ABSTRACT
OBJECTIVE: To evaluate the potential contribution of "extrinsic" contamination of intravenous fluids in hospital bacteremia and infection. DESIGN: Prospective cross-sectional survey of infusate contamination, December 1992 to December 1993. SETTING: A pediatric department (1,500 admissions per year) in a general, urban teaching hospital, serving low-income patients. SAMPLES AND PATIENTS: Infusate samples (0.5 to 1.0 mL) from the injection port used by the staff were taken for cultures from all febrile or septic patients in hospital wards. At least four samples were taken each day; if no febrile or septic patients were available, other patients were sampled at convenience. RESULTS: A 6.8% positive culture rate (87 contaminates in 1,277 infusates) was obtained, without significant differences among the wards. Gram-negative organisms were recovered from 56 samples (62.9%), mainly of the tribe Klebsielleae (56.1%). Coagulase-negative staphylococci were isolated in 30 samples (33.7%). There was no significant difference between the febrile-septic group and the asymptomatic group in the rate of infusate contamination (P = .59). In eight patients, the same organisms were recovered from infusate and blood culture. The overall bacteremia rate was 2.5 per 100 discharges. CONCLUSIONS: Compared to previous reports, higher infusate contamination rates and different organisms (mainly gram-negative) were observed. In hospitals of underdeveloped countries, nosocomial infection control frequently is disregarded. Infusate contamination may be common and could lead to gram-negative bacteremia. In such settings, it seems advisable to perform surveillance studies to identify infusate contamination, because a single infusate contamination could be a signal for an epidemic.
Subject(s)
Bacteremia/etiology , Cross Infection/etiology , Drug Contamination , Gram-Negative Bacterial Infections/etiology , Infusions, Intravenous/adverse effects , Solutions/adverse effects , Child, Preschool , Cross-Sectional Studies , Hospital Departments , Humans , Infant , Infant, Newborn , Infection Control , Infusions, Intravenous/instrumentation , Pediatrics , Prospective StudiesABSTRACT
To identify the statistical methods more frequently used in the medical literature, we reviewed 4,218 papers equivalent to 26 years/periodical. 26% of the papers did not have any statistical method. In the remaining 74% only descriptive (24.5%) or inferential (75.5%) methods were used. Overall there were 3,882 inferential test done. Based on a probability analysis we propose a framework to rationalize the teaching of statistics to improve the decision making process in medicine.
