First live birth before surgical verification of endometriosis-a nationwide register study of 18 324 women

Endometriosis is a chronic inflammatory disease affecting approximately 10% of women of reproductive age ( Zondervan et al. , 2020 ). Even though typical symptoms—dysmenorrhea, pelvic pain, and dyspareunia together with sub/infertility—are well established, common diagnostical delay ranges from 6 to 7 years up to 12 years ( Arruda et al. , 2003 ; Nnoaham et al. , 2011 ; Becker et al. , 2021 ; Taylor et al. , 2021 ). It is therefore likely that endometriosis has an impact on fertility years prior to its eventual diagnosis. According to recent ESHRE guidelines, endometriosis diagnosis can also be based solely on clinical findings, without surgical verification or medical imaging ( Becker et al. , 2022 ). Despite this evident progress in endometriosis diagnostics, surgery remains the gold standard to reliably classify the type of endometriosis, i.e. peritoneal, ovarian, and deep endometriosis ( Leyland et al. , 2010 ). Endometriosis is associated with a range of obstetric interferences, ranging from fertilization to newborn complications ( Leone Roberti Maggiore et al. , 2016 ; Lalani et al. , 2018 ; Horton et al. , 2019 ), but the results are partly inconsistent with a limited number of studies. Little is known about the possible effects of different subtypes of endometriosis on fertility, especially during the years preceding the diagnosis. In addition, data concerning live birth rate (LBR) among all women with endometriosis are scarce. According to studies comprising IVF/ICSI pregnancies, a recent British systematic review and meta-analysis suggests an association between stage III–IV endometriosis (American Society for Reproductive Medicine staging) and reduced LBR ( Horton et al. , 2019 ). A Danish national cohort study found a lower relative risk of childbirth (RR 0.93, 95% CI 0.92–0.95) in women with endometriosis compared to controls ( Hjordt Hansen et al. , 2014 ). However, the study found no difference in the average maternal age at first birth between the cases and controls (26.4 and 26.6 years). The possible impact of surgery on fertility remains enigmatic given the lack of comparative studies concerning conservative versus operative treatment of endometriosis. According to recent review articles, laparoscopic treatment of peritoneal endometriosis may improve spontaneous fertility rate compared to diagnostic laparoscopy only ( Bafort et al. , 2020 ; Hodgson et al. , 2020 ), and there are data that favor surgical management of deep endometriosis to improve fertility ( Darai et al. , 2017 ; Iversen et al. , 2017 ; Roman et al. , 2018 ). In contrast, an Australian review and meta-analysis concluded that there is uncertainty whether operative laparoscopy improves LBR compared to diagnostical laparoscopy ( Leonardi et al. , 2020 ). Caution is also warranted concerning the effects of bowel surgery on fertility outcomes due to the small size of the studies and potential surgical complications ( Vercellini et al. , 2018 ). Even though surgical treatment of ovarian endometrioma leads to diminished ovarian reserve, the impact of surgery on fertility prognosis is inconclusive ( Wu et al. , 2019 ). The aim of our retrospective register-based cohort study was to evaluate, whether endometriosis has an impact on first LBR already years before the surgically verified diagnosis in comparison to women with no endometriosis diagnosis. Our main outcome was incidence rate (IR) and incidence rate ratio (IRR) of first live birth. We compared the age at the first delivery and the number of parous women in each type of endometriosis (ovarian, peritoneal, and deep endometriosis). To our knowledge, this is the first cohort study conducted in such a setting.

