Factors Associated With Severe SARS-CoV-2 Infection

We conducted a national prospective surveillance of children hospitalized with SARS-CoV-2 infection. The study was approved by the Institut National de la Santé et de la Recherche Médicale ethics committee for evaluation, the institutional review board (IRB00003888), and was registered at ClinicalTrials.gov (NCT04336956).

We recruited 60 hospitals throughout France for this study, using the French pediatric bacterial meningitis network.^19,20  We included all pediatric patients with SARS-CoV-2 infection who were hospitalized in one of these centers from February 15 to June 1, 2020. Confirmed SARS-CoV-2 infection was defined by a positive real-time reverse transcriptase polymerase chain reaction (RT-PCR) result for SARS-CoV-2 on a nasopharyngeal swab and computed tomography (CT) scan–based cases were defined by characteristic chest CT scan lesions according to European Centre for Disease Prevention and Control and French National Health Agency case definition.²¹  Radiologic evidence of lesions compatible with COVID-19 were unilateral or bilateral subpleural ground-glass opacities and consolidations, although other etiologies can cause ground-glass opacities on CT, including vaping and interstitial lung infections.^22,23  A CT scan was used to define the site of disease and/or in a few cases the presence of the infection. During the same period, Public Health France, the national French public health agency, counted all hospitalized pediatric SARS-CoV-2 infections cases in France, which allowed for estimating the completeness of our surveillance system. Similar to our surveillance system, the definition of cases counted by Public Health France included a positive RT-PCR result for SARS-CoV-2 or a characteristic chest CT scan lesions.²⁴ 

For each patient, an electronic case report form was prospectively completed on a secure database, and the following data were recorded: demographic characteristics, comorbidities, initial symptoms and clinical signs, biological and microbiologic parameters, radiography findings, treatments, and course during hospitalization.

The main outcome was the proportion of severe disease evolution among all included children with a SARS-CoV-2 infection, defined by the need for either ventilatory or hemodynamic support during hospitalization, or death.¹⁷  Ventilatory support was defined by use of noninvasive ventilation, including high-flow oxygen via nasal cannula, continuous positive airway pressure, and bilevel positive airway pressure or the use of invasive ventilation. We assessed demographic, clinical, and biological factors associated with severe disease progression in children at the acute phase of the SARS-CoV-2 infection. For biological factors, we considered explorations performed at admission. Hypoxemia was defined by oxygen saturation <95%.²⁵  Thrombocytopenia was defined by platelet count <150 g/L.

Secondary outcomes were rate of severe disease by clinical presentation, comorbidities, and age groups (<90 days, 90 days to <1 year, 1 to <5, 5 to <10, 10 to <15, and 15 to <18 years). Among comorbidities, obesity was defined by a BMI >25 and overweight by a weight >+2 SDs for age and sex on the basis of national weight curves and/or a BMI >25.²⁶ 

We described patient characteristics with numbers (percentages) for categorical variables and median (interquartile range [IQR]) for quantitative variables. We assessed the association between these characteristics and severe disease evolution by Fisher’s exact test for categorical variables and the Mann–Whitney U test for quantitative variables. We then used a multivariate logistic regression model with the backward stepwise method to identify factors independently associated with disease severity, estimating odds ratios (ORs) and 95% confidence intervals (CIs). Variables significant at P ≤.20 on univariate analyses were included in the stepwise selection. Quantitative variables associated with severity were transformed into binary variables by using the receiver operating characteristic curve analysis. A 2-sided P value <.05 was considered statistically significant. All statistical analyses involved using R v3.6.1 (http://www.R-project.org).

Since February 15, 2020, 397 hospitalized children were included in the study: 385 had a positive RT-PCR nasopharyngeal test result for SARS-CoV-2, and 12 were CT scan based with radiologic evidence of lesions compatible with COVID-19; see flowchart (Supplemental Fig 3). On the basis of systematic mandatory reporting of any children hospitalized with SARS-CoV-2 infection via Public Health France (1032 cases over the same period), the completeness of our surveillance system was 38.5% (95% CI: 35.5–41.5).

The weekly number of children admitted to hospital with SARS-CoV-2 infection peaked during March 23 to 30, 2020 (ie, 1 week after the national lockdown), and strongly decreased thereafter (Fig 1). This number mirrored the evolution of adults admitted with SARS-CoV-2 infection over the same period in France (http://www.santepubliquefrance.fr). The geographic distribution of cases is displayed in Supplemental Fig 4.

