Rational Empiric Antibiotic Prescription in the ICU

opinions/hypotheses Rational Empiric Antibiotic Prescription in the ICU* Clinical Research Is Mandatory Nina Singh, MD; and Victor L. Yu, MD

The prescribing of antibiotics in the ICU is usually empiric, given the critical nature of the conditions of patients hospitalized there. Appropriate antibiotic utilization in this setting is crucial not only in ensuring an optimal outcome, but in curtailing the emergence of resistance and containing costs. We propose that research in the ICUs is vitally important in guiding antibiotic prescription practices and, therefore, the achievement of above-stated goals. There is wide institutional diversity in the relative prevalence of predominant pathogens and their antimicrobial susceptibilities, and within individual ICUs there exist unique patient populations with varying risks for and susceptibilities to infections and specific pathogens. Appropriate antibiotic prescription practices should be formulated based on surveillance studies and research at individual ICUs; these goals can be accomplished utilizing existing resources. (CHEST 2000; 117:1496 –1499) Key words: antibiotics; ICU; nosocomial pneumonia Abbreviation: MRSA ⫽ methicillin-resistant Staphylococcus aureus

have come to represent the most frequently I CUs identifiable source of nosocomial infections within the hospital, with infection rates and rates of antimicrobial resistance severalfold higher than in the general hospital setting.1–5 Rates of nosocomial infections range from 5 to 30% among ICU patients. Although ICUs generally comprise ⬍ 5% of all hospital beds, they account for 20 to 25% of all nosocomial infections.1–3,6 Pneumonia and bacteremia are currently the most common nosocomial infection in the United States and by far the leading cause of death from nosocomial infections in critically ill patients. A high prevalence of nosocomial infections in ICU patients is associated with high utilization of antibiotics. Although antibiotics represent one of the most frequently prescribed classes of drugs among all hospitalized patients, total antibiotic consumption *From the Veterans Affairs Medical Center (Dr. Singh), Pittsburgh, PA; and University of Pittsburgh (Dr. Yu), Pittsburgh, PA. Manuscript received August 18, 1999; revision accepted December 23, 1999. Correspondence to: Victor L. Yu, MD, VA Medical Center, Infectious Disease Section, University Drive C, Pittsburgh, PA 15240; e-mail: vly⫹@pitt.edu 1496

is approximately 10 times greater in the ICU than in general hospital wards.7 Respiratory infections, suspected or proven, account for nearly one half of all antibiotics used in the ICUs.8 We present data to show that the ecology of potential pathogens, the predominant types of infections, and specific patient populations at risk for infection are unique to individual ICUs. Antibiotic recommendations based on studies performed at a few selected centers may, therefore, have limited widespread applicability and may not be generalizable to all ICU settings. Concepts, but not practices, can be extrapolated from such studies. We propose that research in individual ICUs is essential in guiding antibiotic prescription practices. The spectrum of potential pathogens and the predominant bacterial flora can vary considerably in different ICU settings. Pseudomonas aeruginosa has emerged as the predominant pathogen in cases of nosocomial pneumonia in many ICUs. However, the proportion of the cases of pneumonia that are due to P aeruginosa varies widely from 6 to 31% at various institutions.6,9 –14 Of patients with nosocomial pneumonia in whom a bacteriologic diagnosis was estabOpinions/Hypotheses

