Characteristics of hospitalized children with Omicron BA.5 infection

The final data included 315 Omicron BA.5-infected children and 16,744 febrile children without Omicron infection after exclusion (Fig. 2). The median age was 2.08 years (0.08–14.00 years) in Omicron-infected children and 2.52 years (0.01–18.00 years) in non-Omicron-infected febrile children. There were 122 (36.6%) and 7454 (42.6%) females in the two groups, respectively. The epidemiological characteristics, clinical features, laboratory test results, and radiological findings of the 315 hospitalized children with Omicron infections are summarized in Table 1. Notably, 260 (82.5%) of these hospitalized children (n = 315) were not vaccinated. The epidemiological characteristics and clinical features of 16,744 febrile children with non-Omicron infection were presented in the Supplementary Information (Supplementary Table 1).

Table 1 Epidemiological characteristics, clinical features, laboratory tests, and radiologic findings of children hospitalized with Omicron BA.5 infectionComparison of symptoms in Omicron-infected children and the control group

Three hundred and seven (97.5%) of the 315 children with Omicron infection convulsed during hospitalization, which is 22.7 times more than that (4.3% or 721/16,744) in febrile children without Omicron attending the hospital’s fever clinic during the same period (Fig. 3a). The number of convulsions was significantly higher in Omicron-infected children with convulsions than in non-Omicron-infected children with convulsions (median: 2.0 vs. 1.6, \(P\) < 0.001). Figure 3b illustrates that the body temperatures of Omicron-infected children were significantly higher than those of non-Omicron-infected febrile children (median body temperature: 39.5 vs. 38.6 °C, \(P\) < 0.001). Furthermore, we stratified the temperature data during convulsions for both cohorts and then compared the percentage of patients within groups. Specifically, 3.3%, 35.5%, and 61.2% of the Omicron-infected children had temperatures of ≤ 38.0, 38.1–39.0, and > 39.0 °C during convulsions (\(P\) < 0.001), while for the three groups of non-Omicron-infected febrile children, the percentages were 30.9%, 34.4%, and 34.7% (\(P\) = 0.244), respectively (Fig. 3c). These findings suggest that higher temperatures in Omicron-infected children may partly be responsible for their higher rates of convulsion compared to non-Omicron-infected febrile children during convulsions.

Fig. 3

Comparative analysis between groups of the Omicron-infected children and non-Omicron-infected febrile children. a Comparison of convulsion rates; b Comparison of body temperature during convulsions; c Distribution of body temperatures during convulsions in Omicron-infected children and control children. °C degrees Celsius

Pathogen testing was additionally done after SARS-CoV-2 testing of pharyngeal and nasal swabs from the Omicron-infected and control groups. In the group of hospitalized children infected with Omicron, nine positive RT-PCR tests were detected in 117 cases, and four positive IgM antibody tests were found in 77 cases. In the febrile group of non-Omicron-infected children, 6832 RT-PCR tests were positive in 3431 cases, and 673 IgM antibody tests found 289 positive cases. The pathogens with high positivity rates were, in descending order, MP, RSV, ADV, PIV, HBoV, and RhV.

Within-group analyses of Omicron-infected children

The Mann–Whitney U test revealed that the body temperatures of 307 Omicron-infected children were significantly higher during convulsions than when they were not convulsing (median body temperature: 39.5 vs. 38.2 °C, \(P\) < 0.001) (Fig. 4a). Among the 315 Omicron-infected children, the Mann–Whitney U test showed that 55 vaccinated children had a significantly lower frequency of convulsions than 260 unvaccinated children (mean frequency of convulsions: 1.8 vs. 2.1, \(P\) < 0.001) (Fig. 4b). The vaccination rate was higher among the eight Omicron-infected children without convulsions than in the 307 Omicron-infected children with convulsions (Fisher's exact test; 50.0% vs. 16.6%; \(P\) = 0.034) (Fig. 4c).

