Infection with SARS-CoV-2 during pregnancy and risk of stillbirth: a Scandinavian registry study

Study population

We studied live and stillbirths after 22 completed gestational weeks in Sweden, Denmark and Norway between 1 March 2020 and end of follow-up (27 January 2022 for Sweden and Norway; 31 December 2021 for Denmark). Births were identified through the Swedish Pregnancy Register,18 the Danish National Patient Register (registrations of International Classification of Disease version 10 codes Z38, O80–84 and P95),19 and the Medical Birth Registry of Norway.20 21 The small number of late induced abortions conducted after 22 completed gestational weeks were excluded (n=110 for Denmark; n=23 for Norway; information not available for Sweden). The Danish and Norwegian data included all births nationally, while the Swedish data included 94% of all births in Sweden (in 18 of 21 Swedish regions). We only included singletons and only the first registered birth to each woman during the study period. To avoid oversampling of preterm pregnancies towards end of the study period, we excluded pregnancies without the possibility to reach 42 completed weeks by the end of follow-up. We obtained information on maternal socioeconomic measures, infections with SARS-CoV-2, and vaccination against SARS-CoV-2 from national databases using unique national identification numbers.

Stillbirth

Stillbirth was defined as a fetal death after 22 completed gestational weeks. Gestational age was estimated based on ultrasound for the majority of births (more than 90% in all countries), and on date of last menstrual period when ultrasound estimates were missing. Information on the method of pregnancy dating is available from all registries. We used 22 completed gestational weeks and not 20 gestational weeks because only births occurring after 22 completed gestational weeks were available in the Swedish Pregnancy Register.

SARS-CoV-2 infection

The exposure of interest was a positive test for SARS-CoV-2 after 22 completed gestational weeks up until the day before delivery. Our hypothesis is that infection may increase the risk of fetal death. Thus, infection before gestational week 22 might increase the risk of fetal death prior to 22 completed weeks (miscarriage), which could affect our results when only looking at fetal deaths after 22 gestational weeks (stillbirths). We did not exclude women with infection prior to 22 gestational weeks from the analysis, but adjusted for this as a covariate. In addition, we conducted a sensitivity analysis excluding pregnancies to women infected during pregnancy prior to 22 completed gestational weeks. The beginning of pregnancy was estimated based on the date of birth minus the gestational age in days. Information on laboratory-confirmed PCR positive tests for SARS-CoV-2 was obtained from mandatory reports to SmiNet at the Public Health Agency for Sweden,22 and from the Norwegian Surveillance System for Communicable Diseases for Norway,23 while information on both PCR and antigen positive tests was available from the Microbiology Database at the State Serum Institute for Denmark.24 25 In Denmark, 10% of the positive tests were antigen tests, while 90% were PCR. The testing strategies in the three countries have varied over time. We stopped the follow-up of this study in January 2022, as all three Nordic countries changed their testing strategies around this time, and no longer recommended testing of all symptomatic individuals. More details about the testing strategies across the countries are available in online supplemental file.

Covariates

We obtained information on maternal age at the beginning of pregnancy (continuous), parity (0, 1, 2 or more), educational level (9 years or less, 10–12 years, more than 12 years), household income in tertiles based on the national distribution (first, second and third tertile), living with a partner (yes or no), region of birth (Scandinavia, Other European countries, Middle East/Africa, other/unknown), smoking in pregnancy (yes or no), pre or early-pregnancy body mass index (continuous), pre-existing chronic condition (yes or no) prior to pregnancy and vaccination against SARS-CoV-2 (none, before pregnancy, during pregnancy). A general recommendation for vaccination of all pregnant women was issued in May 2021, in Sweden, August 2021, in Norway and July 2021, in Denmark. Pre-existing chronic conditions prior to pregnancy included hypertension, chronic kidney disease, asthma, cardiovascular disease, thrombosis and diabetes. We also obtained information on pre-eclampsia, gestational diabetes, placental abruption, uterine rupture and shoulder dystocia from the birth and patient registries. These pregnancy complications were not adjusted for in the multivariable model because they may represent potential mediating pathways.

