INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nosocomial pneumonia is a frequent complication among severely ill hospitalized patients, and it remains an important cause of morbidity and mortality (1). In the hospital setting, rapid administration of appropriate antibiotics is important because many nosocomial pathogens cause rapidly progressing pneumonias. Initial empiric antibiotic regimens are usually chosen to cover the major pathogens. Gram-negative bacilli have been repeatedly documented as a major cause of nosocomial pneumonia (4, 5); however, gram-positive bacteria, and, in particular, Staphylococcus aureus, are playing an increasing causative role. In the United States, the results of the National Nosocomial Infections Study (NNIS) have shown that S. aureus is now the leading cause of this disease, followed by Pseudomonas aeruginosa and Enterobacter spp (6); S. aureus is also the leading cause of nosocomial pneumonia in Europe, as shown by a large prevalence study conducted in more than 1,400 Intensive Care Units (ICUs) (7, 8). Consequently, antimicrobial agents active against S. aureus should be included in the empiric therapy of nosocomial pneumonia.

Antimicrobial selection in cases occurring more than 4 d after admission (late-onset pneumonia) remains difficult for at least two reasons. First, multiple organisms frequently grow from cultures of pulmonary secretions in the target population, especially in intubated patients (9); in addition, infections are likely to result from highly resistant organisms, especially in patients who previously received antibiotics (4, 12). In addition to P. aeruginosa, Acinetobacter spp, and multidrug-resistant enterobacteriaceae, methicillin-resistant, S. aureus (MRSA) is prevalent in many institutions (13). Consequently, a recent consensus statement by the American Thoracic Society suggests that patients with severe late-onset hospital-acquired pneumonia receive an antimicrobial regimen effective against both P. aeruginosa and MRSA if MRSA is endemic in the hospital (3). Furthermore, physicians caring for hospitalized patients at risk for acquiring S. aureus are concerned by the recent isolation in Japan and in the United Sates of strains of S. aureus with intermediate in vitro resistance to vancomycin even if such strains do not represent, for the moment, an important clinical problem (17). Concomitantly, the U.S. Centers for Disease Control and Prevention have developed guidelines to encourage appropriate use of vancomycin because of concerns that excessive vancomycin use promotes the emergence and spread of vancomycin-resistant enterococci (20).

Quinupristin/dalfopristin (Synercid) is a semisynthetic parenteral streptogramin consisting of two components that produce a synergistic, in vitro effect against a wide spectrum of gram-positive pathogens. The two major components of quinupristin/dalfopristin are quinupristin (30%), a Group B streptogramin, and dalfopristin (70%), a Group A streptogramin. Streptogramins are members of the macrolide-lincosamide-streptogramin (MLS) group of antibiotics because of similarities in mechanism of action, i.e., inhibition of bacterial protein synthesis. Quinupristin/dalfopristin's spectrum of activity includes Streptococcus pneumoniae (including penicillin-resistant strains), S. aureus (methicillin-resistant and erythromycin- resistant strains), coagulase-negative staphylococci (methicillin-resistant and erythromycin-resistant strains), and Enterococcus faecium (glycopeptide-resistant strains) (21). Quinupristin/ dalfopristin has potent in vitro bactericidal activity against S. pneumoniae and methicillin-susceptible staphylococci but is moderately active against some strains of methicillin-resistant staphylococci as well as against most clinical E. faecium strains. Quinupristin/dalfopristin is not active against Enterococcus faecalis and the majority of gram-negative enteric bacilli or P. aeruginosa.