The clinical characteristics of children are in Table 1. The median age was 16 months (IQR, 51 days–134 months). Overall, 114 (29%) children had comorbidities, mainly chronic respiratory disease (such as asthma, cystic fibrosis, and bronchopulmonary dysplasia), immunosuppression or malignancy, neurologic disorders, and sickle cell disease (Table 1). The main symptoms were fever (n = 300 of 385, 78%), cough (n = 168 of 391, 43%), feeding difficulties (n = 152 of 385, 39%), shortness of breath or dyspnea (n = 119 of 390, 31%), and diarrhea (n = 93 of 392, 24%).

Among the 397 children included, levels of inflammatory parameters were low for most children (median C-reactive protein level [CRP], 5 mg/L [IQR, 2–36], procalcitonin level, 0.15 ng/mL [IQR, 0.1–0.37], neutrophil count, 3.8 g/L [IQR, 2.1–7.1]). A total of 80 of 301 (27%) cases showed lymphopenia and 28 of 313 (9%) showed thrombocytopenia.

Chest radiography findings were abnormal in 70 of 203 (34%) children, with bilateral interstitial infiltrate as the most-frequent lesion (Table 1).

A posteriori and on the basis of the final diagnosis recorded by the physicians and hospital treatments, we distinguished 4 different clinical patterns:

• Pattern 1: paucisymptomatic SARS-CoV-2 infection with no hospital treatment required, admitted for clinical surveillance (n = 148, 37%). Three of these paucisymptomatic children were admitted in February in a PICU to be isolated according to local guidelines, without any other criteria of intensive care requirement.

• Pattern 2: SARS-CoV-2 infection requiring hospital care (n = 158 cases, 40%), with 3 main forms of COVID-19 described: (1) lower respiratory tract infection (LRTI) (n = 95 of 158, 60%), (2) digestive form (n = 24 of 158, 15%), and (3) isolated fever (n = 28 of 158, 18%). We classified children according to the physicians’ clinical assessment describing the severity of each organ involvement (oxygenotherapy requirement, intravenous rehydration, etc). Physicians also defined a primary diagnosis, which allowed for classification in the respiratory or digestive form. Other forms (n = 11 of 158, 7%) are detailed in Supplemental Table 5.

• Pattern 3: multisystem inflammatory syndrome in children (MIS-C) was defined by persistent fever, presence of laboratory markers of inflammation, manifestation of signs or symptoms of organ dysfunction, lacking an alternative diagnosis,²⁷  and, as for the other patients included in this cohort, a positive RT-PCR nasopharyngeal result for SARS-CoV-2 (MIS-C, n = 29, 7%).

• Pattern 4: hospitalization for another disease and COVID-19 was not the final diagnosis (n = 62 cases, 16%). Some of these were admitted children tested for COVID-19 as part of hospital-wide surveillance. The most frequent diagnoses were surgical pathology and urinary tract infection. Each diagnosis is detailed in Supplemental Table 6.

Both the clinical pattern of SARS-CoV-2 infection and the rate of severity were highly different depending on age group (Table 2, Fig 2). Children <90 days of age were the largest group, with 145 patients (37%) of our cohort. Their clinical pattern was specific because most them (59%, 85 of 145) were paucisymptomatic, no MIS-C were observed, and LRTI were diagnosed in 18 cases (12%). In this age group, fever was almost a constant finding (92%). Overall, 4 (3%) children <90 days had a severe disease. Among children <90 days old, 63 of 145 (43%) were <30 days old, among which severe disease was observed in 3 of 63 (5%) cases, including 2 LRTI and 1 other diagnosis: prematurity and Escherichia coli meningitis. In comparison, children between 3 and 12 months old were paucisymptomatic in only 33% (16 of 48) cases, as many as were diagnosed with LRTI (16 of 48). However, patients with severe diseases remained rare, with 3 of 48 cases (6%), and only 1 patient was diagnosed as MIS-C. LRTI was a major diagnosis in all age groups beyond 1 year with 20% to 36% of cases according to age group.