Fig. 4

Within-group comparative analysis of children hospitalized with Omicron infection. a Comparison of body temperature in Omicron-infected children with and without convulsions; b Comparison of the frequency of convulsions in vaccinated and non-vaccinated children infected with Omicron; c Comparison of the rate of vaccination in convulsed and non-convulsed Omicron-infected children. °C degrees Celsius

A multifactorial correlation analysis of 307 Omicron-infected children with convulsions revealed that the convulsion frequency was significantly correlated with vaccination doses using Spearman's rank correlation coefficients (\(P\) < 0.001) (Table 2). Age was negatively but not significantly correlated with the frequency of convulsions (correlation coefficient =  − 0.024, \(P\) = 0.681); older children had relatively fewer frequencies of convulsions (Table 2).

Table 2 Correlation analysis of convulsions in children with Omicron infectionExamination of brain damage

Brain MRI was performed in 112 Omicron-infected children with suspected CNS involvement with peak respiratory frequency > 30 breaths/minute, SpO2 < 94% at rest, and > 2 convulsions among 307 children infected with Omicron and presenting with convulsions. Fifteen of them showed brain damage, with major locations in the thalamus, corpus callosum, frontal lobe, parietal lobe, occipital lobe, and basal ganglia. Figure 5 depicts a patient with thalamic damage. Out of these 15 patients, 14 agreed to undergo biochemical analysis, and 12 exhibited signs of intracranial inflammation. In addition, we reviewed the clinical records and found no vaccination records among these 15 children with brain damage. Among 721 children with fever and convulsions who were non-Omicron-infected, 49 patients with complicated convulsions and suspected brain damage underwent brain examination by MRI or computed tomography (CT), and 43 of them also underwent CSF analysis, which found imaging brain damage in 11 cases and intracranial inflammation in three cases, respectively.

Fig. 5

MRI images of a child with Omicron BA.5 infection, showing damage to the thalamus. a T1WI showed a low signal; b T2WI showed a high signal; c FLAIR water suppression showed a significantly high signal; d DWI showed a significantly high signal; e ADC showed a slightly high signal at the lesion's periphery and a low signal at the center; f DWI showed a slightly increased signal in the splenium of the superior corpus callosum. MRI magnetic resonance imaging, FLAIR fluid attenuated inversion recovery, ADC apparent diffusion coefficient, DWI diffusion weight imaging

WGS of 848 cases sampled in the city by the Guangzhou Center for Disease Control and Prevention showed that all were infected with Omicron BA.5. We performed CSF analysis in the pediatric patients mentioned above to diagnose intracranial inflammation for rapid symptomatic treatment. Due to limited medical resources at the time, no further viral testing of the CSF was performed.

Outcomes of patients

For the 315 children with Omicron infection, palliative treatments were administered immediately upon admission—two critically ill patients among them were admitted to the intensive care unit (ICU). Initial treatments mainly included ibuprofen or acetaminophen for fever, oseltamivir for viral infections, and cefazolin sodium for bacterial infections. Nebulizing cough and sputum-clearing treatments were administered. During convulsions, intravenous midazolam at a dose of 0.2–0.3 mg/kg was administered to sedate children who convulsed for over 5 min. Furthermore, children with high-grade fever were wiped with wet towels to reduce the fever. Patients with severe disease were also administered 400 mg/kg of gamma globulin to nourish the nervous system and boost immunity for 3–5 days.

By December 30, 2022, 255 (81.0%) of the 315 patients had been discharged because they were significantly better, had stable vital signs, and had normal temperatures for more than 24 h. Out of the 12 children diagnosed with intracranial inflammation, 10 were discharged as the lesions had disappeared or shrunk on MRI reexamination. Additionally, three of these children developed motor dysfunction that recovered after approximately 2 weeks of rehabilitation. One of the two children who remained in the hospital had improved imaging features and had been discharged from the ICU, but was still being treated for motor deconditioning. The other child remained in the ICU. On December 28, 2023, we reviewed the follow-up records of these two patients with intracranial inflammation. One of them had improved, but not fully recovered, motor skills accompanied by right lower extremity claudication. Sadly, the other patient died on January 4, 2023, in the ICU.