If quinupristin/dalfopristin were efficacious in the therapy of nosocomial pneumonia, it could address the medical need for alternatives to existing therapies, for example, in the setting of patient allergy or intolerance to glycopeptides and other agents. This prospective, randomized, comparative, open-label, multicenter study was undertaken to evaluate the efficacy and safety of intravenously administered quinupristin/dalfopristin compared with vancomycin in the treatment of nosocomial pneumonia caused, at least in part, by gram-positive bacteria.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Patients were enrolled at 74 centers in France, Germany, Spain, Sweden, and the United States. Patients at least 18 yr of age were eligible if they developed nosocomial pneumonia sufficiently severe to require at least 5 d of parenteral antibiotic therapy. Nosocomial pneumonia was defined by the following signs and symptoms appearing more than 48 h after hospital admission: temperature of at least 38° C or less than 36° C; new or persistent pulmonary infiltrate on a chest radiograph; and at least two of the following: chills, dyspnea, cough, inspiratory or pleuritic chest pain, purulent respiratory secretions, abnormal chest auscultation, altered mental status, leukocytosis of more than 10,000/mm³, or hypoxemia. The decision to include patients was based on clinical findings and positive culture from expectorated sputum or endotracheal aspirate demonstrating > 25 PMN and > 10 squamous epithelial cells/low power field, transtracheal aspirate, bronchoalveolar lavage with a threshold of 10⁴ cfu/ml, protected specimen brush (threshold  10³ cfu/ml), or pleural fluid. A Gram stain was used to include those patients with at least one gram-positive organism.

Patients were excluded if they were pregnant or lactating, had a life expectancy of less than 1 mo, or had pneumonia caused exclusively by organisms other than gram-positive pathogens. Also excluded were patients who had received effective systemic antimicrobial therapy for more than 24 h within 7 d before enrollment, had significant neutropenia (less than 500/mm³), underlying immunocompromising disease (HIV-positive status with a CD4 count < 200/µl, splenectomy) or therapy (patients receiving > 40 mg/d of corticosteroids or other immunosuppressive therapy), or had documented allergy to streptogramin, glycopeptide, or beta-lactam antibiotics. Written, signed, informed consent approved by each center's institutional review board (IRB) was obtained from the patient or his/her legally authorized representative before assignment to a treatment group.

Study Design

Prior to randomization, patients were stratified based on intubation status. Treatment was assigned by telephone using a centralized stratified randomization scheme to minimize any imbalance of treatment assignment within each center and between centers.

Patients were randomized to receive intravenous therapy with either 7.5 mg/kg of quinupristin/dalfopristin every 8 h or 1 g of vancomycin every 12 h. Treatment duration was from 5 to 14 d. Aztreonam was administered at a dosage of 2 g every 8 h to both treatment groups to cover gram-negative organisms. Antibiotic treatments were adjusted according to culture results. Imipenem, at a recommended dosage of 500 mg every 6 h, was permitted if aztreonam-resistant gram-negative pathogens were isolated. If a mixed infection with P. aeruginosa was present then 5 d of treatment with Tobramycin at a dosage of 1.5 mg/ kg every 8 h was recommended.

Definition of Populations

All-treated population. All patients who received any amount of study medication were included in this population.

All-treated population with a baseline pathogen. This population included all patients who received any amount of study medication and had a baseline causative pathogen isolated from a respiratory tract specimen or blood.

Bacteriologically evaluable population. To be bacteriologically evaluable, patients had to have had a baseline causative pathogen isolated from a respiratory tract specimen or blood and receive at least 5 d of study medication to be classified as treatment cure or improvement, or at least 3 d of study medication to be classified as failure. Any patients dying before receiving at least 3 d of study medication were not excluded or classified in the bacteriologically evaluable group but were included in the all-treated group and were classified as failures. Patients also must have completed a test-of-cure assessment (between 7 and 13 d after end-of-treatment in case of cure or improvement or between end-of-treatment and 13 d after the end-of-treatment in case of failure) and not have received any other systemic antibacterial therapy for nonpneumonial infections to which causative pathogens were susceptible before the test-of-cure assessment.

Definition of Efficacy and Safety Parameters

The primary efficacy parameter was clinical response at the test-of-cure assessment in the bacteriologically evaluable population. Major secondary end points included clinical response in the all-treated population, and by-pathogen and by-patient bacteriologic responses in the bacteriologically evaluable population at the test-of-cure assessment.