The characteristics of the women with and without endometriosis at the index day are presented in Table 1. The mean time of follow-up was 15.2 years (SD 6.9) before the surgical diagnosis of endometriosis, set at a median age of 35 years (IQR 30.0–41.4) or index day. Among sub-cohorts, women with ovarian endometriosis were the oldest: their median age was 3.4 and 3.2 years higher compared to women with peritoneal and deep endometriosis, respectively ( P  < 0.001). Characteristics of age and residence matched women in the reference and endometriosis cohorts at the index day. a n (%) unless otherwise stated. IQR, interquartile range. a   The day of hospital discharge following surgical verification of endometriosis. b   Peritoneal versus ovarian: P  < 0.001; deep versus ovarian: P  < 0.001; deep versus peritoneal: P  = 0.12. c   Urban, city; densely populated, area where 200 people or more are living nearby; rural, under 200 people living nearby, usually over 200 m between buildings. d   The highest education level according to the statistics of 2014. e   From the registered inpatient and outpatient hospital visits, diagnosis according to the ICD-9 (1987–1995) and ICD-10 (1996–2012), divided by women who had at least one outpatient or hospital admission within 3 years before the index day. f   Time from 15 years of age until the first live birth, bilateral oophorectomy, sterilization or hysterectomy, or the index day—whichever came first. g   Peritoneal versus ovarian: P  < 0.001; deep versus ovarian: P  < 0.001; deep versus peritoneal: P  = 0.12. Distribution of age groups, birth cohort groups and highest education at the time of surgical verification of endometriosis or corresponding index day are shown in Table 1 . Women with diagnosed endometriosis had higher level of education than the reference women ( P  < 0.001). Women with endometriosis used medical services for gynecological reasons more often than the reference cohort during the three years preceding the index day: 74.8% of women (n = 13 698) in the endometriosis cohort had at least one hospital outpatient (registered since 1998) or inpatient visit (available since 1987) or both due to gynecological reasons, compared with 7.6% (n = 2732) of the reference cohort. Consequently, women with endometriosis had been more frequently diagnosed with infertility (ICD-9 code 628, ICD-10 code N97 as main diagnosis) and with gynecological pain (ICD-9 codes 6250A–6255A, ICD-10 code N94 as main diagnosis) (17.2% and 6.8%) compared with the reference cohort (3.2% and 1.4%), respectively. These diagnoses were most common in women with peritoneal endometriosis ( Table 1 ). The main results are shown in Table 2 . Altogether 65 362 live children were born to 31 081 women during the follow-up of 733 865 person-years. Of the live births, 641 were twins and 6 sets of triplets. There was a statistically significant difference in the mean age at first birth between endometriosis and reference cohort in the selected birth cohorts, 25.8 years (SD 4.7) and 25.7 years (SD 4.5), P  = 0.01. However, the difference is not clinically significant. Fertility outcomes of the women with live born in the reference and endometriosis cohort at index day a before the diagnosis of endometriosis. n (%) unless otherwise stated. IQR, interquartile range. a   The hospital discharge day of the first surgical diagnosis of endometriosis. b   T -test for mean, years (SD) was used: 25.7 (4.5) in the reference and 25.8 (4.7) in the endometriosis cohort. c   Peritoneal versus ovarian: P  < 0.001; deep versus ovarian: P  < 0.001; deep versus peritoneal: P  = 0.27. d   Number of children/parous women, mean (SD). e   Peritoneal versus ovarian: P  < 0.001; deep versus ovarian: P  < 0.001; deep versus peritoneal: P  = 0.17. We found that the first LBR of women with a subsequent surgical diagnosis of endometriosis was significantly lower compared to the reference cohort: IRR of first live birth was 0.51 (95% CI 0.49–0.52) ( Supplementary Fig. S1 ). Increasing age at first delivery was associated with progressively decreasing IRR of first delivery in women with subsequent endometriosis diagnosis ( Table 2 ). Demonstrating the difference between endometriosis and the reference cohort before the index day, and among the sub-cohorts of endometriosis, the cumulative age distributions of first live births are presented in Fig. 2 , showing similar results between ovarian, peritoneal, and deep endometriosis, but lower cumulation of the first live births in the group of other endometriosis. Of the women with other forms of endometriosis, consisting of several types of endometriosis (n = 4884), 1750 (35.8%) had a live birth prior to surgical verification of endometriosis (IR 2.30 [95% CI 2.20–2.41]). IRRs were 0.83 (95% CI 0.78–0.88) compared to women with ovarian endometriosis, 0.84 (95% CI 0.79–0.89) to peritoneal and 0.79 (95% CI 0.72–0.88) to deep endometriosis. Cumulative age distribution of first live births. At left: women before surgical verification of endometriosis (n = 18 324) and the reference cohort (n = 35 793). At right: women before surgical verification of ovarian (n = 6384), peritoneal (n = 5798), deep (n = 1267) endometriosis, and other (n = 4884) endometriosis. To clarify the effect of calendar time on the first delivery, the data are presented according to women’s birth cohorts in Fig. 3 . The difference in cumulative age distribution of first live births between endometriosis and reference cohort is highlighted by ascending birth cohort years, together with a trend of ascending age at first live birth. Respective data of the endometriosis sub-cohorts are presented in Supplementary Fig. S2 , demonstrating the similar effect of calendar time irrespective of the endometriosis subtype. Cumulative age distribution of first live births according to birth cohort of women before surgical verification of endometriosis and the reference women. Top left: birth cohort 1940–1949 (women before surgically verified endometriosis n = 107 and the reference women n = 213), top right: 1950–1959 (n = 2960 and n = 5817), down left: 1960–1969 (n = 6738 and 13 145), and down right: 1970–1979 (n = 8519 and n = 16 618).

The cohort of surgically diagnosed endometriosis during 1987–2012 (n = 49 956) was formed using the Finnish Hospital Discharge Register (FHDR) maintained by Finnish Institute for Health and Welfare ( Saavalainen et al. , 2018 ). In FHDR, diagnoses are recorded using International Statistical Classification of Diseases and Related Health Problems (ICD) revisions ICD-9 in 1987–1995 and ICD-10 since 1996. Surgical procedures are recorded using National League of Hospitals (1986–1996) and Nordic Medico-Statistical Committee (NOMESCO) Classification of Surgical Procedures since 1997. All first endometriosis-associated diagnoses with relevant concomitant procedural codes were accepted. The index day was the day of hospital discharge following the first endometriosis-related surgery. The quality assessment of the surgically diagnosed endometriosis was accurate in 94.6% of cases when data in FHDR were compared with hospital records ( Saavalainen et al. , 2018 ). For each woman with surgically diagnosed endometriosis, two reference women matched for age and residence at the time of index day, who had not been surgically diagnosed with endometriosis until 2012 (n = 98 824) were selected from the Finnish Central Population Register, kept by Digital and Population Data Services Agency. For this cohort study, we included women aged 15–49 years on the index day between 1 January 1998 and 31 December 2012 for this cohort study. The time limitation of the index day was selected to focus on data from a period of more coherent clinical and diagnostical practice. The participants were not to have a history of live birth, bilateral oophorectomy, or hysterectomy before 15 years of age. For the final cohort (n = 18 324), we excluded women born during 1980–1999 due to small birth cohort size at the index day and the birth cohort proximity to the index day. The final cohort of 18 324 women with surgical verification of endometriosis was further divided into four sub-cohorts according to the location of the disease: ovarian (ICD-10 code N80.1; n = 6384), peritoneal (ICD-10 codes N80.2 and N80.3; n = 5789), deep (ICD-10 codes N80.4, N80.5, N80.80, and N80.81; n = 1267), and other forms of endometriosis (ICD-10 codes N80.6, N80.8, N80.89, N80.9, and combined diagnoses; n = 4884). We did not allow any other subsidiary diagnoses of endometriosis in the sub-cohorts of ovarian, peritoneal, and deep endometriosis. The formation of the study cohort is depicted in Fig. 1 . From the final reference cohort (n = 35 793), we excluded women who were clinically diagnosed with endometriosis according to FHDR during 1983–2015 (n = 460) and those with a history of live birth or bilateral oophorectomy before 15 years of age (n = 9). Women aged 15–49 years with surgically verified endometriosis during 1998–2012 in Finland, divided into sub-cohorts according to the type of endometriosis. a) The hospital discharge day of surgical diagnosis of endometriosis. In this study, we linked data from Finnish national registers maintained by the Finnish Institute for Health and Welfare (FHDR: gynecological codes as main diagnosis [ICD-9: 614–629 and ICD-10: N70–N98] and the Register for Sterilizations 1987–2012), from the Digital and Population Data Service Agency (Central Population Register data on live births 1938–2012), and Statistics Finland in 1987–2012 (data on socioeconomical factors). Follow-up started at the age of 15 years and continued until the first live birth, sterilization, bilateral oophorectomy, hysterectomy, or until the surgical diagnosis of endometriosis or corresponding index day—whichever came first. The main outcomes were IR and IRR of first live births prior to surgical verification of endometriosis. In addition, we assessed fertility rate per parous woman (the number of all children divided by the number of parous women in cohort) and maternal age at first live birth in all cohorts before the index day. Descriptive statistics are presented as mean with standard deviation (SD) or median with interquartile range (IQR) for continuous variables and numbers with percentage for categorical variables. The data were compared with Pearson's Chi-squared test or Fisher's exact test for the qualitative variables and the Mann–Whitney U test for quantitative variables. For continuous variables, two-sample independent t -test with 95% CI was performed. To study the IR, the number of the first live births was divided by the accumulated person-years of the specific cohort. The incidence rate ratio (IRR) was a division between the IRs of the cohorts. A P -value of <0.05 was considered statistically significant. The cumulative frequencies were plotted as curves according to age at first live birth, birth cohorts of the participants, or endometriosis sub-cohorts. The statistical calculations were performed using RStudio version 1.4.1103. The study was approved by the ethics committee of the Hospital District of Helsinki and Uusimaa (238/13/03/02/2013). The register-keeping organizations (Finnish Institute for Health and Welfare, Digital and Population Data Service Agency, and Statistics Finland) gave their permission to use their sensitive health data in this study.

We found that the IR of first live birth was reduced by approximately half in women with subsequent surgical diagnosis of endometriosis compared to women with no such diagnosis. Similarly, fertility rate among parous women was significantly lower in women with an eventual surgical diagnosis of endometriosis. We found no differences in IRRs of first live birth between the endometriosis sub-cohorts. We found no clinically significant difference in the age at first birth between the cohorts. To our knowledge, this is the first national cohort study assessing the impact of endometriosis on fertility prior to its surgical verification. The effect of endometriosis on LBR is controversial and understudied. In addition, the possible effect of ART may confound the results, as is the case with operative treatment of endometriosis. A recent Australian population-based cohort study evaluated the risk of adverse pregnancy outcomes in women with endometriosis and women with ART both separately and together. The study found that endometriosis and ART are both independent risk factors for preterm birth, antepartum hemorrhage, placenta previa, and planned birth (caesarean delivery or induced labour) ( Ibiebele et al. , 2022 ). However, they did not assess LBR. A Danish national register-based cohort study addressing both natural and ART pregnancies found a 7% lower overall relative risk for childbirth in women with endometriosis compared to the reference cohort ( Hjordt Hansen et al. , 2014 ). When stratified by the mode of conception, the study found that the chance of delivery improved in both cohorts during the 15-year follow-up due to ART. A population-based retrospective cohort study focusing on the impact of endometriosis on IVF/ICSI outcomes found the LBR similar or even higher in women with endometriosis compared to women with other reasons for infertility ( Senapati et al. , 2016 ). If endometriosis was associated with concomitant infertility diagnoses, the LBR was lower in women with endometriosis compared to those with other reasons for infertility. In contrast to our results, an interview data-based cohort study in the USA assessing the consumed time to achieve pregnancy before surgically diagnosed endometriosis (n = 473) found no significant difference between women diagnosed with endometriosis compared to women with various surgical diagnoses ( Buck Louis et al. , 2016 ). This study did not, however, report unambiguous clinical pregnancy rate or LBR. In a British systematic review and meta-analysis, reduction in LBR was mostly seen in stage III–IV endometriosis studied in women with IVF/ICSI pregnancies ( Horton et al. , 2019 ). In addition, the effect of surgical treatment of endometriosis on LBR is poorly studied ( Bafort et al. , 2020 ). Our results, on