Risk of diabetes and cardiovascular diseases in women with vaginal bleeding before 20 gestational weeks: Danish population-based cohort study

word count: 399 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 2 AB ST R ACT Obj ective: To investigate an association between childbirths affected by vaginal bleeding within 20 gestational weeks and women’s subsequent risks of diabetes and cardiovascular diseases. De s ign: Cohort study in Denmark, 1979-2018. Sett ing: Danish nationwide registries. Part ici p ants : Between 1979 and 2017, we identified 3,085,955 pregnancies among 1,329,331 women. Of these pregnancies, 69,230 ended in a childbirth affected by vaginal bleeding (VB) within 20 gestational weeks; 2,180,855 ended in a VB-unaffected childbirth; 578,355 ended in a termination, and 258,430 ended in a miscarriage. Ma in outcome m easur es : The outcomes were diabetes types 1 and 2, hypertension, ischaemic heart disease, including myocardial infarction, atrial fibrillation or flutter, heart failure, ischaemic and haemorrhagic stroke, coronary artery bypass grafting, and percutaneous coronary intervention. We computed the incidence rates per 10,000 person- years and cumulative incidences of cardiovascular outcomes at up to 40 years of follow-up. Hazard ratios (HRs) with 95% confidence intervals adjusted for age, calendar year, reproductive history, baseline comorbidities, and socioeconomic factors were computed using conventionally and inverse probability of treatment weighted Cox proportional hazards regression.

At the end of the follow-up, among women with VB-affected vs VB-unaffected childbirths, the cumulative risks of the outcome events were 4.5% vs 3.5% for diabetes type . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 3 1; 13.8% vs 11.1% for diabetes type 2; 25.1% vs 21.4% for hypertension; 10.6% vs 8.2% for ischaemic heart disease, including 2.9% vs 2.6% for myocardial infarction; 5.3% vs 4.4% for atrial fibrillation or flutter; 2.3% vs 2.0% for heart failure; 5.1% vs 4.0% for ischaemic stroke; 1.8% vs 1.4% for haemorrhagic stroke; 0.5% vs 0.4% for coronary artery bypass graft; and 2.4% vs 1.9% for percutaneous coronary intervention. Adjusted HRs were 1.2 to 1.3-fold increased for diabetes types 1 and 2, hypertension, ischaemic heart disease, atrial fibrillation or flutter, heart failure, ischaemic and haemorrhagic stroke as well as a coronary artery bypass gr afting . Following women’s first pregnancies and in sensitivity analyses, the HRs remained increased. For comparisons of VB-affected childbirths with terminations, HRs were up to 1.3-fold increased for diabetes and 1.1-fold increased for hypertension, ischaemic heart disease, atrial fibrillation or flutter, and haemorrhagic stroke. HRs were unity when we compared having a VB-affected childbirth with having a miscarriage.

Women’s risks of diabetes types 1 and 2 and multiple cardiovascular outcomes were increased on the relative scale following childbirths affected by VB within 20 gestational weeks when compared with VB-unaffected childbirths and terminations, but not when compared with miscarriages. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 4 W H A T I S A L R EA DY K NOW N O N THI S TOP IC • Vaginal bleeding is a common complication affecting up to 30% of clinically detected pregnancies. • Although an association between miscarriages and greater risk of cardiovascular events is known, the risk of cardiovascular outcomes following a childbirth complicated by vaginal bleeding not ending in a miscarriage is not well investigated. W H A T THI S ST UDY A D D S • This nationwide registry-based cohort study with up to 40 years of follow-up showed that compared with having a childbirth without vaginal bleeding, having a childbirth with vaginal bleeding within 20 gestational weeks was associated with woman’s increased risk of diabetes types 1 and 2, hypertension, ischaemic heart disease, including myocardial infarction, atrial fibrillation or flutter, heart failure, coronary artery bypass grafting, ischaemic and haemorrhagic stroke. • The cardiovascular risks were increased following women’s first pregnancies and when additionally adjusted for smoking and body-mass index. • The long-term cardiovascular risk pro file of women with vaginal bleeding-affected childbirths was similar to that of women with miscarriages. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 5 INTR O D UCTI O N Women’s reproductive health is a predictor of cardiovascular morbidity.[1] Women whose first pregnancy ended in a miscarriage have a 20% greater relative risk of hypertension and type 2 diabetes.[2,3] Furthermore, history of a miscarriage is asso ciated with a 20% increased risk of cardiovascular morbidity in women,[4] history of more than one miscarriage is associated with a 10-20% increased risk of stroke, thromboembolism, ischaemic heart disease, and 30% increased risk of cardiovascular diseases.[1,4,5] Vaginal bleeding (VB) is a pregnancy complication affecting up to 30%[6–8] of clinically recognised pregnancies and is associated with miscarriage.[9] Approximately a third of pregnancies complicated by VB ends in pregnancy loss.[10] At the same time, there are little data on whether VB in pregnancy not ending in a miscarriage is associated with subsequent cardiovascular morbidity. A previous Danish study[11] reported a 25-60% higher risk of multiple cardiovascular outcomes in women following a pregnancy with VB; however, analyses did not control for pre-pregnancy conditions and socioeconomic factors. Comparisons of VB-affected childbirths with alternatively ending pregnancies (terminations or miscarriages) are missing in the literature. The aim of this study was to investigate the association of VB within 20 gestational weeks of a pregnancy ending in a childbirth with the woman’s subsequent risks of diabetes, hypertension, ischaemic heart disease, including myocardial infarction, atrial fibrillation or flutter, heart failure, ischaemic and haemorrhagic stroke, coronary artery bypass grafting (CABG), and percutaneous coronary intervention (PCI). . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 6 M E THOD S Sett ing and design We condu cted a registry-based cohort study using prospectively and routinely recorded data of the Danish registries,[12,13] which are collected in the context of Denmark’s universal tax-supported healthcare. Data from all registries are linkable via a unique personal identifier stored in the Civil Registration System (CRS).[12] It captures the civil and vital status of all residents, country of origin, and immigration and emigration dates. Data on all deliveries are stored in the Medical Birth Registry (MBR) available from 1973.[14] Smoking and body mass index (BMI) in pregnancy became available in the MBR starting in 1991 and 2004, respectively. Hospital encounters and procedures are captured by the Danish National Patient Registry (DNPR).[15] It stores data on medical procedures, inpatient hospital encounters since 1977 and also on outpatient specialist clinic visit s and emergency room visits starting in 1995.[15] The diagnoses recorded in the DNPR are coded according to the international classification of diseases, 8 th revision (ICD-18) in 1977-1993 and 10 th revision (ICD-10) from 1994 onwards, while procedures are coded using the Danish Classification of Surgical Procedures and Therapies before 1996 and Nordic Medico- Statistical Committee Classification of Surgical Procedures afterwards.[15] The psychiatric admissions and outpatient specialist clinic encounters are available in this study starting from 1995 from the Psychiatric Central Research Register.[16] Nationwide data on medication dispensings are available starting in 1995 from the Danish National Prescription Registry[17]. This registry captures th e Anatomical Therapeutic Classification code of the prescribed medication, and the date prescription was redeemed.[17] Data on education,[18] . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 7 employment[19] and income[20] are available from 1980, measured yearly and stored at Statistics Denmark.