Delivery mode and maternal health conditions before and after childbirth: A population-based cohort study from Estonia.

Maire Peters, Katrin Täär, Helle Karro, and Andres Salumets: Conceptualization; Triin Laisk and Reedik Mägi: Data curation; Triin Laisk: Formal analysis; Maire Peters, Katrin Täär, and Triin Laisk: Investigation; Helle Karro, Andres Salumets: Supervision; Katrin Täär, Maire Peters, and Merli Saare: Interpretation of the results; Katrin Täär: Writing – original draft; Maire Peters, Helle Karro, and Triin Laisk: Writing – review and editing; all authors: Approval of the final manuscript.

All biobank participants have signed a broad informed consent form, and the study was carried out under ethical approval no. 1.1‐12/2733 (July 7, 2023) from the Estonian Committee on Bioethics and Human Research and authorized under data release 6‐7/GI/1884 from the Estonian Biobank.

This study was supported by the Estonian Research Council grant PRG1076. The research was conducted using the Estonian Center of Genomics/Roadmap II funded by the Estonian Research Council (project number TT17). This work was also supported by the Estonian Centre of Excellence in Personalized Medicine (CEPM), funded by the Estonian Ministry of Education and Research grant TK214 and by Horizon Europe (NESTOR, grant no. 101120075). Data analysis was carried out in part in the High‐Performance Computing Center of University of Tartu.

In EstBB, data of 121 773 women were available for analysis. After excluding women who had not given birth and women who had delivered via both VD and CS or had multiple delivery, data of 39 183 participants were eligible for further analysis. These women were divided into two groups: only VD (ICD‐10 O80, n  = 30 898) and only CS (ICD‐10 O82, n  = 8285). Of the latter group, 2232 women had delivered exclusively by elective CS (O82.0) and 4294 only by emergency CS (O82.1) (Figure  1 ), the remaining 1759 women had delivered via both – elective and emergency CS. The descriptive characteristics of study groups are given in Table  1 . Women delivering via CS were generally older, had higher BMI, and less deliveries per woman than women in the VD group. Descriptive characteristics of study groups. Abbreviations: SD, standard deviation. Denotes statistically significant difference ( p  < 0.001) between groups. Last recorded BMI. Includes both live and stillbirths. Comparison between the groups showed that the frequencies of 89 ICD‐10 codes differed significantly between the CS and VD groups (Table  S1 ), with CS being associated with a markedly higher number of comorbid conditions than VD (72 vs. 17, respectively) (Figure  2 ). Most of the ICD‐10 codes that showed statistically significant differences between the groups were from the O‐category (conditions related to pregnancy, childbirth, and the puerperium). Of these, 28 codes were more frequent in the CS group and 9 were more frequent in the VD group. These were followed by N‐category codes (diseases of the genitourinary system), with 13 in total—8 more common in the CS group and 4 in the VD group, and Z‐category codes (factors influencing health status and contact with health services), with 9 in the CS and 4 in the VD group. Results for associated ICD‐10 codes for both delivery modes. Logistic regression was used to test the association between ICD‐10 codes and delivery mode. Each triangle on the plot represents one ICD‐10 code and each color stands for a different ICD‐10 diagnosis category. Upward‐pointing triangles indicate ICD codes that are more prevalent in the cesarean section group, while downward‐pointing triangles indicate ICD codes that are more frequent in the vaginal delivery group. In the CS group, higher prevalence of pregnancy‐related health conditions was observed. These included gestational diabetes (O24), hypertensive disorders in pregnancy and preeclampsia (O12–O14), as well as other maternal diseases classified elsewhere but complicating pregnancy, childbirth, and the puerperium (O99), such as anemia (O99.0) and mental disorders (O99.3). In addition, preterm labor (O60) and early puerperal complications—namely, puerperal sepsis (O85), other puerperal infections (O86), and anesthesia‐related complications (O89)—were significantly more frequent among women who delivered by CS (Table  2 , Table  S1 ). Top 20 pregnancy‐ and puerperium‐related comorbidities in cesarean section and vaginal delivery groups. Note : Data are presented as n (%). Odds ratios (OR) and 95% confidence intervals (CI) are derived from logistic regression models adjusted for maternal birth year, body mass index (BMI), and 10 genetic principal components. Abbreviations: CS, cesarean section; VD, vaginal delivery. In the CS group, pregnancy was more likely regarded as high‐risk pregnancy (Z35), whereas in the VD group, ICD‐10 code Z34 (supervision of normal pregnancy) was statistically more common. Women who delivered via CS were more likely to have chronic health conditions, for example, diabetes (E10, E11), hypertensive disorders (I10, I11), obesity (E66), chronic renal diseases (N11, N18), and neurological or psychiatric conditions (G40, G80, G82, F32) (Table  3 , Table  S1 ). ICD‐10 codes representing infertility (N97), assisted reproductive technology (ART, Z31), and endometriosis (N80) were also more prevalent among women in the CS group. Complications classified as long‐term consequences occurred significantly more frequently in the CS group, including chronic urinary diseases (N30), pelvic inflammatory disease (N73), other non‐inflammatory uterine disorders (N85), other genitourinary complications (N99), and incisional hernia (K43). Gastrointestinal conditions potentially related to previous surgery—such as peritoneal adhesions (K66) and paralytic ileus (K56)—were also more frequent among CS patients. In addition, depression (F32), sleep disorders (G47), and chronic pain (R10, R52) occurred more often in the CS group (Table  3 , Table  S1 ). Top 20 comorbidities (excluding O‐category) in comparison between cesarean section and vaginal delivery groups. Note : Data are presented as n (%). Odds ratios (OR) and 95% confidence intervals (CI) are derived from logistic regression models adjusted for maternal birth year, body mass index (BMI), and 10 genetic principal components. Abbreviations: CS, cesarean section; VD, vaginal delivery. We also conducted a sensitivity analysis including smoking status as an additional covariate to assess its impact on the association between delivery mode and health outcomes. The results of this analysis for significant codes from the main analysis are presented in Table  S5 . Overall, for nominally significant ( p  < 0.05) codes, the effect estimate correlation between the two analyses was 0.99, indicating that additional adjusting for smoking does not change the results. Therefore, we did not conduct this additional sensitivity analysis for the other analyses. To assess whether certain diagnoses appeared before or after delivery, we performed additional analysis showing that women who delivered by CS more often had diabetes (E10), depression (F32), and prior ART use (Z31) already before delivery (Figure  3 , Tables  S4 , S6 ). At first, infertility (N97) also showed nominal significance ( p ‐value 0.02) but the association did not persist after FDR correction (FDR p ‐value 0.08). Although some main group diagnoses (e.g., N70, K66, N73) did not reach statistical significance when examined in relation to delivery timing, subgroup analysis (Figure  3 , Table  S6 ) revealed that hydrosalpinx (N70.1) was more common before delivery, whereas peritoneal adhesions (K66.0) and peritoneal adhesions of the female pelvis (N73.6) were more common after CS compared with VD, although the associations with K66.0 were only nominally significant ( p ‐value 0.02, FDR p ‐value 0.06). Diagnosis timing of selected ICD‐10 codes stratified by mode of delivery. All presented comparisons showed statistically significant differences. ICD‐10 codes: E10, type 1 diabetes mellitus; F32, depressive episode; N30.9, cystitis, unspecified; N39.3, stress urinary incontinence; N72, inflammatory disease of cervix uteri; N73.6, female pelvic peritoneal adhesions; N80.6, endometriosis in cutaneous scar; N81, female genital prolapse; N87, dysplasia of cervix uteri; Z31, procreative management; Z35.2, supervision of pregnancy with other poor reproductive or obstetric history. In contrast, VDs were associated with higher rates of perineal lacerations and obstetric trauma (O70, O71), postpartum hemorrhage (PPH, O72), retained placenta (O73), postpartum hemorrhoids, and venous complications (O22, O87) (Table  2 , Table  S1 ). VD was also significantly