<![CDATA[Background Neonatal sepsis remains a diagnostic challenge due to its nonspesific symptoms and signs. Blood culture as the gold standard is still a problem because it takes time, is expensive, and not every health facility is able to perionn.Objective To evaluate the diagnostic accuracy of clinical symptoms, hematologic findings, and C-reactive protein (CRP) in neonatal sepsis.Methods Samples were taken from potentially infected neonates admitted to the Matemal-Perinatal Unit of Sardjito Hospital, between December 1st, 2000 and March 31st, 2001 using at least one of the criteria: prematurity, very low birth weight infants, matemal pyrexia during delivery, premature membrane rupture, or thick, cloudy amniotic fluid. Clinical symptoms, total leukocyte, neutrophil, platelet count, CRP, and blood culture as the gold standard were examined.Results Among 99 neonates enrolled, the sensitivity, specificity, positive and negative predictive value of clinical symptoms were 79.3%, 75.7%, 57.5%, and 89.9%, respectively; leukopenia/leukocytosis were 27.6%, 85.7%, 44.4%, and 74.1%; neutropenia! neutrophilia were 41.4%, 71.4%, 37.5%, and 74.6%; thrombocytopenia were 79.3%, 51.8%, 40.4%, and 85.7%; positive CRP were 58.6%,78.6%,53.1%, and 82.1%. Parallel tests increased the sensitivity up to 89.7%. Specificity, positive and negative predictive value, and likelihood ratio were 44.3%, 40%, 91.2%, and 1.6, respectively. Serial tests increased the specificity up to 88.6%. Sensitivity, positive and negative predictive value, and likelihood ratio were 58.6%, 68%, 83.8%, and 5.1, respectively.Conclusion Clinical sepsis, thrombocytopenia, and CRP are sufficiently accurate as diagnostic tests for sepsis in potentially infected neonates. Parallel tests will increase the sensitivity, while serial tests increase the specificity.]]>
