Design, setting, and population

This retrospective observational study included all patients 0–17 years of age who tested positive for SARS-CoV-2 diagnosis at Bambino Gesù Children Hospital from January to December 2023, regardless of symptoms. During the pandemic, all patients presenting to the emergency department or requiring hospital admission were tested for SARS-CoV-2, irrespective of whether they exhibited symptoms, as part of the hospital’s infection control measures.

To be included, patients needed to have a laboratory confirmed SARS-CoV-2 infection, complete demographic and clinical data available, as well as documented clinical outcomes. Moreover, only those infected with an Omicron variant, confirmed through genetic characterization, were considered. Patients older than 17 years, those lacking reliable demographic or clinical information, or those without retrievable biological samples were excluded from the study. Nasopharyngeal swabs were collected at the time of admission for SARS-CoV-2 testing and sequencing from a proportion of patients. SARS-CoV-2 infection was detected by antigenic or molecular tests. Specifically, the antigenic tests employed were Roche Elecsys SARS-CoV-2 Antigen (Roche Diagnostics GmbH, Mannheim, Germany) and RADT STANDARD F COVID-19 Ag FIA (SD BIOSENSOR, Korea). The molecular tests used were Cepheid Xpert Xpress CoV-2 Plus (Cepheid, Sunnyvale, CA, USA) and SARS-CoV-2 ELITe MGB Kit^® (Elitechgroup, Turin, Italy). Nasopharyngeal swabs with a cycle threshold (Ct)  100 in antigen tests were considered suitable for sequencing. Among 1,510 SARS-CoV-2–positive samples, 576 (38%) met the sequencing threshold, while 934 (62%) did not.

Information on patient demographics and clinical findings were obtained retrospectively from pseudonymized electronic medical records. Comorbidities were categorized based on the systems and functions most directly affected by these conditions.

The severity of COVID-19 was defined on the basis of clinical features, laboratory tests and chest X-ray images²⁸. The following definitions were used: (i) asymptomatic infection, defined as testing SARS-CoV-2 positive but not developing any clinical symptoms; (ii) upper respiratory tract infection, such as rhinitis, pharyngitis, cough, sore throat, runny nose, sneezing, or symptoms of a gastrointestinal tract infection (vomiting, diarrhoea); and (iii) lower respiratory tract infection including clinical signs of bronchitis or pneumonia (with or without signs of gastrointestinal symptoms)²⁹.

Ethics committee statement

The study protocol was approved by the local Research Ethics Committee of Ospedale Pediatrico Bambino Gesù IRCCS (prot. 2384_OPBG_2021) and was conducted under the principles of the 1964 Declaration of Helsinki. Informed consent was waived by the Ethics Committee of Ospedale Pediatrico Bambino Gesù IRCCS following the hospital regulations on observational retrospective studies.

Virus amplification and sequencing

Viral RNAs were extracted from nasopharyngeal swabs by using QIAamp Viral RNA Mini Kit (Qiagen, Hilden, Germany), followed by purification with Agencourt RNAClean XP beads (Beckman Coulter Life Sciences, CA, USA). Both the concentration and the quality of all isolated RNA samples were measured and checked with the Nanodrop (Thermo Fisher Scientific, MA, USA). Amplicons of whole genome sequences of SARS-CoV-2 were generated with a 50 ng viral RNA template, by using CleanPlex SARS-CoV-2 Research and Surveillance Panel (Paragon Genomics, Hayward, CA, USA), QIAseq DIRECT SARS-CoV-2 Kit (Qiagen, Hilden, Germany) and Illumina COVIDSeq Assay (Illumina, San Diego, CA, USA) following manufacturers’ protocol. Libraries were then generated using the Nextera DNA Flex library preparation kit with Illumina index adaptors and sequenced on a MiSeq instrument (Illumina, San Diego, CA, USA) with 2 × 150-bp paired-end reads. Raw reads were trimmed for adapters and filtered for quality (Phred score > 28) using Fastp (v0.23.2)³⁰. Reference-based assembly was performed with BWA-mem (v0.7.17)³¹ aligning against the GenBank reference genome NC_045512.2 (Wuhan, collection date: December 2019)³².

