First birth and total fertility rate in women with surgically verified endometriosis - A nationwide register study of 18 320 women across reproductive life course

Anni Tuominen (A.T.), Liisu Saavalainen (L.S.), Maarit Niinimäki, Mika Gissler, Päivi Härkki (P.H.) and Oskari Heikinheimo (O.H.) contributed to planning and innovating this study. Data linkage and statistical analyses were made by Juuso Saavalainen with the help of L.S. All authors participated in data interpretation. The article draft was composed by A.T. with the help of L.S. under the supervision of P.H. and O.H. O.H. obtained study funding. All authors reviewed and revised this article, and the final version was accepted by all authors.

The ethics committee of the Hospital District of Helsinki and Uusimaa approved the study on March 3, 2013 (238/13/03/03/2013). The registry‐keeping organizations approved data retrieval and linkage: the Finnish Institute for Health and Welfare (THL/5393/14.06.00/2023), Statistics Finland (TK/3123/07.03.00/2023) and Digital and Population Data Services Agency.

The study was funded by the Hospital District of Helsinki and Uusimaa, Finska Läkaresällskapet and the Finnish Cultural Foundation.

The characteristics of women with endometriosis and the reference women are presented in Table  1 . Compared with controls, women with endometriosis had a higher level of education ( p  < 0.001). Women with endometriosis had more often registered infertility (44.0% vs. 10.6%, p  < 0.001) and use of IVF‐treatments (22.8% vs. 4.1%, p  < 0.001) across their reproductive life course than reference women. Most of the infertility diagnoses had been registered prior to first birth in both groups (23.8% vs. 5.4%, p  < 0.001). Characteristics of cohort of women with surgically diagnosed endometriosis according to the types of endometriosis, and of the age and residence‐matched reference cohort. n (%) unless otherwise stated. Abbreviations: IQR, interquartile range; NA, not applicable. p ‐values at middle when compared altogether to the other group/s. Index day = day of hospital discharge following surgical verification of endometriosis. Urban, city; densely populated area where 200 people or more are living nearby; rural, under 200 people living nearby, usually over 200 meters between buildings. First date of infertility‐related registered information diagnosis of infertility (628* in ICD‐9; N97* in ICD‐10) and/or infertility‐related procedures (8117 perturbation of Fallopian tubes; 821* salpingolysis/salpingostomia/resection and anastomosis/reimplantation or microsurgery of Fallopian tubes, ovum application or other infertility‐related procedure; LAA10‐11 ovum pick‐up, LCA30 ovum/embryo transplant, LBF* chromoperturbation, TLW* fresh /transmural/frozen embryo transplant, intracytoplasmic sperm injection) in FHDR, reimbursement of infertility‐related procedure of private health care registered by Social Insurance Institution of Finland, purchase of medication for induction of ovulation or IVF (clomiphene or gonadotropin, excluding leuprolelin) registered by Social Insurance Institution of Finland ‐ reported insemination, ovulation induction or IVF in Medical Birth Register. diagnosis of infertility (628* in ICD‐9; N97* in ICD‐10) and/or infertility‐related procedures (8117 perturbation of Fallopian tubes; 821* salpingolysis/salpingostomia/resection and anastomosis/reimplantation or microsurgery of Fallopian tubes, ovum application or other infertility‐related procedure; LAA10‐11 ovum pick‐up, LCA30 ovum/embryo transplant, LBF* chromoperturbation, TLW* fresh /transmural/frozen embryo transplant, intracytoplasmic sperm injection) in FHDR, reimbursement of infertility‐related procedure of private health care registered by Social Insurance Institution of Finland, purchase of medication for induction of ovulation or IVF (clomiphene or gonadotropin, excluding leuprolelin) registered by Social Insurance Institution of Finland ‐ reported insemination, ovulation induction or IVF in Medical Birth Register. First date of IVF treatment‐related registered information IVF‐related procedure (LAA10‐LAA11, LCA30, TLW*) in FHDR. compensation of infertility‐related procedure of private health care registered by Social Insurance Institution of Finland, purchase of medication for IVF (purchase of gonadotropin together with GnRH agonist/antagonist within 90 days from the gonadotropin purchase) registered by Social Insurance Institution of Finland. reported IVF treatment in Medical Birth Register. IVF‐related procedure (LAA10‐LAA11, LCA30, TLW*) in FHDR. compensation of infertility‐related procedure of private health care registered by Social Insurance Institution of Finland, purchase of medication for IVF (purchase of gonadotropin together with GnRH agonist/antagonist within 90 days from the gonadotropin purchase) registered by Social Insurance