Factors associated with persistent postoperative hyperthermia in pediatric craniocerebral injury: a retrospective cohort study and preliminary risk score Authors: Na Yan, Xiaoyan Zhou, Fujia Zhuang, Xiaoyan Li Published: 2026 ABSTRACT AbstractBackgroundFever represents a common postoperative complication in pediatric patients with craniocerebral injury. The present study aimed to investigate the prevalence and factors independently associated with persistent postoperative hyperthermia in children following craniocerebral injury surgery, thereby providing evidence to inform clinical nursing practice.MethodsA retrospective cohort study was conducted involving pediatric patients who underwent craniocerebral injury surgery. Patients were stratified into two groups: the persistent postoperative hyperthermia group and the non-persistent postoperative hyperthermia group, based on the presence of persistent postoperative hyperthermia (defined as a body temperature ≥ 38.5 °C, sustained or recurrent, for a cumulative duration exceeding 36 h within the first 72 h postoperatively).ResultsA total of 238 pediatric patients who underwent craniocerebral injury surgery were included. The incidence of persistent postoperative hyperthermia was 28.99%. Multivariate logistic regression analysis revealed that age ≤ 5 years (odds ratio [OR] = 3.516, 95% confidence interval [CI]: 2.108–5.854), severe craniocerebral injury (OR = 6.442, 95%CI: 3.285–12.638), elevated postoperative intracranial pressure (≥ 20 mmHg, OR = 4.598, 95%CI: 2.683–7.895), postoperative drainage tube indwelling time ≥ 72 h (OR = 4.002, 95%CI: 2.356–6.817), pulmonary infection (OR = 5.438, 95%CI: 3.126–9.457), and methicillin-resistant Staphylococcus aureus (MRSA) infection (OR = 3.078, 95%CI: 1.756–5.392) were independently associated with persistent postoperative hyperthermia. A preliminary predictive model constructed based on these factors had a sensitivity of 0.754, specificity of 0.879, and Youden index of 0.633 in the cohort, but these performance estimates are likely optimistic and require external validation.ConclusionPersistent postoperative hyperthermia is a prevalent complication in children following craniocerebral injury surgery, which is influenced by multiple factors. The identified factors are independently associated with this outcome, although causality cannot be inferred. The exploratory predictive model, while showing favorable discrimination in this sample, must be externally validated before clinical application.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13741-026-00703-3.Keywords: Postoperative, Fever, Children, Craniocerebral injury, Treatment, Nursing, Care Introduction Craniocerebral injury is a common acute and critical illness in pediatric clinical practice, mainly caused by external factors such as accidental impact, falls from heights, and traffic accidents (Zhong et al. 2025). It is characterized by acute onset, critical condition, and rapid progression, which seriously threatens the life safety and growth of children (Chong et al. 2025). With the continuous advancement of neurosurgical medical technology, surgical treatment has become the core method to save the lives of children with craniocerebral injury and improve neurological prognosis. However, the occurrence of postoperative complications remains a key factor restricting the therapeutic effect, among which fever is one of the most common postoperative complications (Ahmed et al. 2025). Clinical data (Meng et al. 2019; Lin et al. 2021) show that the incidence of fever in children after craniocerebral injury surgery can reach 30% ~ 50%, with the fever duration usually exceeding 24 h and the body temperature often maintaining above 38.5℃. Its pathogenesis is complex, closely related to various factors such as stress response caused by surgical trauma, intracranial infection, central thermoregulatory dysfunction, and absorption fever from postoperative hematoma or necrotic tissue. Persistent fever can significantly increase the oxygen consumption of children's brain tissue, induce or aggravate cerebral edema and increased intracranial pressure, thereby leading to further deterioration of neurological damage (Alsayegh et al. 2024; Lundstedt et al. 2025). At the same time, it may also involve multiple organ functions such as respiratory, circulatory, and digestive systems, which not only prolongs the hospital stay of children but also increases the risk of mortality and long-term sequelae, bringing a heavy physical, psychological, and economic burden to children's families and posing a severe challenge to clinical medical and nursing work (Lang et al. 2021; Cavazzoni and Boakye 2024; Goyal-Honavar et al. 2024). Pediatric patients are in the stage of growth and development, with immature nervous and immune systems and imperfect central thermoregulatory function. Their tolerance to high fever is significantly lower than