Risks of gynecological and breast cancers in women with endometriosis or adenomyosis: A nationwide cohort study

Introduction Endometriosis has been associated with several risk factors for cancer, and population-based research has shown that women with endometriosis have a higher risk of several types of cancer. The aim of this study was to examine the risks of gynecological and breast cancers among women diagnosed with endometriosis or adenomyosis.

A nationwide cohort study was conducted between 1997 and 2018. All women diagnosed with endometriosis or adenomyosis (the first recorded diagnosis during the study period) in Sweden were followed for cancer diagnosis. The reference population was women not diagnosed with endometriosis or adenomyosis during the study period. Data were obtained from the National Patient Register (in patient and out patient specialist care settings), the Swedish Cancer Register, and the Total Population Register. Adjusted standardized incidence ratios with 95% confidence intervals were calculated for cancer. Women with prior gynecological organ removal or mastectomy were not excluded.

A total of 49 133 women were diagnosed with endometriosis or adenomyosis at a mean age of 42.7 years (standard deviation: 9.97 years). Of these, 1784 (3.6%) were diagnosed with gynecological or breast cancer. The overall standardized incidence ratio for gynecological and breast cancer in women with endometriosis or adenomyosis for the whole study period was 1.4 (95% confidence interval, 1.3–1.5) compared to those without. For cancers diagnosed in the same calendar year as endometriosis or adenomyosis (n = 370), the standardized incidence ratios were particularly high for endometrial cancer (standardized incidence ratio 47.0) and ovarian (standardized incidence ratio 39.6) cancer. For cancers diagnosed in subsequent calendar years (n = 1414), the standardized incidence ratios were 0.6 (0.5–0.8) for endometrial cancer, 0.7 (0.5–0.9) for cervical cancer, 1.2 (1.1–1.3) for breast cancer, and 1.9 (1.6–2.2) for epithelial ovarian cancers overall. When looking at ovarian cancer subtypes, the standardized incidence ratios were 1.5 (1.1–2.0), 2.8 (1.7–4.2), and 7.7 (5.4–10.6), for low-grade serous, endometrioid, and ovarian clear cell cancers, respectively.

Women with endometriosis or adenomyosis have a higher subsequent risk of ovarian and breast cancers compared to women without. This may represent important information for clinicians and a foundation for further studies on prophylactic interventions. Abbreviations | | | |---|---| | | | | | | | | | Key message Endometriosis or adenomyosis is associated with persistently increased risks of breast and epithelial ovarian cancers, especially clear cell, endometrioid, and low-grade serous subtypes. 1 INTRODUCTION Endometriosis is an inflammatory disease characterized by lesions of endometrial tissue outside the uterus, mostly occurring on the pelvic peritoneum and ovaries, generating a chronic inflammatory environment that can lead to severe pain and discomfort.1, 2 It affects approximately 10% of women of reproductive age3, 4 and is associated with decreased quality of life5 and infertility.6 The etiology of endometriosis is still unknown but shares some features with cancer, such as resistance to apoptosis, invasion of other tissues and adhesions in the abdomen with subsequent damage to the target organs.7 Endometriosis has also been associated with several risk factors for cancer3 and gene sequencing has demonstrated that about 20% of both epithelial ovarian cancer and deep endometriosis lesions have similar somatic cancer driver mutations.8, 9 Additionally, population-based research has shown that women with endometriosis have a higher risk of several cancer types.7, 10 However, there are several gaps in the identification of morphological subtypes of specific cancers associated with endometriosis. Further research is also needed to elucidate the timing of cancer onset in specific cancer types and subtypes and the long-term risk of cancer development in these women. The aim of the study was therefore to examine the short-term and long-term associations between endometriosis (including adenomyosis) and specific gynecological and breast cancers in Swedish women. 