Outcomes of SARS-CoV-2 and Seasonal Viruses Among Children Hospitalized in Brazil

We analyzed all cases hospitalized with SARI recorded in the Influenza Epidemiologic Surveillance Information System (SIVEP-Gripe). SIVEP-Gripe is a nationwide database established by the Ministry of Health to keep surveillance of SARI in Brazil.⁶  To be registered in the SIVEP-Gripe database, patients must present with a flu-like syndrome and at least 1 of the following criteria: Dyspnea, respiratory distress, oxygen saturation <95% in room air, and cyanosis. For pediatric cases, child-specific symptoms (intercostal retractions, nasal dilation, dehydration, and lack of appetite) are also considered for inclusion in the database. SARI is a notifiable disease in Brazil, and it must be reported to the health authorities within 24 hours of suspicion. The notification is compulsory, and the information is recorded in the SIVEP-Gripe database, which receives data from patients admitted to both public and private hospitals. Upon admission, cases were tested for SARS-CoV-2 and other common seasonal respiratory viral etiologies using reverse transcriptase-polymerase chain reaction or antigen testing. Detailed information regarding the SIVEP-Gripe database, including report form, data dictionary, codes, and all deidentified data, are publicly available at https://opendatasus.saude.gov.br/dataset. Additional information regarding this data set, including the steps of the data retrieving and management, is provided in detail in Supplemental Information 1. The inclusion criterion was pediatric patients (aged <18 years old) registered in the database between February 2020 to February 2023. The information about included and excluded cases is displayed in Fig 1.

SIVEP-Gripe provides clinical and demographic data, including age, sex, ethnicity, country regions, date of symptom onset, date of admission, signs and or symptoms at admission, oxygen saturation, and comorbidities. For analysis, age was categorized into 3 groups (aged <2 years, 2–11 years, and 12–17 years). Additional information regarding data preparation, definitions of covariates, and missing management are detailed in Supplemental Information 2.

The primary exposure of interest was the virus strain identified by the laboratory tests at admission.

The primary outcome was in-hospital mortality. The clinical course was reported regarding respiratory support (none, noninvasive oxygen support, and invasive ventilation), admission to the ICU, and death.

The analysis was conducted in 3 consecutive stages. First, we used summary statistics to describe the clinical and demographic characteristics of the population included in the analysis. For this stage, descriptive data are presented as means (and SDs), medians (and interquartile ranges), counts, and proportions. We used the F test and the χ² test for comparisons between groups when appropriate. In the second step, we evaluated the effect of viral etiology on the survival of children and adolescents with SARI. For this step, we carried out a competing-risk survival analysis, using the cumulative incidence function (CIF)⁷  and the Fine-Gray model⁸  to estimate the cumulative incidence of the primary outcome over time. In this analysis, in-hospital mortality was the primary outcome, and hospital discharge was considered a competing event. Finally, we conducted a multivariable competing-risk survival analysis for each viral strain to identify independent risk factors of death. All models were adjusted by sex, ethnicity, year of admission, region of hospitalization, oxygen saturation at admission, and comorbidities. The results are expressed as adjusted hazard ratios (HRs) and 95% confidence intervals (CIs). All statistical tests were 2-tailed, and statistical significance was set at P < .05. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline for cohort studies.

We accessed data in SIVEP-Gripe, which are already deidentified and publicly available. The study was approved by the Federal University of Minas Gerais institutional review board (Research Ethics Committee, reference number 6.127.414).

Between February 16, 2020, and February 21, 2023, 3 515 224 cases of SARI were reported in the SIVEP-Gripe. Among them, 411 400 patients (11.7%) were <18 years of age at admission. A total of 175 707 cases were excluded. Therefore, the final sample consisted of 235 829 hospitalized pediatric patients (Fig 1). We performed a sensitivity analysis for those cases without information on viral etiology tests (described in detail in Supplemental Information 3).

The demographic and clinical characteristics of the 235 829 pediatric patients hospitalized with SARI according to viral etiology are shown in Table 1. The overall mean age at admission was 3.8 years (SD 4.5). Regarding viral etiology, there was a higher proportion of positive cases for SARS-CoV-2 in adolescents, whereas RSV cases predominate in infants. Regarding clinical outcomes, 54 962 (25.6%) patients were admitted to the ICU, 105 401 (50.63%) required noninvasive oxygen support, 18 472 (8.9%) required invasive ventilation and 7078 (3.0%) individuals had a fatal outcome. Interestingly, although RSV positive cases had proportionally lower oxygen saturation at entry and required more ICU admission and oxygen support, the in-hospital mortality rate was higher for SARS-CoV-2 positive cases (Table 1).

