Despite remarkable strides in our understanding of hospital-acquired infections, the diagnosis and prevention of pneumonia remain controversial [2-7]. Pneumonia is the second most common hospital-acquired infection in the United States, occurring at a frequency of 0.6 to 1.0 episodes/100 hospitalizations and with rates 6-fold to 21-fold higher in mechanically ventilated patients [2, 3]. Pseudomonas aeruginosa and Staphylococcus aureus are common causes of nosocomial pneumonia, but etiologies may vary by patient population, hospital, geographic area, and the technique used for diagnosis [2, 3]. Crude mortality rates range from 20% to 60%, but only about one third of these deaths are directly attributable to the pneumonia [2-4].

Pneumonia must be accurately diagnosed before relevant recommendations for prevention can be developed [2-7]. In this issue, Valles and coworkers [8] report that they used a clinical diagnosis of ventilator-associated pneumonia that was based on a body temperature of 38.3 °C or greater, a leukocyte count less than 4000 cells/mm³ or greater than 12 000 cells/mm³, purulent secretions, and the presence of new and persistent pulmonary infiltrates. Clinical diagnosis was confirmed by bronchoscopy with positive protected specimen-brush culture containing more than 10³ colony-forming units (CFU)/mL, by positive bronchoalveolar lavage fluid culture containing more than 10⁴ CFU/mL, or by a good clinical response to antibiotic therapy. Although this definition may be "ideal" for clinical investigation, the use of routine invasive diagnostic techniques in clinical practice may be impractical.

Most clinicians continue to rely on a clinical diagnosis of ventilator-associated pneumonia because it is convenient, because quantitative bacteriologic findings are often unavailable, and because most patients appear to respond to broad-spectrum antibiotic therapy [6]. Although clinical diagnosis has a high sensitivity, tracheal colonization and inflammation in intubated patients reduce its specificity, particularly in patients with congestive heart failure, pulmonary emboli, or the adult respiratory distress syndrome [8]. In the past decade, bronchoscopy with protected specimen brush or bronchoalveolar lavage has been recommended for diagnosing ventilator-associated pneumonia on the basis of increased specificity; these techniques may reduce the use of antibiotic agents [5]. When performed by experts, bronchoscopy with protected specimen brush and bronchoalveolar lavage have a sensitivity of approximately 70% to 90% (range, 38% to 100%) and a specificity of 70% to 100%. Recent data suggest that "blind" bronchoalveolar lavage and protected specimen brush or quantitative endotracheal aspirates may be useful techniques for diagnosis because they correlate well with diagnosis by bronchoscopy and are less invasive and less expensive than other techniques [6, 7]. Unfortunately, all of these quantitative methods are costly and may be affected by differences in sampling and processing techniques, stage of infection, and previous antibiotic therapy [5, 6]. Until specific information indicates that bronchoscopy with bronchoalveolar lavage or protected specimen brush is superior to clinical diagnosis in reducing antibiotic use, reducing the rates of superinfection, or improving outcome, most clinicians will continue to rely on clinical diagnosis, realizing that it is sometimes more an art than a science.

Aspiration of bacteria colonizing the oropharynx is the major route of bacterial entry into the lung, and leakage of contaminated secretions around the cuff of the endotracheal tube cuff increases the risk for tracheal colonization and ventilator-associated pneumonia [2, 3]. Mahul and colleagues [9] initially reported that the incidence of ventilator-associated pneumonia was lower in 145 patients randomly assigned to receive manual intermittent aspiration of subglottic secretions (13% compared with 29% in patients not receiving aspiration). The data from Valles and coworkers' well-designed clinical trial of 139 patients [8] indicate that compared with control patients who received no intervention, patients assigned to receive continuous aspiration of subglottic secretions had a significantly reduced incidence of ventilator-associated pneumonia (39.6 episodes/1000 ventilator days compared with 19.9 episodes/1000 ventilator days, respectively [relative risk, 1.98; 95% CI, 1.03 to 3.82]). Interestingly, the major effect was on early-onset ventilator-associated pneumonia caused by Haemophilus influenzae, streptococci, and Staphylococcus aureus. Although no differences were noted in crude mortality, duration of ventilation, or duration of intensive care unit stay, the overall attributable mortality was 2.5-fold lower in patients receiving continuous aspiration. Thus, this technique is a simple, safe, and inexpensive prevention strategy that makes sense but unfortunately is not currently available in the United States.

