Infection Control and Hospital Epidemiology

Bloodstream infections (BSIs) are one of the most common healthcare-associated infections and a leading cause of morbidity and mortality. From 1980 to 1990, the incidence of BSIs in U.S. hospitals was estimated to have increased by nearly 70%. In hospitalized patients, BSIs are estimated to account for approximately 10% of all healthcare-associated infections and to have an attributable mortality rate of approximately 15%, making them the eighth leading cause of death in the United States. At most hospitals, surveillance for BSIs is thought to be very complete. After all, most BSIs are diagnosed via blood culture and cultures are obtained from most symptomatic patients. A validation study conducted in National Nosocomial Infections Surveillance (NNIS) system hospitals showed that NNIS system surveillance for BSIs had a positive predictive value of 87%, a sensitivity of 85%, and a specificity of 98.3%. However, as illustrated by the article by Smith et al. in this issue of Infection Control and Hospital Epidemiology, many aspects of our healthcare system and healthcare delivery processes have changed in the past decade, complicating our efforts to conduct surveillance, calculate BSI rates, and implement BSI (and other infection) prevention interventions. These aspects include enormous changes in the way in which and where we provide health care, reductions in clinical personnel, increased use of intravascular catheters (particularly central venous catheters [CVCs]) for a wider variety of reasons and in more clinically diverse populations, reductions in or outsourcing of clinical microbiology services, and failure to integrate our healthcare setting clinical information systems. All of these lead to the potential to have much less robust or complete BSI surveillance systems and thus much less effective BSI prevention and control programs than we think. Let's examine how these changes influence our BSI prevention programs. First, let's examine what has happened to our healthcare delivery system. During the past two decades, there has been a dramatic shift in the delivery of health care from inpatient to outpatient settings. In 1996, nearly 8,000,000 individuals in the United States received medical care in their home; approximately 10% or an estimated 774,113 had at least one indwelling medical device— most of these being intravascular catheters. In that same year, approximately 750,000,000 individuals received care in a physician's office, 75,000,000 visited a hospital emergency department, and 50,000,000 received care at a hospital outpatient clinic. In 1999, approximately 82.5% of the entire U.S. population had more than one visit for receipt of health care to a physician's office, an emergency department, or a clinic or received care at their home. In the "outbreak of BSIs" investigated by Smith et al., this complex system of healthcare delivery complicated the infection control program's surveillance for BSIs and the "outbreak" investigation. The initial perception of an increase in BSIs was based on clinical perception and numerator data. This occurred because denominator data to calculate BSI rates were not readily available, as the patients were being seen in the outpatient clinic and there was no computer database of who (much less what their service or diagnosis was) visited the outpatient clinic, whether those visiting the clinic had a CVC, or whether the catheter was manipulated in the clinic. Initially, because all of the patients had the onset of symptoms as outpatients and were included in a home infusion therapy program, the BSIs were thought to be secondary to home infusion therapy. The majority of patients with BSIs had periodically received home infusion therapy (provided by several companies that did not provide BSI