linked to an increased risk of pelvic organ prolapse (N81), which occurred more often after VD than CS (Figure  3 , Table  S6 ). Although the overall frequency of main code N39 (other urinary system disorders, including various forms of urinary incontinence) did not differ significantly between groups, subcode analysis showed that stress urinary incontinence (N39.3) was more strongly associated with VD and occurred more frequently after VD than CS (Figure  3 , Tables  S3 , S6 ). Interestingly, VD also showed a significant association—with both statistical and temporal alignment—with cervical inflammatory conditions (N72) and cervical dysplasia (N87) (Table  3 , Figure  3 , Table  S6 ). We also compared elective (O82.0) and emergency CS (O82.1) groups to explore the potential underlying reasons for the CS in each group (Figure  4 , Table  S7 ). ICD‐10 codes more commonly associated with elective CS were fetal malpresentation (O32) and maternal care for known or suspected abnormality of pelvic organ (O34). Emergency CS showed statistically significant association with conditions such as abnormal forces of labor (O62), labor and delivery complicated by fetal distress (O68), prematurity (O60), obstructed labor due to fetal malposition or maternal pelvic abnormality (O64, O65), placental abruption (O45), preeclampsia (O14), failed induction of labor (O61), and long labor (O63). Results for associated ICD‐10 codes for elective and emergency cesarean delivery. Logistic regression was used to test the association between ICD‐10 codes and delivery mode. Each triangle on the plot represents one ICD‐10 code and each color stands for a different ICD‐10 diagnose category. Upward‐pointing triangles indicate ICD codes that are more prevalent in the elective cesarean section, while downward‐pointing triangles indicate ICD codes that are more frequent in the emergency cesarean section group.

The current study provides a population‐based comparison of comorbidity profiles associated with CS and VD, using data from the EstBB. Complications can occur regardless of the method of delivery, and it has been well established that CS comprises substantially more risks to the mother compared with VD. Our findings further support this evidence, demonstrating that women who delivered by CS experienced a higher prevalence of several comorbid conditions, including postpartum inflammation (O85, O86), pain (R10, R52), incisional hernias (K43), and PAS spectrum disorders (O43, O44). At the same time, VD was more commonly associated with pelvic floor (N81) and perineal complications (O70) (Table  S1 ). These results emphasize the importance of balancing medical indications with long‐term maternal health considerations when choosing the mode of delivery. According to our study, CS is associated with diagnoses that can be considered short‐term complications like postpartum sepsis, surgical site infections (O85, O86), and anemia (O99.0). Our study did not find an increased risk of PPH (O72) associated with CS, a topic on which previous studies have shown mixed results. Several systematic reviews have identified CS as a clear risk factor for PPH, 20 , 21 whereas some previous studies have reported lower rates of PPH among women undergoing elective CS without labor compared with women entering spontaneous or induced labor. 22 , 23 , 24 In our cohort, PPH was more common after VD than CS (7.7% vs. 4.3%; OR 0.59, 95% CI: 0.52–0.66) (Table  S1 ). However, these findings should be interpreted with caution, as PPH was identified solely based on the ICD‐10 diagnosis code O72, which does note capture the amount of blood loss. Moreover, diagnostic criteria for PPH have varied across studies and clinical guidelines, with traditionally lower blood loss threshold applied for VD (≥ 500 mL) than CS (≥ 1000 mL). 