Statistical analysis

We used Cox regression analysis to evaluate the HR of stillbirth according to SARS-CoV-2 infection. The time axis was gestational age in days, and follow-up started at 22 completed gestational weeks (gestational day 154). The end of follow-up in the Cox model was the gestational day of birth. Infection with SARS-CoV-2 was entered as a time-varying exposure, so women could contribute both unexposed and exposed follow-up time. First, we evaluated the risk according to infection any time after 22 completed gestational weeks. To evaluate whether the risk was different during the first weeks after infection, we estimated the risk of stillbirth in separate analyses using 2, 4 and 6 weeks exposure windows following infection with SARS-CoV-2, and women were considered as unexposed after the end of the specific risk window. In multivariable analyses, we adjusted for age, parity, educational level, household income, living with a partner, region of birth and a time-varying variable of vaccination against SARS-CoV-2. We also adjusted for infection with SARS-CoV-2 during pregnancy prior to 22 completed gestational weeks in the main model. We subsequently adjusted for smoking during pregnancy and early or prepregnancy body mass index in births with this information available (87% of all births). The data from each country were analysed separately, and the results were subsequently meta-analysed using a random-effects model, with heterogeneity estimated using the I² statistic. Country specific results are not shown due to national privacy regulations for presenting small numbers.

In secondary analyses, we evaluated differences in the risk of stillbirth according to a positive test during time periods dominated by different variants of SARS-CoV-2; (Index (prior to 1 February 2021), Alpha (between 1 February 2021 and 30 June 2021) or Delta (between 1 July 2021 and 31 December 2021)). The dates for cut offs for the variants were based on the major circulating variant at the time,26–28 which was similar across the three countries. Finally, we conducted sensitivity analyses excluding women who were infected within 4 weeks prior to 22 completed gestational weeks. In addition, to compare women with the same possibility of being infected throughout pregnancy, we performed a sensitivity analysis restricted to pregnancies with an estimated start after 1 March 2020. These sensitivity analyses were only done for the 4-week risk window analysis. We further conducted a sensitivity analysis defining those who tested positive the last 3 days of pregnancy as unexposed. We also conducted a sensitivity analysis excluding women infected during pregnancy prior to 22 completed gestational weeks.

Analyses were conducted using STATA V.17 (Statacorp) and R V.4.2.1.

Patient and public involvement

No patients were involved in setting the research question or the outcome measures, nor were they involved in developing plans for recruitment, design or implementation of the study. No patients were asked to advise on interpretation or writing up of results. There are no plans to disseminate the results of the research to study participants or the relevant patient community.

Strengths and weaknesses of this study

Important strengths of this study include the population-covering and prospective nature of the registry data, the inclusion of data from three countries, the evaluation of the risk according to dominant variants of SARS-CoV-2, and our ability to adjust for several potential confounding factors. Our study also has limitations. We did not have any information on causes and classifications of stillbirths in the birth registries, including clinical information from autopsies or placental histopathological analysis. Furthermore, no information on SARS-CoV-2 infection of the fetus was available. We were also unable to look at the risk of stillbirth according to the gestational week of infection with SARS-CoV-2 in more detail due to the small number of exposed cases. There were also differences in test strategies and infection rates between countries and during the study period. From Sweden and Norway, we only had information on PCR-positive tests, while from Denmark, information on rapid antigen tests were also available. During the study period, individuals in Denmark who tested positive on a rapid antigen test were recommended to take a confirmatory PCR-test, and 90% of the positive tests from Denmark in this study were PCR, ensuring comparability between the countries. Among women with positive tests, there are likely some asymptomatic cases, as pregnant women exposed to infected individuals were recommended to take a PCR test.29 Also, women with early signs of fetal loss may have been more prone to testing which could inflate risk estimates. However, in the sensitivity analysis, where we defined pregnancies to women who tested positive the last 3 days of pregnancy as unexposed, results were similar. The risk of stillbirth was highest with infection during the Delta period in all three countries, which support that increased testing in threatening stillbirths is less likely to cause the associations. We were unable to distinguish the risk of stillbirths prior to the onset of labour and stillbirths arising intrapartum according to infection with SARS-CoV-2 due to small numbers. We also did not have any information on treatment for SARS-CoV-2, and we could, therefore, not evaluate how it might have impacted the subsequent risk of stillbirth. Finally, we had limited ability to evaluate the proportional hazards assumption due to the small number of exposed cases of stillbirth.

