Introduction

Pregnant persons with SARS-CoV-2 infection are at increased risk of preterm birth [1,2,3,4,5]. However, the relationship between trimester of SARS-CoV-2 infection, illness severity, and preterm birth is not fully understood. SARS-CoV-2 infection is associated with maternal systemic inflammatory response and placental inflammation which may contribute to preterm birth [6]. Placental hypoperfusion can contribute to fetal growth restriction and non-reassuring fetal heart rate tracing which can increase risk of fetal demise and small-for-gestational age [6,7,8,9,10,11].

Observational studies show neonates were delivered earlier to persons with symptomatic, rather than asymptomatic, COVID-19 [2, 12]. Previous studies suggest increased risk of preterm birth among persons with severe COVID-19, however, study populations were limited to those with infection detected at hospital admission for delivery, or included those no longer at risk of preterm delivery (i.e., those with SARS-CoV-2 infection ≥37 weeks gestation) [2, 3, 13, 14]. A study including SARS-CoV-2 infection at any point during pregnancy reported that pregnant persons may be more susceptible to preterm delivery if SARS-CoV-2 infection occurred after 24 weeks gestation; however, the role of illness severity in this association is unclear [15].

Using data from a mother-infant linked longitudinal surveillance cohort, we examined trimester of SARS-CoV-2 infection and COVID-19 severity among pregnant persons and risk of preterm birth, adjusting for maternal age, selected underlying conditions, and pregnancy complications. We also investigated how trimester of infection modifies the relationship between COVID-19 severity and preterm birth. Understanding the association of trimester of SARS-CoV-2 infection and COVID-19 severity with preterm birth may guide strategies to improve uptake in prevention efforts against SARS-CoV-2, such as vaccination. We hypothesize that moderate-to-severe and critical COVID-19 during the third trimester is associated with increased risk of preterm birth.

Methods

The Surveillance for Emerging Threats to Mothers and Babies Network (SET-NET) conducts population-based longitudinal surveillance of pregnant persons and their infants to understand the impact of emerging and reemerging threats. Methods have been described previously [16]. Briefly, state, local, or territorial health departments from 21 jurisdictions in the United States (Arkansas, California [excluding Los Angeles County], Georgia, Houston, Iowa, Los Angeles County, Maryland, Massachusetts, Michigan, Minnesota, Missouri, Nebraska, New Hampshire, New Jersey, New York [excluding New York City], Pennsylvania [excluding Philadelphia], Philadelphia, Puerto Rico, Tennessee, US Virgin Islands, and Washington) collected information on pregnancy outcomes among persons with SARS-CoV-2 infection in pregnancy during January 20-December 31, 2020, and reported these data to CDC by September 3, 2021. CDC receives these data on pregnant persons with positive SARS-CoV-2 infection confirmed by nucleic acid amplification testing (NAAT) during pregnancy through the day of delivery [17].

Reporting jurisdictions collect data on pregnant persons through reporting of pregnancy status in COVID-19 case surveillance, or through linkages of case surveillance data with existing local data systems, such as birth certificates, to determine if infection occurred during a pregnancy. Additional data elements are abstracted from maternal and infant medical records from the birth hospitalization. Of the 21 jurisdictions included in this analysis, trained staff from seventeen participating jurisdictions abstracted medical records for all mother-infant pairs meeting inclusion criteria, while staff from four jurisdictions used a random sampling approach to identify a subset of pregnancies for medical record abstraction. We weighted records abstracted based on the random sample to account for selection probability and nonresponse such that the sampling weights totaled the number of pregnancies meeting the case definition [18].

Preterm birth (<37 weeks gestation) among liveborn infants was the primary outcome. Information about gestational age at birth was obtained from birth certificates or abstracted from medical records. For birth certificate data, obstetric estimate of gestational age was obtained [19]. For medical records, gestational age was estimated from reported information on gestational age at birth or by calculating gestational age from reported estimated delivery date (EDD) or last menstrual period (LMP); abstractors were instructed to prioritize EDD determined by first trimester ultrasound.

We also assessed labor induction and mode of delivery in this analysis. Indications for induction included past due date or post-dates; prelabor rupture of membranes; maternal or labor-related conditions such as pre-eclampsia, maternal respiratory distress, and abnormal labor (failure to progress or descend); and fetal conditions such as non-reassuring antenatal fetal test, oligohydramnios, fetal growth restriction, and poor interval fetal growth. Emergent cesarean deliveries included those indicated for urgent maternal conditions such as maternal decompensation, or fetal conditions such as non-reassuring fetal testing, or uterine rupture. Non-emergent cesarean deliveries included those planned, scheduled, or indicated for reasons such as breech presentation, history of previous cesarean deliveries, failed induction, arrest of labor in the first or second stage.