[21] The cohor t of pr egnanci es aff ec t ed b y vagi n al b le eding e nding in a childbi rt h From the MBR,[14] we assembled the cohort of all pregnancies ending in a live- or stillbirth at the gestational age of at least 28 full weeks before 2004 and 22 full weeks thereafter. Each delivery of multiple pregnancies was counted as one childbirth. Among pregnancies ending in a childbirth, we identified those affected by VB within the first 20 gestational weeks using the DNPR. The index date for the affected pregnancies was the date of the healthcare encounter due to VB. When several VB episodes occurred within 20 gestational weeks of the same pregnancy, the first episode served as the qualifying event (Figure 1). The compar is on coho rts of pr egn anc ies unaf fect ed by vaginal ble eding e nding in a childbirt h, p regn anc ies ending in a t e r minatio n o r miscar riage The pregnancies ending in a childbirth found in the MBR and without a record of VB within 20 gestational weeks in the DNPR formed a comparison cohort of VB-unaffected childbirths and their index date was the date of delivery. We ascertained pregnancies ending in a miscarriage or termination using the primary and secondary diagnoses at any hospital visit. When a miscarriage or pregnancy termination encounter had several dates associated with it (hospital admission, procedure, hospital dischar ge), we selected the earliest of them as the index date (Table S1). A woman with a pregnancy termination encounter characterised by the same diagnostic or procedure code as her previous encounter could re-enter the population of terminations after 180 days. We assumed that the hospital encounter with the same pregnancy-related diagnostic or . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 8 procedure code within 180 days since an earlier such encounter may be related to the complications of the previous event. We applied the same algorithm for miscarriages. We excluded pregnancies of women with history of cardiovascular conditions and diabetes types 1 and 2 (N=78,456) before the index date; childbirths with missing gestational age (N=71,450); and pregnancies with different reproductive events (delivery, miscarriage, termination, or VB) recorded on the same date (N=6,749). Finally, we excluded pregnancies of women whose personal identifier could not be linked with the CRS data, pregnancies with potentially wrongly recorded women’s age (<10 year s), and pregnancies recorded after the woman emigrated from Denmark (N=14,094) (Figure 1). We also ascertained the subset of women having their first pregnancy. From this, we additionally excluded women with a record of pre-existing hypercholesterolaemia and hyperlipidaemia in the DNPR. Dia b et es ty pes 1 an d 2 an d c ar diova s cular outcomes The outcomes were incident diabetes type 1 and type 2, hypertension, ischaemic heart disease, including myocardial infarction, atrial fibrillation or flutter, heart failure, ischaemic and haemorrhagic stroke, CABG, and PCI. We used all available primary and secondary diagnoses recorded in connection with inpatient, outpatient specialist clinic, and emergency room encounters from the DNPR (1979-2018). Diabetes and hypertension outcomes were defined based on the earliest of the hospital-based diagnosis or medication dispensin g. For CABG and PCI, the outcomes were defined using procedure codes from the DNPR (Table S1). Covar i ab les Potential confounding factors were selected using the directed acyclic graph presenting the evidence on the associations or lack of thereof between the measured and unmeasured . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 9 variables in this study (Figure S1). These were women’s age, calendar year of pregnancy end, parity (nulli- or primiparous vs multiparous) as the number of delivered live- and stillborn infants, civil status, employment status, highest attained education, annual women’s income in quartiles; reproductive history (VB diagnosis irrespective of a later delivery record, history of VB within 20 gestational weeks of pregnancy ending in a childbirth, history of a termination or miscarriage), obesity, chronic obstructive pulmonary disease (COPD), chronic kidney and liver diseases, cancer, and psychiatric conditions before the index date. As proxies for the underlying health, we also ascertained the medication use history before the index date of antipsychotics, mood disorders medication, antiepileptics, non-steroid anti-inflammatory drugs [NSAIDs], steroids for systemic use, anti-infectives, and at least one prescription of a beta-blocker, RAS-acting agent, or a diuretic. Stat is t ic al analysis We followed each pregnancy from its index date until the earliest of each of the outcomes, emigration, death, or end of data on 31 December 2018. We constructed the cumulative incidence curves for the outcomes while treating death and emigration as competing risks and computed incidence rates (IR) of the outcomes per 10,000 person-years (PY) among women with VB-affected and VB-unaffected childbirths, terminations, and miscarriages.[22] We used Cox proportional hazards regression to compute the hazards ratios (HRs) for the outcomes of interest with the corresponding 95% confidence intervals (CIs). Different outcomes did not censor one another. To account for different pregnancies in the same woman, we computed the 95% CIs using a robust standard error estimation with su r v i va l R package.[23–25] . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 10 Contrasting women with VB-affected childbirths with each of the comparisons, we computed conventionally adjusted HRs using multivariable Cox regression models and stabilized inverse probability of treatment weighted (IPTW) models for an average treatment effect (ATE), where treatment refers to VB exposure.[26,27] We also computed HRs for average treatment effect in the treated (ATT) using ATT weights[28,29]. We truncated IPTW and ATT weight s at the 1 st and 99 th percentiles.[27] Finally, we repeated all analyses restricting to women’s first pregnancies (Table S2). We evaluated the proportionality of hazards assumptions by using log-minus-log plots and found no deviation of the curves from being parallel for all Cox proportional hazards regression models. We performed several sensitivity analyses. First, we additionally adjusted for smoking among women with childbirths in 1992-2017. Second, we restricted analyses to women with pregnancies in the ICD-10 era (1994-2017) (Table S3). Third, we further adjusted for smoking and pre-pregnancy BMI in the analyses of women with childbirths in 2004-2017 (Table S4). Fourth, we restricted to women’s last recorded pregnancies at the age of 40+ years and additionally adjusted for preeclampsia-eclampsia and placental pregnancy complications history (Table S5). For all analyses we used RStudio[30] and R[31] versions 3.6.0-4.1.0 including the following packages and their dependencies: tidyverse,[32] ggplot2,[33] survival,[24] cmprsk,[34] epiR[35]. All presented counts are masked as mandated by Statistics Denmark.[21] No patient informed consent or ethical approval is required for registry-based research according to Danish legislation. The study was reported to the Danish Data Protection Agency[36] through registration at Aarhus University (record number: AU-2016-051-000001, sequential number 605). . