| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Pulmonary and Critical Care Section (Dr. Shorr), Department of Medicine, Washington Hospital Center, Washington, DC; the Critical Care Department (Drs. Bodi, Rodriquez, and Rello), Joan XXIII University Hospital, Tarragona, Spain; the Critical Care Department (Dr. Sole-Violan), Hospital Negrin, Las Palmas, Spain; and Critical Care Department (Dr. Garnacho-Montero), Hospital Virgen del Rocio, Seville, Spain.
A list of CAPUCI Study Investigators is given in the Appendix.
Correspondence to: Andrew F. Shorr, MD, MPH, FCCP, Pulmonary and Critical Care Medicine, Room 2A-38D, Washington Hospital Center, Washington, DC 20010; e-mail: afshorr{at}dnamail.com
Abstract
Objective: Multiple guidelines exist to aid clinicians in choosing antibiotics to treat patients with severe community-acquired pneumonia (SCAP). Our goal was to assess the impact of following these guidelines, such as those from the Infectious Disease Society of America (IDSA), on the duration of mechanical ventilation (MV).
Design: Analysis of a prospective registry.
Setting: Multiple ICUs in Spain.
Patients: ICU patients with SCAP requiring 
24 h of endotracheal intubation and surviving their ICU course.
Interventions: None.
Measurements and main results: Demographics, comorbid diseases, severity of illness, and process of care variables were recorded. The duration of MV in patients receiving an antibiotic regimen consistent with IDSA guidelines was compared to patients with prescriptions not in accordance with IDSA recommendations. In the cohort (n = 199), Streptococcus pneumoniae was the most frequent pathogen, and unadjusted analysis showed that the duration of MV was longer in persons receiving IDSA-noncompliant regimens (11 days vs 10 days). In a multivariate hazard model, two variables were independently associated with greater durations of MV: development of acute renal failure (hazard ratio, 1.47; 95% confidence interval [CI], 1.02 to 2.12), and prescription of an IDSA-noncompliant regimen (hazard ratio, 1.40; 95% CI, 1.02 to 1.93). Adjusted analysis indicated that patients receiving treatment that was not compliant with IDSA guidelines received MV an added 3 days.
Conclusion: Failure to follow antibiotic recommendations for the treatment of SCAP may increase the need for continuing MV. Conversely, guideline compliance could represent a surrogate marker that captures other aspects of clinical care, rather than be the direct factor leading to better outcomes. Nonetheless, given the costs associated with MV, enhanced guideline compliance may represent a means for improving outcomes and enhancing resource utilization.
Key Words: antibiotics • community-acquired pneumonia • guidelines • mechanical ventilation • outcomes
Community-acquired pneumonia (CAP) is a leading cause of death from infection.12 Furthermore, CAP results in nearly 1 million hospitalizations annually in the United States.12 Over the last decade, multiple risk stratification schemes have been developed to help the clinician determine which patients with CAP require hospital admission and who will likely need ICU treatment.34 Accurate initial risk stratification is key in CAP management, since persons admitted to the ICU with severe CAP (SCAP) face mortality rates approaching 30%.12 Although population-based analyses reveal substantial variation in the proportion of CAP patients admitted to the ICU, those with SCAP consume disproportionate health-care resources relative to hospitalized but less severely ill patients with CAP. For example, 60 to 90% of those with SCAP eventually require mechanical ventilation (MV), a resource-intense intervention.5 Additionally, Angus and colleagues6 estimated that costs of care for SCAP in the ICU are nearly $30,000 per patient (adjusted to 2005 US dollars). Similarly, Kaplan et al,7 in an evaluation of a large Medicare database, concluded that patients with CAP needing MV have an average hospital length of stay (LOS) of approximately 16 days. From a US perspective, these authors7 calculated that the national cost of SCAP among the elderly was $2.1 billion (1997 dollars) annually.