Clinical response. The clinical response was determined by assessing signs and symptoms, and by comparing chest radiographs from the test-of-cure assessment with those obtained at baseline. Clinical responses were classified as: Cure, disappearance of signs and symptoms related to the infection; Failure, requirement for a systemic antibacterial therapy for pneumonia before test-of-cure or requirement for intubation and mechanical ventilation because of clinical worsening after at least 48 h of study treatment (patients who received other antibodies were excluded); Improvement, neither cure nor failure; Indeterminate, no evaluation possible. An indeterminate efficacy response was defined as the inability to assess the signs and symptoms because of lack of information or interference of the assessment by concomitant medical or surgical conditions. Indeterminate response was considered a failure.

By-pathogen bacteriologic response. The by-pathogen bacteriologic response was evaluated from the outcome of all pathogens isolated from baseline specimens or identified from baseline serologic assays, as well as from the outcome for pathogens isolated later during the study. This response was classified as: Eradication, elimination of baseline pathogens; Presumed Eradication, no material available for culture and clinical response of cure or improvement; Satisfactory Reduction, 2-log reduction or greater in the colony counts; Persistence, continued isolation of the causative pathogens without the criteria of satisfactory reduction; Presumed Persistence, no material available for culture and clinical failure; Indeterminate, bacteriologic response to the study drug not evaluate.

By-patient bacteriologic response. The by-patient bacteriologic response was the combination of all by-pathogen bacteriologic responses for each patient.

Safety Parameters

All adverse events or laboratory abnormalities occurring during treatment or within 30 d of discontinuation of the study medication were recorded and assessed for severity and relationship to the study drug by the investigator. Peripheral adverse venous events were assessed separately. Patient survival was recorded during therapy and for 30 d after drug discontinuation.

Statistical Analyses

The target sample size of 350 patients was chosen to ensure with a probability of at least 80% that the lower limit of the 95% confidence interval around the difference between the success rates would not exceed 20% if the treatments were in fact equivalent. This sample size was estimated to provide 80 evaluable patients per group, assuming an evaluability rate of approximately 50%.

The treatments were considered equivalent if the lower limit of the 95% confidence interval for the difference in clinical success rates (quinupristin/dalfopristin versus vancomycin; success rate was defined as cure plus improvement) exceeded a specific value determined by the higher observed response rate, following the recommendations of the 1992 Food and Drug Administration Division of Anti-infectives Drug Products "Points to Consider, Number 2." This value was 20% if the higher response rate was less than 80%, 15% if it was between 80% and 89%, and 10% if the higher response rate was at least 90%.

Efficacy responses and evaluability status were determined by computer algorithms prior to unblinding the study in order to avoid the introduction of bias into the evaluation of this open-label study. For those patients whose evaluability status and/or efficacy response were questionable, a steering committee composed of four investigators reviewed the available data and determined, in a blinded fashion, the evaluability and/or efficacy response.

Stepwise logistic regression analysis was performed to identify significant prognostic factors. Odds-ratios and level of significance were calculated. The distribution (presence or absence) of the risk factors between the two treatment groups was tested using the chi-square test.

The study was a collaborative effort between the investigators and the industrial sponsor. The investigators were responsible for gathering, analyzing, and interpreting the data; writing the manuscript; reporting the results of the study; and submitting the manuscript for publication.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Seventy-four centers contributed to the study, randomizing a total of 304 patients between June 1995 and June 1996. The mean number and range of patients per center is shown in Table 1. Six patients never received the study medication after being randomized; the other 298 patients received either quinupristin/dalfopristin (150 patients) or vancomycin (148 patients). Of these 298 patients, 171 (87 in the quinupristin/dalfopristin group and 84 in the vancomycin group) met the predetermined criteria for bacteriologic evaluability. Reasons for excluding patients from the bacteriologically evaluable population were similar in the two treatment groups (Figure 1). Patients excluded from the bacteriologically evaluable population were included in the analysis in the all-treated population.