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 11 RESU L TS D e sc r i p t i v e ch a r a ct e r i s t i c s Between 1979 and 2017, we identified 3,085,955 pregnancies among 1,329,331 women. Of these, 69,230 (2.2%) ended in VB-affected childbirths; 2,179,940 (70.6%) ended in VB- unaffected childbirths; 578,355 (18.8%) ended in a termination, and 258,430 (8.4%) pregnancies ended in a miscarriage. VB-affected and VB-unaffected childbirths had a similar distribution of women’s age at delivery, parity, civil status, employment, highest attained education, income, history of obesity, COPD, chronic kidney and liver disease, alcohol abuse, and medication utilization (Table 1). VB-affected childbirths were more likely than VB- unaffected childbirths to be of women with a prior miscarriage (26% vs 12%) and a prior psychiatric condition (7.9% vs 6.7%). History of VB in pregnancy was most prevalent in VB- affected childbirths (Table 1). Pregnancies ending in a termination were more likely to be of women of younger age, nulliparous, not married, in the first income quartile and with a prior psychiatric condition (Table 1). Risks and r ates of diabet e s and card i ovas cular ou tcome s During the follow-up of up to 40 years, following VB-affected childbirth there were 1,265 events of diabetes type 1 (IR=8.3, 95% CI: 7.8-8.8 per 10,000 PY); 3,870 events of diabetes type 2 (25.7, 95% CI: 24.9-26.5); 6,485 events of hypertension (43.7, 95% CI: 42.6-44.8); 2,985 events of ischaemic heart disease (19.7, 95% CI: 19.0-20.5), including 745 events of myocardial infarction (4.9, 95% CI: 4.5-5.2); 1,015 events of atrial fibrillation or flutter (6.6, 95% CI: 6.2-7.1); 505 events of heart failure (3.3, 95% CI: 3.0-3.6); 1,295 ischaemic stroke . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 12 events (8.5, 95% CI: 8.0-8.9); 565 haemorrhagic stroke events (3.7, 95% CI: 3.4-4.0); as well as 80 CABG (0.5, 95% CI: 0.4-0.7) and 465 PCI events (2.5, 95% CI: 2.1-3.0) (Table 2). At the end of follow-up, for women with VB-affected vs VB-unaffected childbirths, the cumulative incidences of the outcomes were: 4.5% vs 3.5% for diabetes type 1; 13.8% vs 11.1% for diabetes type 2; 25.1% vs 21.4% for hypertension; 10.6% vs 8.2% for ischaemic heart disease, including 2.9% vs 2.6% for myocardial infarction; 5.3% vs 4.4% for atrial fibrillation or flutter; 2.3% vs 2.0% for heart failure; 5.1% vs 4.0% for ischaemic stroke; 1.8% vs 1.4% for haemorrhagic stroke; 0.5% vs 0.4% for CABG; and 2.4% vs 1.9% for PCI (Figure S2, Table S6). The cumulative incidences of the outcomes at 5, 10, 20, and 30 years of follow-up are reported in Table S6. Hazard ratios for diabetes and th e c a r diovas cular o utcomes When contrasting VB-affected vs VB-unaffected childbirths, there was a greater risk of diabetes type 1 (conventional adjusted HR: 1.22, 95% CI: 1.15-1.29), diabetes type 2 (HR: 1.21, 1.17-1.25) , hypertension (HR: 1.15, 1.12-1.18), ischaemic heart disease (HR: 1.31, 1.26- 1.36) and specifically myocardial infarction (HR: 1.17, 1.09-1.26), atrial fibrillation or flutter (HR: 1.19, 1.12-1.27), heart failure (HR: 1.21, 1.11-1.33), ischaemic (HR: 1.26, 1.19-1.33) and haemorrhagic stroke (HR: 1.27, 1.17-1.39). The risk was also increased for CABG (HR: 1.24, 0.99-1.55) but not for PCI (HR: 1.05, 0.96-1.16). The elevated risks persisted in the analyses restricted to women’s first pregnancies, and HR for PCI was 1.18, 0.98-1.43 (Table 3). The HRs for the ATT effect and conventional HRs were less pronounced than HR for ATE effect adjusted with IPTW. When comparing VB-affected childbirths with terminations, the risks were increased for diabetes type 1 (conventional adjusted HR: 1.28, 95% CI: 1.20-1.37), diabetes type 2 (HR: . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 13 1.34, 1.26-1.42) , hypertension (HR: 1.11, 1.08-1.14), ischaemic heart disease (HR: 1.11, 1.06- 1.16), atrial fibrillation or flutter (HR: 1.10, 1.02-1.18), haemorrhagic stroke (HR: 1.10, 1.00- 1.21), but not for myocardial infarction (HR: 1.00, 0.92-1.09) , heart failure (HR: 1.02, 0.92- 1.12), ischaemic stroke (HR: 1.04, 0.98-1.11), CABG (HR: 1.07, 0.84-1.37) and PCI (HR: 0.96, 0.86-1.06). Following first pregnancies, the associations remained unity for heart failure and atrial fibrillation or flutter and were 1.2 to 1.3-fold increased for diabetes types 1 and 2, hypertension, ischaemic heart disease, ischaemic and haemorrhagic stroke (Table 3). When contrasted with having a pregnancy ending in a miscarriage, having a VB-affected childbirth was not associated with diabetes type 1 or type 2, hypertension, ischaemic heart disease, including myocardial infarction, atrial fibrillation or flutter, heart failure, ischaemic, and haemorrhagic stroke in women (Table 3). Furthermore, no association was found for CABG and PCI (Table 3). The findings were similar when we followed women’s first pregnancies (Table 3). The results remained unchanged after additional adjustment for women’s smoking and BMI in pregnancy in the comparisons of VB-affected with VB-unaffected childbirths and when restricted to women with pregnancies recorded in 1994-2017. In analyses of women’s last childbirths at the age of 40+ years and additionally adjusted for the history of preeclampsia- eclampsia and placental complications, the associations remained for diabetes type 1, hypertension, ischaemic heart disease, and CABG (Table S7-10). . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 14 DI SC U S S I O N Pr i n c ipal f i n dings In this population-based cohort study at the end of up to 40 years of follow-up, the hazards of diabetes types 1 and 2 as well as of cardiovascular outcomes were 1.2 to 1.3-fold increased for women with VB-affected vs VB-unaffected childbirths. The associations persisted in the analyses restricted to first pregnancies and in the sensitivity analyses additionally controlling for smoking and body mass index. We found that the hazards of diabetes types 1 and 2, hypertension, ischaemic heart disease, atrial fibrillation or flutter, and haemorrhagic stroke were 1.1 to 1.3-fold greater in women following VB-affected childbirths vs terminations. No increased risks of the outcomes of interest were observed in the comparisons of women having VB-affected childbirths vs miscarriages. The findings suggested that having a pregnancy complicated by VB despite a childbirth was associated with increased cardiovascular risk. Comp arison with ot her stud i e s An earlier Danish study showed a greater risk of cardiovascular morbidity following pregnancies with VB.[11] In contrast to our study, it used a different definition of VB restricting the exposure window to 12 gestational weeks, the observation ended in 2007, investigated outcomes included overall stroke and overall diabetes types 1 and 2, no comparisons with pregnancies ending in a termination or miscarriage were carried out. Moreover, in contrast to our study, analyses in the previous study were not adjusted for socioeconomic factors and pre-pregnancy morbidity but were adjusted for post-exposure factors such as preterm delivery, prelabour rupture of membranes, foetal growth . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 15 restriction, placental abruption and stillbirth. These could be both mediators and colliders on the pathway from VB in pregnancy to subsequent cardiovascular morbidity in women. Adjusting for post-exposure factors that are colliders could result in inflated associations (Figure S3).[37] Underlying biological mechanisms of VB in pregnancy are not fully known. Markers of systemic inflammation are associated with both miscarriages[38,39] and VB in pregnancy[40–43] and were found to contribute to the development of cardiovascular disease and diabetes type 2.[44] Dislipidemia, insulin resistance, and obesity are risk factors of diabetes type 2 and hypertension, which result in endothelial dysfunction and microvascular inflammation and damage further leading to cardiovascular diseases.[45] The existence of common metabolic and pro-inflammatory pathways via cytokines (IL-1β, IL-6, TNF-α) for VB, miscarriage and vascular pathology could contribute to the explanation of the findings of this study showing an elevated cardiovascular risk in women with a VB-affected childbirth vs women having a VB-unaffected childbirth or a termination but not a miscarriage. Str engths a nd limitat ions of the stu d y One Danish study showed that most women with VB in pregnancy received ultrasound[6] required to establish foetal heart activity at bleeding presentation and exclude miscarriage and ectopic pregnancy. The number of false-positive VBs due to threatened abortion in this study is likely small. At the same time, we were unable to capture episodes of VB not resulting in a hospital-based contact and, therefore, could miss a considerable proportion of affected pregnancies.