that of adults, so the pathophysiological damage caused by persistent hyperthermia after craniocerebral injury surgery is more prominent in children (Lin et al. 2024). At present, clinical nursing for fever in such children mainly relies on conventional symptomatic interventions, including physical cooling (such as warm water sponge bath, ice pack cold compress) and drug cooling (such as application of ibuprofen and acetaminophen) (Lotakis et al. 2023). However, these measures lack precise prediction of high-risk factors for fever and individualized intervention design. Moreover, the clinical identification of core influencing factors for the occurrence of fever is not comprehensive enough, and the timing of intervention is not accurately grasped, resulting in poor cooling effect in some children, prolonged duration of fever, and impact on the overall rehabilitation process (Cavazzoni and Boakye 2024; Lewis et al. 2020). In addition, among existing clinical studies, systematic research on the epidemiological characteristics, screening of associated factors, and effects of targeted nursing interventions for persistent hyperthermia in children after craniocerebral injury surgery is relatively scarce, lacking unified nursing standards and evidence-based basis, which makes it difficult to effectively guide the standardized development of clinical nursing practice. Based on this, this study aims to systematically analyze the current situation of persistent hyperthermia in children after craniocerebral injury surgery and clarify its key associated factors, so as to provide a basis for optimizing clinical treatment and nursing plans. Methods Ethical approval This study was reviewed and approved by the Medical Ethics Committee of our hospital (Approval No.: 202601073–1), and the research implementation strictly followed the Declaration of Helsinki and relevant ethical norms for pediatric clinical research. All guardians of children voluntarily signed written informed consent forms after fully understanding the research purpose, methods, potential risks, and benefits. During the research process, attention was paid to protecting the privacy of children; all clinical data were anonymized and only used for statistical analysis in this study. Study design A retrospective cohort study design was adopted. Children who underwent surgical treatment for craniocerebral injury in the Department of Neurosurgery of our hospital were selected as research subjects, and divided into a persistent postoperative hyperthermia group and a non-persistent hyperthermia group according to whether persistent postoperative hyperthermia occurred after surgery. By retrospectively collecting clinical medical records of children, data on baseline characteristics, disease severity, surgery-related indicators, and postoperative complications were compared and analyzed between the two groups to clarify the current situation and key associated factors of persistent hyperthermia in children after craniocerebral injury surgery, thereby providing a basis for formulating targeted nursing countermeasures. Persistent postoperative hyperthermia was defined as a body temperature ≥ 38.5 °C, sustained or recurrent, for a cumulative duration exceeding 36 h within the first 72 h postoperatively. This threshold was chosen because temperatures ≥ 38.5 °C significantly increase cerebral metabolic demand and are commonly used as an intervention trigger in pediatric neurocritical care. The 36-h cumulative criterion was applied to exclude transient postoperative pyrexia related to surgical stress or resorption of blood products, which typically resolves within 24 h (Ma et al. 2011). Temperature was measured using continuous rectal or axillary thermistor probes in the ICU, and axillary readings were converted to core-temperature equivalents by adding 0.5 °C, in accordance with established pediatric intensive care practice at our institution. Children with preoperative fever, non-infectious fever (e.g., drug fever), and transient fever relieved within 36 h after surgery were excluded. Sample size calculation Referring to the results of previous relevant study (Li et al. 2019; Lou et al. 2015), assuming the incidence of persistent hyperthermia in children after craniocerebral injury surgery was 30%, with α = 0.05 (two-tailed test), β = 0.10, and an allowable error of 5%, PASS 15.0 statistical software was used for sample size estimation. The preliminary calculated minimum sample size required was 228 cases. Considering potential situations such as missing clinical data and non-compliance with inclusion and exclusion criteria, the sample size was appropriately expanded by 10%, and a total of 238 children were finally included as research subjects. Study population Inclusion criteria were defined as follows: (1) children aged 1–12 years who were diagnosed with craniocerebral injury and underwent craniotomy; (2) surgical procedures