Comments to the Scoring System for Pediatric Necrotizing Pneumonia

To the Editors: We read with great interest the study by Piastra and colleagues1 describing a large population of children with necrotizing pneumonia (NP) from several Italian pediatric centers, and we particularly appreciated their effort in proposing a clinical scoring system aimed at supporting early patient stratification based on the risk of prolonged intensive care (PICU) stay and the need for lung surgery. In the postpandemic years, given the increased incidence observed,2 a growing number of children with complicated pneumonia, including NP, have been treated at our Department. This has also been possible because our tertiary-care pediatric center is the largest in Southern Italy and the only one equipped with PICU and a pediatric thoracic surgery unit, thus receiving patients from a wide geographic area. Nevertheless, although periods of PICU admission were sometimes required due to clinical deterioration or postoperative care, our NP patients spent most of their hospitalization in the Pediatric Emergency and Sub-Intensive Care Unit (Sub-PICU), which can provide noninvasive ventilation, continuous monitoring and chest drainage management. Given the high burden in terms of diagnostic and therapeutic efforts, identification of prognostic factors for the severe clinical course of NP is crucial to optimize management. To date, several biomarkers have been identified as risk factors for massive necrosis, but the assessment of their reliability when combined in a clinical score is lacking.3 We retrospectively enrolled 21 children admitted with NP between May 2022 and November 2025 and applied the scoring system proposed by Piastra et al. to assess its reliability in predicting a prolonged PICU stay (>7 days, according to the authors’ definition), and the need for lung surgery. Because most patients spent a few days in the PICU, we modified our outcome parameter by combining the duration of PICU and Sub-PICU stays. The main clinical and laboratory characteristics of the enrolled patients are reported in Table 1. Interestingly, a score ≥6 showed a sensitivity and specificity of 20% and 100%, respectively, in predicting prolonged Sub-PICU/PICU stay, compared with the 78% and 82% reported by Piastra et al. One of the main reasons for this discrepancy is likely the lower rate of Staphylococcus aureus isolation in our population (9.5% vs. the 25% reported by the authors), resulting in a reduction of 2 points in the total score for most patients. Therefore, we slightly modified the proposed score by removing the S. aureus isolation item and replacing it with 2 laboratory parameters known to be markers of a more severe clinical course—hyponatremia (Na <135 mEq/L, 2 points) and hypoalbuminemia (serum albumin <3 g/dL, 2 points) at presentation.3,4 After these modifications, the maximum total score is 16. When applied to our patients using the same cutoff (≥6), the sensitivity and sensibility for predicting prolonged Sub-PICU/PICU stay were 60% and 100%, respectively. Such improvement in sensitivity was not observed when the score was tested for predicting the likelihood of surgical intervention: sensitivity and specificity were 33% and 66.7%, respectively, both with the original and the modified score. TABLE 1 - Characteristics of the Study Population Clinical characteristics* Age (yr) 4.7 (0.2–12.4) Age groups, n (%) ≤1 yr 4 (19) 1–2 yr 1 (4.8) 2–3 yr 2 (9.5) 3–5 yr 5 (23.8) >5 yr 9 (42.9) Sex, n (%) Male 8 (38) Female 13 (62) Comorbidities, n (%) Asthma 1 (4.8) Duration of hospital stay (d)* 15 (5–33) Laboratory findings C-reactive protein at admission (mg/L, reference value 0-5)* 273.9 (63.1–546.9) Identified pathogen, n (%)† Streptococcus pneumoniae 12 (57.1) Mycoplasma pneumoniae 5 (23.8) Staphylococcus aureus 2 (9.5) Streptococcus pyogenes 2 (9.5) Klebsiella pneumoniae 1 (4.8) Viral coinfections 3 (14.3) Interventions Chest drainage, n (%) 14 (66.7) Lung surgery, n (%) 3 (14.3) *Data are presented as median and ranges.†Data from culture and/or PCR on blood, pleural fluid or respiratory samples. Our preliminary data, although limited by the retrospective design and the small sample size, support the need for a critical revision of the proposed score, considering the potential inclusion of additional laboratory biomarkers of lung necrosis. Moreover, the presence of an item strictly related to etiology may compromise its accuracy, given the challenges in reliably identifying pathogens—particularly in patients who have often received antibiotic treatment before admission.