INTRODUCTION

Neonatal sepsis is a significant cause of morbidity and mortality, particularly among preterm infants, and remains a pressing global health concern. Early-onset neonatal sepsis is particularly challenging to diagnose due to its nonspecific clinical presentation, necessitating effective and timely diagnostic tools to reduce adverse outcomes. Traditional methods, such as microbial cultures, are slow and often unavailable in resource-limited settings. This study aimed to develop a robust multivariable risk prediction model tailored to improve early detection of Early Onset Sepsis (EOS) among preterm infants in the Philippines.

We conducted a retrospective analysis at a tertiary hospital in the Philippines using data from 1,354 preterm infants admitted between January 2019 and June 2024. Logistic regression models were employed, and predictors were selected through reverse stepwise elimination. Two scoring methods were developed: one based on beta coefficients divided by standard errors and another standardized to a total score of 100. The models were validated using Receiver Operator Characteristic curve analysis.

Version 1 of the scoring model demonstrated an Area Under the Curve (AUC) of 0.991, with a sensitivity of 90.91% and a specificity of 98.10%. Version 2 achieved an AUC of 0.999, with a sensitivity of 96.4% and a specificity of 92.44%.

The developed models provide a reliable, region specific tool for early detection of neonatal sepsis. Further validation across diverse populations and the integration of emerging diagnostic technologies, such as biomarkers and artificial intelligence, are warranted to enhance their applicability and accuracy.

OBJECTIVE: To analyze the clinical characteristics of elderly patients with sepsis, identify the key factors affecting their clinical outcomes, construct a death risk assessment scale for elderly patients with sepsis, and evaluate its predictive value. METHODS: A retrospective case-control study was conducted. The clinical data of sepsis patients admitted to intensive care unit (ICU) of the First Affiliated Hospital of Wenzhou Medical University from September 2021 to September 2023 were collected, including basic information, clinical characteristics, and clinical outcomes. The patients were divided into non-elderly group (age ≥ 65 years old) and elderly group (age < 65 years old) based on age. Additionally, the elderly patients were divided into survival group and death group based on their 30-day survival status. The clinical characteristics of elderly patients with sepsis were analyzed. Univariate and multivariate Logistic regression analyses were used to screen the independent risk factors for 30-day death in elderly patients with sepsis, and the regression equation was constructed. The regression equation was simplified, and the death risk assessment scale was established. The predictive value of different scores for the prognosis of elderly patients with sepsis was compared. RESULTS: (1) A total of 833 patients with sepsis were finally enrolled, including 485 in the elderly group and 348 in the non-elderly group. Compared with the non-elderly group, the elderly group showed significantly lower counts of lymphocyte, T cell, CD8⁺ T cell, and the ratio of T cells and CD8⁺ T cells [lymphocyte count (×10⁹/L): 0.71 (0.43, 1.06) vs. 0.83 (0.53, 1.26), T cell count (cells/μL): 394.0 (216.0, 648.0) vs. 490.5 (270.5, 793.0), CD8⁺ T cell count (cells/μL): 126.0 (62.0, 223.5) vs. 180.0 (101.0, 312.0), T cell ratio: 0.60 (0.48, 0.70) vs. 0.64 (0.51, 0.75), CD8⁺ T cell ratio: 0.19 (0.13, 0.28) vs. 0.24 (0.16, 0.34), all P < 0.01], higher natural killer cell (NK cell) count, acute physiology and chronic health evaluation II (APACHE II) score, ratio of invasive mechanical ventilation (IMV) during hospitalization, and 30-day mortality [NK cell count (cells/μL): 112.0 (61.0, 187.5) vs. 95.0 (53.0, 151.0), APACHE II score: 16.00 (12.00, 21.00) vs. 13.00 (8.00, 17.00), IMV ratio: 40.6% (197/485) vs. 31.9% (111/348), 30-day mortality: 28.9% (140/485) vs. 19.5% (68/348), all P < 0.05], and longer length of ICU stay [days: 5.5 (3.0, 10.0) vs. 5.0 (3.0, 8.0), P < 0.05]. There were no statistically significant differences in the levels of inflammatory markers such as C-reactive protein (CRP), procalcitonin (PCT), tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ), and interleukins (IL-2, IL-4, IL-6, IL-10) between the two groups. (2) In 485 elderly patients with sepsis, 345 survived in 30 days, and 140 died with the 30-day mortality of 28.9%. Compared with the survival group, the patients in the death group were older, and had lower body mass index (BMI), white blood cell count (WBC), PCT, platelet count (PLT) and higher IL-6, IL-10, N-terminal pro-brain natriuretic peptide (NT-proBNP), total bilirubin (TBil), blood lactic acid (Lac), and ratio of in-hospital IMV and continuous renal replacement therapy (CRRT). Multivariate Logistic regression analysis indicated that BMI [odds ratio (OR) = 0.783, 95% confidence interval (95%CI) was 0.678-0.905, P = 0.001], IL-6 (OR = 1.073, 95%CI was 1.004-1.146, P = 0.036), TBil (OR = 1.009, 95%CI was 