[6] We would also miss a proportion of very early miscarriages. Non- differential misclassification of pregnancies’ status would result in diluted associations. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 16 The diagnoses for miscarriage (spontaneous abortions) in the DNPR were previously validated and showed PPV of 97%[46]. The diagnostic and procedure codes for medical and surgical abortions (pregnancy terminations) were not validated in the DNPR, however, the provoked abortion diagnosis in the Register of Legally Induced Abortions had PPV of 94% and procedure codes, in general, have high validity in the DNPR.[15] The outcome diagnoses were validated with high PPV for diabetes (82-96%),[47] hypertension (92%), PCI (98%),[48] and slightly lower PPV for heart failure (76%),[48] and haemorrhagic stroke (71-80%)[49,50], leading to relatively few false-positive outcomes. The PPV for ischaemic stroke was 71-85%[49,50]; however, since the definition of ischaemic stroke in this study included undefined stroke diagnoses, a larger proportion of false positives may be present. Non-differential misclassification of the outcome is expected to dilute the associations toward or beyond the null value. Owing to the universal healthcare coverage, routine data collection and nearly complete follow-up, the selection bias due to lost-to-follow-up is not expected. The residual confounding by obesity is likely since we had data only on hospital-based diagnoses. Although additional adjustment of analyses for smoking and BMI did not meaningfully shift the HR estimates, the concern of unmeasured and residual confounding remains. An unmeasured confounder associated with VB and with cardiovascular outcomes with the hazard ratio of 1.7-1.8 beyond measured covariables could fully explain away the observed associations according to E-value analysis.[51] This study investigated the total effect of VB during pregnancy on the long-term risk of diabetes type 1 and type 2 and cardiovascular outcomes in women. Although we did not investigate the specific pathways of cardiovascular morbidity in women following a VB- . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 17 affected childbirth, associations with diabetes type 2 and hypertension suggest that metabolic syndrome may be one of them. We cannot exclude that part of the observed association between VB and the outcomes may be mediated via miscarriages of women’s subsequent pregnancies. CO N C LUSI O N Childbirths affected by VB within 20 gestational weeks were associated with an increased relative risk of diabetes types 1 and 2 and multiple cardiovascular outcomes in women when compared with VB-unaffected childbirths and terminations, but not miscarriages. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 18 Cont ribut ors : ED, EHP, HTS, VE designed the study. HTS facilitated the data application process. ED designed the statistical analysis plan. ED conducted the analyses, EHP provided expertise and supervision of the data analysis. All authors participated in the discussion of the results and their critical interpretation. ED drafted the manuscript, organised the writing, and produced a graphical presentation of the results. HTS provided clinical expertise. All authors equally participated in the critical revision of the manuscript and approved its final version. HTS is the guarantor. Funding : Department of Clinical Epidemiology, Aarhus University partaking in studies with institutional funding from regulators and pharmaceutical companies, given as research grants to and administered by Aarhus University. None of these projects is related to the current study. Comp eting inte rests: All authors report no competing interest Ethics approval: not required for registry-based resear ch in Denmark. Data sh a r ing: data is not available for sharing to protect the identity of participants according to Danish legislation. Patien t and p ub li c involveme n t: patients or the public were not involved in the design, or conduct, or reporting, or dissemination of this research. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 19 REFERE NCE S 1 Okoth K, Chandan JS, Marshall T, e t al. Association between the reproductive health of young women and cardiovascular disease in later life: umbrella review. BMJ 2020; 37 1:m3502. doi:10.1136/bmj.m3502 2 Horn J, Tanz LJ, Stuart JJ, et al. Early or late pregnancy loss and development of clinical cardiovascular disease risk factors: a prospective cohort study. BJOG: A n I n ternat io nal J ournal of Obstetrics & Gynaec ology 2019;12 6:33–42. doi:10.1111/1471-0528.15452 3 Harvey RE, Coffman KE, Miller VM. Women-specific factors to consider in risk, diagnosis and treatment of cardiovascular disease. Women s Health (Lo nd Engl ) 2015; 11:239–57. doi:10.2217/whe.14.64 4 Quenby S, Gallos ID, Dhillon-Smith RK, et al. Miscarriage matters: the epidemiological, physical, psychological, and economic costs of early pregnan cy loss. The La nc et 2021; 39 7:1658–67. doi:10.1016/S0140-6736(21)00682-6 5 Oliver-Williams CT, Heydon EE, Smith GC, et al. Miscarriage and future maternal cardiovascular disease: a systematic review and meta-analysis. Hear t 2013; 99:1636–44. doi:10.1136/heartjnl-2012-303237 6 Axelsen SM, Henriksen TB, Hedegaard M, e t a l. Characteristics of vaginal bleeding during pregnancy. Eur opean Journa l o f O bstetr ic s & G ynec ology and Repr o ductive B i o l ogy 1995; 63 :131–4. doi:10.1016/0301-2115(95)02236-8 7 Smits LJM, North RA, Kenny LC, et al. Patterns of vaginal bleeding during the first 20 weeks of pregnancy and risk of pre-eclampsia in nulliparous women: results from the SCOPE study. Acta Obs t etricia et G yn ecologic a Scan dinavic a 2012; 91:1331–8. doi:10.1111/j.1600-0412.2012.01496.x 8 Saraswat L, Bhattacharya S, Maheshwari A, et al. Maternal and perinatal outcome in women with threatened miscarriage in the first trimester: a systematic review. BJ O G 2010; 11 7:245–57. doi:10.1111/j.1471-0528.2009.02427.x 9 Coomarasamy A, Harb HM, Devall AJ, et al. Progesterone to prevent miscarriage in women with early pregnancy bleeding: the PRISM RC T. H ealth Technol As s ess 2020; 24 :1–70. doi:10.3310/hta24330 10 Coomarasamy A, Gallos ID, Papadopoulou A, et a l. Sporadic miscarriage: evidence to provide effective care. The Lancet 2021;397:1668–74. doi:10.1016/S0140- 6736(21)00683-8 11 Lykke JA, Langhoff-Roos J. First trimester bleeding and maternal cardiovascular morbidi ty. Eur J O bstet Gynec ol Repr od Biol 2012;164:138–41. doi:10.1016/j.ejogrb.2012.06.003 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 20 12 Schmidt M, Pedersen L, Sorensen HT. The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 2014;29 :541–9. doi:10.1007/s10654-014-9930-3 13 Laugesen K, Ludvigsson JF, Schmidt M, et al. Nordic Health Registry-Based Research: A Review of Health Care Systems and Key Registries. CLE P 2021; 13:533–54. doi:10.2147/CLEP.S314959 14 Bliddal M, Broe A, Pottegard A, et al. The Danish Medical Birth Register. Eur J Epidemiol 2018; 33 :27–36. doi:10.1007/s10654-018-0356-1 15 Schmidt M, Schmidt SA, Sandegaard JL, et al. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemio l 2015;7 :449–90. doi:10.2147/CLEP.S91125 16 Mors O, Perto GP, Mortensen PB. The Danish Psychiatric Central Research Register. Scand J Publ ic Heal th 2011; 39:54–7. doi:10.1177/1403494810395825 17 Pottegard A, Schmidt SAJ, Wallach-Kildemoes H, et al. Data Resource Profile: The Danish National Prescription Registry. Int J Epidemiol 2017;46:798–798f. doi:10.1093/ije/dyw213 18 Jensen VM, Rasmussen AW. Danish Education Registers. S cand J Public He alt h 2011; 39 :91–4. doi:10.1177/1403494810394715 19 Petersson F, Baadsgaard M, Thygesen LC. Danish registers on personal labour market af filiat ion. Sc a nd J P ubl i c H e al t h 2011;39:95–8. doi:10.1177/1403494811408483 20 Baadsgaard M, Quitzau J. Danish registers on personal income and transfer payments. Scand J Publ ic Heal th 2011; 39:103–5. doi:10.1177/1403494811405098 21 Statistics Denmark. 2021.https://www.dst.dk/en (accessed 30 Jun 2021). 