key: cord-0975469-eprlsp50 authors: Leung, Char; de Paiva, Karina Mary title: Is pregnancy a risk factor for in‐hospital mortality in reproductive‐aged women with SARS‐CoV‐2 infection? A nationwide retrospective observational cohort study date: 2021-12-23 journal: Int J Gynaecol Obstet DOI: 10.1002/ijgo.14066 sha: 2c599d410f3ca073ccf100601f78ad8988dfea28 doc_id: 975469 cord_uid: eprlsp50 OBJECTIVE: To examine the effect of pregnancy on coronavirus disease 2019 (COVID‐19) ‐related in‐hospital mortality in women of reproductive age (between 15 and 45 years), with severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection confirmed with polymerase chain reaction tests, adjusted for factors such as co‐infection and intervention that were not considered in existing literature. METHODS: Data gathered from a nationwide database in Brazil were analyzed using multivariate logistic regression and multivariate Cox regression. Adjusted odds ratios and hazard ratios of independent factors associated with in‐hospital death were calculated. RESULTS: A total of 97 712 women were included in the study. After the adjustment for sociodemographic factors, epidemiologic characteristics, pre‐existing medical conditions, and intervention, pregnant women were found to be associated with lower risk for in‐hospital mortality as well as longer survival time compared with non‐pregnant women. When covariates of intervention were omitted from the analysis, pregnancy did not appear to be a significant factor associated with mortality. CONCLUSION: With the adjustment for intervention that was shown to be an independent factor associated with mortality, pregnancy appeared to have a favorable effect on SARS‐CoV‐2 infection. Given the immunosuppressed state of pregnancy, this finding is in line with the hypothetical protective role of a weaker immune response that inhibits the production of proinflammatory cytokine. LEUNG aNd dE PaIVa database reviews, 3-7 that identified independent factors associated with pregnancy. Findings of these studies suggest similar to higher mortality associated with pregnant women compared with nonpregnant women, after adjustment for sociodemographic factors, symptoms, and pre-existing medical conditions. In Brazil, a number of health information systems have been developed for surveillance including some nationwide ones, such as the Laboratory Environment Manager and the Influenza Epidemiological Surveillance Information System (Sistema de Informação da Vigilância Epidemiológica da Gripe, SIVEP-Gripe). The latter contains the data of patients admitted to hospital or who died without admission and was developed for the surveillance of SARS related to influenza and other respiratory viruses and, upon the outbreak, COVID-19 has been incorporated into the surveillance network. Based on the data gathered from SIVEP-Gripe, the present work identified the independent risk factors associated with COVID-19related in-hospital mortality in reproductive-aged Brazilian women with the aim to investigate whether pregnancy was a predictor of death. The study population was those women registered in SIVEP-Gripe, a publicly available nationwide database managed by the Brazilian Ministry of Health. It has been the primary source of information on COVID-19-related hospital admissions and deaths in Brazil, and has been described elsewhere. 8 Data were gathered from SIVEP-Gripe on September 1, 2021. All cases registered in the database meeting all the following criteria were included in the study: (1) diagnosed with SARS-CoV-2 infection using polymerase chain reaction tests, (2) female, (3) aged between 15 and 45 years inclusively, and (4) recovered or deceased. This means that cases failing to meet any of these criteria were excluded. Furthermore, patients who had not died from SARS-CoV-2 infection were excluded. Data are de-identified and publicly available, and therefore ethics approval is not required in the UK and Brazil. Data of covariates were also gathered from the database, including pregnancy status, sociodemographic factors, clinical characteristics, and intervention, as well as epidemiologic measures. Details of variables collected and treatment for missing data are listed in Table S1 . The statistical analysis consisted of three parts. First, the difference in the summary statistics between pregnant and non-pregnant cohorts was statistically tested. Second, multivariate logistic regression was used to identify independent risk factors for mortality, with the primary aim to see if pregnancy was a predictor of death. The accuracy of the model was assessed by the area under the receiver operating characteristic curve. Furthermore, as a comparison to analyses in the existing literature, the above-mentioned procedure was repeated with covariates on intervention, comorbidities, epidemiologic characteristics, or signs and symptoms omitted. Third, Kaplan-Meier survival curves and Cox regression were used to model two temporal measures: time from admission to recovery and time from admission to death. Details of statistical analyses can be found in Table S2 . All computations were performed using the