included routine neurosurgical operations such as decompressive craniectomy, hematoma evacuation, or intracranial foreign body removal; (3) postoperative hospital stay of ≥ 72 h with complete and traceable clinical medical records; and (4) guardians who provided written informed consent for participation in the study. The cohort comprised children with closed traumatic brain injuries; polytrauma cases were included but represented less than 5% of the sample. Exclusion criteria included: (1) preoperative fever (body temperature ≥ 38.5℃) or confirmed infectious diseases prior to surgery; (2) comorbidity with severe underlying conditions such as severe liver or kidney insufficiency, immunodeficiency diseases, or malignant tumors; (3) death or voluntary discharge from the hospital within 24 h after surgery; and (4) missing key clinical data (e.g., operation time, intracranial pressure monitoring results) that were essential for the study analysis. Data collection A standardized data collection form was established. Two researchers trained uniformly independently extracted clinical medical records of children retrospectively. The extracted content included: (1) Baseline data: age, gender, body mass index (BMI), and presence of congenital heart disease; (2) Disease-related indicators: Glasgow Coma Scale (GCS) was used to assess the severity of craniocerebral injury (mild: 13–15 points, moderate: 9–12 points, severe: 3–8 points); (3) Surgery-related indicators: operation duration, postoperative intracranial pressure monitoring value, and postoperative drainage tube indwelling time; (4) Postoperative complications: occurrence of pulmonary infection, methicillin-resistant Staphylococcus aureus (MRSA) infection, etc.; (5) Fever-related data: postoperative body temperature changes, duration of fever, etc. The institutional protocol for fever management followed a tiered approach beginning with physical cooling measures; acetaminophen was administered if temperature remained above 38.5 °C. Because antipyretic administrations were not systematically recorded, this represents a potential source of variability. After data extraction, a third researcher conducted cross-validation. Disagreements were resolved through consultation or by consulting senior clinicians to ensure data accuracy. Statistical analysis SPSS 26.0 statistical software was used for data processing, and GraphPad Prism 9.0 was used to draw ROC curves. Measurement data were expressed as mean ± standard deviation (x̄ ± s), and independent samples t-test was used for intergroup comparison; count data were expressed as number (percentage) [n (%)], and χ2 test was used for intergroup comparison; rank data (severity of craniocerebral injury) were compared between groups using Kruskal–Wallis H rank sum test. Pearson correlation analysis (for measurement data) or point-biserial correlation analysis (for categorical data) was used to explore the correlation between various factors and persistent postoperative hyperthermia. Variables with P < 0.05 in univariate analysis were included in the multivariate logistic regression model to identify factors independently associated with postoperative persistent hyperthermia. Candidate variables were selected based on univariate significance (P < 0.05). Collinearity was assessed using variance inflation factors (all < 2.0). Continuous variables were dichotomized using clinically relevant cutoffs (age ≤ 5 years, ICP ≥ 20 mmHg, drainage time ≥ 72 h) before modeling. There were no missing data for the included variables. The final model contained 69 events and 6 predictors. Model calibration was evaluated with the Hosmer–Lemeshow test. A preliminary additive prediction model was constructed by assigning 2 points to each associated factor (total score range 0–12). The simplified score was compared with the original logistic regression model; the area under the ROC curve decreased minimally (0.82 vs. 0.84, DeLong test P = 0.32, Supplementary Table S1). ROC curve analysis was used to evaluate the predictive discrimination of the model in the cohort. Sensitivity, specificity, and Youden index under different scoring thresholds were calculated to determine the optimal threshold. A P value < 0.05 was considered statistically significant. Results A total of 238 children who underwent surgery for craniocerebral injury were enrolled in this study, among which 69 cases developed persistent postoperative hyperthermia, with an incidence rate of 28.99%. The results of intergroup comparison of baseline data between the two groups (Table 1) showed that there were statistically significant differences in age, BMI, severity of craniocerebral injury, operation time, postoperative intracranial pressure, postoperative drainage tube indwelling time, and the incidence of pulmonary infection and MRSA infection between the persistent postoperative hyperthermia group and the non-persistent hyperthermia