Study population and setting
This study assessed the relationship between maternal COVID-19 diagnosis and stillbirth from March 2020-September 2021 using an electronic hospital record database that included 736 hospitals across the US. There were 1,249,634 deliveries which met the inclusion criteria of having ICD-10 diagnostic and procedure codes associated with obstetric delivery, ICD-10 codes documenting gestational age of 20 weeks or later, and a discharge date during the study period. Stillbirths were defined as fetal deaths at or after 20 weeks of gestation. Stillbirths, COVID-19 diagnoses and severity, and comorbidities were measured by the presence or absence of the associated ICD-10 diagnosis codes during the hospitalization event. Multivariable Poisson regression models were used to model the relative risk of stillbirth associated with COVID-19, accounting for correlation within facilities and women with multiple-gestation pregnancies. Secondary analyses stratified by time period, before and after the predominance of the Delta variant,COVID-19 severity, and the presence of other underlying comorbidities.
Summary of Main Findings
Among the 1,249,634 deliveries assessed from March 2020-September 2021 (median maternal age 29 years, 53.7% non-Hispanic white, 50.6% with private insurance), there were 21,653 (1.7%) with a documented maternal COVID-19 diagnosis and 1,227,981 (98.3%) without. There were 273 stillbirths (1.26%) among deliveries with COVID-19 diagnoses and 7,881 (0.64%) among those without. COVID-19 was associated with 1.9 times the risk of stillbirth (95% CI 1.69-2.15), independent of maternal age, race, health insurance type, obesity, smoking, diabetes, hypertension, and multiple-gestation pregnancies. The adjusted relative risk associated with COVID-19 was 1.47 (CI 1.27-1.71) and 4.04 (CI 3.28-4.97) when limited to the periods before and after the predominance of the Delta variant, respectively.
The study included data from more than one million deliveries during the COVID-19 in the pandemic, including 3 months of follow-up with Delta as the predominant variant.
Screening protocols for COVID-19 were not universal across hospitals or time and it is possible that women were more likely to be tested for COVID-19 following a stillbirth than a live birth, potentially overestimating an association between the two. Comorbidity ascertainment relied on ICD-10 code documentation at the time of hospitalization and likely only identified a fraction of underlying comorbidities that might differ between the pregnant women with and without COVID-19. Finally, the representativeness of the deliveries in this sample during this time frame to the full source population of deliveries in the US is not assessed.
This analysis included a sufficient sample size to evaluate the risk of stillbirths, a rare outcome. Prior assessments of the risk of stillbirth with COVID-19 were limited by small samples, both in terms of the coverage of the study period and overall number of deliveries.