software R Version 4.1.1. A P value <0.05 was considered significant. A sensitivity analysis was performed following VanderWeele and Ding. 9 E-values, defined as the minimum strength of association, on the risk ratio scale, that an unmeasured confounder would need to have with both the treatment and the outcome to fully explain away a specific treatment-outcome association, conditional on the measured covariates, were reported. Data were gathered from SIVEP-Gripe on September 1, 2021. As Table 1 ). The death rates were 10% and 17% for the pregnant and non-pregnant groups, respectively, and the difference was significant (P < 0.001). However, such a difference might be explained by the difference in other measures. First, the non-pregnant group was generally older, as indicated by the median of age (38 years versus 31 years, P < 0.001). The proportion of Caucasian women by ethnic composition was higher in the non-pregnant cohort (55% versus 47%, P < 0.001) and a higher proportion of Latino (52% versus 44%, P < 0.001) and indigenous (0.3% versus 0.2%, P < 0.001) women was found in the pregnant cohort. With respect to temporal measures, the time from symptom onset to death was 14 and 10 days for the pregnant and non-pregnant cohorts, respectively (P < 0.001). However, the times from symptom onset to admission were 6 and 7 days (P < 0.001) for the two cohorts, respectively; indicative of earlier hospitalization in the pregnant group. Of note, there was no significant difference for time from admission to recovery (P = 0.388). The need for intensive care might be associated with death. Despite the higher rate of intensive care unit (ICU) admission in the | 123 LEUNG aNd dE PaIVa pregnant cohort (29% versus 27%, P < 0.001), the median days in ICU was eight in both cohorts and the difference was not significant (P = 0.836, excluding those not admitted to the ICU). About 48% of the patients in the pregnant cohort required the use of ventilation, significantly lower than the other cohort (60%, P < 0.001). Similarly, the pregnant cohort saw a lower proportion of abnormal chest Xrays (87% versus 92%, P < 0.001). Nevertheless, it is not clear if vaccination and therapeutics play a role in disease severity. The pregnant cohort saw a lower rate of vaccination against SARS-CoV-2 (3% for non-pregnant versus 2% for pregnant, P < 0.001) but higher rates of vaccination against influenza (16% versus 8%, P < 0.001) and the use of antiviral agents (12% versus 8%, P < 0.001). For clinical manifestations, asymptomatic cases were 2.8-fold higher in the pregnant cohort than in the non-pregnant cohort (0.44% versus 0.16%, P < 0.001). Moreover, the pregnant cohort saw fewer lower respiratory tract symptoms including dyspnea (60% versus 70%, P < 0.001) and respiratory discomfort (45% versus 54%, P < 0.001). In line with this, 38% of the pregnant patients had oxygen saturation below 95%, compared with 54% of the nonpregnant patients (P < 0.001). Although this tends to suggest that pregnant patients fared better, the pregnant cohort showed similar upper respiratory tract symptoms. There was no significant difference in the prevalence rate of cough between the two groups (P = 0.989). Pregnant patients had fewer sore throats (22% versus 23%, P < 0.001), but had two-fold higher prevalence of coryza (9% versus 4%, P < 0.001). The death rate between the two groups might also be affected by the underlying medical conditions. About 79% of the pregnant Pregnancy was identified as an independent factor associated with COVID-19-related mortality, as shown in Figure 3 . However, pregnancy appeared to have a favorable effect on SARS-CoV-2 in- for association between pregnancy and covariates are included in Table S1 . Based on the rule of thumb that an E-value of 2 when the outcome is all-cause mortality is considered to provide evidence for robustness to confounding, 10 these E-values, which ranged from 1 to 5.4, suggest that relatively strong confounding assumptions would be needed to eliminate the association between these factors and pregnancy. Results of covariate omission analysis are shown in Figure S1 . The present work identified independent risk factors for mortality in women of reproductive age in Brazil based on the data of nearly 100 000 COVID-19 patients registered in the nationwide database. Existing studies have found higher or similar mortality in pregnant women compared with non-pregnant women, 3-7 demonstrating mixed results. In most studies, [3] [4] [5] 7 pregnancy was treated as the outcome rather than the exposure. From a statistical perspective, the exponential of the coefficient represents the adjusted ORs of pregnancy associated with death. In contrast, the present work examined whether pregnancy was an independent predictor of inhospital death and estimated the effect of pregnancy on COVID-19related in-hospital mortality. The methodology used in the present work makes sense because mortality is the outcome whereas pregnancy is the exposure, resulting in adjusted ORs of death associated with pregnancy. In a study conducted in Mexico, 6 death was treated as the outcome variable. Although varying mortality risk factors between communities can attribute to this finding, the results of the covariate omission analysis showed that findings of pregnancy associated with higher risk for mortality reported in existing literature may be the result of the omission of intervention covariates. Although further investigation is needed to examine this finding, the hypothetical protective role of a weaker immune response where favorable disease course was observed in immunosuppressed hosts may serve as a possible explanation. 