the contrary, show a significant reduction in first LBR irrespective of the endometriosis subtype already prior to its surgical verification. Moreover, we found no significant differences in IRRs of first live birth between endometriosis sub-cohorts. Interestingly, the fertility rate of parous women was lowest in the sub-cohort of ovarian endometriosis compared to those with peritoneal and deep endometriosis, even if women with ovarian endometriosis were the oldest at index day and thus had been fertile in previous years. On the other hand, even if women in the peritoneal sub-cohort had been diagnosed more often with infertility, they still achieved similar first live birth results as other sub-cohorts. These findings might be due to differences in pathophysiology in different forms of endometriosis. Moreover, in the peritoneal sub-cohort, younger maternal age can at least partly compensate for the results. Compared to reference women, women with endometriosis had more outpatient and inpatient visits preceding the index day, offering them a chance to discuss about family planning. The quality of symptoms might lead women with peritoneal disease to seek professional care earlier, leading more often to hospital referral and surgery with subsequent surgical diagnosis of endometriosis, and also registered diagnosis of infertility. Women with other forms of endometriosis together with combined subtypes constituted a heterogenous group. The lower IR of first live birth in that group compared to results in ovarian, peritoneal, and deep sub-cohorts warrants further research; a complex disease might impair fertility more than one specific subtype of endometriosis. In contrast to other studies, the significantly lower overall results of LBR among women with endometriosis in our study might be caused by restricting the follow-up to time prior to surgery. Moreover, the referral cohort of our study was population based, rather than focused on infertile women of other etiology than endometriosis, as in many of the previous studies. Our results concur with the Danish cohort study, where age at first birth was similar between the endometriosis and the control cohort (26.4 versus 26.6 years) ( Hjordt Hansen et al. , 2014 ). According to these results, it seems that even though women with endometriosis have first live birth at the same age as women without, the incidence of first birth remains lower. A general increasing trend in Finland regarding the age at first delivery has most likely affected both the cases and references in our study, as seen in the results. Among all Finnish women, age at first delivery was as low as 23.5 years in 1968, but in three decades, it has approached 28 years and reached 30 years in 2021 (Statistics Finland). A national declining trend in the total fertility rate is similarly evident: a shift from 3.16 (1950) to 1.80 (2012), with a further decline to 1.46 in 2021 (Statistics Finland). It seems that current general knowledge of natural fertility and age-related decrease in female fertility among fertile-aged women is insufficient ( Virtala et al. , 2011 ; Mortensen et al. , 2012 ). Even though the diagnosis of endometriosis may cause distress to the affected women, a delay in diagnosis may complicate family planning and possible infertility treatments further ( Moss et al. , 2021 ). Birth cohort analysis of our study demonstrates differences in the cumulative distribution of first live births. The difference between women with endometriosis and the reference cohort seemed to grow in relation to calendar time: women in the older birth cohorts had higher rates of first live births than women in the younger cohorts. Thus, it seems that together with increasing age at first delivery, the negative impact of endometriosis on fertility leads to progressively lower rates of first live birth already prior to surgical diagnosis. This affects the total fertility rate of women. Besides age and general trends in population, other possible causes to impact the rate of first live birth might be related to lifestyles. In our study, women with endometriosis had higher educational levels than those in the reference cohort. A Nordic register-based study shows that high education among women was formerly rarer and positively associated with nulliparity in Finland. However, the differences in the proportions of nulliparity according to the educational level decreased and vanished from the birth cohort of 1965 onward ( Jalovaara et al. , 2018 ). Even though over half of the women in our study were born after 1965, we assume that the differences in the educational level between the cohorts might lead to an overestimation of the difference of first LBR between the