25 Therefore, variation in diagnostic criteria may have influenced the assignment of the O72 code more often in the VD group. Several studies have addressed the risk and possible risk factors for developing chronic postsurgical pain after CS. The risk of chronic pain is greater with an increasing number of CSs, for women with previous psychiatric conditions and for those who experienced more severe postoperative pain. 26 , 27 , 28 Surgical technique is considered to have an influence on the risk of pain as abdominal wall and uterine incision closure techniques may cause pain due to nerve entrapment, pelvic adhesions, cesarean scar defect (CSD), and scar endometriosis. Adhesions which create traction and fix the uterus to the abdominal wall after CS can be the cause of severe chronic pelvic pain. 17 , 29 , 30 , 31 In our study, chronic abdominal pain (R10, R52) was also among the CS‐associated diagnoses, together with incisional hernia (K43) and bowel obstruction (K56). These complications may also be the result of other abdominal surgery besides CS, or completely unrelated, but compared with the VD group, the increased rates in CS group cannot be overlooked. Same long‐term complications have been previously reported in several studies, including Swedish population‐based cohort study from 2021. 16 However, our study did not find statistically significant difference in the prevalence of some health conditions previously associated with CS, such as miscarriage (ICD‐10 codes O02; O03), ectopic pregnancy (O00), 9 AUB (N92, N93), or dysmenorrhea (N94—pain and other conditions associated with female genital organs and menstrual cycle). 12 Dysmenorrhea or menorrhagia are more likely to occur in women who develop CSD, also known as isthmocele or niche. 12 , 32 , 33 Multiple studies have demonstrated an association between CSD and AUB, with the risk of AUB being significantly higher in women with CSD compared with those without. Dysmenorrhea may also co‐occur with AUB in women with CSD. Consequently, CSD is considered one of the long‐term gynecological complications following CS, contributing to several life‐affecting problems, including AUB, dysmenorrhea, subfertility, ectopic pregnancy, and PAS disorders in subsequent pregnancies. 13 , 14 , 34 , 35 , 36 , 37 We did observe statistically significant association between abnormal placentation (O45) and CS as demonstrated in other studies. Unfortunately, we were unable to assess the prevalence of CSD in our analysis because CSD does not have a specific ICD‐10 code. While we analyzed the prevalence of ICD codes in relation to whether they appeared before or after delivery, we found that infertility and the use of ART were already common prior to CS. This indicates that CS is not strongly associated with the development of infertility afterwards. This has also been observed previously, and our results further support the conclusion that CS has little to no direct impact on future fertility. Instead, the clinical and social factors that lead to the decision to perform CS appear to play a larger role in determining subsequent fertility. 36 , 38 A recent study demonstrated that although women with a prior CS had higher infertility risk, a longer time‐to‐pregnancy also increased the likelihood of having a CS. This bidirectional relationship suggests shared underlying factors, meaning that the procedure itself has little direct effect on fecundability. 38 At the same time, a study conducted in the context of ART showed that in vitro fertilization poorer outcomes are associated more with the presence of CSD than with a previous CS itself. 39 The CS group in our study had higher rates of chronic comorbidities and pregnancy complications like (gestational) diabetes, hypertensive disorders, different mental health, and neurological problems (Table  2 , Table  S1 ). These conditions, if present before or during pregnancy, may influence the choice of delivery method, as they can increase the risk of complications during pregnancy, delivery, and puerperium. Delivery by CS, especially emergency CS, is associated with increased risk for postpartum depression and posttraumatic stress disorder 40 , 41 but women with pre‐existing mental disorders are at higher risk for both—delivery via CS and also psychiatric morbidity after delivery. 42 , 43 VD also carries its own risks, such as perineal trauma, pelvic organ prolapse, and stress urinary incontinence, 44 all of which were confirmed in our study (Tables  2 , 3 and Table  S1 ). Interestingly, we found that cervical inflammatory conditions and cervical intraepithelial neoplasia (CIN) were also more common in the VD group. CIN has not been previously associated with either delivery method as a causative agent. CIN is predominantly caused by persistent human papillomavirus (HPV) infection, the risk being increased by smoking, lower age at first sexual intercourse, multiple sexual partners, and immunosuppression. 