To examine risk factors for, and determine the incidence of, device-associated infections among patients with an implantable vascular access device.

Grady Health System, including a 1,000-bed, inner-city, public, teaching hospital and human immunodeficiency virus (HIV), oncology, and sickle cell clinics in Atlanta, Georgia.

123 consecutive patients who received a PASPort implantable venous access device between January 1 and June 30, 1995.

Retrospective cohort study with follow-up ≥1 year following device implantation.

Underlying illnesses included HIV infection in 66 patients (median CD4 count, 24.4 cells/mm3), malignancy in 51, and sickle cell disease in 6. Mean age of patients was 43.7 years, 50% were male, and 74% were black. Thirty-one (25%) of 123 patients developed a primary or device-associated bloodstream infection (BSI), and 3 of the 31 patients experienced two separate episodes of infection. The overall rate of infection was 1.23 primary BSIs per 1,000 device days. Patients with cancer had a lower rate of infection than those with HIV infection, but the difference was not statistically significant (0.96 vs 1.50 BSIs/1,000 device days; relative risk, 0.58; 95% confidence interval, 0.27-1.26). Subgroup analysis of patients with different malignancies indicated that infection rates differed according to type of cancer, and there was a trend for heterogeneity across the different cancer strata (P=.06). Gram-positive pathogens accounted for 60% of the pathogens recovered. Six (19%) of 31 patients who developed an infection did so within the first 14 days after implantation. In 11 (32%) of the 34 BSIs, the port required removal; two patient deaths were attributed to device-associated bacteremias (0.072 deaths/ 1,000 device days).

Approximately one fourth of patients who had a vascular access device implanted developed a primary BSI, but the overall infection rate (per 1,000 device days) was relatively low, even among those with HIV infection. Primary BSI rates in patients with vascular access devices appeared to differ according to the specific underlying illness.

To evaluate the risk of tuberculosis (TB) transmission to patients potentially exposed to two healthcare providers who worked in outpatient settings for several weeks prior to being diagnosed with acid-fast bacilli smear-positive pulmonary TB.

Potentially exposed patients were notified by letter and television reports of the recommended evaluation for TB infection or disease and availability of free screening at the hospital. Prevalence of infection in the screened patients and the incidence rate of TB over the subsequent 2 years were compared to those of a control group of unexposed outpatients.

An urban inner-city hospital.

1,905 patients with potential exposure to the ill healthcare workers; 487 (25%) presented for evaluation. Controls consisted of 951 unexposed patients.

361 potentially exposed patients had their tuberculin test read; 97 (27%) had a purified protein derivative ≥10 mm. In the comparison group, 148 (25%) of 600 with test readings had a ≥10-mm reaction (risk ratio, 1.18; 95% confidence interval, 0.86-1.60). In multivariate analysis, male gender, non-white race, and older age were significantly associated with a positive tuberculin test; exposure was not. No TB cases were identified during screening. Two years after the exposure, 7 TB cases had been reported to the state registry among 1,905 potentially exposed patients (184 cases/100,000 person-years) versus 4 cases in the comparison group of 951 (210 cases/100,000 person-years).

Evaluation of patients exposed to health-care workers with TB disease in ambulatory settings of an inner-city hospital revealed no evidence of transmission of Mycobacterium tuberculosis due to the exposure.

To determine the proportion of vancomycin orders that are appropriate according to national guidelines and to identify targets for educational messages.

Population-based study of vancomycin use in Oregon during a 3-week period. Survey of pharmacists, prospective flagging of vancomycin orders, and data abstraction from patient charts using standardized forms.

Nonpsychiatric hospitals in Oregon.

Four (6%) of the 66 Oregon hospitals had pharmacy restrictions on initial vancomycin orders. Sixty-four (97%) of the hospitals participated in the study of indications for use; 293 vancomycin orders were reported; 3.8 courses were initiated per 1,000 patient-days. Indications for use were determined for 266 (91%); of these, 159 (60%) were deemed appropriate. Of uses for prophylaxis, empirical treatment of suspected gram-positive infection, and treatment of documented gram-positive infection, 57%, 56%, and 65%, respectively, were appropriate. Of hospitals with <250, 251-475, and >475 licensed beds, 65%, 58%, and 57% of vancomycin orders were appropriate. No single medical specialty accounted for >16% of inappropriate vancomycin use.

Vancomycin was used inappropriately by physicians of many different specialties, in hospitals of all sizes, and in sundry clinical situations. The problem of inappropriate vancomycin use does not lend itself to solution by educational strategies targeted at specific subgroups; restrictions by hospital pharmacies may be required.

To evaluate and compare the risk of long-term central venous catheter (CVC) infection in human immunodeficiency virus (HIV)-infected and cancer patients.

Prospective multicenter cohort study based on active surveillance of long-term CVC manipulations and patient outcome over a 6-month period.

Services of infectious diseases and oncology of 12 university hospitals in Paris, France.

In 1995, all HIV and cancer patients with solid malignancy were included at the time of long-term CVC implantation.

Overall, 31.6% of long-term CVC infections were identified in 32% of 201 HIV and 5% of 255 cancer patients. Most were associated with bacteremia, most commonly coagulase-negative staphylococci. The long-term CVC time-related infection risk was greater in HIV than in cancer patients (3.78 vs 0.39 infections per 1,000 long-term CVC days; P<.001). The independent risk factors of long-term CVC infection were as follows: in HIV patients, frequency of long-term CVC handling and neutropenia; in cancer patients, poor Karnofsky performance status; in both HIV and cancer patients, recent history of bacterial infection. The risk of long-term CVC infection was similar for tunneled catheters and venous access ports in each population.

Prevention of long-term CVC infection should focus first on better sterile precautions while handling long-term CVC, especially in HIV patients who have frequent and daily use of the long-term CVC.