Automazione dell’urinocoltura in campioni da catetere vescicale: criticità applicative ed interpretative dell’uso in routine del laser-nefelometro

Automated urine culture methods have spread throughout Italian clinical laboratories in the last decade.Among available systems, Uro-Quick™ (a laser nephelometric instrument) evaluates the microbial growth in liquid medium, shortening the overall time of laboratory processing. The performances of the instrument have been estimated in patients with indwelling urinary catheter. Urine of patients with CAUTI (Catheter Associated Urinary Tract Infection), because of high dilution and anti-bacterial load, frequently produce working difficulties and misleading interpretation. In 2003-2004 urine specimens coming from patients of S.Filippo Neri Hospital, in Rome (Italy) were examined. Patients with closed-circuit urinary catheter, mainly hospitalized in ICUs, and requesting urine cultures for clinical reasons were enrolled. A total of 100 properly collected specimens, from different patients,were considered. For each, several clinical information (especially antibiotic therapy) were also collected. Urine specimens were examined as follows: 1) microbial count and evaluation of antibacterial load by means of Uro-Quick™ ; 2) microscope (HPF) observation for pyuria in the sediment; direct plate culturing of urine, with 10 μl calibrated loop.A bacteriuria was defined as the growth of 1-2 species, with a count > = 103 CFU/mL; the same value, together with the finding of pyuria (at two degrees: >1 WBC, >5 WBC, respectively), was used for a presumptive definition of CAUTI. A considerable rate of specimens (62%) showed anti-bacterial load (while the reported antibiotic therapy was 54%).The instrumentation revealed a sensibility of 74% and specificity of 98% in finding bacteriuria. However, when negative specimens which had antibacterials were re-classified as positive (for possible presence of inhibited microorganisms), sensibility increased to 95%. On the contrary, considering the presence of >5 WBC in the sediment, sensibility and specificity declined, respectively, to 56% and 85%. In the last case, adjusting the instrumentation results with determination of antibacterial load revealed as an unreliable approach: while sensibility increased at 94%, specificity decreased at 24%. Briefly, Uro-Quick™ revealed good positive predictive value (+PV) in finding bacteriuria, but unsatisfactory negative predictive value (-PV). The system showed anyway low +PV and -PV versus the presumed infection: moreover, these performances were not improved by the simultaneous determination of antibacterial load in specimens.The results suggest the opportunity to set up a separated, not automated, workflow for urine collected from catheter.