key: cord-1026103-g6hqqo50
authors: Magnus, Maria C.; Gjessing, Håkon K.; Eide, Helena N.; Wilcox, Allen J.; Fell, Deshayne B.; Håberg, Siri E.
title: Covid-19 Vaccination during Pregnancy and First-Trimester Miscarriage
date: 2021-10-20
journal: N Engl J Med
DOI: 10.1056/nejmc2114466
sha: d8a0998d5d4c74660b44ccbeac754f7a066ddb86
doc_id: 1026103
cord_uid: g6hqqo50

nan

. Characteristics of women with and without a consultation for "confirmation of pregnancy" (ICPC code W78) in primary care among all births in the Norwegian birth registry in 2020………………………………………………………………………………...9 Table S5 . Odds ratio for Covid-19 vaccination with different types of vaccines in a 5-week window before miscarriage or confirmation of an ongoing pregnancy…………..…………..10 Table S6 . Odds ratio of Covid-19 vaccination in a 5-week window before miscarriage or confirmation of an ongoing pregnancy according to partial or full vaccination status……11 Table S7 . Odds ratio of Covid-19 vaccination in a 5-or 3-week window before miscarriage or confirmation of an ongoing pregnancy with at least 8 weeks follow up after confirmation…………………………………………………………………………….…….12 

The Norwegian national health registries in this study contain registrations of all contacts with health care services, which are mandatory to report, and linked to reimbursement. Diagnostic codes and dates of contact are registered for each individual with a personal identification number (pin), which is issued to all citizens in Norway at birth or immigration for identification and administrative purposes. The pin enables linkages of individual level information across registries. All data were provided by the Covid-19 Emergency Preparedness Register run by the Norwegian Institute of Public Health and were deterministically linked using unique personal identification numbers.

The Emergency Preparedness Register for Covid-19 1 Data in this study were provided through the Emergency preparedness register for Covid-19 ("Beredt C19") administered by the Norwegian Institute of Public Health, according to the Health Preparedness Act §2-4. This registry was established in 2020 to provide authorities with up to date information on prevalence, causal relationships, and consequences of the Covid-19 epidemic in Norway. "Beredt C19" includes information already collected in the healthcare system, national health registries and administrative registers with information about the Norwegian population. The data subjects' right is safeguarded as they can contact the data controller for the different sources included in "Beredt C19" in the usual way this is regulated in Norway.

Through "Beredt C19" we used data from the following sources:

The Norwegian Immunisation Register (SYSVAK) SYSVAK is a register of vaccines in the Norwegian vaccination program, with mandatory registrations of all Covid-19 vaccinations (dates and type of vaccine).

The Norwegian Patient Registry (NPR) 2 The Norwegian Patient Registry includes individual level information on all contacts with specialist health-care services. Information registered includes admission and discharge dates, and diagnostic codes during the hospital stay/outpatient contact. These codes are according to the International Classification of Diseases version 10. We used these codes to identify miscarriages and underlying health conditions (Table S1 and Table S2 ).

The Norwegian Registry for Primary Health Care (KPR) 2 All Norwegian residents are assigned a primary care physician for care, which is universally publicly funded. The physician obtains reimbursements for consultation by reporting the date, the reimbursement code for payment of services, and medical diagnoses for each patient contact. The diagnoses are coded according to the International Classification of Primary Care version 2. We used codes for miscarriage and confirmation of a pregnancy, and codes for underlying health conditions (Table S1 and Table S2 ). 3 There is mandatory reporting of selected infectious diseases to this National Health register. Reporting of all Covid-19 tests is mandatory, and this register contains date of testing and test results.

Administrative data are mandatorily reported to Statistics Norway. We used information from this database on household income in 2018, type of education and years of education completed by 2019.

The Norwegian national birth registry includes information on all pregnancies ending after completion of gestational week 12 or later. The registry includes information on birth outcomes in addition to maternal background characteristics, health during pregnancy, pregnancy outcomes and neonatal health. We used the birth register to exclude fetal losses after week 14.. As registrations in the NPR and KPR do not include the gestational age, we also used data from the birth registry on gestational age at birth for pregnancies completed in 2020 to estimate gestational week of first registration of W78 (confirmation of pregnancy). For pregnancies later registered in the birth registry, most registrations of W78 were made in pregnancy weeks 7-9 ( Figure S1 ). Gestational age in the birth registry is based on routine ultrasound assessments at around 18 gestational weeks provided to all pregnant women, or last menstrual period if ultrasound assessments are lacking.