The number of patients with MIS-C increased in older children age between 1 and 5 years old, 5 to 10 years old, and 10 to 15 years old, with 7 of 46 cases (15%), 9 of 49 cases (18%), and 10 of 65 cases (15%), respectively, but decreased for children age between 15 and 18 years old (only 2 of 44 [5%] MIS-C cases). Antibiotics were the most-used treatment, with a percentage varying from 31% to 65% according to age group. Among children <90 days of age, 40% received antibiotics, despite most of them being paucisymptomatic.

The characteristics described by clinical patterns are provided in Supplemental Table 7, which clearly show a specific pattern for the 29 children diagnosed as MIS-C with high and persistent fever, higher inflammatory syndrome with a median CRP at 179 (105–270), high proportion of severe disease (52%), and a long delay between COVID-19 contact and hospitalization (34 days, IQR: 23–47).

Factors associated with disease severity were assessed for pattern 1 and 2 children because pattern 3 (MIS-C) was not considered an acute SARS-CoV-2 infection and showed a specific profile, and outcomes from children in pattern 4 were not mainly related to SARS-CoV-2 infection. Overall, severe disease was detected in 23 of 306 children (11%, 95% CI: 8%–15%), including 20 patients requiring ventilatory support and 5 with hemodynamic support (Table 3). Numerous characteristics were associated with disease severity on univariate analysis (Table 3), but only 3 were independently associated on multivariate analysis: age ≥10 years (OR: 3.4, 95% CI: 1.1–10.3, P = .034), hypoxemia (OR: 8.9, 95% CI: 2.6–29.7, P = .0004), and CRP ≥80 mg/L (OR: 6.6, 95% CI: 1.4–27.5, P = .012).

We recorded 6 deaths in our surveillance system, plus 1 patient positive for immunoglobulin G anti–SARS-CoV-2 antibody who was enrolled but not included in the main analysis because the case did not meet our inclusion criteria (negative on RT-PCR nasopharyngeal test) (Supplemental Fig 3). These 7 patients were the only deaths reported in children to Public Health France in the context of SARS-CoV-2 infection in France, and some were previously reported.²⁸  Among these 7 children, 3 had a comorbidity (immunodeficiency, leukemia, and encephalopathy). Most had a respiratory failure, with high levels of inflammatory parameters, and the imputability of SARS-CoV-2 infection in the death was high in 5 cases (Table 4).

This national prospective study has allowed for analyzing the clinical spectrum of hospitalized SARS-CoV-2 infection in children, including acute infections and MIS-C, and assessing factors associated with severe clinical evolution. A total of 60 hospitals throughout France participated, which allowed us to obtain a completeness rate of 38.5%. The evolution of pediatric cases over time mirrored that of adult admission reported by Public Health France both temporally and geographically, which suggests that our network adequately captured the epidemiological pattern of hospitalized children with SARS-CoV-2 infections. The strong decrease in cases 2 weeks after the national lockdown highlighted the major impact of these unprecedented measures in reducing the burden of SARS-CoV-2 infection in children.

An important concern in the literature was the specific burden of SARS-CoV-2 infection in young children.^6,29,30  Authors of a report from China⁶  described a higher rate of severe forms among children aged <1 year versus older children, and authors of another report from Italy suggest that infants aged <6 months seem significantly more susceptible to severe forms of the disease, leading to important concerns about this specific age group^6,29,30  and playing a role in recommending closure of most day care centers during lockdown in France. Our results shed particular light on this issue. On the one hand, children <90 days old were the leading group because they accounted for 37% of all children hospitalized, but on the other, 97% (141 of 145) of them had no severe disease and 68% were discharged from the hospital with little care. Fever was the most common symptom (92%, 128 of 139) in this age group, and 40% (58 of 145) received antibiotics. In another words, most of these children were hospitalized owing to risk of bacterial infection because they had fever without a clear source and were <90 days old.³¹  The use of a rapid test at the bedside to identify SARS-CoV-2 would be helpful to improve the care of these children in pediatric emergency departments.

In a recent study, Götzinger et al³²  also suggested that the likelihood of severe forms may be increased in children <1 month old. However, our findings suggest that the rate of severe forms was lowest in very young children and highest for children ≥10 years on both univariate and multivariate analysis. These findings agree with Zachariah et al,¹⁷  who describe the disease course with COVID-19 in 14 children <1 year old, with no case of severe evolution.