45 , 46 , 47 In our study, ICD‐10 code B97.7 (Papillomavirus as the cause of diseases classified elsewhere) was more prevalent in the VD group (16.5%) compared with the CS group (13.4%) but after correction for multiple testing, its p ‐value (1.7 × 10 −5 ) did not reach statistical significance. According to previous research, high parity (5 or more deliveries per woman) 48 can also be regarded as a cofactor for developing or progression of CIN but not an independent factor. The mean number of deliveries per woman was 2.2 in the VD group and 1.9 in the CS group in our study. The main strength of this study is the use of large population‐based data, which provides a substantial number of women for both the CS group and the VD group. Another important strength is the longitudinal nature of the data, allowing us to identify pre‐existing health issues as well as complications that may develop years after delivery. This study also has several limitations. One of the most significant issues is participation bias. Individuals who participate in biobanks may differ from the general population, as they tend to have higher socioeconomic status and are generally healthier and more health conscious. This creates a study population that might not fully reflect real variations in health status and might limit the generalizability of study results. 18 , 49 Biobank data are also subject to inconsistencies and incompleteness. Information gathered from electronic health records or self‐reported questionnaires can vary in quality, and missing data can make certain analyses less reliable. Over time, temporal changes in population behaviors and healthcare practices may further complicate comparisons within long‐term biobanks. In addition, although we adjusted our analysis for several possible confounders, factors not included in this study may still influence the results. As our analysis relies on ICD‐10 codes, the possibility of coding errors cannot be excluded. Some relevant diagnoses may not have been recorded, with clinicians documenting only the code of the main health problem. For older participants, ICD‐10 codes reflecting childhood illnesses may be missing, while younger women in our study population may not have developed certain health conditions yet—such as pelvic organ prolapse, AUB or incisional hernia– or may not have become pregnant again, making it impossible to assess outcomes such as PAS in subsequent pregnancies. Although we observe that certain ICD‐10 codes differ statistically significantly between groups, no definitive conclusions regarding causality can be drawn. The observed associations between CS, VD and a certain diagnosis may be influenced by an unmeasured third factor that contributes to both outcomes. Future analyses focusing on specific outcomes may apply time‐to‐event approaches, such as Cox proportional hazards models, to further characterize the risk trajectories following delivery.

Our analysis, utilizing data from the EstBB and linked registries, provides valuable insight into the health risks associated with CS and VD. As expected, the CS group exhibited a higher prevalence of pre‐existing morbidities, which may influence both the choice of delivery mode and subsequent health outcomes. Women who delivered by CS were also more likely to experience pregnancy‐related complications as well as short‐ and long‐term postpartum morbidities. However, while CS is associated with a greater overall burden of complications, VD also carries its own set of risks, including perineal trauma and long‐term issues like pelvic organ prolapse, cervical inflammatory conditions, and cervical dysplasia. Given the observational design of the study, the identified associations should not be interpreted as causal relationships. Although CS is generally considered safe in contemporary medical practice, both its short‐ and long‐term risks—often not fully understood by patients—should be clearly communicated, particularly to women considering the procedure without medical indications. Counseling should comprehensively address possible complications, and clinical decisions must carefully balance these risks against specific patient needs. Continued research remains essential to further elucidate and mitigate potential long‐term health effects of mode of delivery.