2 MATERIAL AND METHODS 2.1 Study population This nationwide cohort study was conducted between January 1, 1997 (baseline), and December 31, 2018. At baseline the number of women in Sweden was 4 201 715 with a total follow-up of 73 528 574 person-years and a mean follow-up of 17.5 years (standard deviation, SD: ±6.8). 2.2 Data sources The population of eligible women of 18 years and older, was collected from the Total Population Register (managed by Statistics Sweden). Women with endometriosis or adenomyosis were identified through the National Patient Register (The National Board of Health and Welfare) during the study period by having a recorded diagnosis code of endometriosis or adenomyosis (N80.0–9) according to the 10th revision of the International Classification of Diseases (ICD-10).11 Established in 1964, the National Patient Register reached nationwide coverage in 1987, and contains individual-level data on hospital admission and discharge diagnoses.11 From 2001, it was supplemented with data from outpatient specialist care settings.11 The Swedish National Cancer Register was used to collect data on gynecological and breast cancer diagnoses (outcome). Established in 1958, it has been mandatory for clinicians and pathologists to independently report all new cancer diagnoses in patients.12 The diagnosis is morphologically verified in 99% and the completeness of the registry is more than 96.3%.13 ICD-7 codes were used for identifying cancers (Table 1), the subtype of ovarian cancer was defined according to a combination of ICD 7 code 175 and the Systematized Nomenclature of Medicine (SNOMED) codes. | Type of cancers | ICD-7 codes | Cases | % | O | SIR | 95% CI | |---|---|---|---|---|---|---| | Ovarian (all) | 175 | 15 215 | 7.6 | 261 | 3.1 | 2.7–3.5 | | Epithelial (all) | 14 655 | 7.3 | 238 | 3.0 | 2.6–3.4 | | | Clear cell | 751 | 0.4 | 52 | 10.4 | 7.8–13.7 | | | Endometrioid | 1547 | 0.8 | 43 | 5.2 | 3.8–7.0 | | | Low-grade serous | 5379 | 2.7 | 64 | 2.1 | 1.6–2.7 | | | Non-epithelial | 560 | 0.3 | 23 | 4.9 | 3.1–7.4 | | | Cervical | 171 | 9114 | 4.6 | 83 | 1.1 | 0.9–1.3 | | Endometrial | 172, 174 | 29 496 | 14.8 | 240 | 1.8 | 1.5–2.0 | | Other gynecologicala | 176 | 4423 | 2.2 | 27 | 1.6 | 1.0–2.3 | | Breast | 170 | 141 596 | 70.9 | 1173 | 1.2 | 1.2–1.3 | | All | 199 844 | 100.0 | 1784 | 1.4 | 1.3–1.5 | - Note: Bold type: 95% confidence interval does not include 1.00. - Abbreviations: CI, confidence interval; ICD-7, 7th revision of the International Classification of Diseases; O, observed cancer diagnoses in women with endometriosis or adenomyosis; SIR, standardized incidence ratio. - a Vulva, vagina, or unspecified genital location. Data on individual-level sociodemographic factors used in the adjustments were obtained from the Total Population Register. Educational attainment was categorized into two levels: <12 years' or “≥12 years” of education. Region of residence was categorized into “large cities” or outside of large cites in “Southern Sweden” or “Northern Sweden.” Linkage between individual-level data from the different registers was performed using a pseudonymized version of the unique personal identification numbers given to all residents in Sweden.14 2.3 Statistical methods 3 RESULTS A total of 49 133 women were diagnosed with endometriosis or adenomyosis between 1997 and 2018, with a mean age of 42.7 years (SD: ±9.97) and a median age of 43 years. Table S1 shows the distribution of ICD-10 codes. The total follow-up time of women with endometriosis or adenomyosis was 993 863 person-years and the mean follow-up was 20.2 years (SD 4.3). Table S2 includes study population characteristics and person-years of follow-up and number of cancer events in women with and without endometriosis or adenomyosis. A total of 1784 (3.6%) women with endometriosis or adenomyosis were diagnosed with gynecological or breast cancers within the same calendar year (n = 370) or during subsequent calendar years (n = 1414). Table 1 shows that the SIRs were elevated for both epithelial and non-epithelial ovarian cancer, as well as for endometrial cancer and breast cancers in women with endometriosis or adenomyosis