22 Austin PC, Lee DS, Fine JP. Introduction to the Analysis of Survival Data in the Presence of Competing Risks. Cir c ulat ion 2016;133:601–9. doi:10.1161/CIRCULATIONAHA.115.017719 23 Therneau T, Grambsch P. M odel ing S ur v ival Dat a: Ex t end ing the Cox Mode l. New Yor k : Spr inger; 2000 . New York: : Springer 2000. 24 Therneau TM, Thomas L, Elizabeth A, e t a l. survival: Survival A n aly sis . 2021. https://CRA N.R-project.org/package=survival (accessed 6 Apr 2021). 25 Efron B. Nonparametric Estimates of Standard Error: The Jackknife, the Bootstrap and Other Methods. Bi o me t r i k a 1981;68 :589–99. doi:10.2307/2335441 26 Hernán M, Robins JM. Causal inferen ce. 2021. https://www.hsph.harvard.edu/miguel- hernan/causal-inference- book/#:~:text=To%20cite%20the%20book%2C%20please,Chapman%20%26%20Hall%2F CRC.%E2%80%9D (accessed 29 Jul 2021). . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 21 27 Cole SR, Hernan MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol 2008;168:656–64. doi:10.1093/aje/kwn164 28 Sato T, Matsuyama Y. Marginal structural models as a tool for standardization. Epidemiology 2003;14:680–6. doi:10.1097/01.ED E.0000081989.82616.7d 29 Sturmer T, Rothman KJ, Glynn RJ. Insights into different results from different causal contrasts in the presence of effect-measure modification. Pharmacoepide miol Dr ug Sa f 2006; 15 :698–709. doi:10.1002/pds.1231 30 RStudio Team. RStu dio: Int e g rated D e v elopment for R . Boston, MA: : RStudio, PBC 2020. http://www.rstudio.com/ 31 R Core Team. R: A la ngu a ge and environment for s tat is t ic al compu tin g . Vienna, Austria: 2013. http://www.R-project.org/ 32 Wickham H, Averick M, Bryan J, et al. Welcome to the Tidyverse. Jour nal of Ope n Sour c e Soft ware 2019;4 :1686. doi:10.21105/joss.01686 33 Wickham H. ggplot2: Elega nt G ra phi c s for D ata An alysis : Elegant Graph i cs for D ata A naly sis . Springer-Verlag New York 2016. https://ggplot2.tidyverse.org 34 Gray B. cmpr s k: S ub di str ibutio n A naly si s of Com peting Ris ks . 2020. https://CRAN.R- project.org/package=cmpr sk (accessed 31 Mar 2021). 35 Nunes MS and ES with contributions from T, Heuer C, Marshall J, et al. epiR : Tools for t he An alys is of Epidemiologic al Data . 2021. https://CRAN.R-project.org/package=epiR (accessed 29 Jul 2021). 36 Datatilsynet. Datatilsynet/The Danish Data Protection Agency. Datatilsynet/The Danish Data Protection Agency. http://www.datatilsynet.dk/english (accessed 16 Sep 2020). 37 Hernandez-Diaz S, Schisterman EF, H ernan MA. The birth weight ‘paradox’ uncovered? A m J Epidemiol 2006;164:1115–20. doi:10.1093/aje/kwj275 38 Ahmed SK, Mahmood N, Malalla ZH, et al. C-reactive protein gene variants associated with recurrent pregnancy loss independent of CRP serum levels: A case-control study. Gene 2015;56 9:136–40. doi:10.1016/j.gene.2015.05.052 39 Calleja-Agius J, Jauniaux E, Pizzey AR, et al. Investigation of systemic inflammatory response in first trimester pregnancy failure. Human R e pr oduct ion 2012; 27:349–57. doi:10.1093/humrep/der402 40 Calleja-Agius J, Schembri-Wismayer P, Calleja N, et al. Obstetric outcome and cytokine levels in threatened miscarriage. Gyn ec ol Endocrinol 2011;27:121–7. doi:10.3109/09513590.2010.487614 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horváth-Puhó E, Sørensen HT, Ehrenstein V. Diabetes and cardiovascular diseases in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 22 41 Calleja-Agius J, Muttukrishna S, Pizzey AR, et al. Pro- and antiinflammatory cytokines in threatened miscarriages. Amer ic an Jou rnal o f Obs t etrics and Gyne c ology 2011; 20 5:83.e8-16. doi:10.1016/j.ajog.2011.02.051 42 Alijotas-Reig J, Esteve-Valverde E, Ferrer-Oliveras R, et al. Tumor Necrosis Factor-Alpha and Pregnancy: Focus on Biologics. An Updated and Comprehensive Review. Clin Rev A ll er gy Immunol 2017; 53:40–53. doi:10.1007/s12016-016-8596-x 43 Gücer F, Balkanli-Kaplan P, Yüksel M, et al. Maternal serum levels of tumor necrosis factor-α and interleukin-2 receptor in threatened abortion: a comparison with normal and pathologic pregnancies. Fert i lit y and Ster i li t y 2001; 76:707–11. doi:10.1016/S0015- 0282(01)02002-7 44 Willerson JT, Ridker PM. Inflammation as a Cardiovascular Risk Factor. Circ ulati on 2004; 10 9:II–2. doi:10.1161/01.CIR.0000129535.04194.38 45 Petrie JR, Guzik TJ, Touyz RM. Diabetes, Hypertension, and Cardiovascular Disease: Clinical Insights and Vascular Mechanisms. Can J Cardiol 2018;34:575–84. doi:10.1016/j.cjca.2017.12.005 46 Lohse SR, Farkas DK, Lohse N, e t a l. V alidation of spontaneous abortion diagnoses in the Danish National Registry of Patients. C l in Epidemio l 2010;2 :247–50. doi:10.2147/CLEP.S13815 47 Thygesen SK, Christiansen CF, Christensen S, et al. The predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of Patients. BM C Med Res M eth odol 2011; 11 :83. doi:10.1186/1471-2288-11-83 48 Sundbøll J, Adelborg K, Munch T, et al. Positive predictive value of cardiovascular diagnoses in the Danish National Patient Registry: a validation study. BMJ Open 2016; 6 :e012832. doi:10.1136/bmjopen-2016-012832 49 Lühdorf P, Overvad K, Schmidt EB, et al. Predictive value of stroke discharge diagnoses in the Danish National Patient Register. Scand J Publ ic H eal th 2017;45:630–6. doi:10.1177/1403494817716582 50 Wildenschild C, Mehnert F, Thomsen RW, et al. Registration of acute stroke: validity in the Danish Stroke Registry and the Danish National Registry of Patients. C l in Epidemiol 2013; 6 :27–36. doi:10.2147/CLEP.S50449 51 Mathur MB, Ding P, Riddell CA, et al. Web Site and R Package for Computing E-values. Epidemiology 2018;29:e45. doi:10.1097/ED E.0000000000000864 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 23 Figure 1. Flow graph for t he c o hort s asc er tainm ent Table 1. De s cript ive c har ac t eristics of vag in al b leedin g-affect ed c hild birth s , vaginal bleeding- unaff ected pre gnancie s, p regnan cies e n ding in mis car riages , and pregnancie s ending in t erminat ions, Denm ark, 1979-2017 Co mpa r i s on c ohort s Charac t e r i stic VB-a ffected c hildbirt h s , N = 69 ,230 (among 65,395 wo m e n ) VB-unaffected chil dbir ths, N = 2,179 ,940 (among 1 ,172,715 women) Termina tions, N = 578,3 55 (among 4 17,765 women) M i s c arriag es , N = 258,430 ( a mong 219 ,680 women) n % n % n % n % Age a t preg nancy en d, ye ars <20 1,450 2.1 48,980 2.2 89,065 15.0 10,265 4.0 20-24 12,325 18.0 374,385 17.0 139,135 24.0 44,355 17.0 25-29 24,900 36.0 797,360 37.0 121,475 21.0 78,340 30.0 30-34 20,405 29.0 660,015 30.0 109,535 19.0 67,795 26.0 35-39 8,610 12.0 255,870 12.0 82,310 14.0 40,550 16.0 40+ 1,530 2.2 43,330 2.0 36,835 6.4 17,135 6.6 Cale ndar year of pregna ncy 1979-1987 13,555 20.0 379,850 17.0 141,705 25.0 52,485 20.0 1988-1994 19,660 28.0 474,450 22.0 118,945 21.0 73,380 28.0 1995-2002 17,145 25.0 427,135 20.0 120,030 21.0 45,485 18.0 2003-2009 11,730 17.0 471,285 22.0 114,480 20.0 52,135 20.0 2010-2017 7,140 10.0 427,220 20.0 83,195 14.0 34,940 14.0 Pari ty at the i ndex date or a t the d eliv ery da t e for pr eg nancie s endi ng i n a chil dbir th Nulliparous 0 0.0 0 0.0 412,780 71.0 220,920 85.0 Primiparous 32,790 47.0 1,028,400 47.0 45,560 7.9 15,450 6.0 Multiparous 36,440 53.0 1,151,540 53.0 120,015 21.0 22,060 8.5 Number of pregnanc ies at t he index dat e 1 23,075 33.0 923,450 42.0 254,080 44.0 98,490 38.0 2 21,615 31.0 701,650 32.0 116,395 20.0 76,845 30.0 3+ 24,535 35.0 554,840 25.0 207,880 36.0 83,100 32.0 Civi l s tatus Married or in partners hip 40,250 58.0 1,282,555 59.0 197,475 34.0 130,830 51.0 Not married 24,315 35.0 739,130 34.0 244,895 42.0 72,615 28.0 missing 4,660 6.7 158,255 7.3 135,980 24.0 54,985 21.0 Employ ment s tatus Employed 48,755 70.0 1,555,980 71.0 333,625 58.0 178,985 69.0 Retirement or state pension 6,285 9.1 166,125 7.6 70,190 12.0 23,880 9.2 Unemployed 10,430 15.0 325,190 15.0 125,985 22.0 37,975 15.0 missing 3,755 5.4 132,640 6.1 48,555 8.4 17,595 6.8 Highest att a ined educ ation . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 24 Co mpa r i s on c ohort s Charac t e r i stic VB-a ffected c hildbirt h s , N = 69 ,230 (among 65,395 wo m e n ) VB-unaffected chil dbir ths, N = 2,179 ,940 (among 1 ,172,715 women) Termina tions, N = 578,3 55 (among 4 17,765 women) M i s c arriag es , N = 258,430 ( a mong 219 ,680 women) n % n % n % n % Basic education 21,470 31.0 535,195 25.0 248,510 43.0 75,935 29.0 High school or similar 24,985 36.0 818,445 38.0 166,220 29.0 89,975 35.0 Higher education 16,130 23.0 605,295 28.0 73,005 13.0 62,330 24.0 missing 6,645 9.6 221,005 10.0 90,620 16.0 30,190 12.0 Yearly income, quart i les Q1 15,700 23.0 429,805 20.0 216,090 37.0 58,055 22.0 Q2 16,455 24.0 509,415 23.0 127,535 22.0 59,830 23.0 Q3 16,225 23.0 550,835 25.0 88,275 15.0 58,855 23.0 Q4 17,200 25.0 548,745 25.0 85,250 15.0 62,225 24.0 missing 3,645 5.3 141,140 6.5 61,205 11.0 19,465 7.5 Country of origin Denmark 60,610 88.0 1,920,450 88.0 511,065 88.0 227,980 88.0 Other 8,540 12.0 256,795 12.0 67,290 12.0 30,450 12.0 missing 75 0.1 2,695 0.1 0 0.0 0 0.0 Reproductive his tory VB diagnosis history 7,915 11.0 88,980 4.1 32,340 5.6 28,585 11.0 VB within 20 gestational weeks history 3,870 5.6 40,610 1.9 16,645 2.9 8,305 3.2 Previous termination of a pregnancy 15,875 23.0 341,820 16.0 166,515 29.0 52,030 20.0 Previous miscarriage 17,800 26.0 260,360 12.0 55,445 9.6 39,760 15.0 Comorbidities Obesity diagnosis 1,550 2.2 120,565 5.5 13,060 2.3 6,430 2.5 Thyroid disorders 980 1.4 34,380 1.6 6,620 1.1 3,685 1.4 COPD 660 1.0 18,295 0.8 4,885 0.8 2,165 0.8 Chronic kidney disease 125 0.2 3,265 0.1 925 0.2 410 0.2 Chronic liver disease 95 0.1 3,380 0.2 1,015 0.2 405 0.2 Cancer 285 0.4 9,225 0.4 2,415 0.4 1,120 0.4 Psychia t ric comorbidities (1995-20 17) Schizophrenia 360 0.5 9,185 0.4 6,035 1.0 1,490 0.6 Autism spectrum disorders 25 0.1 815 0.1 535 0.1 105 0.1 Neurotic 3,935 5.7 95,655 4.4 44,875 7.8 12,965 5.0 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 25 Co mpa r i s on c ohort s Charac t e r i stic VB-a ffected c hildbirt h s , N = 69 ,230 (among 65,395 wo m e n ) VB-unaffected chil dbir ths, N = 2,179 ,940 (among 1 ,172,715 women) Termina tions, N = 578,3 55 (among 4 17,765 women) M i s c arriag es , N = 258,430 ( a mong 219 ,680 women) n % n % n % n % disorders Eating disorders 480 0.7 13,650 0.6 5,755 1.0 1,640 0.6 Personality disorder 1,435 2.1 32,825 1.5 20,275 3.5 5,195 2.0 Alcohol abuse history 490 0.7 10,925 0.5 11,195 1.9 2,395 0.9 History of any psychiatric condition 5,495 7.9 145,140 6.7 64,955 11.0 19,075 7.4 Comedic ations (1995-2017 ) At least one pres c ript i on any time before t he ind ex d at e Antipsychotic s 905 1.3 26,610 1.2 13,530 2.3 4,035 1.6 Mood disorders medication 4,070 5.9 132,740 6.1 48,170 8.3 16,915 6.5 Antiepil eptics 710 1.0 22,550 1.0 8,050 1.4 2,935 1.1 ADHD 125 0.2 4,540 0.2 3,230 0.6 605 0.2 NSAIDs 18,015 26.0 637,605 29.0 148,140 26.0 68,610 27.0 Aspirin 120 0.2 4,290 0.2 670 0.1 410 0.2 Thyroid disorders 630 0.9 23,535 1.1 4,155 0.7 2,500 1.0 Steroids for systemic use 2,685 3.9 98,465 4.5 19,875 3.4 10,720 4.1 Anti-infectives for systemic use 31,005 45.0 1,144,825 53.0 275,425 48.0 114,910 44.0 At least one pres c ript i on within 30 days before the index date Anti-infectives for systemic use 2,820 4.1 78,160 3.6 32,585 5.6 9,725 3.8 Antiepil eptics 105 0.2 3,335 0.2 1,470 0.3 435 0.2 Antidepressants 250 0.4 7,070 0.3 7,005 1.2 1,610 0.6 CVD medications a 180 0.3 35,515 1.6 1,675 0.3 745 0.3 NSAIDs 220 0.3 1,860 0.1 5,850 1.0 1,690 0.7 a CVD medications: clopidogrel, statins, any of antihypertensives . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 26 Table 2. Incid ence rat es a nd t he num ber of cardiovasc u lar out c om es fo llowing vag in al bleeding- affect ed childbir ths, vagina l b leeding-unaf fect ed pr eg n anc ies, p regn anc ies ending in m i scarr i ages, and pr egna n ci e s endin g in t erminat ions, D e nm ark, 1979-2017 VB-af fected c hild birth s VB-u naffected c hild birth s Ter mination s Mis carr iages Pregnancie s N, e vent D iabet es t yp e 1 All 1,265 27,050 9,020 4,775 First 315 9,570 3,950 1,820 D iabet es t yp e 2 All 3,870 84,230 27,810 14,365 First 1,105 29,870 12,000 5,475 H y per tension All 6,485 146,440 47,995 22,985 First 1,655 48,760 21,440 8,720 Is cha emic heart disease All 2,985 55,250 22,005 9,920 First 680 17,180 9,215 3,600 M yoc ardial inf arct ion All 745 15,405 6,325 2,690 First 170 4,790 2,760 1,010 Atrial fib ril lat ion o r f lutt er All 1,015 21,420 8,830 3,900 First 245 6,910 4,495 1,650 H ear t failur e All 505 10,215 4,770 2,015 First 110 3,335 2,330 800 Is cha emic stroke All 1,295 25,465 10,590 4,430 First 305 8,260 4,595 1,695 H em orrh ag ic stro k e All 565 10,900 4,255 1,805 First 140 3,690 1,840 660 CABG All 80 1,610 815 320 First 20 515 445 150 PCI All 465 10,510 4,205 1,810 First 110 3,170 1,840 680 Inc id ence r ate per 1 0,000 PYs ( 95% C I) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 27 VB-af fected c hild birth s VB-u naffected c hild birth s Ter mination s Mis carr iages Pregnancie s D iabet es t yp e 1 All 8.28 (7.84- 8.75) 6.37 (6.29- 6.44) 7.32 (7.17- 7.47) 8.65 (8.41- 8.90) First 7.16 (6.41- 7.99) 5.95 (5.83- 6.07) 6.70 (6.49- 6.91) 8.46 (8.08- 8.86) D iabet es t yp e 2 All 25.71 (24.91- 26.53) 20.01 (19.88- 20.15) 22.83 (22.56- 23.10) 26.41 (25.98- 26.85) First 25.38 (23.92- 26.91) 18.73 (18.52- 18.95) 20.59 (20.23- 20.96) 25.81 (25.14- 26.50) H y per tension All 43.67 (42.62- 44.74) 35.23 (35.05- 35.41) 39.93 (39.57- 40.29) 42.77 (42.22- 43.32) First 38.36 (36.55- 40.24) 30.89 (30.62- 31.16) 37.30 (36.80- 37.80) 41.60 (40.74- 42.49) Is cha emic heart disease All 19.73 (19.04- 20.45) 13.06 (12.95- 13.17) 18.02 (17.78- 18.26) 18.11 (17.76- 18.47) First 15.47 (14.34- 16.66) 10.71 (10.55- 10.87) 15.77 (15.45- 16.09) 16.84 (16.30- 17.40) M yoc ardial inf arct ion All 4.85 (4.51- 5.21) 3.61 (3.56- 3.67) 5.12 (4.99- 5.25) 4.86 (4.68- 5.04) First 3.78 (3.24- 4.39) 2.96 (2.88- 3.05) 4.67 (4.50- 4.85) 4.68 (4.40- 4.98) Atrial fib ril lat ion o r f lutt er All 6.64 (6.24- 7.06) 5.03 (4.96- 5.09) 7.15 (7.00- 7.30) 7.04 (6.82- 7.27) First 5.48 (4.82- 6.20) 4.28 (4.18- 4.38) 7.62 (7.40- 7.85) 7.63 (7.27- 8.01) H ear t failur e All 3.28 (3.01- 3.58) 2.39 (2.35- 2.44) 3.85 (3.74- 3.96) 3.63 (3.47- 3.79) First 2.47 (2.04- 2.97) 2.06 (2.00- 2.14) 3.94 (3.79- 4.11) 3.69 (3.44- 3.95) Is cha emic stroke All 8.47 (8.02- 8.94) 5.98 (5.91- 6.06) 8.59 (8.43- 8.76) 8.01 (7.78- 8.25) First 6.84 (6.10- 7.64) 5.12 (5.01- 5.23) 7.80 (7.58- 8.03) 7.85 (7.48- 8.23) H em orrh ag ic stro k e All 3.67 (3.38- 2.56 (2.51- 3.44 (3.34- 3.25 (3.11- . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 28 VB-af fected c hild birth s VB-u naffected c hild birth s Ter mination s Mis carr iages Pregnancie s 3.99) 2.60) 3.54) 3.41) First 3.17 (2.68- 3.73) 2.28 (2.21- 2.36) 3.12 (2.98- 3.26) 3.04 (2.81- 3.28) CABG All 0.53 (0.43- 0.66) 0.38 (0.36- 0.40) 0.66 (0.61- 0.70) 0.58 (0.51- 0.64) First 0.45 (0.28- 0.68) 0.32 (0.29- 0.35) 0.75 (0.69- 0.83) 0.70 (0.59- 0.81) PCI All 3.02 (2.75- 3.30) 2.46 (2.42- 2.51) 3.40 (3.30- 3.50) 3.26 (3.12- 3.42) First 2.49 (2.06- 2.99) 1.96 (1.89- 2.03) 3.11 (2.97- 3.26) 3.14 (2.92- 3.39) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 29 Table 3. Haz ard ratios (HRs ) with 95 % confiden ce int ervals ( CIs) for t he as sociations bet we en vagin a l b l e edin g within 20 week s o f ge s t ation and card iova s cul ar ou tcomes, De nmark, 1979-2 018 Com pa r i s ons Cox r eg r e ssion model H R (95% CI) a Diabetes mellitus type 1 All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.22 (1.15-1.29) VB-affected childbirths vs terminations 1.28 (1.20-1.37) VB-affected childbirths vs miscarriages 1.00 (0.93-1.07) VB-affected vs VB-unaffec ted childbirths IPTW 1.26 (1.19-1.34) VB-affected childbirths vs terminations 1.35 (1.25-1.45) VB-affected childbirths vs miscarriages 0.99 (0.91-1.08) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.24 (1.17-1.31) VB-affected childbirths vs terminations 1.27 (1.19-1.36) VB-affected childbirths vs miscarriages 0.92 (0.85-1.00) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.18 (1.05-1.32) VB-affected childbirths vs terminations 1.24 (1.09-1.39) VB-affected childbirths vs miscarriages 0.93 (0.82-1.05) VB-affected vs VB-unaffec ted childbirths IPTW 1.20 (1.07-1.34) VB-affected