Institution of Finland. reported IVF treatment in Medical Birth Register. Peritoneal versus ovarian, p  < 0.001 peritoneal versus deep, p  = 0.112; peritoneal versus combined/other, p  < 0.001. Regarding endometriosis sub‐cohorts, women with ovarian endometriosis were oldest, 37.4 (IQR 31.6–43.5) years at the time of surgical verification of endometriosis. The sub‐cohort of women with peritoneal endometriosis had most often registered infertility (48.9%, p  < 0.001) during the follow‐up, whereas women with ovarian endometriosis had the lowest rate of registered infertility and IVF treatment (36.4% and 18.2%, p  < 0.001). During their reproductive life course, women with endometriosis had a lower rate of first births compared with the reference cohort (68.2% vs. 80.7%, p  < 0.001) (Table  2 ). In addition to birth rate, we present the Kaplan–Meier curves of not giving birth in Figure  2 . To assess the statistical difference between the study cohorts over the reproductive life course, we calculated RMST difference. The RMST of not giving birth from the start of the follow‐up (15 years of age) was 18.9 (18.8–19.0) years in the endometriosis cohort and 15.5 (15.4–15.6) years in the reference cohort, presented in Table  3 . The crude RMST difference was 3.42 (3.26–3.58) years and the adjusted RMST difference was 2.59 (2.42–2.76) years between women with surgically verified endometriosis and the reference cohort, both statistically significant ( p  < 0.001). Figure  3 demonstrates the trend in the crude and adjusted RMST difference between reference and endometriosis cohort from the start until the end of the follow‐up. It shows a persistent and increasing statistical difference between first births of endometriosis and reference cohort, even when adjusted with birth cohort or prior IVF treatment. First births and fertility rates of reference cohort and cohort of women with surgically diagnosed endometriosis with the sub‐cohorts of endometriosis. n (%) unless otherwise stated. Abbreviations: IQR, interquartile range; SD, standard deviation. Peritoneal versus ovarian, p  = 0.002; peritoneal versus deep, p  = 0.006; peritoneal versus combined/other, p  < 0.001. Total FR: number of all born children/number of all women; complete FR: number of all born children/number of women who reached 50 years before 31st of Dec 2019. Peritoneal versus ovarian, p  < 0.001; peritoneal versus deep, p  = 0.68; peritoneal versus combined/other, p  < 0.001. Peritoneal versus ovarian, p  < 0.001; peritoneal versus deep, p  = 0.14; peritoneal versus combined/other, p  < 0.001. Peritoneal versus ovarian, p  < 0.001; peritoneal versus deep, p  = 0.42; peritoneal versus combined/other, p  < 0.001. Peritoneal versus ovarian, p  = 0.003; peritoneal versus deep, p  = 0.12; peritoneal versus combined/other, p  = 0.41. All: person‐years from 15 years of age to first birth, emigration, death, 50 years of age or until the end of 2019; adjusted: person‐years from 15 years of age to first birth, emigration, death, bilateral salpingo‐oophorectomy, hysterectomy, sterilization, 50 years of age or until the end of 2019. Peritoneal versus ovarian, p  < 0.001; peritoneal versus deep, p  = 0.51; peritoneal versus combined/other, p  = 0.26. Peritoneal versus ovarian, p  < 0.001; peritoneal versus deep, p  = 0.02; peritoneal versus combined/othe,r p  < 0.001. Probability of not giving birth. Kaplan–Meier curves and their 95% confidence intervals of not giving birth (i.e. until the first birth) in endometriosis and reference cohort. The start of the follow‐up at point zero is the age of 15 years of the women. Statistical comparison of first births between endometriosis and endometriosis sub‐cohorts versus reference cohort. Crude and adjusted restricted mean survival time (RMST) difference (95% conference intervals) of not giving birth from 15 to 45 years of age (restriction point at 30 years). Adjustments were made for birth cohort (grouped by birth cohorts 1940–1949, 1950–1959, 1960–1969, 1970–1979) and IVF (yes/no) at any time before the first birth/in the end of follow‐up when nulliparous. Difference of not giving birth between endometriosis and reference cohorts. The crude and adjusted a restricted mean survival time (RMST) difference and the 95% confidence intervals assessed by 1‐year steps with restriction point ranging from 0 to 35. a Adjustments were made for birth cohort (grouped by birth cohorts 1940–1949, 1950–1959, 1960–1969, and 1970–1979) and IVF (yes/no) at any time before the first birth/in the end of follow‐up when nulliparous. Women with peritoneal endometriosis had the highest (73.1%), and women with ovarian endometriosis had the lowest proportion of first births (65.2%) (Table  2 and Figure  2 ). Each sub‐cohort of endometriosis had significantly fewer first births compared with the reference cohort ( p  < 0.001). Additionally, each