The Hospital Infection Control Practices Advisory Committee (HICPAC), in conjunction with the Centers for Disease Control and Prevention, recently prepared an updated guideline for the prevention of nosocomial pneumonia [3]. Prevention measures that are supported by well-designed studies or studies deemed effective by experts and that have a good cost–benefit ratio are strongly recommended for all hospitals (category I recommendation). By comparison, category II recommendations are based on suggestive clinical or epidemiologic studies that have a strong theoretical rationale or on definitive studies whose results are not applicable to all hospitals. Prevention measures that lack sufficient data on their usefulness and cost-effectiveness are not recommended. The HICPAC guideline also contains comprehensive background information with referenced recommendations, some of which are discussed below.

Gastric colonization and medications used to prevent stress bleeding have been implicated as risk factors for ventilator-associated pneumonia [10-12]. Antacids, histamine-2 blockers, and sucralfate effectively prevent stress bleeding, but patients randomly assigned to receive sucralfate appear to have lower rates of clinically diagnosed ventilator-associated pneumonia (category II) [10-12].

Enteral feeding may also increase the risk for ventilator-associated pneumonia [3, 5]. Although the proper site, route, and method of tube feeding are controversial, recent data suggest that simply maintaining patients in a semi-upright position appears to reduce reflux, aspiration, and ventilator-associated pneumonia (category I) [3, 13]. In addition, recent data indicate that the use of a nasogastric tube (or nasotracheal tube) increases the patient's risk for nosocomial sinusitis and pneumonia [14]. Inserting gastric and endotracheal tubes orally should reduce the incidence of nosocomial sinusitis and pneumonia.

On the basis of data from several clinical studies, the new HICPAC guidelines recommend that ventilator circuits be changed no more frequently than once every 48 hours rather than every 24 hours (category I). Implied in this recommendation is the understanding that a patient's circuit can be changed less frequently or not at all (category II) [3]. These recommendations emphasize that doing less is sometimes not only better for the patient but may also save millions of health care dollars nationwide [3].

In contrast to many of the simple, low-risk, and cost-effective interventions described above, preventing ventilator-associated pneumonia by the use of prophylactic antibiotic agents for selective decontamination of the digestive tract is more complicated and controversial [15-17]. Although most studies of this technique have shown a significant reduction in bacterial colonization of the oropharynx, the results of its effectiveness in preventing ventilator-associated pneumonia are less compelling, particularly in studies in which the investigators were blinded and pneumonia was clearly defined [17]. In addition, selective decontamination of the digestive tract has had little effect on mortality and may increase the risk for nosocomial infections from gram-positive or multidrug-resistant pathogens [17]. For these reasons, selective decontamination of the digestive tract cannot be recommended for the prevention of ventilator-associated pneumonia [3].

Patients hospitalized in critical care units may acquire nosocomial infections with pathogens such as Staphylococcus aureus and Pseudomonas aeruginosa from the hands of hospital personnel, contaminated devices, or the environment [3, 18]. Compliance of health care workers with proper hand-washing is often poor, and gloves or barrier precautions are not used routinely [3, 18, 19]. Although not covered in the HICPAC prevention guideline, data on the nosocomial transmission of respiratory pathogens such as Mycobacterium tuberculosis, pneumococci, H. influenzae, and Pneumocystis carinii suggest that patients with respiratory tract infection should be appropriately isolated until a diagnosis is established or until respiratory symptoms subside [3, 20-23].

How can we begin to measure the effect of prevention in pounds rather than in ounces? First, we need reliable data based on well-designed studies that have clear end points and that preferably are free from interference or control from the pharmaceutical or the medical device industry. Second, prevention measures must be carefully assessed for cost-effectiveness, risk–benefit ratio, and effect on outcome measures. Third, if preventive measures are effective, they need to be properly implemented and monitored. The continuous aspiration of subglottic secretions, orogastric tubes, and the semi-upright position of the patient are excellent examples of prevention strategies that should be implemented in all hospitals because they are based on sound clinical data and are simple, low-risk, and cost-effective. In contrast, a prevention measure such as selective decontamination of the digestive tract should not be routinely used because the data on its effectiveness in reducing the incidence of ventilator-associated pneumonia are less compelling and because the widespread use of prophylactic antibiotic agents in the critical care setting may substantially add to risk.

The goal for the year 2000 is to continue to develop a rational, scientific basis for preventing nosocomial pneumonia using well-designed clinical trials and molecular techniques to improve our understanding of the epidemiology of ventilator-associated pneumonia, the mechanisms of airway colonization, and measures to enhance the pulmonary response to microbial invasion. As we strive to meet this goal, we should further emphasize prevention through education and training, the development of safer devices, and the proper implementation of the HICPAC prevention guidelines [3]. Only through a comprehensive, multifaceted approach to prevention can we measure the effect in tons rather than in ounces, acknowledging that an ounce of prevention is worth a pound of cure and a ton of treatment.