Introduction

Coronavirus disease 19 (COVID-19) infection has affected public health worldwide, and the pattern of coronavirus variant types have changed rapidly. In South Korea, three epidemics of great magnitude of COVID-19 infection (pre-Delta, Delta, and Omicron periods) were experienced from 2020 to early 2022. Although evidence on the Omicron variant suggests less virulence compared with the Delta strain, it has a greater immune vaccine escape. [1, 2]. The characteristics of the three variants (pre-Delta, Delta, and Omicron) are different, and there is a paucity of information on large study population maternal and neonatal outcomes during each variant period.

Since early 2020, the COVID-19 pandemic has resulted in tremendous public health concerns. In particular, pregnant women have been classified as a vulnerable population to COVID-19 infection from the very start of the pandemic. However, the epidemiological evidence for infection risk during pregnancy on maternal and neonatal outcomes, such as preterm birth, hypertensive disorders in pregnancy, and postpartum hemorrhage of mothers and prematurity, respiratory and neurodevelopmental disorders of neonates, is still undetermined. Firstly, several previous studies were conducted at selected medical institutions, thus creating a risk of selection bias [1, 3, 4]. Secondly, compared to non-Western countries, many studies have been performed in Western countries where health outcomes, quarantine policies, lifestyles, and living environments related to COVID-19 are heterogeneous [1, 5]. Lastly, most of the existing studies examined the association between viral infection and maternal or neonatal outcomes; however, information on the casual relationship between COVID-19 and maternal-neonatal outcomes via advanced analytic approaches is limited [3, 6, 7].

We constructed a national mother-child cohort (NMCC) from January 1, 2020 through March 31, 2022 from the Korea Disease Control and Prevention Agency-COVID19-National Health Insurance Service cohort (K-COV-N cohort), and conducted a thorough analysis to explore our hypothesis that a causal association between COVID-19 infection and maternal and neonatal outcomes exists.

Materials and methods

Study population

We obtained the data of 414,020 mothers and children as a cohort from the mother-child pair database in K-COV-N cohort of the National Health Insurance Service (NHIS) from January 1, 2020 to March 31, 2022 to examine the relationship between maternal and newborn outcomes of COVID-19 infection (Fig 1). Participants were defined by applying two exclusion criteria. First, we excluded participants without demographic information (sex, age, income-level, employment status, residence area, and citizenship). Second, participants were excluded if there was no pregnancy-related information (gestational age, date of delivery, and parity) or if there were multiple fetuses. Therefore, a total of 369,887 mother-child pairs were defined as the subjects of this study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Enrollment of National Mother-Child Cohort (NMCC) for the association of COVID-19 infection during pregnancy with maternal and neonatal outcomes.

https://doi.org/10.1371/journal.pone.0284779.g001

Statistical analysis

To address heterogeneous distribution of covariates (i.e., to find a causal association), we conducted matching methods based on a propensity score (PS). The propensity score was calculated using a logistic regression model including all the covariates previously stated, and the 1:4 greedy nearest neighbor matching technique (caliper = 0.2) without replacement. This was performed to estimate an average treatment effect (in our case, COVID-19 infection). We calculated standardized mean differences (SMD) for distribution of covariates before and after matching between groups (Table 1 and S1–S3 Figs) to establish the matching performance. From the matched samples, we estimated the odds ratio (ORs) and 95% confidence intervals (CIs) to determine the effect of COVID-19 on outcomes using the generalized estimation equation with the “exchangeable” correlation matrix.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics for pregnant women according to COVID-19 variant types before and after propensity score matching.

https://doi.org/10.1371/journal.pone.0284779.t001

Additionally, to examine whether the effects of COVID-19 infection differed between infection periods during pregnancy, we subdivided the outcomes by gestational trimesters: the confirmation in the first trimester (0 to 13 weeks), second trimester (14 to 27 weeks), and third trimester (28 weeks to the end of the pregnancy). An additional subgroup analyses was performed on age group (< 35 vs ≥ 35), employment status, parity, and vaccination status. Statistical analysis as described above was subsequently performed on all sub-groups and sub-periods.