Comparison with previous studies

Our findings support previous studies indicating that SARS-CoV-2 infection increases the risk of stillbirth. The largest study to date from the US PHD-SR, found a twofold increased risk of stillbirth among women with a diagnosis of COVID-19 around the time of delivery.12 A study of 78 centres included in the Spanish Obstetric Emergency Group, including 1347 SARS-CoV-2 PCR-positive pregnant women registered between 26 February 2020 and 5 November 2020 on admission for delivery, and a concurrent sample of PCR-negative mothers, also indicated a greater proportion of stillbirth among women who were positive for SARS-CoV-2 compared with those who were negative (0.7% vs 0.2%, p=0.02).6 Similarly, a population-based study from England which included 342 080 delivering women, of which 3527 were registered in the birth record as being positive for SARS-CoV-2 at the time of delivery, the adjusted relative risk of stillbirth was 2.21 (95% CI 1.58 to 3.11).8

To study differences with viral variants, we used information on the major underlying circulating variant at the time when the woman tested positive. We observed the greatest increased risk of stillbirth among women who tested positive in the Delta dominated period. As our follow-up ended in January 2022, the number of stillbirths was too low to study the risk of stillbirth according to infection with the Omicron variant. Some other studies support that the Delta variant could have a greater impact on the risk of stillbirth compared with other variants. The US (PHD-SR) study found that the risk of stillbirth was greater in women infected with SARS-CoV-2 during the Delta period (adjusted RR 4.04; 95% CI 3.28 to 4.97), and lower in the pre-Delta period (adjusted RR 1.47; 95% CI 1.27 to 1.71).12 A population-based study from Scotland with 9817 women who tested positive for SARS-CoV-2 during pregnancy, also reported a lower rate of stillbirth in the Omicron-dominant period (4.3 per 1000 births) than in the Delta-dominant period (20.3 per 1000 births).16

We observed some differences in the associations across countries as evidenced by the heterogeneity statistics. There was substantially less heterogeneity when examining infection according to the major circulating variant at the time the women were infected with SARS-CoV-2. Statistical heterogeneity may be present due to small numbers in each of the countries, whereas clinical heterogeneity may be caused by national differences in the intensity of the pandemic itself or differences in testing strategies that led to differences in the proportion of positive cases detected across countries. We used random-effects meta-analysis to account for this heterogeneity in the country-combined estimates.

Some mechanistic evidence support that the SARS-CoV-2 virus can influence the risk of stillbirth. A histological analysis of the placenta from 15 women with severe COVID-19 who delivered in the third trimester indicated abnormal vessels and evidence of intervillous thrombi.30 Another study from Sweden of 14 placentas also supported massive perivillous fibrinoid deposition, in addition to intervillositis and tromphoblast necrosis.31 Notably, vascular malperfusion is commonly found in the placenta of stillbirths after COVID-19 infection.32 There is also evidence suggesting that the Delta variant of the SARS-CoV-2 could increase the risk of stillbirth. Histological evidence of one intrauterine fetal demise in an unvaccinated woman with mild symptoms of SARS-CoV-2, suggested that the excessive infiltration of immune cells and cytokines in the placenta due to the Delta variant caused severe placental inflammation and damage, which likely resulted in placental abruption and the demise of the fetus.33 However, the impact of the SARS-CoV-2 on the placenta still remains to be fully understood, as the existing studies are based on very small sample sizes.