Studies of process of care indicate that both the timeliness and appropriateness of antibiotics correlate with survival in CAP.89 Moreover, multiple guidelines from professional medical societies make specific recommendations for antibiotic therapy.12 In several disease states such as congestive heart failure and acute myocardial infarction, failure to comply with national treatment guidelines has been shown to increase both mortality and cost.1011 With respect to CAP, investigators have documented similar relationships. Battleman et al,12 for instance, described a link between both delay in initial antibiotic therapy and failure to prescribe an appropriate antibiotic and prolonged hospital stay in CAP. In SCAP, less is known about the effect of process of care and guideline compliance and their nexus with measures of resource use such as ICU LOS and duration of MV.
To date, there have been no investigations of the interaction between antimicrobial guideline compliance and measures of morbidity for SCAP in the ICU. One could hypothesize that antibiotic guideline compliance in SCAP might actually lead to higher overall hospital costs if it resulted in the use of more broad-spectrum, expensive antimicrobials, or if it decreased hospital mortality and thus prolonged LOS in patients who otherwise would have died soon after admission. Alternatively, abiding by formal antibiotic prescribing recommendations in SCAP could enhance resource use and efficiency by leading to critically ill patients resolving their illness more rapidly. To explore this question, we conducted a secondary analysis of a large, prospective registry of persons with SCAP needing ICU admission and focused on how compliance with the recommendations of the Infectious Disease Society of America (IDSA) affected the duration of MV.
Materials and Methods
Study Overview and Subjects
We conducted a retrospective analysis of a multicenter prospective registry of SCAP. Details of this registry have been described previously.13 Briefly, all patients admitted to one of 33 hospitals in Spain between December 1, 2000, and February 28, 2002, were evaluated for enrollment. To be included in the analysis, patients had to present with a clinical syndrome consistent with CAP (see definition below) and be admitted to the ICU. Patients were admitted to the ICU for either MV or because the clinician believed the individual was otherwise unstable. Admission criteria were not standardized across the multiple participating study sites, nor was the use of noninvasive ventilation. Patients were excluded from the registry if they were admitted from a nursing home, had recently been hospitalized, or met the definition for health-care–associated pneumonia as per the recent American Thoracic Society (ATS)/IDSA guidelines regarding nosocomial pneumonia.14 We additionally excluded persons with SCAP for whom pneumonia, based on the impression of the site investigator, was considered to be a terminal event. Since we focused on the use of MV, we further required that for this analysis subjects receive at least 24 h of MV. Hence, those admitted to the ICU but never needing invasive MV delivered via an endotracheal tube were not included. Additionally, since our objective addressed length of MV we restricted the cohort to individuals surviving the episode of MV. The initial evaluation of this data dealt primarily with mortality.13 In general and in CAP specifically, survivors and nonsurvivors have been shown to have differential patterns of health-care utilization.7 Pooling survivors with eventual nonsurvivors can confound assessments exploring resource use. The timing of death can be influenced by a number of factors, including the aggressiveness with which clinicians seek to limit care. However, persons dying in the ICU may receive overly aggressive care despite its likely futility, or they may receive MV for the entire duration of their illness. These forces act, in turn, to systematically alter how long patients may receive MV. Additionally, since a mortality benefit was shown with guideline compliance in an earlier assessment of this data, we opted to specifically focus on issues among survivors.13 Because of the observational nature of the initial study, local ethics committees waived the requirement for informed consent.
End Points
Duration of MV after initial endotracheal intubation served as the primary end point. This was defined as the time from initiation of MV support to spontaneous breathing. The number of days that the patient was alive and not receiving MV (MV-free days) served as a secondary end point. Again, because of potential confounding due to timing of death in those who eventually died, and because patients dying during their ICU stay tend to consistently consume greater resources than those who survive, we restricted the assessment to patients discharged from the ICU alive.