Results

A total of 369,887 mother-child pair datasets was included as study subjects. Among them were 2,158 (0.58%) cases of COVID-19 infection during pregnancy. This is broken down into 266 (0.07%) cases in the pre-Delta period, 663 (0.18%) cases in the Delta period, and 1,229 (0.33%) cases in the Omicron period (Table 1, S4 and S5 Figs).

In S2 Table, the average (±standard deviation) age of women that were pregnant was 32.4 (4.4) years and the average gestational age at delivery was found to be 38.8 (1.6) weeks. Of those pregnant women, 75% (n = 279,022) were employed and 25% (n = 90,865) were housewives. Furthermore, the cohort was comprised largely of domestic citizens, at 94% (n = 346,510) and foreign citizenship was 6% (n = 23,377). Parity was 58% (n = 215,001) for primiparous women and 42% (n = 154,886) for multiparous women and the rate of cesarean section procedures was 56% (n = 206,805). The percentage of pregnant women with an underlying disease was ~18% (n = 68,372). The vaccination status of the study groups was 75% (n = 277,277) for incomplete vaccination and 25% (n = 92,610) for complete vaccination. After PS matching, the balance between covariate groups was achieved (Table 1 and S1–S3 Figs).

Fig 2 shows the effects of COVID-19 infection during pregnancy on maternal and neonatal outcomes. Overall, the effect of COVID-19 was not observed across all outcomes and periods, except for certain cases (ORs for the pre-Delta period: 1.18, 95% CI: 0.81, 1.72; ORs for the Delta period: 1.11, 95% CI: 0.89, 1.39; ORs for the Omicron period: 0.75, 95% CI: 0.60, 0.93). Moreover, COVID-19 infection increased the risk of PPH in the Delta period (OR: 1.60, 95% CI: 1.00, 2.56) and neonatal outcomes in the pre-Delta period (OR: 2.19, 95% CI: 1.25, 3.86). Even more so, infection with the COVID-19 virus increased the number of admissions to the NICU across all sub-periods (OR of the pre-Delta period: 2.31, 95% CI: 1.31, 4.10; OR of the Delta period: 1.99, 95% CI: 1.47, 2.69; OR of Omicron the period: 2.36, 95% CI: 1.75, 3.18).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Association of COVID-19 infection during pregnancy with maternal and neonatal adverse outcomes using 1:4 propensity score matching.

The arrow indicates when the 95% confidence intervals boundary is beyond the axis limit. In the pre-Delta and Delta periods, the models adjusted for maternal age, sex of child, income, residence area, citizenship, parity, cesarean section, and underlying diseases. The Omicron period, the models adjusted maternal age, sex of child, income, residence area, citizenship, parity, cesarean section, underlying diseases, and vaccination. NA denotes model not applicable. Abbreviations: CI, confidence intervals; OR, odds ratio; NICU: neonatal intensive care unit.

https://doi.org/10.1371/journal.pone.0284779.g002

Fig 3 displays the trimester-specific results. The effect of COVID-19 on overall maternal and neonatal outcomes was not prominent across each trimester in the pre-Delta period, the Delta period and the Omicron period, respectively. Nevertheless, there were several notable trends that should not be overlooked. For example, PPH in the second trimester of the Delta period (OR: 2.26, 95% CI: 1.26, 4.05), neonatal outcomes in the third trimester of the pre-Delta period (OR: 2.73, 95% CI: 1.16, 6.43), and NICU admissions in the third trimester during the pre-Delta (OR: 4.57, 95% CI: 2.11, 9.88), Delta (OR: 3.03, 95% CI: 2.24, 4.11), and Omicron periods (OR: 2.36, 95% CI: 1.75, 3.18) were all positively associated with COVID-19 infection. On the other hand, during the Omicron period, maternal outcomes in the third trimester were negatively associated with COVID-19 infection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Association of COVID-19 infection in trimester of pregnancy with maternal and neonatal adverse outcomes using 1:4 propensity score matching.