cohorts. In addition, the selection bias caused by the surgical diagnosis of endometriosis possibly leads to unbalance of women with other chronic severe diseases, alcoholism, or drug addiction between the cohorts affecting negatively to first LBR in the reference cohort. Moreover, there might have been other factors affecting fertility, such as physical activity and the amount of sleep, that we could not address by means of register-based study. According to previous studies, endometriosis is consistently associated with lean body size ( Vigano et al. , 2012 ; Farland et al. , 2017 ; Liu and Zhang, 2017 ) and the relation was seen also in a Finnish population-based study ( Rossi et al. , 2021 ). Thus, obesity is an unlikely factor influencing lower LBR among women with endometriosis. However, endometriosis is associated with several other co-morbidities ( Gambadauro et al. , 2019 ; Zondervan et al. , 2020 ; Taylor et al. , 2021 ), also affecting mental health and quality of life ( Wang et al. , 2021 ). Together with pelvic pain and dyspareunia, these endometriosis-associated factors most likely complicate intimate relationships and family planning, in addition to the pathophysiological processes of the disease. The strength of our study is the large nationwide cohort with high-quality Finnish national health care registers. In Finland, administrative and health care personnel have a long tradition of collecting population data using standard protocols, based on a personal identity code given to every resident born in Finland since 1967. Well-established and comprehensive registers with continuous data collection enable reliable data linkages. The register quality of Finnish Institute for Health and Welfare has been assessed and the institute has a responsibility to check the accuracy of the collected data ( Gissler et al. , 1995 ). The quality of Finnish health registers has been evaluated in the field of gynecology ( Sund, 2012 ; Heino et al. , 2018 ; Helle et al. , 2022 ). A bidirectional quality assessment of FHDR data considering the first diagnosis of endometriosis in surgery was found reliable ( Saavalainen et al. , 2018 ). In addition, Finnish health care provides easy-access, low-cost treatment for all citizens, regardless of the socioeconomical status, and the diagnostic codes are used for clinical purposes. Our study increases the population-based knowledge of LBR among women with endometriosis and provides a new perspective on fertility without the possible confounding factor of endometriosis surgery. Our study is limited due to its focus on only surgically verified endometriosis, which may cause selection bias ruling out women with conservatively diagnosed and treated endometriosis with milder symptoms. The surgical diagnosis could also have been set during the infertility workup favouring the inclusion of women with infertility in this study. Even though the quality of the endometriosis diagnoses was good and we restricted our study period (era using ICD-10 only), we cannot rule out the possibility of interpersonal variation in diagnostical practice and the possible differences between calendar periods and hospital districts. Moreover, we had no data on pregnancy intentions of the women in our study cohorts. Also, we could not rule out the possible effect of ART and the significant proportion of infertility treatments is provided by private sector health care. In addition, we could not assess the possible impact of coexisting adenomyosis. According to current knowledge, adenomyosis impairs fertility among women with ART and is associated with adverse pregnancy and neonatal outcomes regardless of the mode of conception ( Nirgianakis et al. , 2021 ; Bourdon et al. , 2022 ). Furthermore, differences in the level of education between endometriosis and reference cohorts can possibly confound the results. We focused on the LBR results prior to surgical verification of endometriosis and therefore we do not evaluate LBR of the participants after the surgery or corresponding index day. In conclusion, endometriosis seems to affect parity years before its eventual diagnosis. According to our birth cohort analyses, adverse effects of delayed childbirth were further highlighted in women with subsequent endometriosis diagnosis. We should raise fertility awareness in general, and especially among women with endometriosis due to the chronic nature of the disease. Moreover, evidence-based family planning counseling should be provided to all couples. The impairment of fertility among women with endometriosis should be minimized by offering relevant treatment for endometriosis without delay.

Click here for additional data file. Click here for additional data file.