endometriosis sub‐cohort presented significantly longer time of not giving birth when compared individually to the reference cohort; the adjusted RMST difference was lowest among women with peritoneal and deep endometriosis (1.49 [1.24–1.74] and 1.33 [0.83–1.84]) (Table  3 ). Likewise, comparing RMST of not giving birth between endometriosis sub‐cohorts, women with ovarian and combined/other endometriosis presented significantly longer time of not giving birth than women with peritoneal or deep endometriosis ( p  < 0.001) (Table  S4 ). TFR of all women was significantly lower among women with endometriosis than in reference women (1.33 [SD 1.16] and 1.89 [1.46], p  < 0.001), with a greater difference of CFR of all women (1.27 [1.15] and 1.95 [1.44], p  < 0.001, respectively). Differences were smaller between endometriosis sub‐cohorts (Table  2 ). Results of TFR for all parous women and CFR of parous women showed also significant difference between endometriosis and reference cohort (Table  2 ).

In the present nationwide study, we found that parity in women with surgically verified endometriosis was ~ 10% lower than in reference women when assessed from 15 years of age until the end of fertile age. TFR of all and parous women was also lower among women with endometriosis. Accordingly, infertility was four times and use of IVF over five times more common in women with endometriosis compared with reference women. Significantly lower birth rates were seen across all endometriosis sub‐cohorts compared with the reference cohort. Women with solely peritoneal endometriosis and deep endometriosis had the highest rate of first births. In contrast, women with solely ovarian or combined/other endometriosis had a significantly lower rate of first births than those with peritoneal or deep endometriosis. In addition to birth rate, TFR and CFR were lower in all endometriosis sub‐cohorts compared with the reference cohort, presenting also differences between endometriosis sub‐cohorts. However, there were no clinically significant differences in fertility rates of parous women between endometriosis sub‐cohorts. Among women with endometriosis, CFR was lower than TFR. This finding was most probably due to the lack of infertility treatments among older birth cohorts of women with endometriosis, partly seen in proportions of registered IVF treatment according to birth cohorts. On the contrary, among reference women, the incidence rate of first birth was highest among older birth cohorts of women. LBR of women with endometriosis and women without endometriosis treated with IVF/ICSI has been assessed in two recent meta‐analyses. 11 , 12 A British study found no overall effect of endometriosis on LBR, but reduced LBR was seen in severe forms of endometriosis (stage III–IV, American Society for Reproductive Medicine staging [ASRM]). 11 Similarly, a Chinese study reported no difference in LBR between women with and without endometriosis, but the different endometriosis subtypes or severity of endometriosis were not assessed. 12 The results from women treated with IVF/ICSI cannot be compared, however, to our population‐based results. In addition, the stages of ASRM and the diagnosis of ICD classification cannot be compared. Our results demonstrated an overall lower birth rate and longer time of not giving birth among women with endometriosis, and the difference remained significant also after adjusting for the birth cohort of women and use of IVF before the first birth. The population‐based studies on LBR among women with endometriosis are rare. Our results are in line with the previous Danish nationwide cohort study with reduction in LBR among women with endometriosis compared with reference women and the improvement of LBR due to wider use of infertility treatments during long‐term follow‐up. 14 We also saw a decrease in the RMST difference of not giving birth between women with endometriosis and reference women after adjusting with birth cohort of women and use of IVF before the first birth. In our study, rate of registered infertility and IVF treatments in hospital registers were more common among women with endometriosis than reference women, in particular women with peritoneal and combined/other endometriosis. This not only highlights the association between endometriosis and infertility, but also most probably the need to seek medical attention due to painful symptoms of endometriosis leading possibly to quicker access to infertility treatments, especially IVF, after the surgically verified endometriosis. The rate of infertility and IVF was lowest among women with ovarian endometriosis, perhaps partly due to delayed diagnosis of solely ovarian endometriosis, known to be more often asymptomatic. Supporting this assumption, recent study focusing on sub‐fertile women undergoing assisted reproductive treatment with deep endometriosis lesions or endometrioma visible on transvaginal ultrasound found that 75.8% of the women did not have any previous knowledge about the disease. 