lished, Acinetobacter spp accounted for 25% of the episodes in one report11 and none in others.12,15 Indeed, Acinetobacter was the single most frequently isolated bacteria in patients with ventilator-associated pneumonia in the former study.11 Enterobacter cloacae accounting for 26% of all cases, was the most common pathogen in ventilator associated pneumonia in one study.16 Twenty-seven percent and 11%, respectively, of the cases of ventilator-associated pneumonia in two other studies were also due to Enterobacter spp; only P aeruginosa was isolated more often at these institutions.9,12 By contrast, none of the cases of pneumonia in other ICU studies have been due to Enterobacter.11,15 Methicillin-resistant Staphylococcus aureus (MRSA) has become established as an endemic pathogen in many hospitals and has emerged as a significant cause of nosocomial pneumonia in ICUs at these institutions. Thirty-seven percent of the cases of pneumonia occurring in ICU patients in one report were due to MRSA.15 MRSA and P aeruginosa were the foremost pathogens discovered in cases of nosocomial pneumonia occurring in liver transplant recipients who were being cared for in that ICU.17 Legionella has been documented to be an important nosocomial pathogen for immunocompromised patients in selected studies. Two to 9% of the cases of nosocomial pneumonia in various ICU patients have been shown to be due to Legionella.10,11,15 However, the single most important risk factor is simply whether the ICU potable water system is colonized with Legionella. If it is not, then Legionella will be a consideration only in cases of community-acquired pneumonia at that institution.18 The rates of nosocomial infections in similar types of ICUs can vary 10- to 20-fold among hospitals.19 –21 The frequency and types of infection also vary among different ICUs within the same hospital and between subsets of patients within the same ICU. This is due largely to the unique characteristics predisposing patients to infections in particular ICU populations. Infection rates are generally the highest in burn units (ⱕ 64%) and are the lowest in coronary care or cardiac surgery ICUs (0.5 to 4.7%).1,22–24 Medical and surgical ICUs have an intermediate risk, however, nosocomial infection rates are higher in surgical ICUs (28 to 31%) than in medical ICUs (3.2 to 24%). The relative frequency of predominant infections also varies in different ICUs. Rates of nosocomial pneumonia ranged from 0.5 cases per 100 admissions in the coronary care ICU, to 9 per 100 in the burn unit, with other ICUs ranging between 1.5 and 2.5 cases per 100 admissions in one report.22 In another study, nosocomial pneumonia occurred equally frequently in medical ICU (10%) and surgi-

cal ICU (8%) patients, however, all other major nosocomial infections, except pneumonia, occurred significantly more frequently in the surgical ICU.23 There was a strong trend toward a higher incidence of ventilator-associated pneumonia in cardiothoracic surgery patients; ventilator-associated pneumonia occurred in 9% of the medical ICU patients, in 14% of the surgical ICU patients, and in 22% of the cardiothoracic ICU patients.16 The rate of ventilatorassociated pneumonia was 5.8 cases per 1,000 ventilator-days in the pediatric ICU, 14.5 cases per 1,000 ventilator-days in the surgical ICU, and 18.3 cases per 1,000 ventilator-days in the neurosurgical ICU.20 It should, however, be noted that a “gold standard” for the diagnosis of ventilator-associated pneumonia does not exist. Thus, variability in the criteria and methods of diagnosis of ventilator-associated pneumonia may account partly for the differences in rates of nosocomial pneumonia at different institutions. Primary bacteremia comprised 15% of infections in the neonatal ICU, a rate that is 500% higher than those for the other ICUs at that institution.1 The central line-associated bloodstream infection rate varied from a mean of 4.9 cases per 1,000 central line-days in the surgical ICU to 6.1 cases per 1,000 central line-days in the medical ICU, and to 14.6 cases per 1,000 central line-days in the burn ICU.20 Fifty-one percent of the infections in the coronary care unit, but only 1% of those in the neonatal ICU in the same study, were secondary to a genitourinary source.1 Candida, enteric Gram-negative bacterial, and Pseudomonas infections were more common in patients in medical/surgical ICUs than in patients in other ICUs.1 These data have significant bearing on discerning the likelihood, potential source, and, ultimately, the empiric antimicrobial therapy for infections in a specific ICU population. The size of the ICU has been shown to affect the risk of infection; patients on units with ⱖ 11 beds were at significantly greater risk than those patients on units with ⱕ 5 beds (p ⬍ 0.05).21 Underlying disease and comorbid illness also influence the susceptibility of patients to specific pathogens. In a study in trauma patients, the predominant pathogen causing nosocomial pneumonia was Haemophilus influenzae.25 Trauma patients also were shown to be significantly more likely to develop pneumonia due to Haemophilus or pneumococcus than other patients in another ICU.15 Haemophilus and pneumococcal carriage is common in healthy patients in the community. Trauma patients, unlike patients with chronic illnesses who are admitted to the ICU, were healthy before admission. A lower incidence of Haemophilus and pneumococcal pneumonia in patients with chronic illnesses (and thereCHEST / 117 / 5 / MAY, 2000