The definition of severity should be considered in these comparisons. Although we defined severity as need for ventilatory or hemodynamic support, in line with Zachariah et al,¹⁷  Dong et al,⁶  who described the Chinese cohort, defined severe cases as respiratory symptoms, dyspnea, and hypoxemia,⁶  and Götzinger et al³²  defined severe cases as admission to the PICU. In our cohort, several children <1 year old required transient nasal oxygenotherapy, and several were transferred to the PICU only for surveillance with regards to the very young age, depending on local practices. They evolved favorably in the large majority of cases, without ventilatory support, and have not been considered as severe cases. This information could be useful to guide reopening day care centers and recommend suitable mitigation measures, which are highly difficult to apply in these settings. According to the French Pediatric Society, the educational and social benefits provided by school far outweigh the risks of a possible COVID-19 contamination of children in school environments or in day care centers.³³ 

We also identified several factors associated with severe disease evolution, including hypoxemia and CRP ≥80 mg/L. Few studies assessed factors associated with poor outcomes in pediatric COVID-19. The 2 main studies are a recent US monocentric cohort¹⁷  and a multinational study.³²  Elevated levels of inflammatory parameters and respiratory symptoms were also found as a prognostic factor.³⁴ 

This surveillance system allowed us to identify different clinical profiles of hospitalized children with SARS-CoV-2 infections, ranging from paucisymptomatic forms to MIS-C. The main clinical symptoms reported in our cohort agreed with the literature (fever in 70%–80%, cough in 40%–50%, dyspnea in 30%–40%).^7,17  Among this wide clinical spectrum, the proportion of severe forms was highly variable and greatest for LRTI (21%) and MIS-C (52%). This novel clinical entity has since been widely described in Italy,¹¹  France,^12,13  England,¹⁵  and the United Sates,^14,16  reporting a high rate of PICU transfer (60%–75%) and hemodynamic support requirement (40%–50%).^11,12,15  Our findings confirm that MIS-C is among the most severe forms linked to SARS-CoV-2 infection in children. Our results are in line with previous studies suggesting that MIS-C happens not in the acute phase of the SARS-CoV-2 infection but as a postinfectious phenomenon despite our patients expressing a positive RT-PCR result.

Another issue of concern at the time of school reopening is children with comorbidities, including immunosuppression. Several comorbidities have been associated with poor outcomes in adults,³⁵  including chronic respiratory and cardiac diseases, but these remain unclear for children.^{7,18,28,36,37}  Comorbidities were frequent in our cohort (29%, n = 114), but these comorbidities were not independently associated with severe evolution and accounted for only 22 severe cases of hospitalization on a national scale. These observations may also help policy makers allow for standard school activities for these populations and help physicians to reassure families in this process.

The main strength of our study was the use of a national surveillance system that captured a large part of hospitalized cases of SARS-CoV-2 infections with precise clinico-biological data in France, allowing us to define the rate of severe disease evolution by clinical pattern and factors associated with severe evolution. Our study also has several limitations. First, we decided to include only children with a positive RT-PCR nasopharyngeal result for SARS-CoV-2 or a CT scan revealing radiographic evidence of lesions compatible with COVID-19 and excluded children with positive serology alone. These criteria allowed us to enhance the imputability of SARS-CoV-2 infection in the clinical features observed but may have led to missing some patients with postinfectious disorders such as MIS-C, for which the RT-PCR nasopharyngeal test result may be negative.^11,13  Further studies including MIS-C with negative RT-PCR nasopharyngeal results for SARS-CoV-2 but positive serology are ongoing to appreciate this burden. Second, the clinical spectrum of included patients may depend on the SARS-CoV-2 testing strategy. At the early stage of the epidemic, only patients with respiratory symptoms were tested, and indications became broader when asymptomatic, digestive, and inflammatory forms had been described, which led some hospitals to adopt universal testing at admission.¹⁷  Thus, the proportion of the different forms was likely be influenced by evolving local SARS-CoV-2 testing strategies, and the children in pattern 4 (positive for SARS-CoV-2 but hospitalized for another disease) may be overestimated. However, most of the PICUs throughout France followed the same testing policy, with wide indications for any child requiring PICU admission.