Trimester of infection was determined by gestational age at date of first positive SARS-CoV-2 molecular test. Trimester was categorized as first: <14 weeks, second: 14–27 weeks, and third: 28–36 weeks.

Illness severity categories (asymptomatic infection, mild, moderate-to-severe, and critical), defined in a previous analysis of SET-NET data [20], were adapted from National Institutes of Health (NIH) and World Health Organization (WHO) severity of illness categories [21, 22]. We classified illness severity using a hierarchy prioritizing indications of critical disease first (intensive care unit [ICU] admission, invasive ventilation, complications, such as acute respiratory distress syndrome or extracorporeal membrane oxygenation, and death), followed by moderate-to-severe disease (use of COVID-19 therapies and severe symptoms, such as shortness of breath with fever or cough), followed by mild symptomatic disease and asymptomatic infection. Asymptomatic infection required report of no symptoms; absence of reported symptoms was not sufficient to meet this criterion. Pregnant persons with missing data on these indications were classified as having insufficient information and were excluded from analysis. We conducted a sensitivity analysis and compared the prevalence of preterm birth, underlying conditions, and trimester of SARS-CoV-2 infection among pregnant persons excluded from the analysis due to insufficient information with our included study population.

The current analysis was limited to pregnant persons with laboratory-confirmed SARS-CoV-2 infection occurring at <37 weeks gestation in 2020 and who delivered a live singleton infant. To avoid overestimating the risk of preterm birth from reporting bias, we limited our analysis to pregnancies with at least 47 weeks (43 weeks gestation plus 4 weeks of reporting lag) between calculated LMP (using the gestational dating methods described above) and the end of the reporting period (September 3, 2021).

Unadjusted and adjusted risk ratios (aRR) and 95% confidence intervals (CI) for preterm birth for trimester of SARS-CoV-2 infection (referent = first), illness severity (referent = mild), and illness severity stratified by trimester of infection were estimated by Cox regression with constant time at risk, which estimates prevalence ratios using survey procedures to account for sampling [23]. Mild illness was used as the referent, given the potential detection bias of asymptomatic people presenting for preterm delivery and detected incidentally because of universal screening protocols. Minimum sufficient adjustment variables were determined using a directed acyclic graph (Appendix 1) and included maternal age (in years) as a continuous variable, presence of one or more of the following underlying conditions: pre-pregnancy obesity, chronic lung disease, chronic hypertension, diabetes mellitus type 1 or 2, cardiovascular disease, immunosuppression, and presence of gestational diabetes or hypertensive disorders of pregnancy. We conducted a sensitivity analysis to include adjustment for race/ethnicity to compare any differences in the magnitude or direction of associations. In addition, a time-to-event analysis examined days to delivery after SARS-CoV-2 infection. Kaplan-Meier curves were generated to compare the percentage of persons still pregnant after SARS-CoV-2 infection by illness severity, stratified by trimester of infection. The Wilcoxon test was used to determine differences in days to delivery after infection among persons with mild, moderate-to-severe, or critical COVID-19. Analyses were conducted using SAS 9.4 software and survey procedures to account for jurisdiction, strata, and sampling weights [24]. This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy [25].

Results

Jurisdictions reported 22,278 pregnant persons infected with SARS-CoV-2 between January 20 and December 31, 2020. Of these, 8754 (39.3%) had sufficient information for illness severity classification; however, 27.6% of those with sufficient information were excluded due to one of the following reasons: <47 weeks duration between calculated LMP and the end of the reporting period, multiple gestation pregnancy, did not deliver between 22 and <43 weeks, or infection at ≥37 weeks gestation. Overall, 6336 (28.4%) persons met inclusion criteria (Fig. 1). Pregnant persons were most commonly 25–34 years of age (57.1%), Hispanic or Latino ethnicity (52.2%), enrolled in Medicaid at delivery (50.7%), infected in the third trimester with SARS-CoV-2 (46.5%), and had term deliveries (89.0%) (Table 1). Black, non-Hispanic pregnant persons represented 13.0% of the population. At least one underlying medical condition was reported for 37.5% of pregnant persons, with prepregnancy obesity being the most common condition (36.6%). Gestational diabetes was reported in 12.3% of people and hypertensive disorders of pregnancy were reported in 13.1%. Illness severity varied as follows: 57.3% experienced mild COVID-19, 25.0% moderate-to-severe, 3.5% critical, and 14.2% asymptomatic infection.

Fig. 1: Flow-chart of pregnant persons with SARS-CoV-2 infection included in the study.
figure 1

SARS-CoV-2 severe acute respiratory syndrome coronavirus 2, NAAT nucleic acid amplification testing 1To avoid overestimating preterm birth from reporting bias, we limited to pregnancies with at least 47 weeks (43 weeks gestation plus 4 weeks of reporting lag) between calculated LMP and the end of the reporting period (September 3, 2021). 2We weighted records abstracted based on the random sample to account for selection probability and nonresponse such that the sampling weights totaled the number of pregnancies meeting the case definition.

