key: cord-0681909-rey8khbi
authors: Cojocaru, L.; Noe, M.; Pahlavan, A.; Werzen, A.; Seung, H.; Yoo, Y. C. J.; Tyson, P.; Narayanan, S.; Turan, S.; Turan, O. M.; Chua, J. V.
title: Increased Risk of Severe COVID-19 Disease in Pregnancy in a Multicenter Propensity Score-Matched Study.
date: 2021-06-21
journal: nan
DOI: 10.1101/2021.06.18.21258899
sha: d7ee50dbc1612fb6b2a97cab81b7fb39aa620355
doc_id: 681909
cord_uid: rey8khbi

Background: Respiratory infections have long been associated with higher maternal and perinatal morbidity. Early data did not report an increased risk of SARS-CoV-2 infection or disease severity in pregnancy. However, surveillance data from the Center for Disease Control and Prevention (CDC) indicates a higher risk of severe disease and death in pregnant women with symptomatic SARS-CoV-2 infection, although this data is subject to ascertainment bias. Objective: To explore the association between COVID-19 disease severity and pregnancy in our university-based hospital system using measures such as COVID-19 ordinal scale severity score, hospitalization, intensive care unit admission, oxygen supplementation, invasive mechanical ventilation, and death. Study design: We conducted a retrospective, multicenter case-control study to understand the association between COVID-19 disease severity and pregnancy. We reviewed consecutive charts of adult females, ages 18-45, with laboratory-confirmed SARS-CoV-2 infection in six months between March 1, 2020, and August 31, 2020. Cases were patients diagnosed with COVID-19 during pregnancy, whereas controls were not pregnant at the time of COVID-19 diagnosis. Primary endpoints were the COVID-19 severity score at presentation (within four hours) and the nadir of the clinical course. The secondary endpoints were the proportion of patients requiring hospitalization, intensive care unit admission, oxygen supplementation, invasive mechanical ventilation, and death. Results: A higher proportion of pregnant women had moderate to severe COVID-19 disease at the nadir of the clinical course than nonpregnant women (25% vs. 16.1%, p=0.04, respectively). While there was a higher rate of hospitalization (25.6% vs. 17.2%), ICU admission (8.9% vs. 4.4%), need for vasoactive substances (5.0% vs. 2.8%), and invasive mechanical ventilation (5.6% vs. 2.8%) in the pregnant group, this difference was not significant after the propensity score matching was applied. We found a high rate of pregnancy complications in our population (40.7%). The most worrisome is the rate of hypertensive disorders of pregnancy (20.1%). Conclusions: In our propensity score-matched study, COVID-19 in pregnancy is associated with an increased risk of disease severity and an increased risk of pregnancy complications.

such as influenza, severe acute respiratory syndrome (SARS), and Middle East respiratory 49 syndrome (MERS), that have been associated with higher maternal and perinatal morbidity 1, 2 . 50 Early data did not support an increased risk of SARS-CoV-2 infection or disease severity in 51 pregnancy 3, 4 . However, more recent data from the Center for Disease Control and Prevention 52 (CDC) indicates a higher risk of severe disease and death in pregnant women with symptomatic 53 COVID-19, although this data was limited by ascertainment bias and other limitations of a 54 surveillance study 5, 6 . 55

Despite the incongruence between published studies, pregnancy appears to be associated with 56 increased risk of hospitalization, intensive care unit (ICU) admissions, invasive mechanical 57 ventilation (IMV), extracorporeal membranes oxygenation (ECMO), and death 1, 5, 6 . 58

