Clinical utility of blood cultures drawn from central venous or arterial catheters in critically ill surgical patients.

OBJECTIVE To determine the sensitivity, specificity, and predictive values of cultures done with blood drawn through a central venous or arterial catheter compared with peripheral venipuncture. DESIGN Retrospective cohort study of critically ill surgical patients in whom samples for paired cultures were drawn through a central venous or arterial catheter and peripheral venipuncture. SETTING Tertiary-care, university-affiliated medical center. PATIENTS Two hundred seventy-one patients hospitalized on a surgical and a cardiothoracic intensive care unit between November 1994 and August 1997. INTERVENTIONS None. MEASUREMENTS AND MAIN RESULTS Blinded assessments of culture results done by two physicians were used as the gold standard. Sensitivity, specificity, and positive and negative predictive values were compared for culture of blood from catheters and culture of blood from peripheral venipuncture. Of 499 observations, 426 were catheter-negative/venipuncture-negative, 19 were catheter-positive/venipuncture-positive, 18 were catheter-negative/venipuncture-positive, and 36 were catheter-positive/venipuncture-negative pairs. For catheter draws compared with peripheral venipuncture, sensitivity was 78% (confidence interval [CI], 65% to 90%) and 65% (CI, 50% to 79%) (p = .2), specificity was 95% (CI, 94% to 97%) and 98% (CI, 97% to 99%) (p = .002), positive predictive value was 63% (CI, 51% to 76%) and 78% (CI, 64% to 91%) (p = .1) and negative predictive value was 98% (CI, 96% to 99%) and 97% (CI, 95% to 98%) (p = .3). When central venous specimens as differentiated from arterial catheter specimens were compared with peripheral venipuncture, the difference between positive predictive values reached statistical significance (61% and 82%; p = .04). CONCLUSIONS In critically ill surgical patients, cultures of blood drawn through a catheter are less specific than those obtained from a peripheral venipuncture. Both types of cultures have an excellent negative predictive value. Positive predictive value of cultures of blood drawn through a catheter is low and, when obtained from a central line, statistically less than from a peripheral venipuncture. Additional cultures seem to be necessary for the proper interpretation of a positive culture drawn through a catheter in critical care patients.