30 However, we cannot exclude underreporting of infertility in the ovarian sub‐cohort, especially considering the highest median age at index day in this sub‐cohort. Moreover, women with peritoneal endometriosis presenting the highest infertility rate before the first birth might have had infertility as an indication to index surgery more often than other sub‐cohorts, rising the proportion of women with infertility in part within this sub‐group. In recent cohort studies focusing solely on women treated with IVF/ICSI, endometriosis was associated with lower LBR than women with tubal factor infertility. 15 , 17 These results are challenged by a North American multicenter retrospective study of women undergoing IVF with euploid frozen blastocysts, which found no difference in LBR in women with endometriosis and women with male factor‐related infertility or non‐infertile women treated due to a single‐gene disorder. 10 Similar findings were reported by a Swedish cross‐sectional study and a French retrospective cohort study among women with IVF treatment. 16 , 18 Interestingly, an Indian prospective matched cohort study of IVF‐treated women demonstrated similar LBR between women with endometriosis and tubal factor‐related infertility (OR 0.8, 95% CI [0.5–1.2]), but after the age of 35 years, LBR in the endometriosis group deteriorated (OR 0.5, 95% CI [0.25–0.9]). 31 Considering the endometriosis subtypes, reassuring LBR results have been presented recently when including MAR pregnancies. A Canadian systematic review and meta‐analysis concluded similar LBR in women with ovarian endometriosis managed either operatively or expectantly prior to IVF (OR 0.83, 95% CI [0.56–1.22]). 19 An Italian retrospective study of a cohort of women with untreated rectosigmoid endometriosis found a LBR of 45% including both natural and MAR pregnancies. 20 Focusing on first‐line IVF, a French study reported an even higher (64.4%) LBR among women with unoperated bowel endometriosis. 21 After surgical treatment of rectosigmoid endometriosis, a Danish retrospective study presented an LBR of 53.9% including also MAR pregnancies. 22 In our study, both the RMST of not giving birth and TFR of women through reproductive live course were lowest among women with ovarian and combined/other endometriosis when compared with the reference cohort. As said, women with ovarian endometriosis were the oldest at index day and possibly unable to optimally benefit from the timely diagnosis and use of MAR, as age is known to be the most important single factor affecting fertility. However, women with combined/other endometriosis were approximately at the same age at index day as women with deep endometriosis (34.4 years [IQR 29.9, 40.6] and 34.0 years [IQR 29.4, 40.1]), who achieved the highest rate of first births together with women with peritoneal endometriosis. The lower results among women with combined/other endometriosis might be due to the complexity of the disease. The strengths of our study include the use of reliable and comprehensive national register data. In Finland, the personal identity code number has been used since 1964, and register data collection is a well‐established and continuous part of the healthcare system and Finnish society overall. Register validation studies have showed good quality 32 , 33 , 34 and also the first surgically verified diagnosis of endometriosis was reliable in a bidirectional quality assessment study. 24 A long follow‐up provides a view across the fertile years of women, and the inclusion of both spontaneous and MAR pregnancies allows description of the general population over time. Our study provides further knowledge on the sparsely studied population‐level effects of endometriosis on the birth rate. Our study also has limitations. Endometriosis was verified surgically for women in our study; therefore, our results cannot be generalized to those with a solely clinical diagnosis of endometriosis. Given this is a population‐based register study, we were unable to examine intentions to conceive among these women. However, the proportions of women with infertility and those who underwent IVF prior to first birth reflect this issue partially. Given a cycle‐based register of IVF treatments was lacking in Finland during the study period, these variables were collected from different registers. IVF was uncommon with low success rates in the early 1980s. 35 However, MAR/IVF treatments became more common and effective during the 1990s and accessible for all citizens as it is today. In the current study, we were unable to rule out other reasons for infertility and IVF/ICSI, for example, male factor. Also, with the data used, it was impossible to identify if the pregnancy started due to IVF treatment or by natural conception. Of note, ovarian endometriosis was the largest sub‐cohort in our study. According to recent studies, the prevalence of endometrioma was 20% among symptomatic women undergoing transvaginal ultrasound examination 36 and 10.5% among women with subfertility undergoing assisted reproductive treatment. 