Several physiologic and immunologic changes during pregnancy predispose pregnant women to 59 complications related to COVID-19 pneumonia. Increased oxygen consumption (15-20%) , low 60 functional residual capacity, and increased renal excretion of bicarbonate to compensate for the 61 respiratory alkalosis result in diminished compensatory homeostatic reserve 7, 8 . Additionally, 62 decreased cell-mediated immunity, helper-T-cell numbers, natural killer activity, along with 63 modulatory effects of T helper 17/regulatory T cells, may result in a decreased immunologic host 64 response [9] [10] [11] . Moreover, during pandemics, the unique needs of pregnant women are sometimes 65 overlooked, whether due to inadequate oxygen supplementation 12 , decrease in quality of care 13 , 66 or exclusion from clinical trials 14 . The incidence of COVID-19 is also higher in pregnant women 67 who are Hispanic or non-Hispanic Black. The difference in the prevalence of underlying chronic 68 conditions or social determinants of health may impact the severity of presenting illness or the 69 outcome of infection. 70

To further explore the association between COVID-19 disease severity and pregnancy, we 71 conducted a retrospective case-control study comparing pregnant and nonpregnant women 72 infected with SARS-CoV-2 in our university-based hospital system. We hypothesized that 73 pregnant women with COVID-19 are at an increased risk of disease severity compared with 74

For cases, additional endpoints analyzed were the proportion of patients with adverse maternal 114 outcomes [early pregnancy loss less than 20 weeks (EPL), later pregnancy loss of more than 20 115 weeks (LPL), neonatal demise (ND), fetal growth restriction (FGR), hypertensive disorders of 116 pregnancy (HDP), preterm delivery (PTD) and gestational diabetes (GDM)]. 117

Descriptive statistics were used for frequency, median, and mean. We employed propensity score 119 matching (PSM) to lessen the potential influence of confounding factors and increase the 120 reliability of the results. The PSM was performed using age, race/ethnicity, body mass index, and 121 comorbidities (hypertension, diabetes mellitus, heart disease, chronic pulmonary disease) (Table  122 1). It formed matched sets of cases and controls who share a similar propensity score. Optimal 123 one-to-one matching used the logit of the propensity score as the matching metric. To assess 124 PSM results, the standardized difference for variables included in the propensity score model 125 before and after PSM was provided in Supplementary Figure 1 . 126

The associations between outcomes or laboratory data and pregnancy were measured using a χ 2 127 or Fisher's exact test and Wilcoxon rank-sum test. The associations between characteristics and 128 the outcome, COVID-19 score at the nadir of clinical course, were measured using simple 129 logistic regression. For multivariate analysis, a logistic regression model was conducted to 130 evaluate the main exposure and independent factors. The odds ratio (OR) with 95% CI was used 131 to measure the magnitude of the association. The variance inflation factor (VIF) and correlation 132 coefficients were used to identify multicollinearity. The Hosmer-Lemeshow test, the deviance, 133

and Pearson χ 2 were used to assure the goodness-of-fit of the logistic model. Analyses were 134 performed with SAS software version 9.4 (SAS Institute, Cary, NC). 135

A total of 1,774 patients were identified, of which 637 were excluded due to insufficient 137 information. From the 1,137 patients included in the non-PSM (NPSM) analysis, 189 were 138 pregnant and 948 nonpregnant women. After the PSM algorithm was applied, 180 patients 139 remained in each group. Figure 1 Table 3) . 167

The variance inflation factor (VIF) was calculated for multiple regression to determine the 169 independent risk factors (Supplementary Table 2 ). There were no VIF greater than 2 except no 170 preexisting comorbidities (VIF=4.5), which was excluded from the final model. The backward 171 selection was applied from variables in Supplementary Table 2 (except for no these goodness-of-fit tests denote that the fitted model is correct. We found that pregnancy, age, 176

Hispanic race, BMI, lung disease, and diabetes are independent risk factors for COVID-19 177 severity ( 

The variance inflation factor (VIF) was calculated for multiple regression to determine the 203 independent risk factors (Supplementary Table 3 ). There were no VIF greater than 2 except no 204 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ;  preexisting comorbidities (VIF=6.1), which was excluded from the final model. The backward 205 selection was applied. The overall model was statistically significant likelihood Comparably to the non-propensity-matched population, we found that pregnancy, age, Hispanic 210 ethnicity, BMI, and chronic lung disease are independent risk factors ( 