SNP variants were called with freebayes (v1.3.2)³³ and all SNPs having a minimum supporting read frequency of 2% with a depth ≥ 10 were retained.

Phylogenetic analysis

Consensus sequences were generated using the GitHub freely distributed software vcf_consensus_builder³⁴ considering all SNPs having a minimum read frequency of 40% (high-abundant mutations). SARS-CoV-2 lineages of the obtained consensus sequences were assigned according to Pangolin application (Pangolin v4.1.1)³⁵ and then grouped into seven major sub-lineages (BA.2, JN.1, BA.5, XBB, XBB.1, XBB.1.16, XBB.1.5). All sub-lineages identified are reported in the Supplementary material.

Sequences were aligned using MAFFT v7.475 and manually inspected using Bioedit. The final alignment comprised 458 sequences of 29,164 nucleotides of length. In order to investigate the phylogeny of Omicron clade that affects the paediatric population (0–17 years), a maximum likelihood (ML) phylogeny tree was generated using IQTREE2 (v2.1.3)³⁶ with 1000 bootstrap replicates, using the best-fit model of nucleotide substitution GTR + F + R3 inferred by ModelFinder³⁷. Annotation of the phylogenetic tree, including information about lineages, symptoms, comorbidity and hospitalization was performed with iTOL (v5)³⁸.

Co-infections evaluation

Co-infections were investigated in patients diagnosed with lower respiratory tract infections on samples collected as part of routine clinical care. Viral agents were identified through molecular tests (Allplex Respiratory Panel assay, Seegene Inc., Korea and ARGENE^® HHV6 R-GENE^®, BioMerieux, France) while bacterial pathogens were detected via culture methods.

The presence of co-infections was defined exclusively by the detection of at least one positive result from any diagnostic test, without requiring clinical correlation or additional confirmatory testing.

Statistical analysis

The Likelihood Ratio Test, followed by a logistic regression model that estimated odd ratios (ORs) and corresponding 95% confidence intervals (CIs), was used to compare demographic and clinical findings between general and selected SARS-CoV-2 infected populations.

The Kolmogorov-Smirnov test was used to assess the distribution (normal or non-normal) of the continuous variables. Descriptive statistics were expressed as median (interquartile range, IQR) for continuous data and number (%) for categorical data. To assess for significant differences in patient characteristics and clinical findings, Fisher’s exact test and the Chi-Square test for trend, and the Mann-Whitney and Kruskal-Wallis tests were used for categorical and continuous variables, respectively.

Univariable and multivariable logistic regression models were employed to assess whether the risk of hospitalization was associated with specific SARS-CoV-2 sub-lineages, clinical manifestations, or demographic characteristics. Patients with a previous SARS-CoV-2 infection and those who had already been vaccinated (N = 6) were excluded from these analyses. The primary outcome was hospitalization, with the reference group being patients who were not hospitalized. The covariates included age, Omicron sub-lineage, clinical manifestations and the presence of comorbidity (versus patients without comorbidity). Model stability was evaluated by calculating the events-per-variable (EPV)³⁹, and multicollinearity was assessed using the Variance Inflation Factor (VIF). Model fit was examined with the Cox & Snell and Nagelkerke R² statistics.

For the clinical manifestations, patients were categorized into four groups: those with lower respiratory airway involvement, those with upper respiratory airway involvement, those with only gastrointestinal symptoms, and those who were asymptomatic.

For comorbidities, patients were classified as patients with at least one comorbidity, regardless of type (immunocompromised, cardiovascular disorders, genetic disorders, neurological disorders etc.), and patients without comorbidities. The reference category in the logistic regression models was the presence of at least one comorbidity.