1.000-1.018, P = 0.045), Lac (OR = 1.211, 95%CI was 1.072-1.367, P = 0.002), and IMV during hospitalization (OR = 6.181, 95%CI was 2.214-17.256, P = 0.001) were independent risk factors for 30-day death in elderly patients with sepsis, and the regression equation was constructed (Logit P = 1.012-0.244×BMI+0.070×IL-6+0.009×TBil+0.190×Lac+1.822×IMV). The regression equation was simplified to construct a death risk assessment scale, namely BITLI score. Receiver operator characteristic curve (ROC curve) analysis showed that the area under the ROC curve (AUC) of BITLI score for predicting death risk was 0.852 (95%CI was 0.769-0.935), and it was higher than APACHE II score (AUC = 0.714, 95%CI was 0.623-0.805) and sequential organ failure assessment (SOFA) score (AUC = 0.685, 95%CI was 0.578-0.793). The determined cut-off value of BITLI score was 1.50, while achieving a sensitivity of 83.3% and specificity of 74.0%. CONCLUSIONS Elderly patients with sepsis often have reduced lymphocyte counts, severe conditions, and poor prognosis. BMI, IL-6, TBil, Lac, and IMV during hospitalization were independent risk factors for 30-day death in elderly patients with sepsis. The BITLI score constructed based above risk factors is more precise and reliable than traditional APACHE II and SOFA scores in predicting the outcomes of elderly patients with sepsis.
Humans ; Sepsis/mortality* ; Aged ; Retrospective Studies ; Risk Assessment ; Case-Control Studies ; Prognosis ; Male ; Female ; Intensive Care Units ; Risk Factors ; Aged, 80 and over ; Logistic Models ; Middle Aged

OBJECTIVE: To explore the clinical features of the sepsis in immunocompromised hosts and establish an early warning equation. METHODS: A retrospective study was conducted on sepsis patients admitted to the intensive care unit (ICU) of Binzhou Medical University Hospital from October 2011 to October 2022. General information, infection site, etiology results and drug susceptibility, clinical symptoms, inflammatory indicators, acute physiology and chronic health status evaluation II (APACHE II), sequential organ failure assessment (SOFA), incidence of immune paralysis, and outcome during hospitalization were collected. Based on whether they met the diagnostic criteria for immunocompromised hosts, patients were divided into immunocompromised group and immune normal group. The clinical information of the two groups were compared. Multivariate Logistic regression was used to analyze the risk factors of patients with immunocompromised sepsis and the regression equation model was initially established. Omnibus test and Hosmer-Lemeshow test were used to evaluate the model. RESULTS: A total of 169 patients with sepsis were included, including 61 in the immunocompromised group and 108 in the normal immune group. The top 3 infection sites in the immunocompromised group were bloodstream infection, pulmonary infection and abdominal infection. The top 3 infection sites in the normal immune group were pulmonary infection, bloodstream infection and abdominal infection. The infection rate of Gram-negative bacteria in the immunocompromised group was significantly lower than that in the normal group [49.2% (30/61) vs. 64.8% (70/108), P < 0.05]. The infection rate of Gram-positive bacteria [27.9% (17/61) vs. 13.9% (15/108)] and multidrug-resistant bacteria [54.1% (33/61) vs. 29.6% (32/108)] were significantly higher than those in normal immune group (both P < 0.05). In terms of clinical symptoms, the proportion of fever in the immunocompromised group was significantly lower than that in the immune normal group [49.2% (30/61) vs. 66.7% (72/108), P < 0.05]. Neutrophil count (NEU) and neutrophil percentage (NEU%) in the immunocompromised group were significantly lower than those in the normal immune group. Lymphocyte percentage (LYM%), neutrophil/lymphocyte ratio (NLR), C-reactive protein (CRP), procalcitonin (PCT), APACHE II score, combined shock rate, incidence of immune paralysis, and mortality during hospitalization in the immunocompromised group were significantly higher than those in the normal immune group. Logistic regression analysis showed that NLR, CRP and PCT were risk factors for patients with immunocompromised sepsis (all P < 0.05). The above indicators were used as covariables to construct a Logistic regression equation, that was, Logit (P) = 0.025X₁+0.010X₂+0.013X₃-2.945, where X₁, X₂ and X₃ represent NLR, CRP and PCT respectively. Omnibus test and Hosmer-Lemeshow test show that the model fits well and has certain early warning value. CONCLUSIONS Patients with immunocompromised sepsis have more intense inflammatory response, with Gram-negative bacteria being the predominant pathogen, and a higher incidence of Gram-positive bacterial infections and multi-drug resistant infections. The severity of the disease, in-hospital mortality, the incidence of shock and the incidence of immune paralysis after sepsis were significantly higher. NLR, CRP and PCT were independent risk factors for sepsis in immunocompromised hosts. The regression equation constructed based on this may have early warning significance for patients with immunocompromised sepsis.