1497

fore prior or prolonged hospitalization) may be reflective of the modification of the resident flora of the oropharynx and colonization with more resistant organisms, compared to trauma patients who may still be colonized with Haemophilus or pneumococcus. Neurosurgical patients and patients with craniocephalic trauma were uniquely susceptible to pneumonia due to methicillin-sensitive S aureus in two studies.26,27 MRSA and P aeruginosa, on the other hand, were significant causes of pneumonia in the patients with COPD.26,28 Liver disease and cirrhosis were documented to be risk factors for MRSA pneumonia in the ICU at one institution.15 The time of onset of pneumonia after hospitalization has been shown to influence the incidence and etiology of pneumonia. Variable criteria and cutoffs (ie, ⬎ 48 h, ⬎ 72 h, or ⱖ 4 days) have been proposed to define early-onset vs late-onset nosocomial pneumonia.29,30 The rationale, however, being that prolonged hospital stays increase the likelihood of pneumonia due to more resistant microorganisms. In this regard, time after admission to the hospital may be more relevant than time after intubation as a risk factor for colonization and acquisition of nosocomial infections. We recommend that ICUs assess their institutional trends for the time required for the development of specific types of pneumonia, rather than utilize an arbitrary time criterion. For example, at one ICU,15 the mean time to the onset of nosocomial pneumonia was 4 days for Haemophilus, 36 days for Gram-negative organisms, and 42 days for MRSA (p ⫽ 0.003). These data suggest that even though MRSA was endemic at that institution, vancomycin need not be included routinely in the empiric antimicrobial regimen of patients developing nosocomial pneumonia unless they had been hospitalized for least a month.15 The rationale and arguments forming the basis of our recommendations can be summarized as follows: precise knowledge of the pathogen allows a rational antibiotic selection, an assessment of the likely portal of entry, the normal flora at the site of entry, the clinical setting, and the underlying defect of the host defenses in patients being cared for in that ICU.31 We show that bacterial flora, specific patient population, underlying diseases, and timing of nosocomial pneumonia are sufficiently unique so as to warrant research within individual ICUs to precisely determine the ecology of pathogens specific to that ICU. While therapeutic algorithms from textbooks and consensus approaches to management from expert panels can empower physicians with general concepts, they cannot predict precise antimicrobial choices. Since the adverse impact of inadequate antimicrobial treatment on higher mortality rates in the ICU setting has been amply demonstrated,32 1498

appropriate empiric therapeutic strategies for suspected infections in critically ill patients can contribute to improved outcomes. Recommendations regarding anti-infective regimens based on a computer program that took into consideration epidemiologic information, along with other details relevant to particular patients, led to a significant improvement in the quality of care and to reduced costs.33 While research would be valuable in all ICUs, it is particularly critical for ICUs in tertiary-care centers (where antimicrobial resistance is more likely to be prevalent) and for ICUs caring for subsets of patients with specific underlying illnesses (ie, those likely to have a unique microbial ecology). How then can such research be conducted and should funding be a necessary prelude? We suggest that the implementation of this approach should not require additional institutional resource allocation since infection-control practitioners are already collecting much of the required data. Critical-care providers and intensivists, in conjunction with infection-control practitioners, would, therefore, be the most appropriate personnel to conduct such research in their ICUs. Unfortunately, hospital epidemiology data often are collected haphazardly and in an unfocused fashion that do not enable the ICU clinicians to address the issues listed in Table 1. Thus, the ICU clinician must work in conjunction with the infection-control practitioner to standardize data collection such that data can be used for criticalcare research. A suggested approach to initiating such research in ICUs is outlined in the table. Studies reported from our institution have documented that surveillance for purposes other than infection control is feasible, can be conducted with existing resources, and can generate valuable information.15 The benefits thus accrued are likely not only to impact on patient care

