whether climacteric symptoms had occurred. In addition, healthcare visits with POP diagnoses (ICD‐10 codes N81*) were obtained, and women with a POP diagnosis set prior to hysterectomy were considered as having preceding POP. Concomitant adnexal operations (LA* and LB*) were identified with NCSP codes coexisting with hysterectomy. The incidences of subsequent UI diagnoses (de novo UI) and UI operations after hysterectomy were compared to the prevalences of UI diagnoses and UI operations in the reference group. The incidences of de novo UI and UI operations stratified by the different hysterectomy approaches were assessed, also considering the risks of different UI subtypes. Lastly, we assessed the association between hysterectomy and the risk of de novo UI. IBM SPSS Statistics version 28 was used for the statistical analysis. The differences in continuous variables were analyzed using an independent samples t ‐test or a Mann–Whitney U test, as appropriate, and a chi‐square test was used to analyze the differences in the categorical variables. A two‐sided p ‐value <0.05 was considered statistically significant. The associations between the risk of de novo UI among women with hysterectomy and the prevalence of UI in reference group were analyzed with binary logistic regression models reporting odds ratios (OR) with 95% confidence intervals (CI). For the linear regression analyses, we noted the underlying assumptions of independent observations of dataset, a linear relationship between the independent and dependent variables, and the requirement of no multicollinearity among independent variables. After univariate analysis, we performed multivariate analyses in three separate models, adjusting for parity, BMI, and smoking status (model 1), POP diagnosis (model 2), vaginal delivery (model 3), and level of education (model 4). Finally, we adjusted for all covariates. Cumulative hazard function obtained from Kaplan–Meier analysis was used to estimate the risk of UI in women after hysterectomy in different surgical approaches.