Humans ; Sepsis/immunology* ; Immunocompromised Host ; Retrospective Studies ; Risk Factors ; Intensive Care Units ; Logistic Models ; Male ; APACHE ; Female ; Middle Aged ; Aged

OBJECTIVE: To investigate the clinical characteristics and related risk factors of Clostridium difficile infection (CDI) in intensive care unit (ICU). METHODS: A retrospective study was conducted. Patients with diarrhea admitted to the ICU of the General Hospital of Ningxia Medical University from May 1 to August 30, 2023 were selected. Patients were divided into CDI group and non-CDI group based on the presence or absence of CDI. Clinical data from two groups of patients meeting the criteria were collected and compared, including gender, age, acute physiology and chronic health evaluation II (APACHE II), length of hospital stay, serum lactic acid, parenteral nutrition time, white blood cell count (WBC), procalcitonin (PCT), C-reactive protein (CRP), coagulation indicators, albumin, antibiotic exposure, etc. Multivariate Logistic regression analysis was performed to analyze the risk factors for CDI in ICU diarrhea patients. Receiver operator characteristic curve (ROC curve) was drawn to analyze the predictive value of each index for CDI in diarrhea patients. RESULTS: A total of 24 patients with diarrhea were enrolled, including 9 patients in the CDI group and 15 patients in the non-CDI group. The time of parenteral nutrition in the CDI group was significantly longer than that in the non-CDI group [days: 18.0 (13.5, 19.5) vs. 10.0 (4.0, 18.0)], the serum lactic acid level [mmol/L: 4.40 (3.00, 15.25) vs. 2.50 (1.90, 3.20)] and the ratio of serum lactic acid > 3.9 mmol/L [66.67% (6/9) vs. 6.67% (1/15)] were significantly higher than those in the non-CDI group, with statistical significance (all P < 0.05). Multivariate binary Logistic regression analysis showed that the serum lactic acid level of the patients was an independent risk factor for CDI [odds ratio (OR) = 3.193, 95% confidence interval (95%CI) was 1.011-10.080, P = 0.048]. ROC curve showed that serum lactic acid level had a high predictive value for CDI in ICU patients with diarrhea, and the area under the curve (AUC) was 0.815, respectively. When the cut-off value of serum lactic acid was 3.9 mmol/L, the sensitivity was 66.7% and the specificity was 93.3%. CONCLUSION Patients with diarrhea who have higher serum lactate levels (> 3.9 mmol/L) on admission are at increased risk of developing CDI.