Table 1—Approach to Initiating Mandatory Research in the ICU Research Suggested infections for targeting research efforts Pneumonia Bacteremia Urinary tract infections Focus of research Rates of infections in different ICUs within the same institution Type of predominant infections in the ICU Unique patient populations at risk Underlying diseases and comorbid illnesses Time of onset Predominant pathogens and bacterial flora associated with various infections Antimicrobial susceptibility pattern of microbial agents Changes in epidemiologic trends and antimicrobial susceptibilities over time

Opinions/Hypotheses

and practices but, undoubtedly, on the cost of care, as we have shown.34 By discerning the patient population as risk and the probabilities, patterns, and pathogens associated with specific infections, such studies would ultimately allow rational, focused, and optimal antimicrobial therapy that would be specific to that ICU.

18 19

20

References 1 Brown RB, Hosmer D, Chen C, et al. A comparison of infections in different ICUs within the same hospital. Crit Care Med 1985; 13:472– 476 2 Fraise AP. Epidemiology of resistance in intensive therapy units. J Med Microbiol 1997; 46:447– 449 3 Spencer RC. Epidemiology of infection in ICUs. Intensive Care Med 1994; 20:52–56 4 Fridkin SK, Steward CD, Edwards JR, et al. Surveillance of antimicrobial use and antimicrobial resistance in United States hospitals: project ICARE phase 2. Clin Infect Dis 1999; 29:245–252 5 Archibald L, Phillips L, Monnet D, et al. Antimicrobial resistance in isolates from inpatients and outpatients in the United States: increasing importance of the intensive care unit. Clin Infect Dis 1997; 24:211–215 6 Craven DE, Steger KA. Nosocomial pneumonia in mechanically ventilated adult patients: epidemiology and prevention. Semin Respir Infect 1996; 11:32–53 7 Roder BL, Nielsen SL, Magnussen P, et al. Antibiotic usage in an intensive care unit in a Danish university hospital. J Antimicrob Chemother 1993; 32:633– 642 8 Bergmans DCJJ, Bonten MJM, Gaillard CA, et al. Indications for antibiotic use in ICU patients: a one-year prospective surveillance. J Antimicrob Chemother 1997; 39:527–535 9 A’Court C, Garrard CS. Nosocomial pneumonia in the intensive care unit: mechanism and significance. Thorax 1992; 270:1965–1970 10 Fagon JY, Chastre J, Domart Y, et al. Nosocomial pneumonia in patients receiving continuous mechanical ventilation: prospective analysis of 52 episodes with use of a protected specimen brush and quantitative culture techniques. Am Rev Respir Dis 1989; 139:877– 884 11 Torres A, Aznar R, Gatell JM, et al. Incidence, risks and prognosis factors of nosocomial pneumonia in mechanically ventilated patients. Am Rev Respir Dis 1990; 142:523–528 12 Beck-Sague CM, Sinkowitz RL, Chinn RY, et al. Risk factors for ventilator-associated pneumonia in surgical intensive care unit patients. Infect Control Hosp Epidemiol 1996; 17:374 – 376 13 Sterling TR, Ho EJ, Brehm WT, et al. Diagnosis and treatment of ventilater-associated pneumonia: impact on survival. Chest 1996; 110:1025–1034 14 Dreyfuss D, Mier L, LeBourdelles G, et al. Clinical significance of borderline quantitative protected brush specimen culture results. Am Rev Respir Dis 1993; 147:946 –951 15 Singh N, Falestiny MN, Rogers P, et al. Pulmonary infiltrates in the surgical ICU. Chest 1998; 114:1129 –1136 16 Kollef MH. Ventilator associated pneumonia: a multivariate analysis. JAMA 1993; 270:1965–1970 17 Singh N, Gayowski T, Wagener MM, et al. Pulmonary