compared to those without. The results remained in the sensitivity analysis not including educational level and region of residence (Table S3). The analysis stratified by settings of endometriosis or adenomyosis diagnosis (Table S4) showed few cancer observations and less stable, albeit significant, associations for some cancer types (particularly ovarian subtypes) among women diagnosed with endometriosis or adenomyosis in outpatient settings. Table 2 shows that SIRs were particularly high for ovarian (39.6) and endometrial (47.0) cancers diagnosed in the same calendar year as endometriosis or adenomyosis. For cancers diagnosed in subsequent calendar years, the associations were attenuated but remained elevated for epithelial ovarian (SIR 1.9; 95% CI, 1.6–2.2) and breast (1.2; 1.1–1.3) cancers, with similar findings across <10 and ≥10 years of follow-up. The subsequent risks for endometrial and cervical cancers were lower in women with endometriosis or adenomyosis compared to those without during the study period. | Type of cancer | The same calendar year | Subsequent calendar yearsa | |||||||||| |---|---|---|---|---|---|---|---|---|---|---|---|---| | <10 years of follow-up | ≥10 years of follow-up | All subsequent | |||||||||| | O | SIR | 95% CI | O | SIR | 95% CI | O | SIR | 95% CI | O | SIR | 95% CI | | | Ovarian (all) | 108 | 39.6 | 32.5–47.8 | 85 | 1.8 | 1.5–2.3 | 68 | 1.9 | 1.5–2.4 | 153 | 1.9 | 1.6–2.2 | | Epithelial (all) | 92 | 36.2 | 29.2–44.4 | 80 | 1.8 | 1.5–2.3 | 66 | 2.0 | 1.5–2.5 | 146 | 1.9 | 1.6–2.2 | | Clear cell | 15 | 100.0 | 55.8–165.3 | 24 | 8.8 | 5.7–13.2 | 13 | 6.2 | 3.3–10.6 | 37 | 7.7 | 5.4–10.6 | | Endometrioid | 21 | 61.8 | 38.2–94.6 | 15 | 2.9 | 1.6–4.8 | 7 | 2.6 | 1.0–5.3 | 22 | 2.8 | 1.7–4.2 | | Low-grade serous | 20 | 23.5 | 14.4–36.4 | 21 | 1.4 | 0.8–2.1 | 23 | 1.6 | 1.0–2.5 | 44 | 1.5 | 1.1–2.0 | | Non-epithelial | 16 | 84.2 | 48.0–137.1 | 5 | 1.8 | 0.6–4.3 | 2 | 1.2 | 0.1–4.3 | 7 | 1.6 | 0.6–3.2 | | Cervical | 31 | 9.0 | 6.1–12.8 | 40 | 0.8 | 0.6–1.1 | 12 | 0.5 | 0.2–0.8 | 52 | 0.7 | 0.5–0.9 | | Endometrial | 156 | 47.0 | 39.9–55.0 | 39 | 0.6 | 0.4–0.8 | 45 | 0.7 | 0.5–0.9 | 84 | 0.6 | 0.5–0.8 | | Other gynecologicalb | 1 | 2.1 | 0.0–11.9 | 14 | 1.64 | 0.9–2.8 | 12 | 1.4 | 0.7–2.5 | 26 | 1.5 | 1.0–2.3 | | Breast | 74 | 2.5 | 2.0–3.2 | 619 | 1.2 | 1.1–13 | 480 | 1.2 | 1.1–1.3 | 1099 | 1.2 | 1.1–1.3 | | All | 370 | 9.4 | 8.5–10.4 | 797 | 1.2 | 1.1–1.2 | 617 | 1.1 | 1.0–1.2 | 1414 | 1.1 | 1.1–1.2 | - Note: Bold type: 95% confidence interval does not include 1.00. - Abbreviations: CI, confidence interval; O, observed cancer diagnoses; SIR, standardized incidence ratio. - a Does not include same calendar year. - b Vulva, vagina, or unspecified genital location. Table 3 shows that the risk for subsequent ovarian cancer was significantly elevated among women under 60 years of age with endometriosis or adenomyosis compared to those without during the study period, particularly for epithelial ovarian cancer, and especially for clear cell cancer. The risk of breast cancer was significantly increased among all age groups, except among women aged 50–59 years, whereas the risk for endometrial cancer was lower among women aged ≥60 years. | Type of cancer | 18–50 years of age | 50–59 years of age | 60–69 years of age | ≥70 years of age | |||||||| |---|---|---|---|---|---|---|---|---|---|---|---|---| | O | SIR | 95% CI | O | SIR | 95% CI | O | SIR | 95% CI | O | SIR | 95% CI | | | Ovarian (all) | 50 | 2.7 | 2.0–3.6 | 64 | 1.9 | 1.5–2.5 | 29 | 1.3 | 0.9–1.9 | 10 | 1.3 | 0.6–2.4 | | Epithelial (all) | 48 | 2.9 | 2.2–3.9 | 61 | 1.9 | 1.5–2.5 | 27 | 1.2 | 0.8–1.8 | 10 | 1.3 | 0.6–2.4 | | Clear cell | 19 | 16.7 | 10.0–26.1 | 14 | 6.4 | 3.5–10.8 | 3 | 2.5 | 0.5–7.3 | 1 | 3.5 | 0.0–19.8 | | Endometrioid | 7 | 2.9 | 1.2–6.0 | 9 | 2.7 | 1.2–5.1 | 4 | 2.5 | 0.7–6.6 | 2 | 3.5 | 0.3–12.7 | | Low-grade serous | 10 | 2.0 | 0.9–3.6 | 14 | 1.2 | 0.6–2.0 | 15 | 1.7 | 0.9–2.7 | 5 | 1.5 | 0.5–3.5 | | Non-epithelial | 2 | 1.0 | 0.1–3.5 | 3 | 1.8 | 0.3–5.3 | 2 | 3.3 | 0.3–12.3 | 0 | || | Cervical | 37 | 0.8 | 0.6–1.1 | 12 | 0.7 | 0.4–1.3 | 2 | 0.3 | 0.0–1.1 | 1 | 0.4 | 0.0–2.1 | | Endometrial | 13 | 0.8 | 0.5–1.5 | 38 | 0.7 | 0.5–1.0 | 22 | 0.5 | 0.3–0.7 | 11 | 0.6 | 0.3–1.0 | | Other gynecologicala | 11 | 3.4 | 1.7–6.1 | 5 | 0.9 | 0.3–2.0 | 2 | 0.4 | 0.0–1.6 | 8 | 2.6 | 1.1–5.1 | | Breast | 331 | 1.2 | 1.0–1.3 | 378 | 1.1 | 1.0–1.2 | 285 | 1.2 | 1.1–1.4 | 105 | 1.5 | 1.2–1.8 | | All | 442 | 1.2 | 1.1–1.3 | 497 | 1.1 | 1.0–1.2 | 340 | 1.1 | 1.0–1.2 | 135 | 1.3 | 1.1–1.6 | - Note: Bold type: 95% confidence interval does not include 1.00. - Abbreviations: CI, confidence interval; O, observed subsequent cancer diagnoses; SIR, standardized incidence ratio. - a Vulva, vagina, or unspecified genital location. Table S5 shows that the most frequently diagnosed subtypes of ovarian cancers in women with endometriosis or adenomyosis were low-grade serous (n = 64, 24.5%), followed by clear cell (n = 52, 19.9%) and endometrioid (n = 43, 16.5%) ovarian cancers. 4 DISCUSSION In this nationwide study of 49 133 women diagnosed with endometriosis or adenomyosis, 1784 (3.6%) were diagnosed with gynecological or breast cancer (1997–2018). In the same calendar year of diagnosis, women with endometriosis or adenomyosis had a 40-fold higher risk of ovarian cancer and a 47-fold higher risk of endometrial cancer compared to women without these diagnoses. Throughout follow-up, the risks for subsequent epithelial ovarian cancer among women with endometriosis or adenomyosis were nearly twice that of women without endometriosis or adenomyosis diagnosed during the study period, whereas the risk of endometrial cancer was approximately half as high. Several studies have described the coexistence of endometriosis and gynecological cancers, particularly ovarian cancer.17 Therefore, our findings of high risks of gynecological cancers diagnosed in the same calendar year as endometriosis or adenomyosis were expected. However, we did not expect the high magnitude of the association, with SIRs ranging from 9.0 for cervical to 100.0 for ovarian clear cell cancer. Explanations for these associations may include that endometriosis or adenomyosis is diagnosed as a result of investigations prompted by suspicion of gynecological cancers or vice versa, or as an unexpected finding during surgery for such cancers. An earlier Swedish study18 found an overall SIR for ovarian cancer of 1.4 (95% CI, 1.2–1.7) during follow-up (1969–2001) and 2.2 (1.4–3.4) after 10–15 years of follow-up after a diagnosis of endometriosis or adenomyosis. Our study found SIRs of similar magnitude for ovarian cancer during more recent years (1997–2018) and includes data from specialized out patient care settings, where women with endometriosis or adenomyosis are also diagnosed. A study on national register data from Finland (1987–2012) also observed a similar overall SIR for ovarian cancer (1.8; 95% CI, 1.5–2.1), and similar SIRs for ovarian endometrioid (3.1; 95% CI, 2.2–4.4) and clear cell (5.2; 95% CI, 3.2–7.9) cancers.19 The same study also observed, as in our study, that the overall SIR for ovarian cancer remained elevated >10 years from the diagnosis of endometriosis or adenomyosis (1.9; 95% CI, 1.5–2.4).19 Additionally, a recent population-based study from Utah, United States,20 observed that ovarian cancer risk was higher among women with endometriosis compared to those without, with a crude hazard ratio of 3.6 (95% CI, 3.1–4.3) for ovarian cancer overall, and 10.9 (95% CI, 6.0–19.7) for ovarian clear cell cancer, 7.9 (95% CI, 5.5–11.2) for ovarian endometrioid cancer, and 7.3 (95% CI, 2.2–24.6) for ovarian low-grade serous cancer. The mean age of women in this study (42.7 years) was somewhat higher than in a Finnish study (36.4 years)19 and in the study from Utah (36 years).20 The reason for this difference may reflect a difference in diagnostic procedures or in indications for surgical procedures. Moreover, results from a recent Mendelian randomization study showed increased risks of clear cell and endometrioid ovarian cancers in women with endometriosis: odds ratio 2.0 (95% CI 1.7–2.5) and 1.5 (95% CI, 1.3–1.7), respectively.21 These findings also align with an earlier meta-analysis of previous studies.17 Although the association between endometriosis and ovarian cancer seems to be well established, the lifetime risk of ovarian cancer in women with endometriosis is less than 3%.10 