group (all P < 0.05). However, no statistically significant differences were observed in gender composition ratio or the prevalence of congenital heart disease between the two groups (P > 0.05). Specifically, children in the persistent hyperthermia group were younger, had lower BMI, a significantly higher proportion of severe craniocerebral injury, longer operation duration, significantly elevated postoperative intracranial pressure, prolonged drainage tube indwelling time, and notably higher incidences of postoperative pulmonary infection and MRSA infection compared with the non-persistent hyperthermia group (Table 1). Table 1 Baseline data of children after craniocerebral injury surgery and comparison between the persistent postoperative hyperthermia group and the non-persistent hyperthermia group (n = 238)VariablesPersistent postoperative hyperthermia group (n = 69)Non-persistent hyperthermia group (n = 169)t/χ2pAge (y)3.85 ± 1.726.28 ± 2.358.963 < 0.001BMI (kg/m2)16.25 ± 2.1817.03 ± 2.462.1350.034Male/female38/3195/740.1260.723Congenital heart disease8(11.59%)12(7.09%)1.5820.208Craniocerebral injury grade*28.754 < 0.001- Mild5(7.25%)68(40.24%)- Moderate23(33.33%)76(44.97%)- Severe41(59.42%)25(14.79%)Operation time (min)158.62 ± 32.45126.38 ± 28.767.892 < 0.001Postoperative intracranial pressure (mmHg)22.58 ± 3.6216.85 ± 2.9712.365 < 0.001Postoperative drainage tube placement time (h)86.35 ± 18.7252.68 ± 15.3414.258 < 0.001Pulmonary infection32(46.38%)25(14.79%)31.682 < 0.001MRSA infection15(21.74%)8(4.73%)18.953 < 0.001*Craniocerebral injury grade was evaluated by Glasgow Coma Scale (GCS): Mild (GCS 13–15 points), Moderate (GCS 9–12 points), Severe (GCS 3–8 points)BMI Body mass index, MRSA Methicillin-resistant Staphylococcus aureus Baseline data of children after craniocerebral injury surgery and comparison between the persistent postoperative hyperthermia group and the non-persistent hyperthermia group (n = 238) *Craniocerebral injury grade was evaluated by Glasgow Coma Scale (GCS): Mild (GCS 13–15 points), Moderate (GCS 9–12 points), Severe (GCS 3–8 points)BMI Body mass index, MRSA Methicillin-resistant Staphylococcus aureus The results of correlation analysis (Table 2) indicated that persistent postoperative hyperthermia was moderately negatively correlated with children’s age (r = −0.486, P < 0.001) and weakly negatively correlated with BMI (r = −0.138, P = 0.034). It was positively correlated with the severity of craniocerebral injury, operation time, postoperative intracranial pressure, postoperative drainage tube indwelling time, pulmonary infection, and MRSA infection. Among these, the correlations with the severity of craniocerebral injury (r = 0.623), postoperative drainage tube indwelling time (r = 0.615), and postoperative intracranial pressure (r = 0.578) reached a moderate to strong level (all P < 0.001). In contrast, no significant correlation was found between congenital heart disease and persistent postoperative hyperthermia (r = 0.095, P = 0.208), suggesting that this factor may not be a major influencing factor for persistent postoperative hyperthermia (Table 2). Table 2 Correlation analysis between persistent postoperative hyperthermia and various factors in children after craniocerebral injury surgeryVariablesrpAge (y)−0.486 < 0.001BMI (kg/m2)−0.1380.034Congenital heart disease0.0950.208Craniocerebral injury grade (severity)0.623 < 0.001Operation time (min)0.352 < 0.001Postoperative intracranial pressure (mmHg)0.578 < 0.001Postoperative drainage tube placement time (h)0.615 < 0.001Pulmonary infection0.492 < 0.001MRSA infection0.386 < 0.001The correlation coefficient of categorical variables was calculated by point-biserial correlationBMI Body mass index, MRSA Methicillin-resistant Staphylococcus aureus Correlation analysis between persistent postoperative hyperthermia and various factors in children after craniocerebral injury surgery The correlation coefficient of categorical variables was calculated by point-biserial correlation BMI Body mass index, MRSA Methicillin-resistant Staphylococcus aureus Variables with statistically significant differences in univariate analysis were included in the multivariate logistic regression model, and the variable assignment criteria are detailed in Table 3. The regression analysis results (Table 4) revealed that age ≤ 5 years, severe craniocerebral injury (with mild injury as the reference), increased postoperative intracranial pressure (≥ 20 mmHg), postoperative drainage tube indwelling time ≥ 72 h, complicated pulmonary infection, and MRSA infection were factors independently associated with persistent postoperative hyperthermia in children after craniocerebral injury surgery (all P < 0.05). Among these, children with severe craniocerebral injury had the strongest association (OR = 6.442, 95% CI: 3.285 ~ 12.638), followed by those with complicated pulmonary infection (OR = 5.438, 95% CI: 3.126 ~ 9.457) and increased postoperative intracranial pressure (OR = 4.598, 95% CI: 2.683 ~ 7.895), indicating that the aforementioned factors have important associations with the occurrence of persistent postoperative hyperthermia. Given the retrospective design and the possibility that some of these variables (e.g., infections) were diagnosed during the same postoperative window as the fever, they should be interpreted as associated factors rather than causal risk factors. The Hosmer–Lemeshow test indicated good model calibration (P = 0.41) (Table 4). Table 3 Variable assignment for multivariate logistic regression of factors associated with persistent postoperative hyperthermia in children after craniocerebral injury surgeryFactorsVariables​AssignmentPersistent high feverYYes = 1, No = 0AgeX1 ≤ 5y = 1, > 5y = 0Craniocerebral injury gradeX2Severe = 2, Moderate = 1, Mild = 0Postoperative intracranial pressureX3Increased (≥ 20 mmHg) = 1, Normal (< 20 mmHg) = 0Postoperative drainage tube placement timeX4 ≥ 72 h = 1, < 72 h = 0Pulmonary infectionX5Yes = 1, No = 0MRSA infectionX6Yes = 1, No = 0MRSA Methicillin-resistant Staphylococcus aureus Variable assignment for multivariate logistic regression of factors associated with persistent postoperative hyperthermia in children after craniocerebral injury surgery MRSA Methicillin-resistant Staphylococcus aureus Table 4Logistic regression analysis of factors associated with persistent postoperative hyperthermia in children after craniocerebral injury surgeryVariablesβWaldOR95%CIpAge ≤ 5y1.25818.6323.5162.108 ~ 5.854 < 0.001Craniocerebral injury grade (severe vs mild)1.86325.3686.4423.285 ~ 12.638​ < 0.001Postoperative intracranial pressure increased1.52621.4594.5982.683 ~ 7.895 < 0.001Postoperative drainage tube placement time ≥ 72 h1.38519.7624.0022.356 ~ 6.817 < 0.001Pulmonary infection1.69223.8545.4383.126 ~ 9.457 < 0.001MRSA infection1.12512.3683.0781.756 ~ 5.3920.001Reference category for craniocerebral injury grade: Mild.MRSA Methicillin-resistant Staphylococcus aureus Logistic regression analysis of factors associated with persistent postoperative hyperthermia in children after craniocerebral injury surgery Reference category for craniocerebral injury grade: Mild. MRSA Methicillin-resistant Staphylococcus aureus A preliminary predictive scoring model for persistent postoperative hyperthermia was constructed based on the 6 independently associated factors identified by multivariate logistic regression analysis, and a simplified equal-weight scoring method was used to quantify the risk (each factor was assigned 2 points, with a total score ranging from 0 to 12 points; see Table 5 note for scoring criteria). It must be noted that pulmonary infection and MRSA infection are typically diagnosed during the same 72-h postoperative window as the outcome of persistent hyperthermia, and thus they are not true baseline predictors but rather early clinical events that co-occur with or immediately precede the fever definition. The score therefore reflects a combination of pre-existing and early postoperative factors, and its application is intended for risk stratification once the early postoperative clinical picture has emerged, not for preoperative prediction. ROC curve analysis (Fig. 1) showed that the model had acceptable discrimination in the cohort. By calculating the sensitivity, specificity, and Youden index at different scoring thresholds (Table 5), the optimal threshold was determined to be 6.0 points, with a corresponding Youden index of 0.633, sensitivity of 0.754, and specificity of 0.879. It should be emphasized that this model was derived and evaluated in the same sample without internal validation (e.g., bootstrapping) or external validation; thus, reported performance is likely optimistic. The score should be regarded as an exploratory risk stratification tool and is not yet ready for clinical application. Table 5 Sensitivity and specificity of the preliminary prediction model for persistent postoperative hyperthermia in children after craniocerebral injury surgery in the cohort at different score thresholdsTotal scoreSensitivitySpecificityYouden Index1.01.0000.0000.0002.00.9860.1850.1713.00.9420.3560.2984.00.8970.5820.4795.00.8250.7630.5886.00.7540.8790.6337.00.6820.9250.6078.00.5360.9680.5049.00.3250.9860.31110.00.1580.9950.15311.00.0421.0000.04212.00.0001.0000.000The scoring criteria of the prediction model refer to the 6 associated factors included in Table 4, with 2 points assigned to each factor. The total score ranges from 0 to 12 points (Age ≤ 5y = 2 points, Severe craniocerebral injury = 2 points, Postoperative intracranial pressure increased = 2 points, Postoperative drainage tube placement time ≥ 72 h = 2 points, Pulmonary infection = 2 points, MRSA infection = 2 points). The optimal threshold is 6.0 points (Youden Index = 0.633) Sensitivity and specificity of the preliminary prediction model for