Full size image
Table 1 Demographic and infection characteristics of pregnant persons with SARS-CoV-2 infection by trimester of infection and disease severity.
Full size table

Among 862 infants born preterm (11.0%), 5.5% were extremely preterm (22–27 weeks), 9.6% were early preterm (28–31 weeks), 9.7% were moderately preterm (32–33 weeks), and 75.2% were late preterm (34–36 weeks) (Table 1). Among persons who delivered preterm, most had mild COVID-19 (39.7%), were infected with SARS-CoV-2 in the third trimester (56.9%) (Table 2) and were infected at least 15 days before delivery (58.5%) (data not shown).

Table 2 Risk ratios for preterm birth among pregnant persons with SARS-CoV-2 infection for trimester of infection and COVID-19 severity.
Full size table

Persons infected with SARS-CoV-2 in the third trimester were at increased risk of preterm birth compared with those infected in the first trimester (aRR: 1.8; 95% CI: 1.3–2.7) (Table 2). Compared with pregnant persons who experienced mild COVID-19, those with critical COVID-19 (aRR: 4.6; 95% CI: 3.0–6.9) or asymptomatic infection (aRR: 2.6; 95% CI: 1.7–3.9) were at increased risk of preterm birth. Pregnant persons experiencing moderate-to-severe COVID-19, compared with mild, did not have an increased risk of preterm birth (aRR: 1.0; 95% CI: 0.7–1.5).

When stratified by trimester of infection, risk of preterm birth was elevated for pregnant persons with critical COVID-19 (aRRs 4.0–4.6) in the second and third trimesters and with asymptomatic infection in all three trimesters (aRRs 2.1–3.4) (Table 2). The association between preterm birth and critical COVID-19 in the first trimester was not estimated due to small cell sizes. Among infections occurring during the second or third trimester, critical COVID-19 was associated with shorter time from infection to delivery, compared with mild COVID-19 (p < 0.001 and p = 0.02, respectively) (Fig. 2). The median number of days to delivery from SARS-COV-2 infection for pregnant persons with critical COVID-19 was 86 among those infected in the second trimester, and 13 among those infected in the third trimester.

Fig. 2: Timing from infection to delivery among pregnant persons with SARS-CoV-2 infection by severity of COVID-19 and trimester in which infection occurred.
figure 2

SARS-CoV-2 severe acute respiratory syndrome coronavirus 2, COVID-19 Coronavirus disease 2019 Dotted lines indicated median time from infection to birth at each level of COVID-19 severity. A SARS-CoV-2 infection occurring in the first trimester. B SARS-CoV-2 infection occurring in the second trimester. C SARS-CoV-2 infection occurring in the third trimester.

Full size image

Nearly two-thirds of pregnant persons did not have an induction of labor (65.7%), regardless of disease severity or trimester of infection (Table 3). Among those induced, the most common reason for induction was maternal condition (58.3%). Where mode of delivery was known (96.7%), most persons delivered vaginally (64.7%), whereas 35.3% delivered by cesarean. Among those who delivered by cesarean, non-emergent was reported in 71.3%, and emergent was reported in 28.7%. Rates of cesarean deliveries remained similar by trimester of infection and for asymptomatic and mild COVID-19. Nearly half of those with moderate-to-severe or critical COVID-19 delivered by cesarean, with most critical COVID-19 cesarean deliveries as emergent (76.2%). Among persons with critical COVID-19 who delivered by emergent cesarean, the most reported reason for emergent cesarean was maternal condition (75.4%).

Table 3 Labor induction and mode of delivery among pregnant persons with SARS-CoV-2 infection during pregnancy by trimester of infection and disease severity.
Full size table

Discussion

In this analysis of pregnant persons with SARS-CoV-2 infection from 2020, those with critical COVID-19 in the second or third trimester had four times the risk of preterm birth compared with pregnant persons with mild COVID-19. Moderate-to-severe COVID-19 in any trimester was not associated with increased risk of preterm birth. Critical COVID-19, compared with mild, was associated with shorter time to delivery when infection occurred during the second or third trimester. Pregnant persons with asymptomatic infection in any trimester had increased risk of preterm birth. Although we observed an increased risk of preterm birth for pregnant persons with critical COVID-19 or asymptomatic infection in the second trimester, confidence intervals were wide, suggesting that associations be interpreted with caution.