View larger version (33K): [in this window] [in a new window]   Figure 1.   Flow of participant and reasons for withdrawals. Definition of abbreviations: Q/D = quinupristin/dalfopristin; V = vancomycin; a = prohibited antibiotic prior treatment; b = concomitant prohibited antibiotic; c = inadequate duration of treatment; d = indeterminate response; e = no valid baseline pathogen; f = test-of-cure assessment too early or too late; g = insufficient signs and symptoms at baseline; h = hospitalized < 48 h prior to onset of signs and symptoms.

Overall, quinupristin/dalfopristin-treated and vancomycin-treated patients were comparable at baseline with respect to demographic, prognostic, and risk factors (Tables 2a and 2b). This was true for the all-treated population (overall and intubated at baseline), the all-treated population with baseline pathogens (monomicrobial and polymicrobial pneumonia), and the bacteriologically evaluable population (overall and intubated at baseline). In the all-treated population intubated at baseline, bilateral pneumonia was somewhat more frequent in the quinupristin/dalfopristin group (46.4%) than in the vancomycin group (31.5%; p = 0.023), and bacteremic pneumonia was more frequent in the vancomycin group (17.6%) than in the quinupristin/dalfopristin group (6.3%; p = 0.009). In the bacteriologically evaluable population, bacteremic pneumonia was more frequent in the vancomycin group (20.2%) than in the quinupristin/dalfopristin group (8.0%; p = 0.022). Pneumonia that was gram-positive alone (24 in the quinupristin/ dalfopristin group versus 23 in the vancomycin group), gram-negative alone (27 in the quinupristin/dalfopristin group versus 20 in the vancomycin group), or mixed gram-positive and gram-negative (36 in the quinupristin/dalfopristin group versus 41 in the vancomycin group) was similarly distributed across both groups.

Pathogenic Organisms

A total of 173 pathogens were identified in each treatment group. Causative gram-positive pathogens were similarly distributed between groups in the all-treated and in the bacteriologically evaluable populations (Table 3). In the all-treated population, S. aureus represented 39.9% of isolates in the quinupristin/dalfopristin group and 39.9% in the vancomycin group. In the bacteriologically evaluable population, S. aureus represented 38.5% and 39.6%, respectively. In the all-treated population, 162 of 346 baseline pathogens (47.0%) were isolated by invasive procedures, including protected specimen brushes, distal protected specimens, and bronchoalveolar lavages. Similar findings were observed in the bacteriologically evaluable population (data not shown).

Treatment

In the all-treated population, the mean duration of treatment was 10.1 ± 4.0 d in the quinupristin/dalfopristin group and 9.5 ± 4.1 d in the vancomycin group (p = 0.662). All patients were treated with aztreonam. In the quinupristin/dalfopristin group, 20 patients were treated with imipenem and 17 patients were treated with tobramycin; in the vancomycin group, 17 patients were treated with imipenem and 21 patients were treated with tobramycin. The distribution of treatments with aztreonam, imipenem, and tobramycin was comparable for both treatment groups. In the bacteriologically evaluable population, nine (10.3%) patients in the quinupristin/dalfopristin group and 12 (14.2%) patients in the vancomycin group received imipenem or tobramycin to which the gram-positive organisms isolated at baseline were susceptible. Study medication was administered only via a central venous catheter in 83 (53.3%) quinupristin/dalfopristin-treated and 91 (61.5%) vancomycin-treated patients (p = 0.28).

Efficacy

Analysis of clinical and bacteriologic responses in the bacteriologically evaluable population and the all-treated population demonstrated that quinupristin/dalfopristin produced a success rate equivalent to that of vancomycin (Table 4).

Clinical response. In the bacteriologically evaluable population, the clinical success rate in the quinupristin/dalfopristin treatment group was not different from that in the vancomycin group (56.3 versus 58.3%; 95% CI, 16.8 to 12.8). Equivalence between the treatment groups was also demonstrated in the bacteriologically evaluable population intubated at baseline (54.2% in the quinupristin/dalfopristin group versus 53.7% in the vancomycin group; 95% CI, 16.1 to 17.0).

