| Objective:To study the clinical characteristics of pediatric sepsis, explore the influential factors of sepsis severity; develop and validate a new scoring model to stratify the severity of pediatric sepsis, thus assisting the diagnosis and treatment of sepsis in children.Methods:②The study was retrospectively performed at Children’s hospital in Hunan province. Children who presented to the pediatric intensive care units (PICU) with sepsis were recruited and randomly divided into training group and validation group.￢ariables associated with the severity of sepsis were selected by student’s t-test, Mann-Whitney U-test, and Pearson’s chi-square test. Stepwise discriminant analysis was used to develop the Fisher discriminate model for pediatric sepsis. A simplified scoring model was generated using weightings defined by the discriminate coefficients. The discriminant ability of the model was tested by sensitivity, specificity and receiver operating characteristic (ROC) curves.Results:②Basic characteristics:During the2-year study period (March2011to March2013),634patients admitted to PICU met the criteria of sepsis or severe sepsis, with476patients placed in training group and158patients in validation group. Of the634patients,231(36.4%) had severe sepsis.424(66.9%) patients improved or recovered,49(7.7%) gave up treatment and were discharged,161(25.4%) patients died. The average age was1.26±1.92years (range from1month to14years) and65.3%were male. There were no statistical difference in gender, age, length of hospital stay, mechanical ventilation and the proportion of patients with severe sepsis between the training group and validation group.②Model development:The discriminant analysis showed that prothrombin time (>14s), D-dimer (positive), total bilirubin (>6μmol/L), serum total protein (<60g/L), uric acid (>350μmol/L), PaO2/FiO2ratio (<300), myoglobin (>90μg/L) were associated with severity of sepsis. These seven variables were assigned with values of4,3,3,4,3,3,3points respectively based on the standardized canonical discriminant coefficients. The total score was23point. Patients with higher scores had higher risk of severe sepsis and a significantly higher fatality rate.③Model validation:The area under the receiver operating characteristic curve (AUC) for Fisher discriminate model was0.816(95%CI:0.771～0.861) in the training group and0.836(95%CI:0.765～0.907) in the validation group. We had the best sensitivity (0.76) and specificity (0.73) in training group at cut-off value0.24, the discriminate result remained well both in the validation group (sensitivity and specificity were0.78and0.75, respectively) and in the total sample (sensitivity and specificity were0.76and0.74, respectively). And for simplified scoring model, the AUC was0.800(95%CI:0.753～0.846) in the training group and0.825(95%CI:0.750～0.899) in the validation group. When the cut-off point was14, the model showed the best performance to discriminate severe sepsis with a sensitivity of0.67and specificity of0.81in the training group. The discriminate result remained well both in the validation group (sensitivity and specificity were0.71and0.80, respectively) and in the total sample (sensitivity and specificity were0.67and0.80, respectively).Conclusion:①The severity of pediatric sepsis is associated with prothrombin time, D-dimer, total bilirubin, serum total protein, uric acid, oxygenation index, and myoglobin levels within24hours of admission.②The present Fisher discriminant model and simplified scoring model both included7variables. For the simplified scoring model, prothrombin time (>14s), D-dimer (positive), total bilirubin (>6umol/L), serum total protein (<60g/L), uric acid (>350λmol/L), PaO2/FiO2ratio (<300), myoglobin (>90μg/L) were assigned with values of4,3,3,4,3,3,3, respectively.②The proposed disease severity scoring model for pediatric sepsis was simple and reliable, which has important clinical significance in evaluating the severity of pediatric sepsis and predicting its progress. |
PDF Full Text Request