The arrow indicates when the 95% confidence intervals boundary is beyond the axis limit. NA denotes model not applicable. Abbreviations: CI, confidence intervals; OR, odds ratio; NICU: neonatal intensive care unit.

https://doi.org/10.1371/journal.pone.0284779.g003

Fig 4 presents the results of the subgroup analysis by age groups, employment status, parity, and vaccination status. Unfortunately, we were unable to establish a consistent pattern to determine the risk differences between sub-populations. Based on the point estimate, however, fully vaccinated pregnant women revealed a lower rate of NICU admissions in the Omicron period, compared to mothers who did not have a second vaccine. In addition, pregnant women aged 35 or older generally showed a higher risk of PPH compared to their younger counterparts.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Subgroup analysis of association of COVID-19 infection during pregnancy with adverse outcomes (neonatal outcome, postpartum hemorrhage, and admission to NICU) using 1:4 propensity score matching.

Abbreviations: CI, confidence intervals; OR, odds ratio; NICU: neonatal intensive care unit.

https://doi.org/10.1371/journal.pone.0284779.g004

Finally, our sensitivity analysis showed consistent results (S6–S9 Figs) despite different modelling specifications and covariates being considered in the analysis.

Discussion

To the best of our knowledge, our study is the first NMCC to use data covering the different epidemic periods from the Pre-delta to the early phase of the Omicron variant. In addition, with prospective national cohort data, this study examined the effects of COVID-19 infection on maternal and neonatal outcomes in Korea. Our study indicates that the effect of COVID-19 was sporadically observed in neonatal outcomes and PPH across the sub-periods of COVID-19 variants. Additionally, COVID-19 infections increased admissions to the NICU during the entire variant study periods.

Previous studies have suggested uncertain or mixed results on the effect of COVID-19 infection during pregnancy on the risk of pregnancy-related outcomes (preeclampsia, preterm birth, miscarriage, and perinatal death) [1, 2, 5, 7, 12–16]. In a prior multinational study, a substantial increase in the risk of pregnancy-related complications in women with COVID-19 diagnoses was noted. However, as the authors acknowledged, there were significant differences in the risk of severe maternal morbidity and mortality among developed and less developed nations, as well as the risk of reporting bias [3]. Our study included the entire population of pregnant women in Korea and considered multiple demographic, socioeconomic, and medical histories that could potentially have inherent bias. Thus, with this nationwide cohort, we propose that our study and therefore the results, eliminate any estimate bias and are robust.

Nevertheless, our findings on PPH (positive risk estimates in the second trimester in the Delta sub-periods) are consistent with a previous meta-analysis [6]. Recent literature suggests that COVID-19-induced thrombocytopenia might inhibit physiologic coagulation leading to postpartum bleeding [17–19]. Furthermore, it is also speculated that COVID-19 infection with the Delta variant with relatively strong virulence levels might affect placental development and subclinical placental infection in mid-gestation. This can lead to disturbances to normal hemostasis after delivery, physiologically achieved by complete placental separation and uterine myometrial contraction that directly compresses the avulsed vessels on the implantation site, and can in turn lead to PPH [10, 20]. In the third trimester, particularly towards the end of the pregnancy and close to birth, obstetricians or maternal-fetal-medicine specialists in most of the medical institutions in Korea favored timely Cesarean section to vaginal delivery in COVID-19 infected pregnant women to minimize transmission and as a result, the risk of PPH may be reduced [11, 21, 22]. Additionally, although we included all the delivery cases observed in Korea, our results on PPH were not statistically significant. Because of this, we surmised that the number of infected mothers (2,158 cases) might be insufficient to examine and appropriately determine any causal association between COVID-19 infection and PPH. Therefore, further studies should be implemented.

Despite statistical insignificance, the relative frequency of NICU admissions was higher in the COVID-19 infected group than in the COVID-19 non-infected group during all three periods, in accord with the literature [3, 23]. On December 31, 2020, the Korean Society of Pediatric Infectious Diseases and the Korean Society of Neonatology updated guidelines for the management of newborn infants born to mothers with suspected or confirmed COVID-19 infection, including precautions for healthcare workers, neonatal resuscitation, neonatal isolation and medical care, breastfeeding, testing for COVID-19, and mother/child contact [24]. The national response by government and academia to COVID-19 cases in newborns in Korea was very efficient. This timely policy might have increased the number of NICU admissions but overall it helps to prevent adverse maternal and neonatal outcomes.

