This study was designed as a retrospective case-control study. Infants under 3 months of age diagnosed with late-onset GBS sepsis and admitted to the Children's Hospital of Soochow University between January 1, 2007, and December 31, 2022, were included as study subjects. Infants with severe congenital malformations and/or inherited metabolic disorders were excluded. This study adheres to the ethical standards approved by the Medical Ethics Committee of the Children's Hospital of Soochow University (Approval No. 2022CS010). All data were anonymized, and no biological samples or personally identifiable information were involved. The requirement for informed consent from the guardians was waived with the approval of the Ethics Committee. All study procedures were performed according to the ethical standards in the Declaration of Helsinki.

Patients were divided into a complication group and a non-complication group based on the presence or absence of sepsis-related complications. The diagnostic criteria for late-onset GBS sepsis and associated complications were based on previously published standards (7). Specifically, the diagnosis required the presence of relevant clinical signs and symptoms plus microbiological confirmation via a positive GBS culture obtained from a sterile body cavity (e.g., blood or cerebrospinal fluid). Late-onset GBS sepsis was defined as GBS sepsis with an onset age of ≥7 days.

Complications primarily included septic shock, disseminated intravascular coagulation (DIC), primary or metastatic suppurative inflammation or abscesses (such as hepatic abscess, pericarditis, purulent meningitis, septic arthritis, and osteomyelitis). Septic shock: sepsis-induced circulatory and tissue hypoperfusion with persistent hypotension requiring vasoactive support to maintain adequate mean arterial pressure (e.g., dopamine >5 μg/kg·min or any dose of dobutamine, norepinephrine, or epinephrine), together with clinical or laboratory evidence of hypoperfusion. GBS-related purulent meningitis is characterized by clinical symptoms of purulent meningitis, CSF findings consistent with the diagnosis, and the isolation of group B Streptococcus from CSF and/or blood cultures. DIC: an acquired coagulopathy characterized by systemic activation of coagulation leading to widespread microvascular thrombosis, consumptive depletion of coagulation factors and platelets, and secondary hyperfibrinolysis, presenting with bleeding tendency and microcirculatory failure. Osteomyelitis: diagnosis is based on compatible clinical features (limb/joint pain with fever), supportive laboratory abnormalities (elevated ESR and leukocytosis), imaging evidence of deep soft-tissue swelling and/or bony destruction, and confirmation of infection by aspiration or operative drainage of pus. Final diagnosis integrates clinical, laboratory, imaging, and microbiological data. All the above diagnoses are made with reference to Zhu Futang Practice of Pediatrics and Practical Neonatology (7, 8). Abnormal peripheral white blood cell count (including leukocytosis and leukopenia), thrombocytopenia, and pathological jaundice were defined according to the criteria outlined in Zhu Futang Practice of Pediatrics and Practical Neonatology (7, 8). Neonatal hypoalbuminemia was defined as a serum albumin level of < 25 g/L (9).

All clinical data were retrieved from the hospital's electronic medical record system. Information collected included the infant's basic characteristics (sex, birth weight, age at admission), perinatal factors (gestational age at delivery, mode of delivery, presence of clinical chorioamnionitis, and premature rupture of membranes), initial clinical manifestations, and laboratory findings within 24 h of admission. Laboratory evaluations encompassed white blood cell (WBC) count, platelet count, C-reactive protein (CRP), procalcitonin (PCT), liver and renal function tests, coagulation profile, arterial blood gas analysis, blood culture, and cerebrospinal fluid (CSF) examination.

The Pediatric Sequential Organ Failure Assessment (pSOFA) score (10): was calculated on the first day of admission for all infants. The pSOFA score evaluates six organ systems: respiratory, coagulation, liver, cardiovascular, neurological, and renal, with each system scored from 0 to 4. The total pSOFA score was obtained by summing the individual scores of each system on day 1 of admission. The Pediatric Logistic Organ Dysfunction Score 2 (PELOD-2) (11): was used to assess dysfunction in five systems: central nervous, cardiovascular, renal, respiratory, and hematologic. Each system is scored from 0 to 6, and the total PELOD-2 score was calculated by summing the individual system scores on the first day of admission (for infants transferred from another hospital, scoring was based on the first admission day at the referring hospital). A high pSOFA/PELOD-2 score was defined as a score exceeding the optimal cutoff value determined by the receiver operating characteristic (ROC) curve analysis for predicting the occurrence of complications in infants with late-onset GBS sepsis in this study.