Slightly above 40% of pregnant women in our study experienced pregnancy complications. The 219 median gestational age at delivery was 32 weeks. Hypertensive disorders of pregnancy were the 220 most common complication (20.1%), followed by preterm delivery (10.6%), gestational diabetes 221 mellitus (11.1%), and fetal growth restriction (3.2%). The early pregnancy loss rate was 2.2%, 222 and the late pregnancy loss 1.6%. There were no cases of neonatal demise. 223

Pregnancy was associated with an increased risk of COVID-19 severity at the nadir of clinical 226 course compared to nonpregnant patients. This was statistically significant even after adjusting 227 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ;  for baseline demographics and preexisting medical conditions. While there was a higher rate of 228 hospitalization, ICU admission, need for vasoactive substances, and IMV in the pregnant group, 229 this difference was not statistically significant after the propensity score matching was applied. It 230 is possible that this was due to the reduction in the sample size once the propensity score 231 matching was applied, as there was still a trend for higher rates of hospitalization, ICU 232 admission, and IMV in the pregnant group. Nonetheless, pregnancy showed an increased risk of 233 COVID-19 severity compared with a control group with well-balanced demographic and 234 preexisting medical condition characteristics. 235

In congruence with other studies, the Hispanic population appears to be disproportionately 236 affected. We found a high rate of pregnancy complications in our population (40.7%). The most 237

concerning is the rate of hypertensive disorders of pregnancy (20.1%). While we do not have a 238 pregnant COVID-19 negative control group, the rate of HDP is significantly higher when 239 compared to historical rates (6.9%) from the same state 19 . 240

Our data is in congruence with the CDC report and supports that pregnant women are at an 242 increased risk of severe COVID-19. As previously mentioned, several physiologic 7, 8 , and 243 immunologic changes 9-11 in pregnancy could contribute to an increased risk of respiratory 244 infection severity. However, it is not clear if the increased risk of HDP is a consequence of 245 COVID-19 infection or if these patients are predisposed to both conditions. A higher cytokine 246 levels that are seen with both HDP 20 and COVID-19 21 , as well as different susceptibility to both 247 conditions with angiotensin converting enzyme 2 polymorphism, might be in favor of the latter. 248

Pregnant women are at an increased risk of COVID-19 disease severity. Similarly, age, increased 250 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; BMI, Hispanic ethnicity, and chronic lung disease are independent risk factors. Though, based 251 on the confidence interval close to 1, the size effect of age and BMI may be small. We 252 recommend close surveillance of symptomatic pregnant women with COVID-19, especially in 253 the presence of other risk factors such as Hispanic ethnicity and chronic lung disease. Moreover, 254 we need to prioritize resources and ensure access to new treatment strategies and preventive 255 measures (e.g., vaccination) in this population. 256

One lesson learned during the current and previous pandemics is that despite an increased risk of 258 disease severity of infectious diseases in pregnancy, there is a lack of an equitable response in 259 drug development, vaccine studies, enrollment in clinical trial or development of monitoring and 260 management protocols that meet the unique needs of pregnant women 14, 22 . These deficiencies, 261 along with the increased risk of disease severity in pregnancy, further increase the risk of 262 pregnancy outcomes. We should advocate for pregnant women to determine their eligibility and 263 entry into research studies and contribute to the development of pregnancy-specific guidelines. 264

Treatment and vaccine studies should include pregnant women early in the trials as opposed to in 265 the subsequent phases. In addition, further research is required to elucidate the relationship 266 between COVID-19 in pregnancy and pregnancy complications. 267