Among persons with critical COVID-19, we observed higher frequency of emergent cesarean deliveries due to maternal condition, and maternal condition as the most common reason for labor induction. This is consistent with previous studies in which preterm birth was increased among mothers whose worsening COVID-19 symptoms led to obstetric intervention [26, 27]. Studies reported an increased risk of medically-indicated preterm birth among pregnant persons with COVID-19, and a higher prevalence of medically-indicated preterm birth among persons with severe-critical COVID-19 [2, 14].

Preterm birth rates are higher among non-Hispanic Black persons than for White non-Hispanic, Asian, or Hispanic persons in the United States [28, 29]. We conducted a sensitivity analysis by including race/ethnicity in our models, and adjusted risk ratios did not differ.

Findings from our study can be used to counsel patients about the importance of vaccination in order to reduce risk of preterm birth. Pregnant persons are at increased risk for severe COVID-19 [4, 30, 31], and our findings suggest that those with critical COVID-19 are at higher risk of preterm birth. Our study did not find increased risk of preterm birth among persons with moderate-to-severe COVID-19; however, it is important to note that the grouping of moderate and severe disease may mask any increased risk of preterm birth among persons with severe COVID-19. Our study was unable to distinguish between moderate and severe COVID-19, as the surveillance system does not collect data on use of supplemental oxygen.

Little is known about the impact of asymptomatic SARS-CoV-2 infection on pregnancy outcomes or placental pathology. In this study, asymptomatic persons had increased risk of preterm birth, however this may be attributed to screening bias, residual confounding, or subclinical processes impacting gestation. Among asymptomatic persons infected with SARS-CoV-2 prior to 37 weeks gestation, the proportion of preterm deliveries was high (20%). Pregnant persons may have been diagnosed upon admission for delivery, however, to assess risk of preterm birth, we excluded those with infection > =37 weeks gestation, potentially overestimating the risk of preterm delivery among asymptomatic persons. However, screening at admission does not account completely for the increased risk, as asymptomatic persons infected in the first trimester also had elevated risk of preterm birth. Further research is needed to understand whether this finding could be due to subclinical processes that may contribute to preterm birth. Placental inflammation in the setting of COVID-19 has been described and has been proposed as a possible mechanism for adverse pregnancy outcomes (including stillbirth) and may be part of the pathway for preterm delivery among pregnant people with COVID-19. Given a lack of systematic testing and reporting of placental testing and pathology, we are unable to assess the impact of otherwise asymptomatic infection on the placenta in this surveillance network.

It is possible that the increased risk of preterm birth seen in pregnant persons with critical COVID-19 works through the mechanism of systemic inflammation and hypercoagulability. Among pregnant persons with severe COVID-19, an association between COVID-19 and higher rates of preeclampsia has been reported [7, 32]. It is unclear if COVID-19 causes a preeclampsia-like syndrome or if SARS-CoV-2 infection results in increased risk for preeclampsia. Systemic inflammation and hypercoagulability are features seen both in nonpregnant COVID-19 patients [33] and in preeclampsia [8, 10, 11].

Strengths of this study include the ability to define COVID-19 severity using data abstracted from medical records. In addition, analysis was limited to those at risk of preterm birth (i.e., SARS-CoV-2-infection <37 weeks), and to those who had at least 47 weeks between LMP and end of reporting period to prevent overestimating the risk of preterm birth.

Study findings are subject to at least five limitations. First, a large proportion of pregnant persons (49%) could not be classified for COVID-19 severity due to insufficient information and were excluded from analysis. However, a sensitivity analysis revealed that prevalence of preterm birth, underlying conditions, and trimester of SARS-CoV-2 infection among excluded persons were similar to our included population. Second, data were gathered from sources that varied across jurisdictions, which might have led to misclassification of variables of interest, such as preterm birth and trimester of infection due to differing sources of gestational age. However, we implemented a conservative gestational dating approach to avoid overestimating preterm births [34, 35]. Third, mild COVID-19 or asymptomatic infection may be under ascertained or misclassified as such patients may not have sought care and, therefore, had no medical record. Fourth, our population captures infections occurring in 2020 when COVID-19 vaccinations were not yet available and before circulation of Delta (B.1.617.2) and Omicron (B.1.1.529) variants. Fifth, SET-NET data do not capture history of preterm delivery, nor whether the preterm delivery was medically indicated.

This study describes a large population-based cohort of persons with SARS-CoV-2 infection during pregnancy and preterm birth. Critical COVID-19 occurring in the second or third trimester was associated with increased risk of preterm birth and delivery occurred sooner after infection compared with mild COVID-19. In addition to being at increased risk for severe COVID-19 [4, 30, 31], pregnant persons with SARS-CoV-2 infection are at increased risk for preterm birth [36], and these data further suggest the association is particularly evident for those who experience critical illness. Individuals who are pregnant or planning to become pregnant should be vaccinated against COVID-19 to mitigate serious illness, death, and adverse pregnancy outcomes such as preterm birth [37].

Disclaimer

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.