Comparability of clinical response between treatment groups was also observed in different subsets of patients based on demographic variables, presenting severe conditions considered as prognostic and risk factors (Table 5). In the bacteriologically evaluable population intubated at baseline, multiple organisms were isolated from 31 of 72 (43%) quinupristin/dalfopristin-treated patients and from 34 of 67 (51%) vancomycin-treated patients. Clinical success rates were 54.8% with quinupristin/dalfopristin and 47.1% with vancomycin in cases of polymicrobic infection, and 53.7 and 60.6% in cases of monomicrobic infection. In addition, analysis of early-onset (less than 5 d after hospital admission) and late-onset (5 d or more after admission) pneumonia in the bacteriologically evaluable population intubated at baseline failed to demonstrate any significant difference in the clinical success rates (66.6% in the quinupristin/dalfopristin group and 70.0% in the vancomycin group for early onset pneumonia; 50.0 and 46.8%, respectively, for late onset pneumonia).

Clinical success rates for patients with selected gram-positive baseline pathogens in the bacteriologically evaluable population were similar among patients with documented quantitative cultures at baseline (data not shown). Clinical success rates among bacteriologically evaluable patients with MRSA pneumonia were also similar between the treatment groups, although success rates were somewhat lower among patients with MRSA pneumonia regardless of whether they were randomized to quinupristin/dalfopristin or to vancomycin (Table 6).

Bacteriologic response. Bacteriologic responses in the bacteriologically evaluable population were comparable between treatment groups for the entire population as well as for subgroups defined by type of infection (monomicrobic versus polymicrobic), type of pathogen, and presence of bacteremia (data not shown).

Multivariate analysis. A stepwise logistic regression analysis was performed on the bacteriologically evaluable population. The variables tested in the model were: study drug treatment group, age, sex, race, APACHE II score, congestive heart failure, smoking status, chronic lung disease, diabetes mellitus, ventilation status, intubation status, ICU admission at enrollment, multilobar pneumonia, pleural effusion, time between onset of signs and symptoms and treatment, and respiratory rate. Six variables were associated with a decreased likelihood of clinical success in both treatment groups: age older than 65 yr (odds ratio [OR], 0.45; p = 0.0001), chronic lung disease (OR, 0.38; p = 0.021), diabetes mellitus (OR, 0.22; p = 0.012), mechanical ventilation for longer than 5 d during the study (OR, 0.44; p = 0.025), multilobar pneumonia (OR, 0.45; p = 0.033), and bacteremic pneumonia (OR, 0.37; p = 0.062). Treatment regimen assignment was not an independent factor associated with clinical or bacteriologic failure.

Decreasing in vitro susceptibility to quinupristin/dalfopristin and vancomycin. Three bacteriologically evaluable patients, two in the quinupristin/dalfopristin group and one in the vancomycin group, exhibited decreasing susceptibility of the causative S. aureus to the respective study drug during the study.

Safety

Adverse events. The number of patients with nonvenous and venous adverse events was comparable between the two groups. In the quinupristin/dalfopristin group, 145 (96.7%) patients with nonvenous adverse events and 36 (24.0%) patients with venous adverse events were observed, whereas the incidence was 138 (93.2%) and 29 (19.6%) in the vancomycin group, respectively (Table 7). Related venous adverse events exhibited a tendency to be more frequent in the quinupristin/ dalfopristin group than in the vancomycin group (28 and 16, respectively; p = 0.056).

Premature discontinuation from study treatment. Treatment with the study drug was discontinued prematurely because of nonvenous adverse events for 20 (13.3%) of the 150 quinupristin/dalfopristin patients and 14 (9.5%) of 148 patients in the vancomycin group (p = 0.293). Study discontinuation because of venous adverse events was observed in three patients from the quinupristin/dalfopristin group.