Humans ; Retrospective Studies ; Risk Factors ; Intensive Care Units ; Clostridium Infections ; Clostridioides difficile ; Male ; Female ; Middle Aged ; Aged ; Diarrhea/microbiology* ; Logistic Models ; ROC Curve ; Adult

OBJECTIVE: To investigate the predictive value of inflammatory indicator and serum cystatin C (Cys C) for the prognosis of patients with sepsis-associated acute kidney injury (SA-AKI). METHODS: A prospective observational study was conducted. Patients with SA-AKI admitted to the intensive care unit (ICU) of the General Hospital of Ningxia Medical University from January 2022 to December 2023 were selected as the study subjects. General patient data, sequential organ failure assessment (SOFA), acute physiology and chronic health evaluation II (APACHE II), inflammatory indicator, and serum Cys C levels were collected. The 28-day survival status of the patients was observed. A multivariate Logistic regression model was used to analyze the risk factors affecting the poor prognosis of SA-AKI patients. Receiver operator characteristic curve (ROC curve) was plotted to evaluate the predictive efficacy of each risk factor for the prognosis of SA-AKI patients. RESULTS: A total of 111 SA-AKI patients were included, with 65 patients (58.6%) in the survival group and 46 patients (41.4%) in the death group. The SOFA score, APACHE II score, interleukin-6 (IL-6), procalcitonin (PCT), hypersensitive C-reactive protein (hs-CRP), and serum Cys C levels in the death group were significantly higher than those in the survival group [SOFA score: 15.00 (14.00, 17.25) vs. 14.00 (11.00, 16.00), APACHE II score: 26.00 (23.75, 28.00) vs. 23.00 (18.50, 28.00), IL-6 (ng/L): 3 731.00±1 573.61 vs. 2 087.93±1 702.88, PCT (μg/L): 78.19±30.35 vs. 43.56±35.37, hs-CRP (mg/L): 266.50 (183.75, 326.75) vs. 210.00 (188.00, 273.00), serum Cys C (mg/L): 2.01±0.61 vs. 1.62±0.50, all P < 0.05]. Multivariate Logistic regression analysis showed that SOFA score [odds ratio (OR) = 1.273, 95% confidence interval (95%CI) was 1.012-1.600, P = 0.039], IL-6 (OR = 1.000, 95%CI was 1.000-1.001, P = 0.043), PCT (OR = 1.018, 95%CI was 1.002-1.035, P = 0.030), and Cys C (OR = 4.139, 95%CI was 1.727-9.919, P = 0.001) were independent risk factors affecting the 28-day prognosis of SA-AKI patients. ROC curve analysis showed that the area under the curve (AUC) of SOFA score, IL-6, PCT, and Cys C in predicting the 28-day prognosis of SA-AKI patients were 0.682 (95%CI was 0.582-0.782, P = 0.001), 0.753 (95%CI was 0.662-0.843, P < 0.001), 0.765 (95%CI was 0.677-0.854, P < 0.001), and 0.690 (95%CI was 0.583-0.798, P = 0.001), respectively. The combined predictive value of these four indicators for the prognosis of SA-AKI patients were superior to that of any single indicator, with an AUC of 0.847 (95%CI was 0.778-0.916, P < 0.001), a sensitivity of 95.7%, and a specificity of 56.9%. CONCLUSION The combination of SOFA score, IL-6, PCT, and Cys C provides a reliable predictive value for the prognosis of SA-AKI patients.
Humans ; Acute Kidney Injury/mortality* ; APACHE ; C-Reactive Protein ; Cystatin C/blood* ; Interleukin-6/blood* ; Logistic Models ; Predictive Value of Tests ; Procalcitonin/blood* ; Prognosis ; Prospective Studies ; Risk Factors ; ROC Curve ; Sepsis/mortality*

OBJECTIVE: To investigate the relationship between mean perfusion pressure (MPP) and the risk of sepsis-associated acute kidney injury (SA-AKI) and its prognosis, and to determine the optimal cut-off value of MPP for predicting SA-AKI. METHODS: A retrospective cohort study was conducted. The clinical data of adult patients with sepsis were collected from the Medical Information Mart for Intensive Care-IV 2.2 (MIMIC-IV 2.2) database. The patients were divided into two groups based on the occurrence of SA-AKI. Baseline characteristics, vital signs, comorbidities, laboratory indicators within 24 hours of intensive care unit (ICU) admission, and clinical outcome indicators were collected. Mean MPP was calculated using the average values of mean arterial pressure (MAP) and central venous pressure (CVP), MPP = MAP-CVP. Cox regression models were constructed, relevant confounding factors were adjusted, and multivariate Logistic regression analysis was used to investigate the associations between MPP and the risk of SA-AKI as well as ICU death. The predictive value of MPP for SA-AKI was evaluated using receiver operator characteristic curve (ROC curve) analysis, and the optimal cut-off value was determined. RESULTS: A total of 6 009 patients were ultimately enrolled in the analysis. Among them, SA-AKI occurred in 4 755 patients (79.13%), while 1 254 patients (20.87%) did not develop SA-AKI. Compared with the non-SA-AKI group, the MPP in the SA-AKI group was significantly lowered [mmHg (1 mmHg≈0.133 kPa): 62.00 (57.00, 68.00) vs. 