Covariates
Using standardized data recording tools, patient demographics, chronic health state, process of care, antimicrobial administration, and severity of illness were recorded. Specific comorbidities of interest included the presence of obesity, diabetes mellitus, COPD, cardiac disease, neurologic illness, chronic renal and liver disease, malignancy, immunosuppression, and HIV seropositivity. To evaluate the process of care, time from patient presentation to each of the following was assessed: antibiotic administration, drawing of blood for cultures, pulse oximetry, and arterial blood gas (ABG) analysis. With respect to antimicrobial therapy, we determined if the medications administered were in accordance of the recommendations of the IDSA, if they were active against the culprit pathogen based on in vitro sensitivity testing, and if the clinician later had to alter the initial antibiotic regimen. During the time of this study, the IDSA recommended ceftriaxone, cefotaxime, ampicillin-sulbactam, or piperacillin-tazobactam in combination with a fluoroquinolone or macrolide.1 We measured severity of illness using the APACHE (Acute Physiology and Chronic Health Evaluation) II score.15 We further noted if either shock or acute renal failure developed and if the patient had concomitant bacteremia or empyema. Finally, we assessed if rapid radiograph spread of pulmonary infiltrates evolved.
Definitions
All definitions were chosen prospectively as part of the original registry project. CAP specifically was defined as an acute lower respiratory tract infection. Three criteria had to be met for a CAP diagnosis: (1) presence of an acute pulmonary infiltrate on the chest radiograph and compatible with a pneumonic process, (2) a clinical examination consistent with CAP, and (3) acquisition of the infection in a non–health-care setting (eg, the patient could not be hospitalized or a patient in either a long-term care facility or nursing home). We employed previously published guidelines to identify patients with COPD,16 and considered neurologic compromise to be present if there was an alteration of consciousness (eg, confusion).13 Any patient with either a primary immunodeficiency syndrome or a secondary immunodeficiency due to radiation therapy or use of cytotoxic agents, chemotherapy, or corticosteroids (daily dose equivalent to > 20 mg of prednisolone for at least 2 weeks) or AIDS was classified as "immunosuppressed." Patients requiring vasopressors for > 4 h despite adequate fluid resuscitation were labeled as having shock, while we considered acute renal failure to be present if the urine output measured < 20 mL/h or if the urine output was < 80 mL > 4 h. Rapid radiographic spread required an increase in the size of the pulmonary opacities by > 50% at 48 h after presentation.
One external investigator (an attending level physician with training in critical care) reviewed all antibiotic prescriptions to evaluate if these were consistent with IDSA recommendations. Appropriate antibiotic therapy was defined as treatment with at least one agent to which all recovered isolates were susceptible in vitro (or were expected to be susceptible for Pneumocystis jiroveci or Legionella pneumophila).
Sensitivity Analysis
Because the IDSA guidelines generally exclude patients known to be immunosuppressed, we completed a second analysis. In this instance, we excluded the patients meeting the predefined definition for immunosuppression (see above) and reanalyzed our primary end point. The purpose of this was to determine if and how the inclusion of immunosuppressed patients affected our observations.
Statistics
We compared continuous variables with Student t test for parametrically distributed variables or with the Mann-Whitney U test for nonparametrically distributed data. Categorical variables were analyzed with either a 
2 test or a Wilcoxon rank-sum test, as appropriate. To assess the probability of remaining on MV after the diagnosis of CAP as a function of IDSA guideline compliance, we created survival curves in accordance with the methods of Kaplan and Meier and employed a log-rank test. All tests were two tailed, and a p value 
0.05 was assumed to represent statistical significance.
To determine the independent impact of guideline observance on duration of MV on SCAP, we created a Cox proportional hazards model with days of MV as the outcome variable. After confirming that the proportional hazard assumption was not violated, we entered all factors that were significant in the univariate analysis at the 0.20 level into the model. Candidate variables were removed in a stepwise fashion if p > 0.10. Colinearity was assessed via correlation matrices. Ninety-five percent confidence intervals (CIs) around significant hazard rations are also reported. All analyses were conducted using statistical software (SPSS 11.0; SPSS; Chicago, IL).