Also, from the trimester-specific analyses, there was a positive association between maternal infection in the third trimester and neonatal admission to the NICU. It is hypothesized that the risk of adverse outcomes in neonates who survive and acquire transferred immunity to COVID-19 via the placenta in the first and the second trimesters are at no greater risk than those neonates birthed by non-infected mothers during the same trimesters of pregnancy. Flannery et al. stated that the cord blood antibody concentrations correlated with the duration between onset of infection and delivery, and the transfer ratios increased with increasing time [25]. In addition, in Rathberger’s study, a longer period of time between maternal infection and birth appeared to favor the transfer of antibodies to the fetus [26]. This was especially prominent in the third trimester, as on the verge of delivery and birth, the risk of NICU admission was significantly higher in COVID-19 infected groups during the pre-Delta, Delta and Omicron periods. This event is most likely due to insufficient time to obtain immunity and a lack of COVID-19-oriented intensive care as preemptive and preventive measures against the pandemic. During the pre-Delta period, when the systematic medical care for COVID-19 infection was not fully available, the risk of adverse neonatal outcomes, as well as NICU admissions, increased in the COVID-19 infected group.

Based on the large cohort, we investigated the different impact of COVID-19 infection on pregnant women and neonates by sub-populations. Based on the point estimates, we observed several differences. Firstly, during the pre-Delta period, the risk of adverse neonatal outcomes and NICU admission was more pronounced in women of age < 35 years than in women of age ≥35 years; however, this pattern diminished in the Delta and Omicron periods. Secondly, the effect of infection on NICU admission was more prominent in employed mothers in the Delta period, and yet, this trend dissipated in the Omicron period. Nevertheless, across all sub-population results, we could not find any statistically significant differences in risk estimates. Therefore, these results on sub-populations should be interpreted tentatively.

Vaccination during pregnancy is a common way to prevent morbidity from infectious diseases, such as influenza and pertussis. However, COVID-19 vaccines were not explicitly studied in pregnant women and offspring, and so vaccinations for pregnant women were not offered in Korea in October 2021. Although the effect of vaccination was limited in our study, there were few pregnant women who were fully vaccinated during the pre-Delta and the Delta periods. During the Omicron period, the risk of NICU admission was higher in offspring of unvaccinated mothers than offspring of vaccinated mothers. This positive effect of vaccination in pregnant women has been continually recognized in much of the literature [4, 27, 28]. From a study carried out in Israel between January and April 2021, full records of COVID-19 vaccination status and third trimester of pregnancy showed a significant reduction in the risk of composite adverse neonatal outcomes among vaccinated women [29]. In addition, we cautiously speculate that vaccination might play a partial but important role in the negative association between maternal outcomes in the third trimester during the Omicron period and COVID-19 infection.

The main strengths associated with this study are that it provides novel data related to maternal and neonatal outcomes associated with COVID-19 infection during the pre-Delta, Delta and Omicron waves of the pandemic. Moreover, the fact that this nationwide cohort has a relatively large population size enabled this study to include sub-population and trimester-specific analyses as well as a COVID-19 variant-specific analysis. However, only the time of infection was used to indicate the predominant variant due to unavailable variant sequencing data. The Omicron-wave cohort was limited to the third trimester of pregnancy; hence we cannot rule out the possibility of a different course of first or second-trimester pregnancies infected with the Omicron strain. Finally, in this study, we focused on the main maternal outcomes in late gestation; therefore, further studies focusing on early pregnancy losses should be performed in the future.

Conclusions

To conclude, we found little evidence of the effect of COVID-19 on maternal and neonatal outcomes during the study period. However, we discovered possible evidence that suggests a relationship between COVID-19 infection in the second trimester and PPH, and the association of COVID-19 with increased NICU admissions, which can burden the healthcare system. Therefore, we propose that the timely and successful policies of government and academia in response to COVID-19 infections in newborns in Korea caused an increase in NICU admissions, but nonetheless, prevented adverse maternal and neonatal outcomes simultaneously.

Supporting information

Acknowledgments

This study was conducted as part of the public-private joint research on the COVID-19 co-hosted by the KDCA and the NHIS. This study used the database of the KDCA and the NHIS for policy and academic research. The research number of this study is KDCA-NHIS-2022-1-531.