Cesarean section (CS) is the most common gynecological surgery globally. In the past 30 years, cesarean deliveries have become increasingly common in developed countries—in Europe, the rate has more than doubled, rising from 11.2% in 1990 to 26% in 2019 and in some countries even more than 50% of deliveries are via CS. 1 , 2 In Estonia, the rate of CS increased from 6.4% in 1992 to 21% in 2007 and has remained similar since. 3 If performed for the right indication, CS is a potentially lifesaving procedure for both the mother and the newborn. The World Health Organization (WHO) stated in 1985 that CS rates up to 15% reflect the proportion of deliveries with clear medical indications in which CS reduces maternal, neonatal, and infant mortality and morbidity but beyond that number, there is an inevitable burden of adverse outcomes. 4 Based on currently available scientific evidence, the European Association of Perinatal Medicine and the European Midwives Association recommended in 2024 that optimal CS rates should range between 15% and 20% to ensure the best outcomes for mothers and babies. 5 , 6 The rise in CS rates is influenced by a complex interplay of factors, including changes in clinical practice, within the healthcare system, and various demographic and socioeconomic trends. Key contributors to the increasing number of primary CSs include indications such as non‐reassuring fetal status, labor arrest, suspected fetal macrosomia, and CSs that are not medically indicated, often referred to as “cesarean on maternal request.” 7 , 8 Nonmedical requests for CS may stem from maternal anxiety or fear of labor pain, psychosomatic or psychiatric conditions, concerns about vaginal or perineal trauma, or worries about potential complications for the baby. Additional influences may include a desire to deliver on a specific date or particular social, cultural, or religious preferences. For gynecologists, contributing factors may involve concerns about legal consequences or litigation related to adverse outcomes of vaginal deliveries (VDs), and in some settings, financial incentives—especially in private hospitals. 9 , 10 , 11 It has been well established that giving birth via CS has more complications compared with VD. Short‐term complications of CS are postoperative infections, hemorrhage, anemia, organ injury, and anesthesia‐related adverse events. 1 , 4 , 5 Previous CS is also associated with an increased risk of repeated CS, uterine scar dehiscence and rupture, placenta accreta spectrum disorders (PAS), ectopic pregnancy, preterm birth, etc. 9 , 12 , 13 , 14 Moreover, recent studies suggest that mode of delivery could influence future cardiovascular disease risk. 15 Considering efforts for personalized disease risk prediction, the association between mode of delivery and future comorbidities should be evaluated in more detail. 15 Research as well as patient counseling has focused more on the pregnancy‐related and short‐term adverse events of CS rather than long‐term complications affecting women's everyday life, for example, chronic pain, abnormal uterine bleeding (AUB), fertility problems, (scar) endometriosis, surgical adhesions, and bowel obstruction. 13 , 14 , 16 , 17 Because these complications may only become apparent years after CS, it is essential to have and use robust scientific data when informing patients—especially those without a clear medical indication for CS—about these potential risks. Our primary objective was to compare health data of women who have delivered by CS and those who have delivered vaginally, using the Estonian Biobank (EstBB) database. We aimed to identify the diagnoses associated with each mode of delivery, as well as the health issues that may lead to the preference of one delivery method over the other.

The authors report no conflicts of interest.

For this historical cohort study, we used a large dataset from the EstBB, which is a population‐based biobank with health information for over 200 000 participants in the latest data freeze. 18 All biobank participants have signed a broad informed consent form. Information on International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD‐10) codes is obtained via regular linking with the Estonian Health Insurance Fund (EHIF) and other relevant registries, including the Estonian National Health Information System, the Estonian Cancer Registry, the Myocardial Infarction Registry, the Population Registry, and the Causes of Death Registry. The availability of linked data varies by registry. While electronic health records are available from 1993 onwards, the majority of diagnoses originate from 2003 and later. More comprehensive linkage coverage across multiple registries has been achieved since approximately 2010, with some registries (e.g., EHIF treatment and prescription data, cancer screening data, and Myocardial Infarction Registry) becoming available or more consistently linked in subsequent years (2013–2023). Therefore, for older biobank participants, the completeness of electronic health data may be limited. An overview of the temporal coverage of the different registries has been described previously. 