The study was approved by The Regional Committee for Medical and Health Ethics of South/East Norway (reference number 141135), which waived the need for consent from participants in this registry-based study.

We identified all women registered with a miscarriage in the first trimester or a confirmation of an ongoing pregnancy between February 15 th and August 15 th , 2021 through Norwegian registries. We used three mandatory National health registries to identify pregnant women: The Norwegian Registry for Primary Health Care, the Patient Registry, and the Medical Birth Registry of Norway. Miscarriages were identified by ICPC-2 code W82 ("Spontaneous abortion") in the primary care database Care or ICD-10 codes O02 ("Other abnormal products of conception"), and O03 ("Spontaneous abortion") in the Patient Registry, while the birth registry provided information on miscarriage occurring between 12 and 14 completed gestational weeks (Table S1 ). Primary care visits for confirmation of ongoing pregnancies were identified in the primary care database by the ICPC-2 code W78 ("Confirmed pregnancy"). The index date was the first registration of a miscarriage or confirmation of an ongoing pregnancy. Induced abortions were not included in this study.

We defined first-trimester miscarriage as a spontaneous fetal loss that occurred before 14 completed gestational weeks. The birth registry contributed information on miscarriages which occurred during gestational weeks 13 and 14. We used the patient registry and primary care database to identify miscarriages that occurred before 12 completed weeks. The administrative codes used to identify miscarriages in these databases are shown in Table  S1 . None of these codes include induced abortions, which were excluded from our analyses. Since registered miscarriages in the patient registry and the primary care database do not include information on gestational length, we ensured they did not occur within a pregnancy registered in the birth registry. We considered miscarriages to have occurred before 12 completed gestational weeks when they were not registered in the birth registry, in which registrations start at gestational week 12. In order to ensure that two separate registrations in the patient registry or primary care database did not reflect the same pregnancy, we also required at least 42 days between the day of a completed pregnancy in the birth registry and the registration of a miscarriage in the patient registry, or at least 90 days between the day of a completed pregnancy in the birth registry and registration of a miscarriage in primary care. Women in primary care have more follow-up visits after normal pregnancies, and so a longer interval is required. This process should identify all recorded miscarriages in Norway occurring prior to 14 completed gestational weeks. The date of the occurrence of the miscarriage contains some degree of uncertainty, since the fetal demise could have occurred within a few days prior to the registration in the healthcare system. Notably, we were only able to evaluate miscarriages which resulted in contact with health-care services, and could not include miscarriages which occurred before a woman realized she was pregnant or which did not require medical care.

Among all completed pregnancies in the birth registry in 2020 with known last menstrual period date, approximately 41% could be identified as having a date of primary-care confirmation of pregnancy (ICPC-2 code W78). Not all pregnant women receive this confirmation as many women use home pregnancy testing only. However, among all registered births in the birth registry in 2020, characteristics of women with this code for confirmation of an early pregnancy in primary care were similar to those without this code (Table S4) , when comparing characteristics registered at birth. We used the date of clinical confirmation or miscarriage as the index date for analysis.

Covid vaccination is not recommended in Norway in the first trimester of pregnancy, except in women with high risk conditions. 5 Table S2 ). To be included as a covariate, a diagnostic code must have been noted on at least two health-care visits between January 2017 and the end of the study period. To account for potential influence of unmeasured seasonal and other time-related factors, we also adjusted for calendar month of the index date of miscarriage or confirmed pregnancy.

