Although a direct relationship between nosocomial infection and mortality in intensive care unit (ICU) patients has not always been demonstrated formally, it is possible to conclude that nosocomial infections increase the risk of death in critically ill patients. A more precise analysis indicates that: 1) this effect is highly probable for pneumonia, doubtful for bacteremia, and uncertain for urinary tract infection; 2) risk increases with duration of stay in the ICU; 3) bacterial etiology modifies the risk; and 4) this effect is stronger in less severely ill patients, probably because the severity of underlying disease remains the most significant factor.

Our highest priority should be to increase compliance with hand hygiene guidelines. It is food for thought that in several studies the group with the lowest compliance in intensive care settings is intensive care physicians. It is a reasonable assumption that they are role models for the rest of the ICU staff. Studies on practical and efficient means to increase compliance with hand hygiene guidelines and to influence behavior surely are needed more than are elaborate and sophisticated studies on the effects of handwashing.

We performed a meta-analysis to evaluate whether stress ulcer prophylaxis with histamine-2-receptor antagonists or antacids reduces mortality in critically ill patients. It appears that sucralfate results in a lower incidence of nosocomial pneumonia than either antacids or histamine-2-receptor antagonists. With respect to mortality, strong trends favored sucralfate over both antacids and histamine-receptor antagonists.

Morbidity and mortality due to infection acquired either before or after admission to the intensive care unit is still a major problem. A limited range of potentially pathogenic microorganisms (PPM) are involved, and infection with these PPM usually follows a predictable pattern. Selective decontamination of the digestive tract (SDD) should be investigated further in patients with primarily curable diseases who are not infected on admission to intensive care but become infected there.

Topical antimicrobials may complement traditional infection control measures and hasten the control of an outbreak. When a suitable nonabsorbable antimicrobial regimen can be found, its use during outbreaks caused by Enterobacteriaceae colonizing the digestive tract appears especially attractive, whether the infection occurs via primary endogenous colonization or exogenous colonization. In the latter circumstance, however, topical antimicrobials are not a substitute for isolation precautions and careful handwashing, which remain insufficiently adhered to in ICUs. Topical antimicrobials should be used as part of a global strategy in which isolation precautions remain of primary importance. It is likely, but unproven by well-designed studies, that nasal decontamination of patient carriers during outbreaks due to methicillin-resistant Staphylococcus aureus can be useful to control the outbreak. In this instance, however, skin and wound disinfection appears as an essential complementary measure, and careful attention should be given to extranasal sites of colonization with MRSA, where resistance to the topical agent can emerge. Whatever the circumstances and setting, prolonged use of topical antimicrobials during protracted outbreaks is less effective and carries a major risk of resistance to the topical agent(s) used against the outbreak strains. In such circumstances, careful microbiologic monitoring of colonization is mandatory.

Risk factors related to operating rooms include patient-associated risks, the operating room environment, ventilation systems, cleansing and sterilization, and operating room personnel. Although constantly debated, surgical wound infection surveillance with appropriate feedback to surgeons is one of the few effective measures that helps reduce surgical infection rates, and we strongly recommend its use. We also recommend the further study of other potential components of effective infection control programs for surgical patients.

Bacteria are found in 90% of surgical incisions, regardless of surgical technique or environment. Initially, the number of pathogens is low, but proliferation is facilitated by favorable local conditions and weakened host defense mechanisms. The principles of presurgical administration of antibiotics are well defined, but a few controversies persist. An ideal antimicrobial agent would not induce bacterial resistance in pathogenic organisms, would penetrate tissues effectively, would have a long enough half-life so that a single injection could guarantee protection throughout the operation, would have a low toxicity, would not interfere with anesthetics, would be easy to administer, and would be cost-effective.

The potential for bacterial resistance probably existed prior to the arrival of humans on earth and bacterial populations isolated before the antibiotic era surely contained antibiotic-resistant organisms. Antibiotic resistance has undergone an explosive development following the introduction of antibiotics in medical practice and in agriculture, and there is no doubt that the higher prevalence of bacterial resistance is closely related to human activities. Strict infection control policies limit the risk of patient-to-patient transmission of resistant as well as susceptible bacteria.