To our knowledge, this is the first study to evaluated pregnancy impact on COVID-19 severity 269 using a PSM control. Our study's strengths are a relatively large sample size for both pregnant 270 and nonpregnant groups and a well-matched propensity score population, which decreases the 271 impact of confounding factors. Lastly, our population is predominantly Hispanic and non-272

Hispanic Black, which has been disproportionately affected by COVID-19 6 

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ;  understanding risk factors and outcomes in this population. 274

The limitations of our study are mainly related to the retrospective design. A certain proportion 275 of patients were excluded by design from the data analysis for PSM. Since these patients would 276 have belonged to the nonpregnant control group, this exclusion is unlikely to have influenced the 277 PSM results. There might have been differential testing practices or test availability between 278 pregnant and nonpregnant populations or disadvantaged groups. Even though we excluded 279 non-COVID-19 related hospitalizations, the threshold for hospitalization might have been lower 280 in the pregnant group. Likewise, the pregnancy goal for oxygen saturation of greater than 94% 281 might have resulted in lower thresholds for oxygen supplementation. 282

Lastly, the study is likely underpowered to detect a difference in rare events such as the need for 283 renal replacement therapy, extracorporeal membrane oxygenation, and death. In addition, while 284 we controlled for the most common comorbidities, the sample size of the PSM group was 285 prohibitive in terms of performing the PSM for all comorbidities. Therefore, our study results 286 may not be generalizable to other populations. 287

In our propensity score-matched study, COVID-19 in pregnancy is associated with an increased 289 risk of disease severity and an increased risk of perinatal complications. 290 Tables, Figures, and Supplementary Tables   Table 1 . Baseline characteristics of non-propensity matched and propensity-matched populations (1:1) . 1.0 IQR: interquartile range; n: number; BMI: body mass index; PMH: past medical history; HTN: hypertension; DM: diabetes mellitus. *Frequency missing: 9 (race/ethnicity, non-pregnant), 134 (BMI, non-pregnant), 9 (BMI, pregnant) in the non-propensity matched. † Excluded the subjects with missing values (see Figure 1 ). 17(85) 9(90) IQR: interquartile range; n: number. ALT: alanine aminotransferase; AST: aspartate aminotransferase; CRP: C-reactive protein; CXR: chest x-ray; CT: computer tomography. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021.

;

Potential Maternal and Infant Outcomes from (Wuhan) Coronavirus 2019-nCoV Infecting Pregnant Women: Lessons from SARS, MERS, and Other Human Coronavirus Infections

Pandemic 2009 influenza A(H1N1) virus illness among pregnant women in the United States

Clinical Characteristics of Pregnant Women with Covid-19 in Wuhan, China

Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study

Update: Characteristics of Symptomatic Women of Reproductive Age with Laboratory-Confirmed SARS-CoV-2 Infection by Pregnancy Status -United States

Characteristics of Women of Reproductive Age with Laboratory-Confirmed SARS-CoV-2 Infection by Pregnancy Status -United States

Pneumonia in pregnancy

Pneumonia in pregnancy

Respiratory complications of pregnancy

Pneumonia in pregnancy

The T helper type 17/regulatory T cell paradigm in pregnancy

Intensive care management of varicella pneumonia

COVID-19 Barriers to Care for Pregnant Patients in Prolonged Isolation

Protection by Exclusion: Another Missed Opportunity to Include Pregnant Women in Research During the Coronavirus Disease 2019 (COVID-19) Pandemic

The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies

Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support

The REDCap consortium: Building an international community of software platform partners

Remdesivir for the Treatment of Covid-19 -Final Report

Evaluation of US State-Level Variation in Hypertensive Disorders of Pregnancy

Maternal Serum Cytokine Concentrations in Healthy Pregnancy and Preeclampsia

Protected to death: systematic exclusion of pregnant women from Ebola virus disease trials

Acknowledgments: none.

Severity score* Reference = 0-1

Crude OR (95%CI)

Crude OR (95%CI)

Yes vs. no