This national surveillance system allowed for describing the broad clinical spectrum of children hospitalized with acute SARS-CoV-2 infections and defining the risk of severe disease for each clinical form. Infants <3 months old were the main age group requiring hospitalization, but most of them expressed a mild form of COVID-19. The independent factors associated with increased risk of severe disease may help clinicians manage this infection in children.

In addition to the authors, the following investigators participated in the PANDOR study: Anxionnat Raphaël, Auburtin Brigitte, Azemar Benjamin, Barrey Catherine, Belgaid Abdelmalek, Benkaddouss Sakina, Benoist Grégoire, Bensaid Philippe, Biran Valérie, Bordet Jeanne, Blanc Philippe, Bost-Bru Cécile, Boussicault Gérald, Brehin Camille, Brouard Jacques, Cantais Aymeric, Casha Paul, Caurier Bernard, Charbonnier Lorelei, Charreton Claire, Chenel Claude, Chognot Didier, Cohen Laure, Dabbagh Dana, De Feraudy Yvan, Delestrain Céline, Dessioux Emmanuelle, Do-Cao Jéremy, Dommergues Marie-Aliette, Drouard Anne, Dubos François, Dupré Charlène, Durrmeyer Xavier, Elmerich Florence, Enchery Sophie, Fruscione Sophie, Girard Florent, Gouraud François, Gruss Richard, Guerder Margaux, Guglielmino Patricia, Hentgen Véronique, Hoeusler Valentine, Huet Fréderic, Idier Charlotte, Jany Bérangère, Javouhey Etienne, Kalach Nicolas, Kom Rémi, Kutsia Ekaterina, Larakeb Anis, Latry Vanessa, Le Stradic Camille, Leboucher Bertrand, Lemaire-Weber Aurélie, Lepage Jean-Marie, Leteurtre Stéphane, Loeile Camille, Louf Sylvie, Maakaroun-Vermesse Zoha, Mandelcwajg Alexis, Marin Mariana, Marques Cécilia, Martha Sandrine-Anne, Martinat Laurence, Masserot Caroline, Mathivon Laurence, Meslin Pauline, Mestre Nathalie, Milcent Karen, Mizzi-Rozier Marie, Monceaux Françoise, Monfort Marie, Mougenot Emmanuelle, Moulène Eric, Florence Moulin, Nattes Elodie, Odorico Sophie, Orcel Marie-Clothilde, Oualha Medhi, Ouziel Antoine, Peigne Chantal, Pellegrino Béatrice, Petit Isabelle, Pinquier Didier, Pons Charlotte, Rabier Valérie, Robert Blandine, Roche Christine, Roullaud Sylvie, Roybet Danielle, Rozental Jonathan, Sanchez Stéphane, Schneider Jean-Marc, Soto Bertrand, Szulc Marie, Thach Caroline, Thevenet Léa, Tran Antoine, Ursulescu Nicoleta, Valentin Cécile, Verney Cindy, Vignaud Olivier, Walser Sébastien, Yangui Mohamed-Amine, Zimmermann Brigitte.

We thank our clinical research assistants: Bommenel Tchadie, Bouquignaud Christelle, Bouvy Emilie, Fleury Marjory, Kara Sofia, Khalil Mirna, Madpathage Senanayake Videsheka, Vassal Mélanie, Vasseur Valérie. We are grateful to Isabelle Ramay, Claire Prieur, Marine Borg, Aurore Prieur, Laura Meyet and Jéremy Levy from ACTIV, Association Clinique et Thérapeutique Infantile du Val-de-Marne, Créteil, France. We are grateful to Cecile Hoffart, Maxime Brussieux from Clinical Research Center, Centre Hospitalier Intercommunal de Créteil, Créteil, France. We are grateful to Denise Antona, Public Health France. We are grateful to Pauline Scherdel and Pr Martin Chalumeau, Inserm UMR 1153. We are grateful to Pr David Skurnik, INSERM U1151-Equipe 11. We are grateful to every and each microbiologic laboratory staff who performed SARS-COV2 RT-PCR in each center.

CI

confidence interval

CRP

C-reactive protein level

CT

computed tomography

IQR

interquartile range

LRTI

lower respiratory tract infection

MIS-C

multisystem inflammatory syndrome in children

OR

odds ratio

RT-PCR

real-time reverse transcriptase polymerase chain reaction