Results
The initial cohort comprised 529 patients with SCAP. Of these patients, 148 patients died, while 182 patients (174 never ventilated) required < 24 h of MV. This left a final study population of 199 persons. The median age of this group was 63 years (range, 18 to 84 years), and approximately 70% were male. Overall, the median APACHE II score was 18 (range, 14 to 44). Table 1 demonstrates the distribution of pathogens isolated in the population. A culprit organism was identified in 112 patients (56.3%). Streptococcus pneumoniae accounted for nearly one half of infections, followed by Haemophilus influenzae. Pseudomonas aeruginosa and other Gram-negative organisms rarely were responsible for SCAP.
|
Nearly 60% of patients were treated with antibiotics in accord with the IDSA guidelines. Among participating hospitals, the median rate of guideline compliance was 62.2% (range, 20 to 100%). Five institutions had compliance rates of less than 40%, but these centers contributed only 15% of the entire study population. Variability in the rates of guideline adherence, however, across the centers was not statistically different (p = 0.347).
In 22.1% of patients, monotherapy was prescribed. Among these 44 patients, the most common treatments were amoxicillin-clavulanate (n = 19) and levofloxacin (n = 16). For patients initially administered multiple agents (n = 155), use of a nonpseudomonal third-generation cephalosporin (eg, ceftriaxone, cefotaxime) and a macrolide (eg, erythromycin, clarithromycin) was common. This regimen was employed in 56.7% of the combination therapy population. Antipseudomonal cephalosporins, carbepenems, and antipseudomonal penicillins were administered infrequently (n = 8), as were aminoglycosides (n = 12).
As shown in Table 2, there were several differences between patients receiving IDSA-compliant regimens and patients whose antibiotic prescription was not IDSA consistent. Although there were no distinctions in the age and gender distributions of the cohorts, patients with treatment varying from the guidelines were more likely to be tobacco users or to utilize illicit IV drugs. With respect to comorbidities, there was a statistical trend toward malignancy being less frequent in individuals in the IDSA-compliant group. Both immunosuppression and HIV, however, occurred more often in patients whose anti-infective therapy varied from the official recommendations. For example, underlying immunosuppression was 4.92 times (95% confidence interval [CI], 1.54 to 15.71) more common in the population that was not guideline observant. Certain process of care measures additionally differed between the cohorts. There was no difference in the median time to administration of antibiotics, but a delay in antibiotic administration (> 4 h) occurred in only 31.1% of patients receiving a regimen inconsistent with IDSA recommendations, compared to 54.0% of other patients (p = 0.067). Timely drawing of blood for cultures was infrequent. This was accomplished in approximately 30% of those in the IDSA-compliant group vs 39.3% of others (p = 0.108). Importantly, the appropriateness of initial antibiotic therapy was similar between the two groups and exceeded 80% in each case. In terms of severity of illness, the groups were similar. Ventilator-associated pneumonia occurred at similar rates in the two groups (7.0% in guideline concordant subjects vs 8.0% in the remainder, p = 0.794).
|
|
|
|
Discussion
This secondary analysis of a large observational study of critically ill patients requiring MV for SCAP reveals that failure to follow formal guidelines for antibiotic prescribing adversely affects morbidity and health-care utilization. Noncompliance with the IDSA guidelines prolongs the duration of MV by 3 days. Moreover, the impact of deviating from the guideline is independent of multiple other known predictors of outcome in SCAP. As a corollary, our findings indicate that enhanced rates of guideline compliance can have a significant effect on resource utilization.
Although multiple investigators have studied predictors of mortality in CAP and in SCAP, there are few studies addressing either predictors of the duration of MV in this disease or the potential importance of antibiotic guidelines on resource use. Meehan and colleagues8 illustrated the significance of process of care on mortality in CAP. Their analysis focused on an elderly population and included relatively few patients needing ICU care.8 Menendez et al,17 in an assessment of guideline adherence in CAP among 1,200 patients, observed that antibiotic guideline adherence reduced the risk for death by nearly half. Unlike our project, however, neither of these investigations focused on a cohort of persons at high risk of death in the ICU or examined the relationship between guideline compliance and resource utilization.