15 Although the maternal immune system protects the mother from infection, it is suppressed for successful pregnancy, so as to tolerate the fetus, which is considered as "foreign". 16 In immunosuppression, on the one hand, inhibition of viral replication by the immune system is restricted. On the other hand, immunosuppression can inhibit the production of proinflammatory cytokines, for example interleukin-6, which has been found to be an adequate predictor of severe COVID-19 17 because of its role in cytokine release syndrome and acute respiratory distress syndrome. 18 In fact, some studies have suggested that immunosuppression during pregnancy could protect the mother from the cytokine release syndrome. [19] [20] [21] [22] Nevertheless, this is not to say that immunosuppression is beneficial to SARS-CoV-2 infection. In fact, the present work found immunocompromise associated with increased risk for mortality ( Figure 3 ). Taken together, how immunosuppression is related to COVID-19related mortality is dependent on the type of immunosuppression, as pointed out in some systematic reviews. [23] [24] [25] Looking beyond pregnancy, the present work also demonstrated measures that can influence the risk of mortality. As addressed, intervention played a key role in mortality risk. Vaccination against SARS-CoV-2, for instance, reduced the risk of mortality although the insignificant OR might be attributed to the small sample size, which accounted for only 3% (2899 out of 97 712) of the study cohort. For influenza-related intervention, it was shown that vaccination appeared to offer some beneficial effects but the use of antivirals did The authors would like to thank Marcus Lacerda of Fundação de Medicina Tropical Dr Heitor Vieira Dourado for his advice. The authors have no conflicts of interest. Char Leung was responsible for study design, data collection, data analysis, data interpretation, and drafting the manuscript. Karina Mary de Paiva was responsible for literature review and data interpretation. Code available upon reasonable request. 2019 in pregnancy: living systematic review and metaanalysis Impact of COVID-19 on pregnancy Characteristics of women of reproductive age with laboratory-confirmed SARS-CoV-2 infection by pregnancy status -United States Update: Characteristics of symptomatic women of reproductive age with laboratoryconfirmed SARS-CoV-2 infection by pregnancy status -United States Severity of illness by pregnancy status among laboratory-confirmed SARS-CoV-2 infections occurring in reproductive-aged women in Colombia Pregnant women with SARS-CoV-2 infection are at higher risk of death and pneumonia: propensity score matched analysis of a nationwide prospective cohort (COV19Mx) The risk of clinical complications and death among pregnant women with COVID-19 in the Cerner COVID-19 cohort: a retrospective analysis Clinical characteristics and risk factors for death among hospitalised children and adolescents with COVID-19 in Brazil: an analysis of a nationwide database Sensitivity analysis in observational research: introducing the E-value Technical considerations in the use of the E-value The unacceptable tragedy of maternal mortality associated with COVID-19: (re)politicization of women's health and rights and the position of Brazilian nursing The tragedy of COVID-19 in Brazil: 124 maternal deaths and counting Laboratory biomarkers predicting COVID-19 severity in the emergency room Effect of socioeconomic inequalities and vulnerabilities on health-system preparedness and response to COVID-19 in Brazil: a comprehensive analysis How is immunosuppressive status affecting children and adults in SARS-CoV-2 infection? A systematic review The maternal immune system during pregnancy and its influence on fetal development Association between IL-6 and severe disease and mortality in COVID-19 disease: a systematic review and meta-analysis SARS-CoV2 and immunosuppression: a double-edged sword Do pregnant women have protective immunity against COVID-19? Immune response to COVID-19 during pregnancy Endogenous control of inflammation characterizes pregnant women with asymptomatic or paucisymptomatic SARS-CoV-2 infection Is pregnancy an immunological contributor to severe or controlled COVID-19 disease? COVID-19 in immunocompromised patients: a systematic review of cancer, hematopoietic cell and solid organ transplant patients Does type of immunosupression influence the course of Covid-19 infection? SARS-CoV-2 (COVID-19) in patients with some degree of immunosuppression Influenza vaccination and the risk of COVID-19 infection and severe illness in older adults in the United States