persistent postoperative hyperthermia in children after craniocerebral injury surgery in the cohort at different score thresholds The scoring criteria of the prediction model refer to the 6 associated factors included in Table 4, with 2 points assigned to each factor. The total score ranges from 0 to 12 points (Age ≤ 5y = 2 points, Severe craniocerebral injury = 2 points, Postoperative intracranial pressure increased = 2 points, Postoperative drainage tube placement time ≥ 72 h = 2 points, Pulmonary infection = 2 points, MRSA infection = 2 points). The optimal threshold is 6.0 points (Youden Index = 0.633) The ROC curve for predicting persistent postoperative hyperthermia after craniocerebral injury surgery in the cohort Discussion The results of this study showed that the incidence of persistent postoperative hyperthermia in children after craniocerebral injury surgery was 28.99% (69/238), which is basically consistent with the incidence of 30% ~ 50% reported in previous clinical studies, further confirming that persistent hyperthermia is a common complication in children after craniocerebral injury surgery. This incidence rate suggests that clinical practice should attach great importance to postoperative temperature management, especially in neurosurgical postoperative nursing, where the prevention and early intervention of hyperthermia should be regarded as key work contents. Persistent hyperthermia is more harmful to pediatric patients; it not only increases the oxygen consumption of brain tissue and induces cerebral edema but also may amplify neurological damage through inflammatory responses, prolong the hospital stay, and increase the risk of long-term sequelae (Lang et al. 2021; Guo et al. 2025). Therefore, the incidence rate and associated factors identified in this study provide critical epidemiological evidence for the establishment of targeted clinical prevention and control systems. Multivariate logistic regression analysis revealed that age ≤ 5 years (OR = 3.516) and severe craniocerebral injury (OR = 6.442) were the strongest independently associated factors for persistent postoperative hyperthermia. In addition, correlation analysis showed a moderate to strong positive correlation between the severity of craniocerebral injury and hyperthermia (r = 0.623), which is closely related to the physiological characteristics of children and the pathological mechanism of the disease. Young children have immature nervous systems and imperfect central thermoregulatory functions, resulting in poor tolerance to stress stimuli such as surgical trauma and infection, making them more prone to thermoregulatory disorders (Goyal-Honavar et al. 2024; Simunovic-Erpusina et al. 2024). In contrast, children with severe craniocerebral injury are often accompanied by impaired hypothalamic-pituitary axis function, which disrupts the central thermoregulatory mechanism (Cammisa et al. 2025). Meanwhile, the cascade effect of inflammatory responses induced by cerebral ischemia and hypoxia further exacerbates the occurrence and persistence of hyperthermia (Yanai Milshtein et al. 2024). Furthermore, children with severe craniocerebral injury usually require longer surgical duration and postoperative drainage, which indirectly increases the risk of infection, forming a "synergistic effect" for the development of hyperthermia. This also explains why the incidence of hyperthermia in this group is significantly higher than that in children with mild injuries. Increased postoperative intracranial pressure, postoperative drainage tube indwelling time ≥ 72 h, pulmonary infection, and MRSA infection were all independently associated with persistent hyperthermia. Among these, pulmonary infection (OR = 5.438) and increased postoperative intracranial pressure (OR = 4.598) exhibited particularly strong associations. However, these variables must be interpreted with caution. In this retrospective study, the exact temporal relationship between infection diagnosis, ICP elevation, and the onset of hyperthermia could not always be determined. These factors may represent direct causes of fever, consequences of a severe clinical trajectory, or markers of underlying injury severity, rather than true pre-event predictors. Elevated postoperative intracranial pressure, which is likely a marker of severe brain injury and disturbed cerebral physiology, can directly induce hyperthermia by affecting the central thermoregulatory pathway and aggravating cerebral edema and inflammatory responses (Zhang et al. 2025). Prolonged drainage tube indwelling time increases the probability of retrograde infection, especially infections caused by drug-resistant bacteria such as MRSA, leading to uncontrollable infections and subsequent persistent hyperthermia (Li 2020; Li et al. 2025). As a common postoperative complication, pulmonary infection-induced infectious fever overlaps with central high fever, significantly