Competing interestsThe authors declare that they have no competing interests.Authors’ contributionsSJC, CCK and FOV performed genotyping Vorinostat and statistical analysis. SJC drafted the manuscript. SS, CEM, RJOD, NPD, NP, DWC, JAB, TNW and JAS enrolled patients, collected samples and data, and defined phenotypes. DWC, JAS and AVSH coordinated the study. SJC, CCK, FOV, AR, AW, DWC, JAB, TNW, JAS and AVSH contributed to the conception and design of the project. All authors read and approved the final manuscript.AcknowledgementsThe present study was supported by the Wellcome Trust, UK. SJC is a Wellcome Trust Clinical Research Fellow and is supported by the NIHR Biomedical Research Centre, Oxford. CCK is a scholar of the Agency for Science, Technology and Research (A-STAR), Singapore and member of the MBBS-PhD programme, Faculty of Medicine, National University of Singapore.

AR is supported by the EU FP6 GRACE grant and the Academy of Finland. DWC is supported by the NIHR Biomedical Research Centre, Oxford. JAS is funded by the Wellcome Trust. TNW is funded by the Wellcome Trust, European Network 6 BioMalpar consortium Project and the MalariaGen Network funded by Bill and Melinda Gates. AVSH is a Wellcome Trust Principal Fellow. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. This paper is published with the permission of the director of the Kenya Medical Research Institute.
Of 581 evaluable patients, 136 (23%) had bacteremia. The median age was 66 years (interquartile range 46 to 78 years) and 219 (38%) were male.

We evaluated three different models: a clinical model including seven bed-side characteristics, the clinical model plus PCT, and a PCT only model. The diagnostic abilities of these models as reflected by area under the curve of the receiver operating characteristic were 0.71 (95% confidence interval (CI): 0.66 to 0.76), 0.79 (95% CI: 0.75 to 0.83) and 0.73 (95% CI: 0.68 to 0.77) respectively. Calculating corresponding sensitivity and specificity for the presence of bacteremia after each step of adding a significant predictor in the model yielded that the PCT > 0.25 ��g/l only model had the best diagnostic performance (sensitivity 0.95; 95% CI: 0.89 to 0.98, specificity 0.50; 95% CI: 0.46 to 0.55).

Using PCT as a single decision tool, this would result in 40% fewer blood cultures being taken, while still identifying 94 to 99% of patients with bacteremia.The TTP of E. coli positive blood cultures was linearly correlated with the PCT log value; the higher the PCT the shorter the TTP (R2 = 0.278, P = 0.007).ConclusionsPCT accurately predicts the presence of bacteremia and bacterial load in patients with febrile UTI. This may be a Batimastat helpful biomarker to limit use of blood culture resources.IntroductionUrinary tract infection (UTI) is one of the most common infectious diseases.