This article focuses on the selection and interpretation of diagnostic tests, emphasizing the importance of understanding how their mathematical parameters affect the information they provide in various settings. The utility and limitations of sensitivity, specificity, predictive value, and receiver operating characteristic (ROC) curves are discussed using catheter-related bloodstream infections as an example. ROC curves have been used for selecting optimal cutoff values for a positive result and for selecting among several alternative diagnostic tests. For example, 16 different tests have been proposed for diagnosis of catheter-related bloodstream infection; ROC analysis provides an effective way to determine which test offers the best overall performance.

The project goal was to decrease excessive vancomycin use. Interventions included an educational chart note the first day of therapy, followed by pharmacists discussing the need for continued therapy with patients' physicians. Empirical vancomycin use improved from 20% to 90% compliance with guidelines within 6 months of the intervention

From 1988 to 1998, we identified nine patients with Candida guilliermondii fungemia. Four of the five patients with nosocomial infection died, while all of the non-nosocomial cases survived, even though one half of them (2/4) did not receive any treatment Nosocomial C guilliermondii fungemia is often associated with poor outcome despite aggressive antifungal therapy.

To strengthen guidelines for vancomycin use, practical guidelines were developed. A prospective survey was conducted of all patients receiving vancomycin during two 1-month periods, 1 year apart, during which significant improvements were noted. Practical guidelines may contribute to appropriateness of vancomycin use, serve as educational tools, and facilitate improved surveillance.

Physicians and clinical employees at a children's hospital were surveyed to compare their tuberculosis (TB) screening and immunization statuses. Failure to offer screening and immunization services to non-employee physicians was associated with lower rates of reported immunity to several vaccine-preventable diseases and with markedly lower rates of TB screening.

We studied risk factors for nosocomial infections among 500 critically ill children who were admitted to a pediatric intensive care unit from August 1994 through August 1996 and who were prospectively followed until death, transfer, or discharge. Age, gender, postoperative state, length of stay, device-utilization ratio, pediatric risk of mortality score, and total parenteral nutrition were the risk factors studied. Through multivariate analysis, we identified three independent risk factors for nosocomial infection: device-utilization ratio (odds ratio [OR], 1.6; 95% confidence interval [CI95], 1.10-2.34), total parenteral nutrition (OR, 2.5; CI95, 1.05-5.81) and length of stay (OR, 1.7; CI95, 1.31-2.21).