childbirths vs terminations 1.27 (1.10-1.47) VB-affected childbirths vs miscarriages 0.91 (0.80-1.03) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.18 (1.06-1.32) VB-affected childbirths vs terminations 1.12 (0.99-1.27) VB-affected childbirths vs miscarriages 0.91 (0.80-1.02) Diabetes mellitus type 2 All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.21 (1.17-1.25) VB-affected childbirths vs terminations 1.24 (1.19-1.29) VB-affected childbirths vs miscarriages 1.01 (0.97-1.06) VB-affected vs VB-unaffec ted childbirths IPTW 1.24 (1.20-1.28) VB-affected childbirths vs terminations 1.29 (1.24-1.35) VB-affected childbirths vs miscarriages 1.01 (0.97-1.06) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.22 (1.18-1.26) VB-affected childbirths vs terminations 1.21 (1.17-1.26) VB-affected childbirths vs miscarriages 0.93 (0.89-0.98) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.34 (1.26-1.42) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 30 Com pa r i s ons Cox r eg r e ssion model H R (95% CI) a VB-affected childbirths vs terminations 1.31 (1.23-1.40) VB-affected childbirths vs miscarriages 1.02 (0.96-1.09) VB-affected vs VB-unaffec ted childbirths IPTW 1.34 (1.26-1.43) VB-affected childbirths vs terminations 1.37 (1.27-1.49) VB-affected childbirths vs miscarriages 1.01 (0.94-1.08) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.35 (1.27-1.43) VB-affected childbirths vs terminations 1.23 (1.15-1.32) VB-affected childbirths vs miscarriages 1.00 (0.93-1.07) Hypertens i on All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.15 (1.12-1.18) VB-affected childbirths vs terminations 1.11 (1.08-1.14) VB-affected childbirths vs miscarriages 1.02 (0.98-1.05) VB-affected vs VB-unaffec ted childbirths IPTW 1.19 (1.16-1.22) VB-affected childbirths vs terminations 1.16 (1.12-1.20) VB-affected childbirths vs miscarriages 1.02 (0.98-1.06) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.17 (1.14-1.20) VB-affected childbirths vs terminations 1.09 (1.06-1.12) VB-affected childbirths vs miscarriages 0.98 (0.94-1.02) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.20 (1.14-1.26) VB-affected childbirths vs terminations 1.23 (1.17-1.30) VB-affected childbirths vs miscarriages 1.09 (1.03-1.15) VB-affected vs VB-unaffec ted childbirths IPTW 1.23 (1.17-1.29) VB-affected childbirths vs terminations 1.16 (1.09-1.24) VB-affected childbirths vs miscarriages 1.01 (0.96-1.07) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.21 (1.15-1.27) VB-affected childbirths vs terminations 1.10 (1.04-1.16) VB-affected childbirths vs miscarriages 1.06 (1.01-1.12) Ischaemic heart disease All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.31 (1.27-1.36) VB-affected childbirths vs terminations 1.11 (1.06-1.16) VB-affected childbirths vs miscarriages 1.06 (1.01-1.11) VB-affected vs VB-unaffec ted childbirths IPTW 1.42 (1.36-1.47) VB-affected childbirths vs terminations 1.17 (1.11-1.23) VB-affected childbirths vs miscarriages 1.07 (1.01-1.13) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.35 (1.30-1.40) VB-affected childbirths vs terminations 1.11 (1.06-1.16) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 31 Com pa r i s ons Cox r eg r e ssion model H R (95% CI) a VB-affected childbirths vs miscarriages 1.02 (0.97-1.08) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.34 (1.25-1.45) VB-affected childbirths vs terminations 1.26 (1.16-1.37) VB-affected childbirths vs miscarriages 1.17 (1.08-1.27) VB-affected vs VB-unaffec ted childbirths IPTW 1.40 (1.30-1.52) VB-affected childbirths vs terminations 1.24 (1.13-1.37) VB-affected childbirths vs miscarriages 1.09 (1.00-1.19) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.36 (1.26-1.47) VB-affected childbirths vs terminations 1.08 (0.99-1.17) VB-affected childbirths vs miscarriages 1.11 (1.02-1.21) Myoc ardial infarc t ion All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.17 (1.09-1.26) VB-affected childbirths vs terminations 1.00 (0.92-1.09) VB-affected childbirths vs miscarriages 0.94 (0.85-1.03) VB-affected vs VB-unaffec ted childbirths IPTW 1.26 (1.17-1.36) VB-affected childbirths vs terminations 1.02 (0.92-1.12) VB-affected childbirths vs miscarriages 0.93 (0.83-1.04) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.20 (1.12-1.29) VB-affected childbirths vs terminations 1.01 (0.93-1.10) VB-affected childbirths vs miscarriages 0.93 (0.83-1.04) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.18 (1.02-1.38) VB-affected childbirths vs terminations 1.14 (0.97-1.35) VB-affected childbirths vs miscarriages 1.04 (0.88-1.23) VB-affected vs VB-unaffec ted childbirths IPTW 1.23 (1.06-1.44) VB-affected childbirths vs terminations 1.04 (0.86-1.27) VB-affected childbirths vs miscarriages 0.95 (0.80-1.14) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.20 (1.03-1.40) VB-affected childbirths vs terminations 0.96 (0.82-1.14) VB-affected childbirths vs miscarriages 1.00 (0.85-1.19) At rial f i b r ill ation or flut te r All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.19 (1.12-1.27) VB-affected childbirths vs terminations 1.10 (1.02-1.18) VB-affected childbirths vs miscarriages 1.00 (0.92-1.09) VB-affected vs VB-unaffec ted childbirths IPTW 1.26 (1.19-1.35) VB-affected childbirths vs terminations 1.04 (0.96-1.13) VB-affected childbirths vs miscarriages 0.98 (0.89-1.08) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 32 Com pa r i s ons Cox r eg r e ssion model H R (95% CI) a VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.23 (1.16-1.31) VB-affected childbirths vs terminations 1.05 (0.97-1.12) VB-affected childbirths vs miscarriages 0.94 (0.86-1.04) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.15 (1.01-1.31) VB-affected childbirths vs terminations 1.05 (0.92-1.21) VB-affected childbirths vs miscarriages 1.01 (0.88-1.16) VB-affected vs VB-unaffec ted childbirths IPTW 1.25 (1.10-1.42) VB-affected childbirths vs terminations 0.84 (0.71-1.00) VB-affected childbirths vs miscarriages 0.85 (0.73-0.98) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.19 (1.04-1.35) VB-affected childbirths vs terminations 0.89 (0.77-1.02) VB-affected childbirths vs miscarriages 0.95 (0.83-1.09) Hea r t failure All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.21 (1.11-1.33) VB-affected childbirths vs terminations 1.01 (0.92-1.12) VB-affected childbirths vs miscarriages 0.96 (0.85-1.08) VB-affected vs VB-unaffec ted childbirths IPTW 1.30 (1.19-1.43) VB-affected childbirths vs terminations 1.00 (0.89-1.13) VB-affected childbirths vs miscarriages 0.94 (0.82-1.08) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.25 (1.14-1.37) VB-affected childbirths vs terminations 1.00 (0.91-1.11) VB-affected childbirths vs miscarriages 0.94 (0.82-1.07) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.10 (0.91-1.33) VB-affected childbirths vs terminations 1.03 (0.84-1.25) VB-affected childbirths vs miscarriages 1.02 (0.83-1.25) VB-affected vs VB-unaffec ted childbirths IPTW 1.18 (0.97-1.42) VB-affected childbirths vs terminations 0.88 (0.70-1.11) VB-affected childbirths vs miscarriages 0.86 (0.70-1.06) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.12 (0.92-1.35) VB-affected childbirths vs terminations 0.83 (0.68-1.02) VB-affected childbirths vs miscarriages 0.96 (0.78-1.18) I s ch a e m i c s t r o k e All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.26 (1.19-1.33) VB-affected childbirths vs terminations 1.04 (0.98-1.11) VB-affected childbirths vs miscarriages 1.02 (0.94-1.09) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 33 Com pa r i s ons Cox r eg r e ssion model H R (95% CI) a VB-affected vs VB-unaffec ted childbirths IPTW 1.33 (1.26-1.41) VB-affected childbirths vs terminations 1.05 (0.98-1.13) VB-affected childbirths vs miscarriages 1.01 (0.93-1.10) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.29 (1.22-1.36) VB-affected childbirths vs terminations 1.05 (0.99-1.12) VB-affected childbirths vs miscarriages 0.98 (0.90-1.07) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.24 (1.10-1.39) VB-affected childbirths vs terminations 1.16 (1.03-1.31) VB-affected childbirths vs miscarriages 1.08 (0.96-1.23) VB-affected