The data were analyzed using SPSS version 26.0. For continuous variables with a normal distribution, results were expressed as x̄ ±s, and comparisons between groups were performed using the independent samples t-test. For non-normally distributed continuous variables, data were presented as median (P₂₅-P₇₅) and compared using the Mann–Whitney U-test. Categorical variables were expressed as frequencies and percentages, and group comparisons were conducted using the χ² test. Multivariate logistic regression analysis was conducted to identify independent risk factors for complications. Construct the ROC curve to evaluate the predictive efficacy of clinical variables for the complications. All clinical variables considered as potential predictors were recorded preceding the diagnosis of the adverse outcome. A P-value < 0.05 was considered statistically significant.

The general characteristics of the two groups are summarized in Table 1. There were no significant differences between the complication and non-complication groups in terms of sex, prematurity, birth weight, small for gestational age, maternal age, fever before delivery, mode of delivery, chorioamnionitis, meconium-stained amniotic fluid, asphyxia at birth, premature rupture of membranes ≥18 h, or breastfeeding (P > 0.05). However, the gestational age was significantly lower in the complication group compared to the non-complication group (P < 0.05).

The clinical manifestations of the two groups are presented in Table 2. Compared with the non-complication group, the complication group had significantly higher incidences of seizure and bulging anterior fontanelle, as well as higher fever temperatures and lower PaO₂/FiO₂ ratios (P < 0.05). There were no significant differences between the two groups in the proportions of patients presenting with fever, decreased responsiveness, jaundice, tachypnea, poor feeding, somnolence, irritability, muscle tone abnormalities, mottling of the skin, cyanosis, and vomiting (P > 0.05).

Laboratory findings and severity scores within 24 h of admission are presented in Table 3. Compared with the non-complication group, the complication group showed significantly higher rates of hypoalbuminemia, concurrent positive blood and CSF cultures, elevated creatinine and blood urea nitrogen levels, higher pSOFA and PELOD-2 scores, and greater proportions of patients with high pSOFA and high PELOD-2 scores. In contrast, pH and serum albumin levels were significantly lower in the complication group (P < 0.05). There were no significant differences between the two groups in terms of WBC count, leukocytosis, leukopenia, platelet count, thrombocytopenia, PCT, CRP, fibrinogen, serum sodium, serum potassium, or lactic acid levels (P > 0.05).

ROC curve analysis revealed that the optimal cut-off value for both the pSOFA and PELOD-2 scores in predicting complications in infants with late-onset GBS sepsis was 3.50. The area under the curve (AUC) was 0.784 (95% CI: 0.700–0.860) for the pSOFA score and 0.751 (95% CI: 0.660–0.840) for the PELOD-2 score. At this cut-off, the sensitivity and specificity were 76.70% and 78.00% for pSOFA, and 73.30% and 75.60% for PELOD-2, respectively.

Multivariate logistic regression was performed to identify independent predictors of complications in infants with late-onset GBS sepsis. Variables were selected a priori based on clinical relevance and significant intergroup differences, including creatinine, urea nitrogen, PH, albumin, hypoalbuminemia, high pSOFA score, and high PELOD-2 score. The analysis of 101 complete cases (60 events; EPV = 8.571) yielded a statistically significant model (Omnibus test: χ² = 36.755, P < 0.001; −2 Log Likelihood = 71.376) with good calibration (Hosmer-Lemeshow: χ² = 6.031, P = 0.644). The model's explanatory power was substantial (Nagelkerke R² = 0.501), with an overall accuracy of 82.10%. The absence of multicollinearity was confirmed (all VIF < 3.00, Tolerance > 0.40). It was ultimately found that the following factors were identified as being independently associated with adverse complications in infants with late-onset GBS sepsis: high PELOD-2 score (OR: 3.615, 95% CI: 1.027–12.727, P = 0.045), high pSOFA score (OR: 9.354, 95% CI: 1.989–43.984, P = 0.005), high creatinine levels (> 0.78 mg/dL) (OR: 1.130, 95% CI: 1.053–1.213, P = 0.001) and hypoalbuminemia (OR: 1.486, 95% CI: 1.010–2.186, P = 0.044). Details are shown in Table 4.

The ROC curve (Figure 1) for predicting complications in infants with late-onset GBS sepsis was constructed using a combination of high PELOD-2 score, high pSOFA score, high creatinine levels, and hypoalbuminemia. The results showed that the optimal cut-off value for predicting complications was 0.021, with an AUC of 0.858 (95% CI: 0.782–0.934), a sensitivity of 77.78%, and a specificity of 82.61%. The overall predictive accuracy was 80.22%, the positive predictive value was 81.40%, the negative predictive value was 79.17%, and the Youden index was 0.604. [Note: The ROC curve analysis for serum creatinine levels identified an optimal cut-off value of 0.78 mg/dL for predicting complications in infants with late-onset GBS sepsis. This cut-off yielded an AUC of 0.710 (95% CI: 0.609–0.812), with a sensitivity of 71.67% and a specificity of 75.61%].