The relation between antibiotic control and resistance was assessed by review of selected journal articles from 1988 through 1994. Most studies of control or monitoring do not report susceptibility pattern as an outcome measure. Several aspects of bias and confounding prevent more than analysis of temporal association between antibiotic use restriction and resistance pattern. However, in a few institutions there has been an increase in susceptibility to antimicrobials following intensive control or monitoring. Moreover, in a few hospitals, intensive antibiotic control for selected drug-organism pairs was associated with a high prevalence of susceptibility, and the proportion susceptible fell abruptly when control or monitoring was relaxed or removed. These examples, coupled with the recent emergence of new resistant organisms, justify attempts to determine the value of intensive antibiotic control as a preventive practice. Because these studies were performed in single institutions, their power to distinguish associations was poor. Cooperative multicenter studies are needed in which selection and classification biases are addressed prospectively, and in which confounding factors are controlled.

An increasing range of antibacterial compounds is being used for nonclinical purposes, especially in the fields of animal husbandry and fish farming. As in human medicine, exposure to antibiotics has lead to the emergence of antibiotic-resistant bacteria in animal populations. The potential impact of antibiotic use in animals on human health and the management of clinical infections in humans is discussed in light of growing evidence to suggest that “new” resistance genes and multiresistant pathogens with increased pathogenicity are emerging in food animals as a direct consequence of antibiotic exposure.

Molecular typing methods are highly discriminatory for strains that cannot be separated by phenotypic methods. They are especially suited to monitoring outbreaks in children's hospitals. In many cases, molecular typing has improved the understanding of the mechanism of nosocomial acquisition of organisms by allowing distinction between endogenous and exogenous infections. Among exogenous infections, it has distinguished between individual and epidemic strains, thus differentiating cross-infection from independent acquisition.

An increasing proportion of the U.S. population resides in nursing homes (NHs). No surveillance system exists for infections in these facilities. To determine the incidence and types of infections in NH residents, and to identify predictors of death among residents with infections, we initiated a surveillance system at 13 NHs in California during a 6-month period from October 1989 through March 1990. The study included 1,754 residents, among whom 835 infections were identified during the study period. The most common infections were urinary tract infections (UTIs; 286, 34.2%), respiratory tract infections (RTIs; 259, 31%), and skin infections (150,17.9%). Of the 259 residents with respiratory tract infections, 69 (27%) had pneumonia. Antimicrobials were prescribed for 646 (77%) of the infectious episodes. Residents with pneumonia were more likely to die than residents with other infections (4 of 69 versus 12 of 766; P=0.04). Symptoms of altered body temperature (fever, hypothermia, chills) and change in mental status also were associated with an increased risk of a fatal outcome (10 of 260 versus 6 of 575; P= 0.01) and (7 of 127 versus 9 of 708; P= 0.004). This study suggests that the most common infections among NH residents are UTIs, RTIs, and skin infections. Pneumonia, symptoms of fever, and mental status changes all were associated with increased mortality. The frequency of infections among NH residents and their impact on resident outcome highlights the need for infectious disease surveillance in this population.

Preventive bacilleCalmette-Guérin (BCG) vaccination, together with case finding and effective chemotherapy, has formed an integral part of the tuberculosis (TB) control program in most countries. In some low-incidence countries the balance of prevention has been more on the side of chemoprophylaxis than of BCG vaccination. The time clearly has come when the strategy of mass BCG vaccination no longer is indicated medically, nor is it cost-effective. The pros and cons of the programs need to be critically evaluated against the present epidemiological background, taking into account the facts that TB, the killer disease, is recovering strength, human immunodeficiency virus infection is on the increase, and multidrug-resistant TB has changed the outcome of this previously fully curable disease.

Although no longer appropriate for mass programs, BCG vaccination still should be considered for the protection of selected risk groups in low-incidence countries. The overall efficacy may be of the order 50% to 80%, but the variation is great. Therefore, further research urgently is needed on the effectiveness of BCG as an intervention in local TB programs.

We conducted a retrospective review of nosocomial outbreak investigations conducted by the Hospital Infections Program, Centers for Disease Control and Prevention, from January 1991 through March 1994. Selected outbreaks have demonstrated the utility of molecular methods such as plasmid analysis, plasmid restriction endonuclease analysis, ribotyping, restriction fragment polymorphism, pulsed-field gel electrophoresis, and polymerase chain reaction in confirming the clonality of the outbreak and in confirming the source of the outbreak implicated in the epidemiologic investigation. These data show that molecular typing of isolates is particularly useful when combined with epidemiologic investigations of nosocomial outbreaks.