vs VB-unaffec ted childbirths IPTW 1.29 (1.15-1.45) VB-affected childbirths vs terminations 0.98 (0.84-1.14) VB-affected childbirths vs miscarriages 0.96 (0.85-1.10) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.26 (1.12-1.41) VB-affected childbirths vs terminations 1.05 (0.92-1.19) VB-affected childbirths vs miscarriages 1.05 (0.93-1.20) H e m or r h a gi c s tr o ke All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.27 (1.17-1.39) VB-affected childbirths vs terminations 1.10 (1.00-1.21) VB-affected childbirths vs miscarriages 1.05 (0.94-1.18) VB-affected vs VB-unaffec ted childbirths IPTW 1.37 (1.25-1.49) VB-affected childbirths vs termin ations 1.05 (0.93-1.17) VB-affected childbirths vs miscarriages 1.04 (0.92-1.18) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.30 (1.19-1.41) VB-affected childbirths vs terminations 1.12 (1.02-1.24) VB-affected childbirths vs miscarriages 1.05 (0.92-1.19) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.30 (1.10-1.54) VB-affected childbirths vs terminations 1.23 (1.03-1.47) VB-affected childbirths vs miscarriages 1.19 (0.99-1.43) VB-affected vs VB-unaffec ted childbirths IPTW 1.36 (1.15-1.62) VB-affected childbirths vs terminations 1.12 (0.90-1.40) VB-affected childbirths vs miscarriages 1.11 (0.91-1.35) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.32 (1.11-1.56) VB-affected childbirths vs terminations 1.18 (0.98-1.42) VB-affected childbirths vs miscarriages 1.18 (0.98-1.42) CAB G All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.24 (0.99-1.55) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 34 Com pa r i s ons Cox r eg r e ssion model H R (95% CI) a VB-affected childbirths vs terminations 1.07 (0.84-1.37) VB-affected childbirths vs miscarriages 1.01 (0.76-1.36) VB-affected vs VB-unaffec ted childbirths IPTW 1.36 (1.08-1.70) VB-affected childbirths vs terminations 0.97 (0.72-1.30) VB-affected childbirths vs miscarriages 1.00 (0.71-1.41) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.28 (1.02-1.60) VB-affected childbirths vs terminations 1.04 (0.80-1.33) VB-affected childbirths vs miscarriages 1.02 (0.73-1.41) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.31 (0.83-2.04) VB-affected childbirths vs terminations 1.07 (0.67-1.71) VB-affected childbirths vs miscarriages 0.99 (0.61-1.60) VB-affected vs VB-unaffec ted childbirths IPTW 1.35 (0.86-2.12) VB-affected childbirths vs termin ations 0.94 (0.55-1.59) VB-affected childbirths vs miscarriages 0.84 (0.51-1.37) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.31 (0.84-2.04) VB-affected childbirths vs terminations 0.73 (0.46-1.18) VB-affected childbirths vs miscarriages 0.92 (0.57-1.48) PCI All p r eg nancie s VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.05 (0.96-1.16) VB-affected childbirths vs terminations 0.96 (0.86-1.06) VB-affected childbirths vs miscarriages 0.93 (0.82-1.05) VB-affected vs VB-unaffec ted childbirths IPTW 1.14 (1.04-1.25) VB-affected childbirths vs terminations 0.99 (0.88-1.12) VB-affected childbirths vs miscarriages 0.92 (0.80-1.05) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.09 (0.99-1.19) VB-affected childbirths vs terminations 0.97 (0.87-1.07) VB-affected childbirths vs miscarriages 0.92 (0.80-1.05) F irs t p regnanc ies VB-affected vs VB-unaffec ted childbirths Conve nt iona l 1.18 (0.98-1.43) VB-affected childbirths vs terminations 1.17 (0.96-1.43) VB-affected childbirths vs miscarriages 1.05 (0.85-1.29) VB-affected vs VB-unaffec ted childbirths IPTW 1.23 (1.02-1.49) VB-affected childbirths vs terminations 1.11 (0.88-1.41) VB-affected childbirths vs miscarriages 0.98 (0.79-1.22) VB-affected vs VB-unaffec ted childbirths ATT- we ighted 1.20 (0.99-1.45) VB-affected childbirths vs terminations 1.01 (0.82-1.24) VB-affected childbirths vs miscarriages 1.01 (0.82-1.24) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint Dudukina E, Horvá th-Puhó E, S ørensen HT, Ehrenstein V., Diabetes and cardiovascular disea ses in women with pregnancies complicated by vaginal bleeding: Danish population-based cohort study 35 CA BsG, coro nary arter y b y p as s graft ; P CI, per cuta neo us coro na r y in ter venti o n a Ad j ust e d f or w om en’s a g e a t pregn a n c y end , calen da r year, pa ri ty , c ivil s t a t us, e mplo yment, hig he s t att ai ned educ a t io n, i nco me in ye ar-s p ecific q uartil es, r e pro d u c ti v e hist o r y ( previ ous e pi s odes of vagi nal b le edi n g , termina ti ons, mis carria g es), co m or bi di t ies (chro nic pul mon ary d ise a s e , chr on i c kid ney dis ea s e, chr on i c l i ve r dis e as e, t hyr oi d , ca ncer, r h eum a tic c o n d iti on s, t h y r o id c o ndit io n s , a lco ho l ab us e, an d i n divi dual c l as s e s of p sy c hi a t r i c co nd it io ns a v a i l ab l e s ta r t ing in 199 5 ) ; m ed i c at i on u se ( av a i l a bl e s ta r t ing in 199 5 ) a ny ti me b e fo r e the i n de x (a nt ips y c h otics , m edic a ti o n f or moo d d is ord e rs, N S AIDs , as piri n , t hy roi d disor de rs medica ti o n , stero id s f o r s y s t emic use, a nt i -i nfect ives for s y s temic use) an d 3 0 days bef ore t he ind ex d a te ( anti-i n f ect ive s f o r sy st em i c u s e , an tid ep r e s sa nt s , a ny o f c lop ido g re l , st a ti n s , an ti hy p e r ten s iv es , NS A ID s) . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint 1,490,389 records: Still- and live-births in Denmark, 1973-1996: 1,483,364 Livebirths 7025 Stillbirths 2,829,027 records: Still- and live-births in Denmark, 1973-2017 2,816,212 Live-births 12,815 Stillbirths Including 89,723 infants born from multi-fetal delivery N, women = 1,443,429 1,338,638 records: Still- and live-births in Denmark, 1997-2017: 1,332,848 Livebirths 5790 Stillbirths One identifier randomly selected for each multi-fetal delivery is kept (N=44,458) 2,783,747 2,771,570 Live births 12,177 Stillbirths 92,806 records: Vaginal bleeding hospital diagnoses within 20 gestational weeks N, women = 67,555 Excluded 402,068 Records outside of the study period (<1979) 2,381,679 records: Pregnancies ending in still- and live births, 1979-2017 N, women = 1,240,002 3,408,135 Eligible pregnancies N, women = 1,411,869 203,528 records: Vaginal bleeding hospital diagnoses, the DNPR (1977-2018); N, women = 120,697 747,237 records: Hospital diagnoses for pregnancy terminations, the DNPR (1977-2018); N, women = 453,396 646,224 records: Distinct records for pregnancy terminations, the DNPR (1977-2018); N, women = 465,813 241,974 records: Hospital procedure records for pregnancy terminations, the DNPR (1977-2018); N, women = 183,804 414,405 records: Hospital diagnoses for miscarriages, the DNPR (1977-2018); N, women = 243,743 287,426 records: Distinct primary and secondary diagnoses at hospital encounters for miscarriages, the DNPR (1977-2018); N, women = 243,743 Excluded 322,182 records: 14,511 Repeated episodes of vaginal bleeding in pregnancy 71,450 Unknown gestational age 6,749 Encounters for different reproductive events (delivery, miscarriage, termination, vaginal bleeding) registered on the same date 67,923 Dates of encounters for reproductive events (delivery, miscarriage, termination, vaginal bleeding) are outside of the study period ( 2017) 68,999 Excluded from unaffected childbirths comparator cohort due to vaginal bleeding episode 78,456 History of any outcome of interest (CVD, hypertension or diabetes types 1 and 2) before the index date 10,911 Personal identifier not found in the CRS or age of < 10 at the date of reproductive event 3,183 Emigrated before the index date Final study population (clouded) 3,085,955 pregnancies (1979-2017) among 1,329,331 distinct women 2,179,940 Pregnancies unaffected by vaginal bleeding ending in childbirths among 1,172,715 women 69,230 Pregnancies affected by vaginal bleeding within 20 gestational weeks and ending in a childbirth among 65,395 women 578,355 Pregnancies ending in a termination among 417,765 women 258,430 Pregnancies ending in a miscarriage among 219,680 women . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 18, 2022. ; https://doi.org/10.1101/2022.03.18.22272466doi: medRxiv preprint