65.00 (60.00, 70.00), P < 0.01], and the ICU mortality was significantly increased [11.82% (562/4 755) vs. 1.59% (20/1 254), P < 0.01]. Three Cox regression models were constructed: model 1 was unadjusted; model 2 was adjusted for gender, age, height, weight and race; model 3 was adjusted for gender, age, height, weight, race, heart rate, respiratory rate, body temperature, hemoglobin, platelet count, white blood cell count, anion gap, HCO₃^-, blood urea nitrogen, serum creatinine, Cl^-, Na⁺, K⁺, fibrinogen, international normalized ratio, blood lactic acid, pH value, arterial partial pressure of oxygen, arterial partial pressure of carbon dioxide, sequential organ failure assessment score, Charlson comorbidity index score, use of vasopressors, mechanical ventilation, and urine output. Multivariate Logistic regression analysis showed that when MPP was treated as a continuous variable, there was a negative correlation between MPP and the risk of SA-AKI in model 1 and model 2 [model 1: odds ratio (OR) = 0.967, 95% confidence interval (95%CI) was 0.961-0.974, P < 0.001; model 2: OR = 0.981, 95%CI was 0.974-0.988, P < 0.001], and also a negative correlation between MPP and the risk of ICU death (model 1: OR = 0.955, 95%CI was 0.945-0.965, P < 0.001; model 2: OR = 0.956, 95%CI was 0.946-0.966, P < 0.001). However, in model 3, there was no significant correlation between MPP and either SA-AKI risk or ICU death risk. when MPP was used as a multi-categorical variable, in model 1 and model 2, referring to MPP ≤ 58 mmHg, when 59 mmHg ≤ MPP ≤ 68 mmHg, as MPP increased, the risk of SA-AKI progressively decreased (OR value was 0.411-0.638, all P < 0.001), and the risk of ICU death also gradually decreased (OR value was 0.334-0.477, all P < 0.001). ROC curve showed that MPP had a certain predictive value for SA-AKI occurrence [area under the ROC curve (AUC) = 0.598, 95%CI was 0.404-0.746], and the optimal cut-off value was 60.5 mmHg. CONCLUSION MPP was significantly associated with the risk of SA-AKI, with an optimal cut-off value of 60.5 mmHg, and also demonstrated a significant correlation with the risk of ICU death.
Humans ; Acute Kidney Injury/physiopathology* ; Retrospective Studies ; Sepsis/physiopathology* ; Middle Aged ; Prognosis ; Male ; Female ; Aged ; Risk Factors ; Intensive Care Units ; Adult ; Logistic Models ; Proportional Hazards Models

OBJECTIVE: To identify early potential risk factors for hospital-acquired pneumonia (HAP) in patients with mild traumatic brain injury (mTBI), construct a risk prediction model, and evaluate its predictive efficacy. METHODS: A case-control study was conducted using clinical data from mTBI patients admitted to the neurosurgery department of Changzhou Second People's Hospital from September 2021 to September 2023. The patients were divided into two groups based on whether they developed HAP. Clinical data within 48 hours of admission were statistically analyzed to identify factors influencing HAP occurrence through univariate analysis. Least absolute shrinkage and selection operator (LASSO) regression analysis was employed for feature selection to identify the most influential variables. The dataset was divided into training and validation sets in a 7:3 ratio. A multivariate Logistic regression analysis was then performed using the training set to construct the prediction model, exploring the risk factors for HAP in mTBI patients and conducting internal validation in the validation set. Receiver operator characteristic curve (ROC curve), decision curve analysis (DCA), and calibration curve were utilized to assess the sensitivity, specificity, decision value, and predictive accuracy of the prediction model. RESULTS: A total of 677 mTBI patients were included, with 257 in the HAP group and 420 in the non-HAP group. The significant differences were found between the two groups in terms of age, maximum body temperature (MaxT), maximum heart rate (MaxHR), maximum systolic blood pressure (MaxSBP), minimum systolic blood pressure (MinSBP), maximum respiratory rate (MaxRR), cause of injury, and laboratory indicators [C-reactive protein (CRP), procalcitonin (PCT), neutrophil count (NEUT), erythrocyte sedimentation rate (ESR), fibrinogen (FBG), fibrinogen equivalent units (FEU), prothrombin time (PT), activated partial thromboplastin time (APTT), total cholesterol (TC), lactate dehydrogenase (LDH), prealbumin (PAB), albumin (Alb), blood urea nitrogen (BUN), serum creatinine (SCr), hematocrit (HCT), hemoglobin (Hb), platelet count (PLT), glucose (Glu), K⁺, Na⁺], suggesting they could be potential risk factors for HAP in mTBI patients. After LASSO regression analysis, the key risk factors were enrolled in the multivariate Logistic regression analysis. The results revealed that the cause of injury being a traffic accident [odds ratio (OR) = 2.199, 95% confidence interval (95%CI) was 1.124-4.398, P = 0.023], NEUT (OR = 1.330, 95%CI was 1.214-1.469, P < 0.001), ESR (OR = 1.053, 95%CI was 1.019-1.090, P = 0.003), FBG (OR = 0.272, 95%CI was 0.158-0.445, P < 0.001), PT (OR = 0.253, 95%CI was 0.144-0.422, P < 0.001), APTT (OR = 0.689, 95%CI was 0.578-0.811, P < 0.001), Alb (OR = 0.734, 95%CI was 0.654-0.815, P < 0.001), BUN (OR = 0.720, 95%CI was 0.547-0.934, P = 0.016), and Na⁺ (OR = 0.756, 95%CI was 0.670-0.843, P < 0.001) could serve as main risk factors for constructing the prediction model. Calibration curves demonstrated good calibration of the prediction model in both training and validation sets with no evident over fitting. ROC curve analysis showed that the area under the ROC curve (AUC) of the prediction model in the training set was 0.943 (95%CI was 0.921-0.965, P < 0.001), with a sensitivity of 83.6% and a specificity of 91.5%. In the validation set, the AUC was 0.917 (95%CI was 0.878-0.957, P < 0.001), with a sensitivity of 90.1% and a specificity of 85.0%. DCA indicated that the prediction model had a high net benefit, suggesting practical clinical applicability. CONCLUSIONS The cause of injury being a traffic accident, NEUT, ESR, FBG, PT, APTT, Alb, BUN, and Na⁺ are identified as major risk factors influencing the occurrence of HAP in mTBI patients. The prediction model constructed using these parameters effectively assesses the likelihood of HAP in mTBI patients.
Humans ; Risk Factors ; Case-Control Studies ; Logistic Models ; Healthcare-Associated Pneumonia/epidemiology* ; Brain Injuries, Traumatic/complications* ; Male ; Female ; ROC Curve ; Pneumonia/etiology* ; Middle Aged ; Adult

OBJECTIVE: To analyze the factors influencing the development of refeeding syndrome (RFS) in patients with sepsis and its impact on clinical prognosis. METHODS: A retrospective case-control study method was used to collect the clinical data of patients with sepsis admitted to the intensive care unit (ICU) of Qingdao Municipal Hospital from December 2018 to December 2023. The patients were divided into RFS and non-RFS groups according to whether RFS occurred, and the basic data, nutritional status and assessment scale, laboratory indicators, nutritional intake, medical history and prognosis were compared between the two groups. Binary multifactorial Logistic regression analysis was used to screen the influencing factors of the occurrence of RFS in patients with sepsis. RESULTS: A total of 544 patients with sepsis were finally enrolled, of whom 250 did not develop RFS and 294 developed RFS, with an incidence of 54.0%. Compared with the non-RFS group, the patients in the RFS group had lower body mass index (BMI), albumin, prealbumin, baseline electrolytes (serum phosphorus, serum potassium, and serum magnesium), creatinine-height index, and protein intake, and had higher nutritional risk screening 2002 (NRS2002) score, sequential organ failure assessment (SOFA) score, calorie intake, and the proportions of feedings during the 48 hours of ICU admission, history of diabetes and septic shock. Binary multifactorial Logistic regression analysis showed that BMI [odds ratio (OR) = 0.910, 95% confidence interval (95%CI) was 0.857-0.947, P < 0.001], SOFA score (OR = 1.166, 95%CI was 1.085-1.254, P < 0.001), albumin (OR = 0.946, 95%CI was 0.902-0.991, P = 0.019), baseline serum phosphorus (OR = 0.343, 95%CI was 0.171-0.689, P = 0.003), baseline serum potassium (OR = 0.531, 95%CI was 0.377-0.746, P < 0.001), creatinine-height index (OR = 0.891, 95%CI was 0.819-0.970, P = 0.008), caloric intake (OR = 1.108, 95%CI was 1.043-1.178, P = 0.001), protein intake (OR = 0.107, 95%CI was 0.044-0.260, P < 0.001), and feedings during the 48 hours of ICU admission (OR = 0.592, 95%CI was 0.359-0.977, P = 0.040) and septic shock (OR = 0.538, 95%CI was 0.300-0.963, P = 0.037) were independent influence factors on the occurrence of RFS in septic patients. Of the 544 patients, 267 died at 28 days, with a mortality of 49.1%. The 28-day mortality of patients in the RFS group was significantly higher than that in the non-RFS group [54.4% (160/294) vs. 42.8% (107/250); χ² = 7.302, P = 0.007]. 544 patients had a length of ICU stay of 20 (17, 24) days. The patients in the RFS group had a significantly longer length of ICU stay than that in the non-RFS group [days: 20 (17, 25) vs. 19 (17, 23); Z = -2.312, P = 0.021]. CONCLUSIONS The incidence of RFS in septic patients is high. Factors influencing the occurrence of RFS in septic patients include BMI, SOFA score, albumin, baseline serum phosphorus, baseline serum potassium, caloric intake, protein intake, feeding within 48 hours of ICU admission, and septic shock. RFS prolongs the length of ICU stay and increases the 28-day mortality in patients with sepsis.
