• Shun Yuan Guo, Yin Zhou, Qing Feng Hu, Jiong Yao, and Hong Wang.
• Department of Infection Disease (YZ, HW), Zhejiang Provincial People's Hospital, Hangzhou, China; Department of Bacterial Laboratory, Institute of Clinical Laboratory Diagnostics (QFH, JY), Zhejiang Provincial People's Hospital, Hangzhou, China; and Department of Neurology (SYG), Zhejiang Provincial People's Hospital, Hangzhou, China.
• Am. J. Med. Sci. 2015 Jun 1; 349 (6): 499-504.

BackgroundPrediction of the species of pathogen among patients with sepsis within hours would be helpful in accelerating proper treatment. As a potential method of shortening the time to identification, this study considered the usefulness of measuring procalcitonin (PCT) to predict blood culture (BC) results.MethodsThe authors retrospectively analyzed the data of patients with a diagnosis of sepsis in their hospital from December 2012 to December 2013. The authors analyzed all diagnostic episodes consisting of BC and PCT concentration. The diagnostic performance of PCT to predict gram-negative bacteremia was tested using a receiver operative characteristic curve. Logistic regression was constructed using the presence of gram-negative bacteria as the dependent variable.ResultsA total of 262 diagnostic episodes met the inclusion criteria. According to BC classifications, a significantly higher value of PCT was observed in bloodstream infections caused by gram-negative bacteria (26.7 ng/mL, 0.09-188.3) than that in bloodstream infections caused by gram-positive bacteria (0.84 ng/mL, 0.05-18.79) or Candida spp. (1.12 ng/mL, 0.07-49.68). A cutoff value of ≥ 3.39 ng/mL for PCT showed a sensitivity of 80%, a specificity of 71%, a positive predictive value of 35%, a negative predictive value of 91% and an area under the curve of 0.73 for gram-negative bacteremia identification by BC. Among the 122 diagnostic episodes with positive BC results, a cutoff value of ≥ 6.47 ng/mL for PCT yielded a sensitivity of 74%, a specificity of 81%, a positive predictive value of 82%, a negative predictive value of 75% and an area under the curve of 0.81 for gram-negative bacteremia identification.ConclusionsPCT may represent a useful tool for differentiating gram-positive from gram-negative bloodstream infection with a significantly higher PCT level indicating gram-negative bacteremia.

23 keywords...

hide…