Mortality. There were 70 deaths in the all-treated population while patients were receiving study drug or within 30 d after therapy. Of these deaths, 38 were in the quinupristin/dalfopristin group and 32 were in the vancomycin group (mortality rates: 25.3 and 21.6%, respectively; p = 0.45). One death in the quinupristin/dalfopristin group and none in the vancomycin group were possibly related to study medication, as assessed by the investigators.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study compared the efficacy and safety of quinupristin/ dalfopristin versus vancomycin in the treatment of hospital-acquired gram-positive pneumonia. Quinupristin/dalfopristin in a dose of 7.5 mg/kg intravenously every 8 h was found to produce a clinical success rate equivalent to that of vancomycin in bacteriologically evaluable patients (primary efficacy parameter of the trial) as well as in the all-treated population. This result was consistently observed in subgroups of the bacteriologically evaluable population: patients intubated at baseline, patients with early-onset and late-onset pneumonia, and patients with monomicrobic and polymicrobic pneumonia. Response rates in both treatments groups were relatively low for the modest numbers of patients with pneumonia due to methicillin-resistant strains of S. aureus. The frequencies of nonvenous adverse events and mortality rates were comparable between treatment groups. Quinupristin/dalfopristin treatment was associated with a somewhat higher frequency of related venous adverse events.

Among the strengths of the present study were the frequent use (47%) of quantitative culture techniques, which enhanced the quality of diagnostic criteria and bacteriologic data, including distal protected specimen, bronchoalveolar lavage, and protected specimen brush; stratification at enrollment by intubation status at baseline; and determination of efficacy responses and evaluability status, either algorithmically or by an independent study steering committee, before unblinding of the randomization (27). This design provided a well-characterized, relatively homogeneous population of hospitalized patients with clinically and bacteriologically well-documented nosocomial pneumonia. It is noteworthy that a large proportion of patients were enrolled in intensive care units and in trauma centers and that 220 (74%) of 298 of the all-treated population were intubated at baseline.

The ATS (3) guidelines for the treatment of nosocomial pneumonia indicate that initial therapy be empirical using a combination of antimicrobial treatments. Because our study was designed to evaluate patients with gram-positive nosocomial pneumonia, we chose aztreonam as the first-line companion agent to treat gram-negative pathogens because of its specific activity against these organisms. Investigators were authorized to discontinue aztreonam for nosocomial pneumonia caused exclusively by gram-positive organisms and to replace aztreonam by imipenem with or without tobramycin in the case of nosocomial pneumonia due to aztreonam resistant pathogens. Both aztreonam and imipenem are recommended drugs to treat nosocomial pneumonia in the ATS guidelines, despite epidemiologic evidence of gram-negative resistance in some units.

To our knowledge, this trial represents the largest therapeutic study of gram-positive nosocomial pneumonia performed to date (28). This study focused on patients with suspected or proven gram-positive nosocomial pneumonia, and gram-positive pathogens, both associated with other pathogens or alone, accounted for nearly 60% of the total pathogens identified in the all-treated and the bacteriologically evaluable populations. As a consequence, S. aureus was the most frequently identified pathogen at baseline, comprising almost 40% of total identified isolates.

The primary end point of this study was the clinical response in the bacteriologically evaluable population. The study results demonstrate the equivalence of quinupristin/dalfopristin and vancomycin in treating patients with gram-positive nosocomial pneumonia. Interestingly, such an equivalence was found for the all-treated population as well as for patients intubated at inclusion (the stratification criterion). It was also found in all tested subgroups of patients defined by the time of occurrence (early-onset and late-onset pneumonias), by the number of pathogens involved (monomicrobic and polymicrobic pneumonias), by the use of invasive diagnostic techniques, and by the type of causative gram-positive pathogens (S. aureus and S. pneumoniae). Similar results were observed in these same subgroups for by-pathogen bacteriologic response. Finally, the number of MRSA cases was relatively small, but the clinical success rates are comparable between quinupristin/dalfopristin and vancomycin groups (6/20 versus 8/18; 14.4 [44.9-16.1]).