Figure 2 shows the cases hospitalized with SARI over time in patients <18 years of age with laboratory-confirmed viral etiology, stratified by age group and year of hospitalization. The first laboratory-confirmed SARS-CoV-2 infection was reported in Brazil on February 16, 2020. Figure 2 shows that SARS-CoV-2 infection became prevalent in all age groups around the 14th epidemiologic week of 2020 and maintained a prevalence >80% throughout 2020. During the second year of the pandemic, SARS-CoV-2 infection was still prevalent in children, but RSV became dominant in infants around the 10th epidemiologic week. During the third year of the pandemic, with the emergence of the omicron variant, SARS-CoV-2 infection became predominant again, regardless of age, until around the ninth epidemiologic week, when RSV returned to predominate in infants (Fig 2).

According to the competing-risk survival analysis, the overall estimated probability of death of the entire cohort was 3.2%. Figure 3 illustrates the CIF according to viral etiology. The estimated probability of a fatal outcome at 30 days of hospitalization according to the viral strain was 6.5%, 3.4%, 2.9%, 2.3%, 2.1%, and 1.8%, for SARS-CoV-2, coinfection, adenovirus, influenza, other viruses, and RSV, respectively. For those cases with a negative test at admission, the probability of death was estimated as 2.4% (Fig 3).

In the competing-risk univariate analysis, children with a positive test for SARS-CoV-2 had a hazard of death 3 times higher compared with individuals with a negative test (HR, 3.3; 95% CI, 3.1–3.5). Cases with coinfection (HR, 1.7; 95% CI, 1.2–2.4) and a positive test for adenovirus (HR, 1.4; 95% CI, 1.2–1.7) also had a significantly increased death hazard compared with the reference category. On the other hand, cases with a positive test for RSV (HR, 0.57; 95% CI, 0.51–0.65) and rhinovirus (HR, 0.79; 95% CI, 0.69–0.94) had a relatively reduced hazard of death during hospital stay (Fig 4).

Figure 5 shows the multivariable competing risk analysis models of death risk factors for each of the main viral strains (Fig 5A–D). For RSV-positive cases, the following covariates were associated with fatal outcome, age (12–17 years), Indigenous ethnicity, admission in Northeast or North regions, oxygen saturation <95% at admission, and presence of comorbidities (Fig 5A). For influenza-positive cases, age (12–17 years), admission in Northeast or North regions, oxygen saturation <95%, and presence of comorbidities (Fig 5B). For adenovirus cases, covariates associated with the primary outcome were Indigenous ethnicity, admission in Northeast or North regions, oxygen saturation <95%, and presence of comorbidities (Fig 5C). For SARS-CoV-2–positive cases, age (<2 years and 12–17 years), Indigenous ethnicity, admission in Northeast or North regions, oxygen saturation <95%, and presence of comorbidities were risk factors of death. Of note, admission after 2021 was increasingly protective against death for cases positive for SARS-CoV-2 (Fig 5D).

In this population-based study conducted over the 3 years of the COVID-19 pandemic in Brazil, we analyzed a nationwide database of hospitalized children and adolescents with SARI and confirmed viral etiology. To our knowledge, this report is the first large-scale study assessing competing risks of death in children and adolescents with SARI associated with SARS-CoV-2 and other seasonal respiratory viruses. After accounting for competing events and confounding factors, patients infected with SARS-CoV-2 had the highest risk of death compared with those with a negative test. Positive tests for adenovirus, influenza, and coinfections were also associated with an increased risk of death. Other risk factors for death included hospitalization in the poorest regions of Brazil, low oxygen saturation at admission, and presence of comorbidities.