21

22

23

24

25

26

27 28 29

30 31 32

33

34

infiltrates in liver transplant recipients in the intensive care unit. Transplantation 1999; 67:1138 –1144 Yu VL. Prevention and control of Legionella: an idea whose time has come. Infect Dis Clin Pract 1997; 6:420 – 421 Weber DJ, Raasch R, Rutala WA. Nosocomial infections in the ICU: the growing importance of antibiotic-resistant pathogens. Chest 1999; 115(suppl):34S– 41S National Nosocomial Infections Surveillance System. National Nosocomial Infections Surveillance (NNIS) report: data summary from October 1986-April 1997, issued May 1997; a report from the NNIS System. Am J Infect Control 1997; 25:447– 487 Vincent JL, Bihari DJ, Suter PM, et al. The prevalence of nosocomial infection in intensive care units in Europe: results of the European Prevalence of Infection in Intensive Care (EPIC) Study. JAMA 1995; 274:639 – 644 Wenzel RP, Thompson RL, Landry SM, et al. Hospitalacquired infections in intensive care unit patients: an overview with emphasis on epidemics. Infect Control 1983; 4:371–375 Craven DE, Kunches LM, Lichtenberg DA, et al. Nosocomial infection and fatality in medical and surgical intensive care unit patients. Arch Intern Med 1988; 148:1161–1168 Daschner FD, Frey P, Wolff G, et al. Nosocomial infections in intensive care wards: a multicenter prospective study. Intensive Care Med 1982; 8:5–9 Miller EH, Caplan ES. Nosocomial Hemophilus pneumonia in patients with severe trauma. Surg Gynecol Obstet 1984; 159:153–156 Rello J, Torres A, Ricart M, et al. Ventilator-associated pneumonia by S. aureus, comparison of methicillin-resistant and methicillin-sensitive episodes. Am J Respir Crit Med 1994; 150:1545–1549 Espersen S, Gabrielsen J. Pneumonia due to S. aureus during mechanical ventilation. J Infect Dis 1981; 144:19 –23 Rello J, Torres A. Microbial causes of ventilator-associated pneumonia. Semin Respir Infect 1996; 11:4 –31 Pingleton SK, Fagan FY, Leeper KV. Patient selection for clinical investigation of ventilator-associated pneumonia, criteria for evaluating diagnostic techniques. Chest 1992; 102(suppl):S553–S556 Langer M, Cigado M, Mandelli M, et al. Early onset pneumonia: a multicenter study in intensive care units. Intensive Care Med 1987; 13:342–346 Yu VL, Stoehr GP, Starling RC, et al. Empiric antibiotic selection by physicians, evaluation of reasoning strategies. Am J Med 1991; 301:165–172 Kollef MH, Sherman G, Ward S, et al. Inadequate antimicrobial treatment of infections: a risk factor for hospital mortality among critically ill patients. Chest 1999; 115:462– 474 Evans RS, Pestotnik SL, Classen DC, et al. A computerassisted management program for antibiotics and other antiinfective agents (see comments). N Engl J Med 1998; 338: 232–238 Singh N, Rogers P, Atwood CA, et al. Short-course empiric therapy for suspected nosocomial pneumonia: a proposed solution for indiscriminate antibiotic prescription for pulmonary infiltrates in the intensive care unit. Paper presented at: 36th Annual Meeting of the Infectious Diseases Society of America; November 14, 1998; Denver, CO; abstract 86

CHEST / 117 / 5 / MAY, 2000

1499