prolonging the duration of fever (Yue and Yan 2023). Thus, several of the identified factors may cluster in the most severely injured children receiving intensive care, and the model may predominantly reflect overall postoperative complexity rather than specific causal pathways. This result suggests that clinical practice should strengthen postoperative intracranial pressure monitoring, optimize drainage tube nursing, and enhance infection prevention and control to potentially reduce the burden of hyperthermia from multiple dimensions. The scoring prediction model constructed in this study based on 6 independently associated factors demonstrated acceptable discrimination in the cohort, as confirmed by ROC curve analysis. When the optimal threshold was set at 6.0 points, the sensitivity was 0.754, the specificity was 0.879, and the Youden index was 0.633. The equal weighting of 2 points per factor was a deliberate simplification for clinical convenience; the minimal loss in discrimination compared to the full logistic regression model is shown in the supplement. However, the model was neither internally validated (e.g., bootstrapping) nor externally validated; therefore, performance is likely optimistic, and the score remains exploratory. Adopting a simple scoring method, this model does not require complex detection equipment and, if validated, could quickly identify children potentially at higher risk for persistent postoperative hyperthermia, providing a convenient tool for the implementation of individualized clinical interventions. This score is intended for postoperative risk stratification once the early clinical picture has emerged, typically 24–48 h after surgery, when ICP data and preliminary microbiology results become available. It is not a preoperative screening tool. For children with a score ≥ 6.0 points, clinical practice may consider initiating enhanced monitoring and preventive measures in advance, such as shortening the drainage tube indwelling time, strengthening infection prevention and control, and optimizing temperature management protocols, thereby aiming to achieve "early identification and early intervention" to potentially reduce the incidence and duration of hyperthermia. Meanwhile, the construction of this model also provides new insights for improving the risk assessment system in neurosurgical postoperative nursing. Several important confounders were not captured in this study and should be considered when interpreting the results. Sedation, analgesia, and mechanical ventilation can significantly influence thermoregulation and may blunt or modify the febrile response. These variables were not consistently extractable from the medical records, and their absence may confound the observed associations. Other unmeasured factors, such as postoperative nutritional status, transfusion requirements, and ICU length of stay, could also contribute to the clustering of hyperthermia with severe illness. These limitations highlight the need for prospective studies with comprehensive data collection. Based on the identified associations and combined with clinical practice needs, potential targeted clinical treatment, nursing, and management strategies may be proposed as hypotheses for future investigation, though they require prospective validation. As an exploratory approach, the prediction model established in this study could be used to stratify postoperative children. Those with a score ≥ 6.0 points could be included in a high-risk monitoring protocol, receiving 1–2 hourly intensive temperature monitoring and dynamic assessment of intracranial pressure and infection indicators to achieve stratified risk management. For young children with severe injuries, an individualized temperature intervention plan featuring "physical cooling as the main approach and drug cooling as a supplement" may be considered, such as cold compress with a head ice cap combined with warm water sponge bath, to avoid the impact of potent antipyretic drugs on liver and kidney functions (Lehrnbecher et al. 2023; Wuni et al. 2025; Ohns 2025). For central hyperthermia, mild hypothermia therapy can be used in combination to precisely control the body temperature between 36.5 ~ 37.5℃ (Kucuk 2025). In terms of infection prevention and control, strict sterile operation could be implemented, the drainage tube indwelling time may be shortened when clinically feasible (preferably within 72 h), and respiratory care (such as regular turning and tapping the back, early respiratory care bundles, nebulization inhalation) could be strengthened to reduce the risk of pulmonary infection. For children at risk of MRSA infection, early screening for drug-resistant bacteria may be conducted, and preventive anti-infective treatment may be considered if necessary (Hu et al. 2025). In the regulation of postoperative physiological indicators, continuous monitoring of intracranial pressure is essential (Lee et al. 2025). Measures such as rational use of dehydrating agents and position management may be considered to control intracranial pressure < 20 mmHg, thereby potentially reducing the inducement of central hyperthermia. Additionally, optimizing the surgical process to shorten the operation duration may reduce the stress stimulation of surgical trauma on children. This study still has certain limitations that should be considered. Firstly, the retrospective cohort study design may lead to selection bias, and incomplete or missing clinical data records may affect the accuracy of the results. Crucially, temporality between candidate predictors (e.g., infections) and the onset of hyperthermia could not be rigorously established; therefore, no causal inferences should be drawn from these associations. Secondly, the study samples were all from a single center, lacking multi-center large-sample validation, so the generalizability of the exploratory prediction model needs to be further confirmed. The model was optimized and tested on the same cohort, which yields overly optimistic performance estimates; bootstrapping or split-sample internal validation and independent external validation are essential before any clinical use. Thirdly, important confounders such as sedation, analgesia, mechanical ventilation, and postoperative nutritional status were not considered in the study, which may result in the omission of relevant variables and residual confounding. Fourthly, because pulmonary infection and MRSA infection are diagnosed contemporaneously with the hyperthermia outcome, their inclusion in the model limits its utility for pre-event risk prediction. The model may be more accurately viewed as a composite severity index identifying children who are already on a complicated postoperative trajectory. Its clinical value lies in flagging high-risk cases for intensified management once early signs become apparent, rather than in anticipating fever before it develops. Future prospective studies that capture the exact timing of infection onset relative to fever will be necessary to disentangle these relationships. Finally, the effect of nursing interventions was not prospectively verified, and the practical application value of the proposed strategies needs to be further confirmed by subsequent clinical studies. Future research can adopt a prospective multi-center design, expand the sample size, improve variable collection, explicitly differentiate central from infectious fever, document the timing of all complications relative to fever onset, and validate the effectiveness of the prediction model. Interventional studies are also needed to further optimize the prevention and control strategies for persistent postoperative hyperthermia, thereby providing a more solid evidence-based basis for postoperative nursing in pediatric neurosurgery. Conclusion In summary, this retrospective single-center study found that persistent postoperative hyperthermia occurred in 28.99% of children after craniocerebral injury surgery. Age ≤ 5 years, severe craniocerebral injury, elevated postoperative intracranial pressure, postoperative drainage tube indwelling time ≥ 72 h, complicated pulmonary infection, and MRSA infection were independently associated with this outcome. The preliminary additive scoring system showed acceptable discrimination in the cohort (sensitivity 0.754, specificity 0.879 at a threshold of 6.0 points), but these estimates are optimistic, and the tool requires rigorous internal and external validation before clinical implementation. Prospective studies that disentangle the temporality and etiology of fever are needed to refine risk stratification and inform targeted management strategies. Supplementary Information Supplementary Material 1. Acknowledgements None. Authors’ contributions Na Yan: Conceptualization, Methodology, Software Xiaoyan Zhou: Data curation, Writing—Original draft Fujia Zhuang: Visualization, Investigation Xiaoyan Lin: Supervision, Software, Validation, Writing—Reviewing and Editing All authors have read and approved the final manuscript. Funding This study did not receive any funding in any form. Data availability The data associated with the paper are not publicly available but are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate In this study, all methodologies were conducted in compliance with the Helsinki Declaration. The study has been reviewed and approved by the ethics committee of Children’s Hospital of Nanjing Medical University (approval number: 202601073–1). And written informed consents had been obtained from all the parents or guardians of children. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Na Yan, Xiaoyan Zhou and Fujia Zhuang contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The data associated with the paper are not publicly available but are available from the corresponding author on reasonable request.