The COVID-19 pandemic has dramatically affected the epidemiology of acute respiratory infections. The measures taken to mitigate the pandemic have affected the spread of respiratory viruses, like RSV, influenza, and parainfluenza, leading to a decrease in infections and hospitalizations.^9 –13  During the analyzed period from 2020 to 2023, we observed an apparent alternating trend in dominance between SARS-CoV-2 and RSV in infants, whereas for adolescents, SARS-CoV-2 was the leading cause of hospitalization. Some studies have reported an RSV surge when physical distancing measures were eased, and schools reopened, resulting in a shift in the peak season and increased infections in older children.^14 –17  Interestingly, in our cohort, a positive test for RSV was associated with an overall comparatively lower risk of death in our cohort. However, we must emphasize that RSV was related to 327 deaths in Brazil in the period analyzed, and RSV-associated morbidity and mortality is a substantial health care burden worldwide.^18 –21 

Additionally, staggered COVID-19 vaccination campaigns may have affected the circulation and transmission of the SARS-CoV-2.^5,22  Interestingly, we observed that hospitalization after the second year of the pandemic conferred a significant protection against death, which may be related to a range of factors, including a steep learning curve regarding COVID-19 and the vaccine programs. In this regard, our previous studies have shown that, despite limited protection against SARS-CoV-2 infection, the vaccine schedules conferred significant protection against hospitalization and death.^23 –25 

Our analysis showed that, among all viral etiologies, SARS-CoV-2 had the highest hazard of death among children and adolescents hospitalized with SARI. Compared with subjects with a negative test at admission, positive cases for SARS-CoV-2 had a crude HR of 3.3 (95% CI 3.1–3.5). Importantly, we also found in the competing-risk survival analysis that not only SARS-CoV-2 infection but also adenovirus, influenza, and cases with coinfection had a significant hazard of death. Notably, in the United States, an epidemiologic analysis of all infectious and respiratory diseases from 2019 to 2022 revealed that COVID-19 ranked the first cause of death in children and adolescents, followed by influenza and pneumonia.²⁶  In a French nationwide, population-based study, Piroth and colleagues³  showed an overall in-hospital mortality higher in patients with COVID-19 than in those with influenza (16.9% vs 5.8%). Of note, in adolescents, the in-hospital mortality was 10 times higher for COVID-19 than for influenza.³  In adults, Ludwig et al²⁷  found an overall in-hospital mortality more than twofold higher in COVID-19 versus influenza (14% vs 6%) in Germany. Using a similar methodology to our study, Portmann and colleagues⁴  reported that the adjusted HR for in-hospital death for the SARS-CoV-2 omicron variant versus influenza was 1.54 (95% CI 1.18–2.01) in a cohort of hospitalized patients in Switzerland. Interestingly, regarding coinfection, Agathis et al²⁸  reported that respiratory virus codetections increased disease severity among children <5 years of age hospitalized with SARS-CoV-2 infection.

In the second step of our analysis, we evaluated risk factors of in-hospital mortality among pediatric patients with SARI according to the viral etiology. As expected, the factors related to death varied among the virus strains. However, of particular interest, in the competing-risk multivariable analysis, the following 3 covariates were significantly associated with death in all models: Presence of oxygen saturation <95% at admission, hospitalization in the poorest regions of the country (Northeast and North), and presence of comorbidities. Our analysis confirmed previous findings regarding factors associated with in-hospital mortality in children with SARI. The presence of any chronic medical condition was identified as a risk factor, consistent with several previous COVID-19 pediatric studies.^29 –38  Comorbidities have been shown to increase the severity of clinical outcomes and the risk of ICU admission, extended hospitalization, and mortality in pediatric influenza cases, as well.³⁹  The importance of hypoxemia as a predictor of mortality in children with acute lower respiratory infections has been demonstrated in a meta-analysis conducted before the COVID-19 pandemic.⁴⁰  Furthermore, our data revealed that children and adolescents hospitalized in less-developed regions of Brazil had a higher hazard of death compared with those in the wealthiest region. Health disparities, including factors like race, poverty, and access to health care, have been strongly associated with COVID-19 outcomes in several studies.^41,42  Brazil is a middle-income country with significant disparities in the social determinants of health, including differences in access to and quality of public health services, as a low coverage of ICU in North and Northeast regions.^43,44 