30 However, given the high proportion of women with isolated endometrioma in our study, there might be an effect of underreporting peritoneal endometriosis when encountered together with ovarian endometrioma during surgery. In addition, at the beginning of our study's index day period (1998–2012), the tendency to perform surgery on small endometriomas was more common compared with modern guidelines for the surgical treatment of endometriosis. This effect might in part raise the rate of surgically diagnosed isolated ovarian endometriosis. The surgery was merely used for the diagnosis of endometriosis in our study; we were unable to study the impact of various surgical procedures conducted during index surgery on fertility. In addition, we could not examine the role of concomitant adenomyosis on birth rate. According to recent studies, adenomyosis most probably affects fertility. 37 , 38 , 39 , 40 Moreover, our study covers a several decades‐long follow‐up period with evolving clinical practice in diagnostics, surgical techniques, and infertility treatments. Therefore, the results can only be generalized with caution. It is worth noting that we were unaware whether reference women would have had a surgically confirmed diagnosis of endometriosis after 2012 or a clinically diagnosed endometriosis after 2015 or had undiagnosed endometriosis.As women with endometriosis had a higher education level, they might seek professional help for their symptoms or infertility more easily, leading to detection bias.

Women with surgically verified endometriosis had fewer first births, and the TFR of parous women was lower throughout their reproductive life when compared with reference women in this nationwide setting. According to our results, women with peritoneal endometriosis had the highest, and women with ovarian and combined/other endometriosis had the lowest birth rate. The role of MAR and age of women cannot be underestimated, which emphasizes the importance of timely diagnosis of endometriosis and appropriate fertility counseling and treatment planning.

The association between endometriosis and infertility is commonly accepted, and the mechanisms behind the phenomenon are complex. 1 However, data concerning the impact of endometriosis on birth rate are limited, especially among different subtypes of endometriosis (peritoneal, ovarian and deep endometriosis). In addition, this field of research is challenged by confounding factors, that is, use of medically assisted reproduction (MAR) and lack of randomized studies. Moreover, the possible role of endometriosis surgery in enhancing fertility is inconclusive. 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 The birth rate of endometriosis sufferers has mostly been studied among women seeking help for infertility. Two recent meta‐analyses of women treated with IVF/ICSI concluded that there are no overall differences in live birth rate (LBR) between women with and without endometriosis. 11 , 12 On the contrary, a narrative synthesis of systematic reviews and meta‐analyses concluded a reduction in LBR among women with endometriosis; however, it emphasized both the qualitative and quantitative heterogeneity of the included studies. 13 A Danish population‐based study showed a reduced LBR among women with endometriosis (RR 0.93, 95% CI [0.92–0.95]), even though the results improved during long‐term follow‐up due to the use of IVF. 14 Results from the cohort studies considering the impact of endometriosis on LBR are inconclusive. 10 , 15 , 16 , 17 , 18 However, considering ovarian and deep endometriosis, reassuring LBR results have been presented recently when including MAR pregnancies. 19 , 20 , 21 , 22 We previously found a reduction in the first LBR among women with endometriosis during the years preceding the eventual surgical diagnosis of endometriosis compared with reference women (IRR 0.51, 95% CI [0.49–0.52]) with no difference between the endometriosis subtypes. 23 In the present study, our aim was to assess the first birth rate and total fertility rate (TFR) across the entire reproductive life course among women with endometriosis in a nationwide setting. In addition, we assessed the statistical difference of first births between the study cohorts with restricted mean survival time (RMST) of not giving birth from the 15 years of age until the age of 45 years. The same outcomes were reported among the sub‐cohorts of peritoneal, ovarian, deep and combined/other forms of endometriosis.