Therefore, prophylactic salpingo-oophorectomy in women with endometriosis for risk reduction does not seem to be justified. Moreover, screening has not been shown to improve survival for ovarian cancer in general,22 although this has not been studied in women with endometriosis. Further research is therefore needed to examine surveillance and prophylactic options for ovarian cancer in women with endometriosis. The use of oral contraception as a means of reducing ovarian cancer risk could be a viable option, particularly since oral contraception is also a recommended treatment for endometriosis.23 Previous studies have not demonstrated a statistically significant increase in risk of endometrial cancer in women with endometriosis17, 19, 24; however, these studies did not stratify the follow-up time as in our study. We found that the overall SIR for endometrial cancer was significantly elevated in women with endometriosis or adenomyosis compared to those without (SIR 1.8), driven by the strong association within the same calendar year of diagnosis (SIR 47.0). In contrast, for cancers diagnosed during subsequent calendar years after endometriosis or adenomyosis diagnosis the SIR was 0.6 (95% CI, 0.5–0.8) and remained low (SIR 0.7) ≥10 years of follow-up. These findings suggest that endometriosis or adenomyosis does not confer an inherent long-term increased risk of endometrial cancer. Furthermore, the observed risk reduction for subsequent endometrial cancers may be attributed to the protective effects of hormonal treatments (contraception) for endometriosis or adenomyosis or hysterectomy due to endometriosis or adenomyosis. Similar associations were also observed between endometriosis or adenomyosis and cervical cancer. A decreased risk of cervical cancer in women diagnosed with endometriosis has also been observed by others,10, 17, 19, 25 and may be due to hysterectomy as a part of the treatment for endometriosis or more frequent visits to gynecologists, where earlier stages of cervical dysplasia are identified and treated. A previous study also found a lower incidence of cervical carcinoma in situ among women with endometriosis19 hypothesizing that decreased sexual activity due to endometriosis' pain could be a reason for this observation.19 We also found an increased risk of breast cancer among women with endometriosis or adenomyosis compared to those without (SIR 1.2; 95% CI 1.2–1.3), which persisted throughout follow-up and may be due to hormonal treatment for endometriosis.23 This is consistent with findings from a meta-analysis on 20 previous studies (summary relative risk 1.0; 95% CI 1.0–1.1) for breast cancer in women with endometriosis compared to those without.17 Women with endometriosis should be informed about the elevated breast cancer risk, enabling them to make informed decisions about breast cancer screening which in Sweden is recommended to start at 40 years of age.26 The estimated prevalence of endometriosis is approximately 10%3, 4 among women of reproductive age. We identified 49 133 (1.2%) women with a registered endometriosis or adenomyosis diagnosis during a study period of 22 years in a population of approximately 4.2 million women, suggesting underreporting of the condition. As 91.7% of the codes for the endometriosis diagnosis were three-digit ICD codes (N80) and only 3.9% were recorded as adenomyosis, it was not possible to distinguish how adenomyosis may have influenced the results. The study cohort may consist of women with more severe symptoms or those who underwent abdominal surgery for other reasons. Women with cancer-related symptoms are more likely to receive investigations revealing undiagnosed endometriosis or adenomyosis, or vice versa. Detection bias may also persist for several years after endometriosis or adenomyosis diagnosis, as women with persistent symptoms often undergo diagnostic procedures, which can lead to earlier cancer detection. Confounding by medications, infertility, and cancer risk factors, such as smoking, obesity, or cervical HPV status may exist and needs to be addressed in future research. We did not exclude women diagnosed with endometriosis or adenomyosis before the study period, or those with prior