18 For the analysis presented in this paper, we used the EstBB data up to the end of 2023, with delivery dates spanning through 1953–2023. However, diagnoses dating before 2004 are mostly based on self‐reported information. Between 2004 and 2023, a total of 73 133 deliveries were identified among EstBB participants, compared with 278 638 deliveries recorded nationally in Estonia during the same period according to Estonian Health statistics and health research database. 19 This corresponds to approximately 26% of all deliveries during the study period and reflects the broad population coverage of the EstBB, which includes approximately 20% of the Estonian adult population. Only female biobank participants who had given birth were included in the analysis. Participants were divided into two groups—women with one or more CS (ICD‐10 code O82) and women with one or more VD (ICD‐10 code O80), the latter serving as the reference group. Women with a history of both CS and VD as well as those with twin or higher order multiple pregnancies (ICD‐10 code O84) were excluded (Figure  1 ). Flowchart describing the formation of study groups. Women with a history of both CS and VD, as well as women with twin or higher order multiple pregnancies, were excluded. Using individual level data in the EstBB, we performed an analysis to find ICD‐10 main codes associated with the ICD‐10 code O80 (VD) and the ICD‐10 code O82 (CS) (Figure  1 , Table  S1 ). We excluded diagnoses related to exogenous factors (poisoning, accidents, injuries, etc. in the ICD‐10 S, T, U, V, W, X, and Y chapters). As several ICD‐10 main codes have heterogeneous subgroup codes reflecting different medical conditions within the same main code, we also performed additional subcode analysis to a selection of main ICD‐10 codes to see if there is a certain subcode which affects the main code level results substantially (Tables  S2 , S3 ). Demographic characteristics were compared using the t ‐test and chi‐square test for continuous and categorical variables, respectively. Logistic regression was used to assess associations between delivery mode and health conditions. The models were adjusted for maternal birth year, body mass index (BMI), and 10 genetic principal components. These principal components are summary variables derived from genome‐wide genetic data that capture the major axes of genetic variation across individuals, reflecting differences in ancestry. They were included to account for potential population stratification and to reduce false associations due to ancestry or regional differences. We also conducted a sensitivity analysis including smoking status as an additional covariate to assess its impact on the associations between delivery mode and health outcomes. Resulting effect estimates and standard errors were converted to odds ratios (OR) and confidence intervals (CI). Because we applied a hypothesis‐free approach, thousands of ICD‐10 diagnosis codes were tested individually in several separate analyses, and a very stringent Bonferroni‐corrected significance threshold was applied to reduce the risk of false positive findings. The Bonferroni‐corrected significance threshold was based on the total number of statistical tests performed across all ICD‐10 analyses. In total, 8979 tests were conducted, including analyses of 1416 main ICD‐10 codes in two separate comparisons (CS vs. VD and elective vs. emergency CS) and 6147 ICD‐10 subcodes included in the subcode‐level analyses. Accordingly, the statistical significance threshold was adjusted to p  < 5.5 × 10 −6 . Results were visualized using the PheWAS library in R 4.3.2. We compared whether specific selected diagnoses (listed in Table  S4 ) occurred more frequently before or after the index event (CS or VD) using a two‐sample proportion test (prop. test) in R. This analysis was restricted to women up to 45 years of age to ensure a sufficient coverage of electronic health records. Data processing was performed using the dplyr 1.1.4, tidyr 1.3.1., and readr 2.1.5. packages. For each ICD‐10 code of interest, we calculated counts of “before” and “after” diagnoses within the CS and VD groups. The prop. test was then applied to evaluate differences in proportions. When any expected count was below 5, we used Fisher's exact test (fisher.test) instead of the chi‐square approximation. While Bonferroni correction was applied in the primary ICD‐10 association analyses, false discovery rate (FDR) adjustment was considered more appropriate for these exploratory secondary analyses due to less conservative nature of the approach. Accordingly, p ‐values were adjusted for multiple testing using the Benjamini–Hochberg correction (FDR p ‐value < 0.05). All tests were two‐sided, and results are reported with 95% CIs. Analyses were performed in R 4.3.2.

Table S1: List of all statistically significant ICD‐10 codes for comparison between cesarean section and vaginal delivery. Table S2: ICD‐10 subcodes used in cesarean section and vaginal delivery groups analysis. Table S3: List of all statistically significant ICD‐10 subcodes comparing cesarean section and vaginal delivery. Table S4: ICD‐10 codes and subcodes used in two‐sample proportion test. Table S5: List of all statistically significant ICD‐10 codes for comparison between cesarean section and vaginal delivery, adjusted for smoking. Table S6: ICD‐10 diagnoses with statistically significant differences in frequency before versus after delivery, by mode of delivery. Table S7: List of all statistically significant ICD‐10 codes for comparison between emergency and elective cesarean sections.