A comparative assessment of two conservative methods for the diagnosis of catheter-related infection in critically ill patients

Abstract

To assess the utility of two in situ techniques, differential time to positivity (DTP) and semiquantitative superficial cultures (SQSC) for diagnosing catheter-related bloodstream infection (CR-BSI) in critically ill adults. This was a prospective cohort study in patients with suspected CR-BSI arising from a short-term arterial catheter (AC) or a central venous catheter (CVC). On suspicion of CR-BSI, devices were removed. Blood, skin, catheter tip and hub cultures were taken. Infection rates were compared against the diagnosis of CR-BSI using matched tip and blood cultures. Of 120 episodes of clinically suspected CR-BSI in 101 patients examined, 9 (7.5 %) were confirmed as CR-BSI. Validity values (95 % CI) for the diagnosis of CR-BSI arising from both AC and CVC for DTP were: sensitivity 44 % (15-77 %), specificity 98 % (93-100 %), positive predictive value (PPV) 67 % (24-94 %), negative predictive value (NPV) 96 % (90-98 %), positive likelihood ratio (LR+) 25 (5-117), negative likelihood ratio (LR-) 0.6 (0.3-1.0), diagnostic odds ratio (DOR) 44 (7-258), and accuracy 94 % (92-98 %). Validity values (95 % CI) for SQSC were: sensitivity 78 % (41-96 %), specificity 60 % (50-69 %), PPV 14 % (6-26 %), NPV 97 % (89-99 %), LR+ 1.9 (1.0-2.3), LR- 0.4 (0.1-1.3), DOR 5.1 (1.1-19), and accuracy 61 % (51-69 %). DTP combined with SQSC improved sensitivity and NPV to 100 % whilst the DOR increased to 25.8 (95 % CI 3-454). CR-BSI can be ruled out by undertaking DTP and SQSC concurrently for both ACs and CVCs with 100 % sensitivity and NPV.