Humans ; Retrospective Studies ; Sepsis/complications* ; Prognosis ; Refeeding Syndrome/etiology* ; Case-Control Studies ; Intensive Care Units ; Male ; Nutritional Status ; Female ; Risk Factors ; Middle Aged ; Logistic Models ; Body Mass Index ; Aged

OBJECTIVE: To explore the relationship between serum sodium level and the risk of delirium in patients with sepsis. METHODS: Based on the Medical Information Mart for Intensive Care-IV (MIMIC-IV), adult patients with sepsis in the intensive care unit (ICU) were enrolled. The serum sodium level prior to the onset of sepsis during hospitalization was used as the exposure variable. Delirium was assessed using the ICU-confusion assessment method (ICU-CAM) as the primary outcome. Patients were divided into delirium and non-delirium groups based on the occurrence of delirium. The relationship between serum sodium level and delirium risk was described using restricted cubic spline (RCS) to determine the optimal reference range for serum sodium. Logistic regression analysis was used to evaluate the effect of blood sodium levels on delirium in sepsis patients. Subgroup analyses were performed to explore potential interactions and further validate the robustness of the results. Receiver operator characteristic curve (ROC curve) analysis was performed to assess the predictive value of serum sodium level for delirium occurrence in patients with sepsis. RESULTS: A total of 13 889 patients with sepsis were included, of which 4 831 experienced delirium. The maximum and mean serum sodium values were significantly higher in the delirium group compared to the non-delirium group, while there were no statistically significant differences in terms of initial and minimum serum sodium values between the two groups. Compared with the non-delirium group, the delirium group had a higher mortality and longer hospital stay. The RCS curve showed that a "U"-shaped relationship between serum sodium level and delirium risk in patients with sepsis, with the optimal reference range for average serum sodium was 135.3-141.3 mmol/L. Group based on this reference range, compared to the group with 135.3 mmol/L ≤ serum sodium ≤ 141.3 mmol/L, the delirium incidence and mortality were significantly higher, and the hospital stay was longer in the groups with serum sodium < 135.3 mmol/L and serum sodium ≥ 141.3 mmol/L [delirium incidence: 36.92%, 40.88% vs. 31.22%; 28-day mortality: 23.08%, 20.15% vs. 13.39%; 90-day mortality: 30.75%, 24.81% vs. 18.26%; in-hospital mortality: 19.53%, 17.48% vs. 11.61%; ICU mortality: 14.35%, 14.05% vs. 9.00%; hospital length of stay (days): 10.1 (6.1, 17.7), 9.4 (5.4, 17.0) vs. 8.9 (5.5, 15.4), length of ICU stay (days): 3.7 (2.1, 7.1), 4.0 (2.1, 8.9) vs. 3.2 (1.9, 6.8); all P < 0.01]. Logistic regression analysis showed that, in the initial model and each factor-adjusted models, compared to the reference group with 135.3 mmol/L ≤ serum sodium < 141.3 mmol/L, serum sodium < 135.3 mmol/L increased the risk of delirium in septic patients by 21% to 29% [odds ratio (OR) was 1.21-1.29, all P < 0.01], while serum sodium ≥ 141.3 mmol/L increased the delirium risk by 28%-52% (OR was 1.28-1.52, all P < 0.01). Subgroup analyses based on gender, age, race, diuretic use, and sequential organ failure assessment (SOFA) score revealed there was no significant interactions between subgroup variables and serum sodium, and the results supported that both serum sodium < 135.3 mmol/L and serum sodium ≥ 141.3 mmol/L were risk factors for delirium in septic patients. ROC curve analysis showed that the area under the curve (AUC) for predicting delirium in septic patients based on serum sodium was 0.614, with a cut-off value of 139.5 mmol/L yielding a specificity of 67.5% and sensitivity of 50.9%. CONCLUSIONS The risk of delirium in patients with sepsis is associated with serum sodium level in a "U"-shaped manner. Both high and low serum sodium levels are associated with increased risk of delirium, higher all-cause mortality, and prolonged hospital stays in patients with sepsis. Abnormal serum sodium levels may have predictive value for sepsis-associated delirium and could serve as an early biomarker for identifying delirium in septic patients, although further validation is needed.