Using logistic regression, we estimated odds ratios (OR) and 95% confidence intervals (CI) for Covid-19 vaccination within 5-week and 3-week windows prior to the index date. Multivariable analyses adjusted for maternal age, country of birth, marital status, educational level, household income, parity, health care professional, underlying risk conditions for covid-19, previous positive test for SARS-CoV-2, and calendar month. Age was entered as a continuous variable, while other covariates were categorical and entered using the appropriate number of dummy variables. For the few variables with missing information, we used "unknown" as a separate category; we did not conduct multiple imputation because the unknown category was considered to be a unique group. For example, those with unknown education are likely to have completed their education abroad, and those with unknown income level are likely to not have permanent residence in Norway. We subsequently stratified our model by type of vaccine, and whether the vaccination was first or second dose. We also conducted sensitivity analyses restricted to health-care personnel to explore the potential role of unmeasured confounding. Women who are health professionals were more likely to be vaccinated and their vaccination behavior was less likely to be influenced by other unmeasured characteristics. The Hosmer and Lemeshow's goodness-of-fit test also indicated that our model fit the data well (p>0.4). The values for the Pregibon (1981) Delta-Beta influence statistic were all below 0.09. Only 1% of the observations had a value >0.01. Sensitivity analyses excluding this one percent of the observations yielded similar results. As some pregnancies which were confirmed towards the end of the study period could have ended in a miscarriage after follow-up, we also conducted sensitivity analyses restricting to pregnancies with at least 8 weeks follow up after first registration. This analysis was restricted to pregnancies with an index date prior to June 20 th , 2021. The results were similar to what was observed in the main analysis (Table S7 ).

All analyses were conducted using Stata version 16 (Statacorp, Texas).

Limitations include lack of information on gestational age at the time of registrations. We could not confirm the exact gestational week that vaccination during pregnancy occurred. However, since we knew the date of vaccination and the index date, our approach was to estimate odds of vaccination in the five weeks or three weeks prior to the index date. It is unlikely that many miscarriages were registered before reaching 5-6 weeks of gestation, thus a window of five weeks prior to index date likely to reflect a dose received during pregnancy. As described above, pregnancies registered after gestational week 14 in the birth registry were excluded, which should ensure that miscarriages in this study occurred between before week 14.

Second, based on data on completed pregnancies from 2020, we were likely able to include only approximately 40% of ongoing pregnancies in our control group. We were able to show, based on 2020 data, that underlying background characteristics among completed pregnancies with and without this code were similar (Table S4) , which reduces concern about differential selection of controls with different characteristics introducing potential selection bias.

Third, there are limited data on potential confounders in the registries, and unmeasured factors which are related to both vaccination and miscarriages may result in unmeasured confounding, In particular, unmeasured lifestyle characteristics such as smoking and body-mass index, are not captured in the Norwegian national health registries. Still, adjustment for a range of factors which are mandatory to register in the national data bases, including income and educational level, did not have a large impact on estimates. Also, analyses restricted to health care personnel yielded similar results. This group of women were prioritized and urged to receive vaccination, although the general recommendations for vaccination during pregnancy (i.e. to be vaccinated in 2. and 3. Trimester) was also the recommendations for health care personnel. The women in this study were likely to be vaccinated before they knew they were pregnant. Since health care workers were more likely to be vaccinated, their vaccination status is less likely to be influenced by other characteristics and unmeasured factors. Still, it is possible that unmeasured factors might mask or attenuate an increased risk with vaccination, if vaccinated women had a higher prevalence of unmeasured factors associated with a lower risk of miscarriage, or a lower prevalence of a factor associated with a higher risk of miscarriage relative to unvaccinated women. Previous studies looking at influenza vaccination have described this potential "healthy vaccination" bias. 6, 7 I05, I06, I07, I08, I09, I2, I31, I32, I34, I35, I36,  I37, I39, I40, I41, I42, I43, I46, I48, I49, 

Emergency preparedness register for COVID-19 (Beredt-C19)

The Norwegian Patient Registry and the Norwegian Registry for Primary Health Care: Research potential of two nationwide health-care registries

Norwegian institute of Public Health. Norwegian Surveillance System for Communicable Diseases (MSIS)

Updated vaccine recommendations for pregnant women

Does influenza vaccination improve pregnancy outcome? Methodological issues and research needs

Risk of Preterm or Small-for-Gestational-Age Birth After Influenza Vaccination During Pregnancy: Caveats When Conducting Retrospective Observational Studies