Bedside prediction scoring for emergent diagnosis of late onset neonatal sepsis

ABSTRACT

@#<p style="text-align justify;"><strong>BACKGROUND:</strong> Late Onset Neonatal Sepsis (LONS) or nosocomial sepsis has a significant mortality and morbidity that leads to overtreatment. Overtreatment happens when antibiotics are frequently started and/or shifted, eventually leading to increasing antimicrobial resistamce in NICU.<br /><strong>OBJECTIVE: </strong>To validate bedside nosocomial sepsis scoringdeveloped by Okascharoen in 2005.<br /><strong>METHODOLOGY:</strong> All neonates admitted in NICU suspected of LONS were enrolled. Using Okascharoe scoring, subjects were scored based hypotension/poor perfusion, abnormal body temperature, respiratory insufficiency, complete blood count, and length of umbilical catheter use. Growth of organisms during blood culture is considered positive outcome and is considered confirmed sepsis.<br /><strong>RESULTS:</strong> Of the one-hundred-nineteen (119) subjects included in the analysis, 59 were confirmed sepsis and 60 were LONS negative. Subjects with confirmed sepsis had more events of hypotension/poor perfusion (p < 0.001; -0.141, -0.438), thrombocytopenia (p 0.000; -0.169, -0489), and prolonged umbilical catheter usage (p 0.014; -0.051, -0.311). The ROC curve has an AUC of 0.753 (p < 0.001; 0.664-0.842), which means a randomly chosen neonate with LONS will have a higher predicted score than a neonate without LONS. The sensitivity of this tool was 0.92 (0.82-0.97) and specificity of this tool was 0.32 (0.21-0.46) in this setting. The positive LR =1.35 (1.12-1.64) while the negative LR = 0.26 (0.10-0.65)<br /><strong>CONCLUSION:</strong> This scoring is a valid tool that can be used in point-of-care scoring for antibiotic stewardship in a neonate with suspected sepsis.<br /><strong>RECOMMENDATION</strong> it is recommended that a score > 5 be used to be predictive of late onset sepsis, and this would have sensitivity of 83.3%, specificity of 61%, positive predictive value of 68.5% and a negative predictive value of 78.3%</p>