Humans ; Delirium/etiology* ; Sepsis/complications* ; Sodium/blood* ; Intensive Care Units ; Risk Factors ; Male ; Middle Aged ; Female ; Aged ; Logistic Models ; Adult

OBJECTIVE: To identify the optimal machine learning algorithm for predicting post-stroke epilepsy (PSE) within one year following acute stroke, establish a nomogram model based on this algorithm, and perform external validation to achieve accurate prediction of secondary epilepsy. METHODS: A total of 870 acute stroke patients admitted to the emergency department of Xiang'an Hospital of Xiamen University from June 2019 to June 2023 were enrolled for model development (model group). An external validation cohort of 435 acute stroke patients admitted to the Fifth Hospital of Xiamen during the same period was used to validate the machine learning algorithms and nomogram model. Patients were classified into control and epilepsy groups based on the development of PSE within one year. Clinical and laboratory data, including baseline characteristics, stroke location, vascular status, complications, hematologic parameters, and National Institutes of Health Stroke Scale (NIHSS) score, were collected for analysis. Nine machine learning algorithms such as logistic regression, CN2 rule induction, K-nearest neighbors, adaptive boosting, random forest, gradient boosting, support vector machine, naive Bayes, and neural network were applied to evaluate predictive performance. The area under the curve (AUC) of receiver operator characteristic curve (ROC curve) was used to identify the optimal algorithm. Logistic regression was used to screen risk factors for PSE, and the top 10 predictors were selected to construct the nomogram model. The predictive performance of the model was evaluated using the ROC curve in both the model and validation groups. RESULTS: Among the 870 patients in the model group, 29 developed PSE within one year. Among the nine algorithms tested, logistic regression demonstrated the best performance and generalizability, with an AUC of 0.923. Univariate logistic regression identified several risk factors for PSE, including platelet count, white blood cell count, red blood cell count, glycated hemoglobin (HbA1c), C-reactive protein (CRP), triglycerides, high-density lipoprotein (HDL), aspartate aminotransferase (AST), alanine aminotransferase (ALT), activated partial thromboplastin time (APTT), thrombin time, D-dimer, fibrinogen, creatine kinase (CK), creatine kinase-MB (CK-MB), lactate dehydrogenase (LDH), serum sodium, lactic acid, anion gap, NIHSS score, brain herniation, periventricular stroke, and carotid artery plaque. Further multivariate logistic regression analysis showed that white blood cell count, HDL, fibrinogen, lactic acid and brain herniation were independent risk factors [odds ratio (OR) were 1.837, 198.039, 47.025, 11.559, 70.722, respectively, all P < 0.05]. In the external validation group, univariate logistic regression analysis showed that platelet count, white blood cell count, CRP, triacylglycerol, APTT, D-dimer, fibrinogen, CK, CK-MB, LDH, NIHSS score, and cerebral herniation were risk factors for PSE one year after acute stroke. Further multiple logistic regression analysis showed that APTT and cerebral herniation were independent predictors (OR were 0.587 and 116.193, respectively, both P < 0.05). The nomogram model, constructed using 10 key variables-brain herniation, periventricular stroke, carotid artery plaque, white blood cell count, triglycerides, thrombin time, D-dimer, serum sodium, lactic acid, and NIHSS score-achieved an AUC of 0.908 in the model group and 0.864 in the external validation group. CONCLUSIONS The logistic regression-based prediction model for epilepsy one year after acute stroke, developed using machine learning algorithms, showed optimal predictive performance. The nomogram model based on the logistic regression-derived predictors showed strong discriminative power and was successfully validated externally, suggesting favorable clinical applicability and generalizability.
Humans ; Machine Learning ; Stroke/complications* ; Nomograms ; Epilepsy/etiology* ; Algorithms ; Male ; Female ; Logistic Models ; Middle Aged ; Aged ; Risk Factors ; Bayes Theorem