The success rates (55 to 60%) in both treatment groups were lower than might be expected. The comparator clinical success rate had been anticipated to be 75%. Several reasons explain this low success rate. First, the population of patients studied in this trial was severely ill, with a high percentage of patients hospitalized in ICUs, intubated, and treated with mechanical ventilation, and with significant underlying disease, as confirmed by an overall high APACHE II score and the high overall mortality rate around 25% (31). Second, although the addition of a new systemic antimicrobial treatment before the test-of-cure assessment required the patient to be considered a failure, as is usual in such trials evaluating antimicrobial agents (26), such a criterion merits careful examination in ICU patients who are likely to have persistent colonization of the lower airways with the same pathogen or with a "superinfecting" pathogen, whereas the clinical signs and symptoms of pneumonia appear to have improved or resolved (32). Finally, it must be noted that comparably low response rates have been reported in other recent studies (35, 36). Further confirmation of equivalence of the two drugs was provided by the results of multivariate analyses of factors as independent contributors to an untoward outcome. The factors identified by the model are all known to be of significant prognostic value in severely ill patients: age, severity of underlying medial conditions, duration of mechanical ventilation, or severe pneumonia such as multilobar or bacteremic pneumonia (3, 37). These analyses failed to identify study drug treatment assignment as an independent prognostic factor.

The number of patients with at least one nonvenous adverse event was comparable across the two groups. However, a higher percentage of patients reported a nonvenous adverse event with a possible relationship to the study medication in the quinupristin/dalfopristin group than in the vancomycin group. This difference appeared to be predominantly due to a greater incidence of gastrointestinal (diarrhea, vomiting, anorexia) and skin events in the quinupristin/dalfopristin group. Arthralgias and myalgias have been observed in other studies of quinupristin/dalfopristin (38), but were rarely documented in the current study, possibly because of sedation associated with the patients' intubated status. The severity of adverse events was similar in the two groups. The overall mortality rate was comparable in the two groups, and it appeared correlated with patient severity of illness and the severity of pneumonia. In the same way, the overall percentage of all-treated patients with peripheral venous intolerance at least once during the study treatment was comparable between groups. Laboratory adverse events of note in other studies of quinupristin/ dalfopristin have included hyperbilirubinemia (primarily the conjugated fraction), postulated to result from competitive inhibition of excretion.

While it is well known that widespread vancomycin use promotes emergence of vancomycin-resistant organisms, most notably enterococci, as with any antimicrobial, excessive use of quinupristin/dalfopristin may limit its future utility. The impact of quinupristin/dalfopristin on antibiotic resistance patterns in the hospital setting is less clear and requires further evaluation; however, data from clinical studies indicate that the rate of emerging resistance is comparable to that seen with other classes of antimicrobial agents. The present study compared only vancomycin and quinupristin/dalfopristin. The relative clinical benefits of other agents that retain activity against some vancomycin-resistant organisms, including trimetroprim-sulfamethoxazole, tetracyclines, rifampin, and aminoglycosides, are not known. Decisions regarding therapy of gram-positive infections must include consideration not only of efficacy, but also toxicity, selection of resistant organisms, and other factors.

In summary, this study demonstrated that quinupristin/dalfopristin is equivalent to vancomycin for treating patients suspected of having nosocomial pneumonia caused, at least in part, by gram-positive pathogens. Patients with S. aureus pneumonia, and specifically patients with pneumonia caused by MRSA, had comparable clinical response rates. However, for nosocomial pneumonia cased by MRSA, additional data from future studies would be of interest. Besides factors relating to the patient's condition, the major risk factors for clinical failure related to pneumonia with either regimen were multilobar involvement and bacteremic lung infection. Quinupristin/dalfopristin represents a new and effective alternative therapeutic approach in hospitalized patients with gram-positive nosocomial pneumonia.