gynecological organ removal or cancer diagnosis, which may be limitations that also need to be addressed in future research. Moreover, as an observational study, causality cannot be inferred. Another limitation is that we used calendar year of endometriosis and adenomyosis, and cancer diagnoses and these findings were interpreted as coexistence of endometriosis and cancer. There are also several strengths in our study that balance the limitations. It is based on high-quality national register data for both endometriosis and cancer diagnoses, that is, not self-reported. We included data from outpatient specialist care settings (2001–2018), capturing more women diagnosed with endometriosis or adenomyosis than previous studies relying on inpatient data. However, the study period with outpatient data was shorter and included few observations, with less stable, albeit significant, SIR estimates for some cancer types. The use of a pseudonymized personal identification number ensured virtually no loss to follow-up. Lastly, our findings were consistent with previous studies. 5 CONCLUSION In this nationwide study of 49 133 women diagnosed with endometriosis or adenomyosis between 1997 and 2018 in Sweden, 370 were also diagnosed with gynecological or breast cancers in the same calendar year as endometriosis and 1414 in subsequent years. Women with endometriosis or adenomyosis had a higher risk of ovarian, endometrial, and breast cancers compared to women without endometriosis or adenomyosis. For most cancers, the increased risks were confined to the same calendar year as endometriosis or adenomyosis diagnosis, with particularly high risks of endometrial and ovarian cancers. During subsequent calendar years of follow-up, the risks of endometrial and cervical cancers were lower in women with endometriosis or adenomyosis compared to those without during the study period, whereas the elevated risk persisted for breast and ovarian cancers, including ovarian clear cell carcinoma, endometrioid carcinoma, and low-grade ovarian serous carcinoma subtypes. These findings indicate that clinicians should have a higher index of suspicion for gynecological or breast cancers in women diagnosed with endometriosis. Whether the persisting increased risk of subsequent breast and epithelial ovarian cancers represents an opportunity for earlier cancer detection or prophylactic interventions remains to be studied. AUTHOR CONTRIBUTIONS LM—Conceptualization; Formal analysis; Writing—original draft. XL—Data curation; Formal analysis; Writing—reviewing and editing. FH—Conceptualization; Formal analysis; Writing—reviewing and editing. FJ—Conceptualization; Formal analysis; Writing—reviewing and editing. AD—Conceptualization; Formal analysis; Writing—reviewing and editing. KS—Conceptualization; Formal analysis; Writing—reviewing and editing. CB—Conceptualization; Formal analysis; Writing—reviewing and editing; Supervision. FUNDING INFORMATION This work was supported by non-commercial research grants, that is, governmental funding granted to Filip Jansåker (Alf-YF funding) for clinical research within the Swedish National Health Services, Region Skåne (Sweden). The funding sources had no role in the study design; the collection, analysis, and interpretation of data; the writing of the report; or the decision to submit the paper for publication. CONFLICT OF INTEREST STATEMENT All authors state that they have no conflict of interest. ETHICS STATEMENT This population-based national registry study was a non-intervention study based on pseudonymized secondary data obtained from the Swedish authorities and approved by the Ethical Review Board, Lund, Sweden (Dnr. 2012/795 (February 6, 2013) and later approved amendments). DATA AVAILABILITY STATEMENT This study made use of several national registries and, owing to legal concerns, data cannot be made openly available. Further information (including accessibility to these data) regarding the nationwide registries is available from the Swedish National Board of Health and Welfare (https://www.socialstyrelsen.se/en/statistics-anddata/registers/) and Statistics Sweden (https://www.scb.se/en/). The code used in the analysis can be provided upon reasonable request.