Clinical Prediction Model Development and Validation for the Detection of Newborn Sepsis, Diagnostic Research Protocol

BACKGROUND: Neonatal sepsis is a leading cause of sickness and death in the entire world. Diagnosis is usually difficult because of the nonspecific clinical symptoms and the paucity of laboratory diagnostics in many low- and middle-income nations (LMICs). Clinical prediction models may increase diagnostic precision and rationalize the use of antibiotics in neonatal facilities, which could lead to a decrease in antimicrobial resistance and better neonatal outcomes. Early detection of newborn sepsis is critical to prevent serious consequences and reduce the need for unneeded drugs. OBJECTIVE: The aim is to develop and validate a clinical prediction model for the detection of newborn sepsis. METHODS: A cross-sectional study based on an institution will be carried out. The sample size was determined by assuming 10 events per predictor, based on this assumption, the total sample sizes were 467. Data will be collected using a structured checklist through chart review. Data will be coded, inputted, and analyzed using R statistical programming language version 4.0.4 after being entered into Epidata version 3.02 and further processed and analyzed. Bivariable logistic regression will be done to identify the relationship between each predictor and neonatal sepsis. In a multivariable logistic regression model, significant factors (P< 0.05) will be kept, while variables with (P< 0.25) from the bivariable analysis will be added. By calculating the area under the ROC curve (discrimination) and the calibration plot (calibration), respectively, the model’s accuracy and goodness of fit will be evaluated.