Some aspects of our results deserve further discussion. As previously mentioned, our findings highlight that some risk factors for death are shared among all strains of the virus. However, it should be noted that age was associated with the primary outcome in almost all models, except the adenovirus model. We previously reported that age as a risk factor for COVID-19–related death was “U” shaped, with infants and adolescents at higher risk for a worse outcome.^30,33,45  In the present analysis, we found that adolescents were at higher risk of death than the reference group (aged 2–11 years) in all virus models, except those who tested positive for adenovirus. Unexpectedly, infants (aged <2 years) were at an increased risk of death only in cases positive for SARS-CoV-2, but not for seasonal viruses. Similarly, indigenous ethnicity was also strongly associated with death in all models, except the flu model. Our previous studies also showed that indigenous ethnicity was a risk factor for COVID-19–related death. Therefore, we cannot rule out the possibility that the relatively small sample size of seasonal viruses may have hampered the analysis, increasing the likelihood of false-negative results in some of these models. However, despite these issues possibly related to study power, we must emphasize that our results highlight more susceptible subgroups of pediatric patients, whose characteristics may need to be considered in future epidemics, regardless of viral etiology.

The strength of this study is its large cohort size, which allowed comparisons of clinical characteristics, outcomes, and risk factors for death in children and adolescents with SARI caused by SARS-CoV-2 and other respiratory viral infections. However, our study has several limitations. First, a major limitation is the lack of essential information on viral etiology tests for many patients, most labeled in the database as “awaiting test results.” To address this issue, a sensitivity analysis was performed for patients with an unknown viral etiology. In this analysis, there was no discernible pattern in the clinical outcomes in cases with unknown viral etiology, except for cases labeled as inconclusive, with a mortality rate similar to the cases with positive SARS-CoV-2 test results. Secondly, another relevant issue concerns the heterogeneity of testing for other viruses in Brazil during the pandemic period. The testing strategies varied widely between Brazilian regions, depending on test availability. During periods of high demand, the viral testing panel provided by the Ministry of Health focused on the most prevalent viruses; namely, SARS-CoV-2, influenza, and RSV. Therefore, the true prevalence of other viruses such as adenovirus may have been underestimated in our cohort. Although all models were controlled for region and year of admission, we cannot rule out the possibility that this may have introduced bias into the analysis. For example, some patients with severe clinical conditions or immunosuppressive diseases may have been investigated in more detail for the presence of adenovirus, and this characteristic may have influenced survival analysis. Third, information on vaccination status was not included in the analysis because of the heterogeneity of vaccination campaigns among different age groups. This decision was made because many pediatric cases were not vaccinated, because vaccine schedules were not available for children and adolescents in 2020 and most of 2021. Finally, we must consider that, despite the size of the cohort included in the analysis, the study setting may limit the generalizability to other populations. Consistent with our previous analyses of successive COVID-19 pandemic waves in Brazil,^30,33,45  the mortality rate (6.1%) from SARS-CoV-2 infection in children was much higher than that in developed countries, which reported an overall mortality rate of <1% in hospitalized patients.⁴⁶  We found that these findings are mainly related to the lack of access to health care in regions of the country with poor socioeconomic conditions, especially in vulnerable ethnic groups such as indigenous individuals. In this sense, we believe that these peculiar aspects of a middle-income country like Brazil may hinder the generalization of our findings to a certain extent. Therefore, further comparative studies on the outcomes of SARS-CoV-2 and seasonal viruses in diverse scenarios are required.

In this analysis based on a national multicenter registry that obtained information on various clinical aspects of children and adolescents hospitalized with SARI of laboratory-confirmed viral etiology, we found that SARS-CoV-2 infection had the highest mortality rate among virus strains. Our analysis showed heterogeneous viral circulation among different age groups during the 3 years of the pandemic in Brazil. Risk factors for death varied between different strains of the virus. However, an increased risk of death was uniformly associated with comorbidities, hospitalization in the poorest regions of Brazil, and low oxygen saturation at admission. Presumably, SARS-CoV-2 will eventually become an endemic seasonal pathogen, but little is known about how SARS-CoV-2 will behave in this new environment in terms of equilibrium dynamics with seasonal viruses. In this context, our findings highlight the importance of monitoring and understanding viral epidemiology in the post–COVID-19 era to establish timely preventive measures to lessen the severity of upcoming seasonal outbreaks of respiratory viruses.

We thank all frontline health care workers for their impressive efforts to tackle the COVID-19 pandemic in Brazil. All data from the SIVEP-Gripe were systematically collected by these frontline health care workers.

CI

confidence interval

CIF

cumulative incidence function

COVID-19

coronavirus disease 2019

HR

hazard ratio

RSV

respiratory syncytial virus

SARI

severe acute respiratory infection

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2