The authors have stated explicitly that there are no conflicts of interest in connection with this article.

The cohort formation of first surgically verified endometriosis during 1987–2012 ( n  = 49 956) from the Care Register for Health Care (CRHC), and age and residence‐matched reference women has been described previously. 24 Index day was the hospital discharge day after the first surgically verified diagnosis of endometriosis with correspondent day in the reference cohort. In our previous study, we formed a sub‐cohort of fertile‐aged women (15–49 years) at index day between January 1, 1998 and December 31, 2012 ( n  = 18 324) excluding those with first birth, hysterectomy or bilateral salpingo‐oophorectomy (or unilateral done twice) before the age of 15. This was done to include only the era of International Statistical Classification of Diseases and Related Health Problems 10th revision (ICD‐10) for more coherent clinical practice. In addition, we excluded women born during 1980–1999 due to small cohort size and proximity to index day limiting the fertile years. 23 Women with surgically verified endometriosis were further divided into sub‐cohorts according to the location of the disease: ovarian (ICD‐10 code N80.1), peritoneal (ICD‐10 codes N80.2 and N80.3) and deep endometriosis (ICD‐10 codes N80.4, N80.5, N80.80 and N80.81) allowing no subsidiary diagnosis of endometriosis. For this study, we complemented our previous cohort data with the Medical Birth Register and registers from the Social Insurance Institute of Finland, together with a data update until the end of 2019 from the CRCH, Register for Sterilization, and Statistics Finland for comprehensive information regarding births, history of infertility‐related treatments, and sociodemographic background information. We excluded nine women from our previous cohort due to inaccurate fertility‐related data (four women from the endometriosis and five women from the reference cohort). In addition to our previous study, we analyzed the sub‐cohort of combined/other forms of endometriosis in detail. Our final study cohort comprised 18,320 women with surgically verified endometriosis, with sub‐cohorts of peritoneal ( n  = 5786), ovarian ( n  = 6519), deep ( n  = 1267) and combined/other forms endometriosis ( n  = 4748) (Figure  1 ). A possible subsidiary code of adenomyosis (ICD‐10: N80.0) was accepted. Characteristics of the sub‐cohort of combined/other forms of endometriosis are presented in Supporting Information Table  S1 . The reference cohort included altogether 35,788 age‐nd residence‐matched women at the time of the index day, without surgical diagnosis of endometriosis until 2012. We excluded also women with a clinical diagnosis of endometriosis during 1983–2015, and those with a history of birth, hysterectomy, sterilization, or bilateral salpingo‐oophorectomy before 15 years of age, or inaccurate fertility‐related data (Figure  1 ). Flow chart of the study cohorts. a Index day: The hospital discharge day of surgical diagnosis of endometriosis and the correspondent day in the reference cohort. To define the proportion of women with infertility before the first birth, we gathered recorded diagnosis of infertility (ICD‐9: 628*; ICD‐10: N97*) and selected infertility‐related procedures (8117, 821*, LAA10‐LAA11, LCA30, LBF*, TLW10‐TLW12, TLW14 [national classification until 1996, NCSP NOMESCO Classification for Surgical Procedures since 1997]) from the CRCH together with register data maintained by the Social Insurance Institution of Finland concerning first medication purchase of clomiphene citrate, gonadotropin ([excluding serum gonadotropin and follitropin delta], GnRH agonists [excluding leuprolelin]), and insurance compensation of private sector healthcare (infertility‐related procedures) and recorded information concerning MAR in Medical Birth Register (Table  S2 ). As there was no specific IVF register during the study period, we identified women with a history of IVF from the different registers used. These included selected procedures (LAA10‐LAA11, TLW10‐TLW12, TLW14, and LCA30) recorded in the CRHC, register of compensations of private sector procedures and/or selected medication purchase data (gonadotropin together with GnRH agonist/antagonist within 90 days from the first gonadotropin purchase) maintained by the Social Insurance Institution of Finland, and/or a recorded procedure of IVF in the Medical Birth Register (Table  S2 ). We recorded the first date of which ever above‐mentioned registered event and from the private section where only Jan 1st of the year was registered instead of the specific date of the treatment. We linked Finnish national register data, maintained by the Finnish Institute of Health and Welfare (CRHC [in‐patient visits 1987–2019 and out‐patient visit 1998–2019], Register