Incidence of Pulmonary and Venous Thromboembolism in IVF Pregnancies After Fresh and Frozen Embryo Transfer

A population-based cohort study with data prospectively collected in the nationwide registries in Sweden. Through the national registries the investigators collect data on all women with a child born in Sweden between 1992-2012 and these will be linked to data from the registries by the use of the unique personal identity number.

Study population All women who gave birth at the age of 15-50 years to their first child from the 1st of January 1992 until the 31st of December 2012 comprise the study population. Women will be categorized as either giving birth after IVF or natural conception pregnancies. Data on whether IVF was performed by the use of fresh or frozen embryo transfer will be retrieved.

The registries The national registries in Sweden are considered to have good coverage and validity in general and the Swedish Medical Birth Register (MBR) by the National Board of Health and Welfare in Sweden includes annually 97-99.5 % of all child births in Sweden, up until 2005 all live child births and stillborn from week 28 and thereafter including all child births from week 22. From the MBR and the Swedish national quality registry of assisted reproductive technology, the Q-IVF, the investigators collect information on pre-pregnancy and pregnancy variables. From the Patient registry (PR) ICD codes for PE and overall VTE will be retrieved and from the Swedish register of Education completed years at school and finally from the Swedish Cause of Death Register data on causes of deaths .

Time periods - pregnancy length, the trimesters and postpartum. The follow-up period is from the estimated start of pregnancy until day 42 after delivery, i.e. six weeks postpartum or until an event of PE or other VTE occurred before the end of that period. When determining the trimesters, the best estimation of pregnancy length from the MBR that estimates the days of pregnancy by primary ultrasound will be used or when this is missing by calculation of the first day of the last period. Start of pregnancy is defined as the date of delivery minus the precalculated best estimation of the pregnancy length found in the MBR. Regarding trimesters and postpartum the first trimester is defined to comprise the start of pregnancy, day 0 until day 90, the second trimester from day 91 until day 181 and the third trimester from day 182 until three days before delivery date. Delivery and postpartum period is defined to comprise two days before until six weeks after delivery date, since labor often starts 1-2 days before the actual delivery. Women with pregnancy length exceeding 300 days are to be excluded, considered as extreme values.

Exposures - IVF with fresh or frozen embryo transfer The exposures are either an IVF pregnancy with fresh embryo transfer performed directly after ovarian stimulation or an IVF pregnancy with frozen embryo transfer, which was thawed and transferred in a later, non-stimulated cycle. These exposures will be compared to a control group of non-exposed women which consists of all pregnant women not listed in the MBR or the IVF-registries and thus considered to be spontaneously conceived, here referred to as natural pregnancy.

Outcome - first incident pulmonary embolism or overall venous thromboembolism The outcome is the occurrence of the first incident PE or first incident VTE during pregnancy or postpartum. All women with pre-pregnancy events are excluded.

The diagnoses of PE and DVT will be defined in the PR with data from 1987 on national inpatient care and from 1997 also outpatient and diagnoses based on the International classification of diseases, ICD. The investigators will use the 8th edition, ICD-8 (1969-1986), for previous diagnoses and exclusions of pre-pregnancy venous thromboembolism and the ninth, ICD-9 (1987-1996), and the tenth edition, ICD-10 (1997--), for both previous and incident diagnoses.

The investigators will use all diagnosis codes for PE (ICD-8: 450.01-03, 450.09, 673.98-99; ICD-9: 415B, 673C; ICD-10: I26.0, I26.9, O882), DVT (ICD-8: 451.00, 451.98-99, 671.01-02, 671.08-09; ICD-9: 451B; ICD-10: I80.1-3, I80.8-9, O22.3, O87.1), portal vein thrombosis (ICD-8 453.09; ICD-9: 452; ICD-10: I81.9), vena cava thrombosis (ICD-8; ICD-9: 453C; ICD-10: I82.2), renal thrombosis (ICD-8; ICD-9: 453D; ICD-10: I82.3), cerebral vein thrombosis (ICD-8: 321.00, 321.09; ICD-9: 325; ICD-10: O22.5, O87.3 and diagnoses codes for other localizations of DVT or emboli (ICD-8: 453.09; ICD-9: 453W, 453X, 671F; ICD-10: I82.8-9, O87.9).

Other predictors or potential confounding factors. The investigators will adjust for potential confounding factors; age, calendar year of delivery, body mass index (BMI), multiple birth, smoking, country of birth and education level. Age is one of the strongest known risk factors for VTE. BMI of 30 or more is a known risk factor for VTE. Another risk factor for VTE is parity, which is one of the reasons why the investigators in this study choose to study only the first child birth in all women in contrast to a previous study where women were included with their first IVF pregnancy. Thus, that study also included women who had given birth to a child before without known IVF with adjustments performed in the statistical analysis for parity.

The continuous variables are to be categorized as follows, age at delivery (<25, 25-29, 30-34, ≥35 years), pre-pregnancy BMI (<25, 25-29 or ≥30 kg/m2), educational level recorded as number of school years (≤9 years -compulsory school, 10-12 years - upper secondary school, >12 years -University level), pre-pregnancy cigarette smoking status (yes/no). Country of birth is either Sweden or other country.

Statistical analyses:

Baseline characteristics of women are to be reported as frequencies and percentages for categorical variables and as median and interquartile range for continuous variables.

Cox regression models are to be used to estimate hazard ratios (HRs) and 95% confidence intervals (CI) in order to assess the association between exposure (IVF with fresh or frozen embryo transfer versus the referent group with natural conception) and each of the two study outcomes. First, HRs are to be estimated under a proportional hazard assumption for the entire pregnancy duration, including the postpartum period. Then, this assumption will be relaxed by allowing the HRs to vary over the different trimesters and the postpartum period by means of a time-dependent Cox regression model. Models are to be adjusted for potential confounders as described above.