of Sterilizations [1987–2018], and Medical Birth Register [1987–2019]), the Digital and Population Data Service Agency (Central Population Register Data on live births 1938–2012), the Social Insurance Institute of Finland (medicine purchases for infertility treatment, and insurance compensations due to private sector IVF‐treatment 1996–2019) and from Statistics Finland (socioeconomic factors, emigration and death 1987–2019). From the CRHC, we gathered dates of the fertility‐related surgical procedures and diagnoses: sterilization (LGA* in 2018–2019 to complement data from Register for Sterilization until 2018), hysterectomy, bilateral salpingo‐oophorectomy (or unilateral salpingo‐oophorectomy done twice) and diagnostic codes for infertility. Registers, time periods, and retrieved information are described in Table  S3 . Kaplan–Meier censored follow‐up time started at the age of 15 years and was continued until first birth, hysterectomy, bilateral salpingo‐oophorectomy (or unilateral done twice [the date of the latter procedure]), sterilization, emigration, age of 50 years, death or 31st December 2019, whichever came first. In case of total and complete fertility rate (CFR) of all and parous women, the follow‐up continued until 31st December 2019. Our primary outcome was first birth rate. While first birth rate indicates overall first birth frequency, RMST provides insight into how long individuals remained without a first birth. 25 To analyze the statistical difference of first births between women with surgically diagnosed endometriosis and women without endometriosis, we used crude and adjusted RMST difference of not giving birth, that is, until the first birth. Results were reported according to endometriosis sub‐cohorts. Adjustable variables were birth cohort of women and registered IVF at any time before the first birth or before the end of follow‐up with nulliparous women. As secondary outcomes, we report (TFR) of all women (births per all women) and CFR among women achieving 50 years of age during follow‐up. TFR and CFR was calculated separately for parous women. The categorical variables of our study were compared using Pearson's Chi‐squared test and the continuous variables using two‐tailed t ‐test. For survival analysis, we used Kaplan–Meier estimator to evaluate the time of not giving birth between the endometriosis and reference cohort as well as to the sub‐cohorts. Cox's regression model showed a time‐delayed effect between the endometriosis and reference cohort and crossing survival curves in the analysis of sub‐cohorts, indicating time‐varying hazards. Therefore, the method was rejected due to the violation of the proportional hazard assumption, which was seen in statistical tests and in the graphical examination of scaled Schoenfeld residuals and log–log survival plots. The RMST and RMST difference with 95% confidence intervals (CI) were used as an estimate to analyze the statistical differences between cohorts, 25 , 26 similarly to previous studies in the field of gynecology. 27 , 28 This approach compares a defined area of the survival curve up to a specific time point. In this study, it interprets the average survival time of not giving birth, that is, to the first birth. The time point ( t ) was chosen to be at 45 years (30 years of follow‐up), based on the clinical knowledge of reproduction and the availability of IVF treatments in Finland restricted to women under 40 years in public healthcare. The covariates were chosen by their availability and by existing clinical knowledge to be adjusted. The birth cohort was determined by the woman's birth decade (1940–1949, 1950–1959, 1960–1969, 1970–1979); the decade of 1970 was used as a reference level. The IVF was used as a binary variable (yes/no) and was defined as any registered IVF existing any time before the first birth, or until the end of follow‐up, if not giving birth. 29 All analyses were performed using R Statistical Software (version 4.3.1; R Core Team 2023, Vienna, Austria), with additional packages (survival, survminer, survRM2) for survival analysis. 25

Table S1. Detailed information on endometriosis diagnoses comprising the sub‐cohort of Combined/Other endometriosis. Table S2. Registers used to search information regarding recorded infertility and IVF‐related information among women with surgically verified endometriosis 1998–2012 and reference women. The first date of whichever registered event was recorded. Table S3. Registers with time periods of the collected data used in the study. Table S4. Statistical comparison of first births between endometriosis sub‐cohorts. Crude and adjusted restricted mean survival time (RMST) differences (95% confidence intervals) of not giving birth between the sub‐cohorts of endometriosis from 15 to 45 years of age (restriction point at 30 years).