Similarly, multiple studies1819 document the need for initially appropriate antibiotics in a number of disease states ranging from severe sepsis to ventilator-associated pneumonia. In SCAP specifically, prior work has indicated that predictors of death include severity of illness (eg, APACHE II score), development of organ failure, and presence of shock.12 As with CAP in general, researchers12 have also documented the relationship between antibiotic selection and mortality in SCAP. For example, earlier work13 with this data set suggested that the risk for death increased by 66% if the clinicians did not follow the IDSA prescribing recommendations. Our work builds on these prior efforts in that we specifically controlled for an interaction between guideline compliance and the appropriateness of antibiotics. After adjusting for appropriateness of antibiotics, guideline compliance remained an important correlate of health-care use.
In our univariate analysis comparing patients treated with IDSA-consistent antibiotics, several patient characteristics (eg, illicit IV drug use, neurologic disease) identified patients less likely to be treated with a guideline-concordant regimen. It may be that physicians consciously act differently when approaching these types of patients because they perceive that these individuals are unique. In other words, they actively conclude that these patients are at risk for specific types of pathogens and that therefore the IDSA guidelines might not apply. In persons with neurologic disease, one might be so concerned about aspiration that they might determine there would be no need for coverage of "atypical pathogens." Alternatively, it may reflect the belief, either consciously or subconsciously, that they can, in some subpopulations, accurately determine the pathogen. This is of course a misperception, since multiple studies12 document that practitioners routinely cannot predict the specific organism responsible for a given infection. Then again, it may be that these subpopulations do merit a different approach to antibiotics. This hypothesis, however, would require formal testing in a larger clinical trial of more homogenous populations.
Why might guideline compliance positively affect outcomes? On one level, complying with a guideline could improve outcomes through either increasing rates of initially appropriate antibiotic therapy or via enhancing process of care. Some studies812 document that guidelines certainly have this effect. In our cohort, both of the measures of process and rates of initially correct antibiotic prescribing were similar. However, as is seen in many pneumonia studies, we only identified a culprit organism approximately one half the time. Therefore, we cannot exclude the possibility that guideline compliance may have had its effect by actually altering appropriate coverage rates. Alternatively, guideline compliance may only be a surrogate marker for quality of care overall. In other words, hospitals that follow guidelines may simply do a better job at caring for patients with SCAP. The lack of a center-specific affect in our multivariate model, nonetheless, suggests this is not likely the case. More generally, though, others have also found that guideline compliance seems to have an independent effect on decreasing mortality in CAP. Mortensen and coworkers20 found that after controlling for multiple confounders, including not only disease severity but also after addressing process of care and antibiotic appropriateness, failure to adhere to a national antibiotic guideline increased the risk for death nearly six-fold.
The present analysis is novel in that it expands on many earlier efforts by underscoring the effect of antibiotic recommendations on the duration of MV. Little is known about factors associated with longer durations of MV in SCAP. Rather, much attention has centered on predicting the initial need for MV. In an analysis of 696 patients comparing the severity measures proposed by the ATS (the "modified" criteria) to those of the British Thoracic Society, Ewig et al21 concluded the modified ATS rule performed superiorly, with an overall accuracy of 90%. A more extensive analysis by Angus et al6 encompassing > 1,300 subjects also determined that the ATS criteria had the highest predictive resolution; the area under the receiver operating characteristic curve for the ATS criteria equaled 0.74. Neither of these reports, however, explored the subsequent duration of MV. Again, none of these researchers explored how any of these risk stratification tools performed at forecasting the length of MV once it is instituted.
Extrapolated to a single hospital or to a national health-care system, our observations suggest major improvements in outcome can be achieved. Financially, patients who acquire SCAP disproportionately account for many of the costs of care associated with CAP. In the United States, on average, patients needing MV for SCAP are hospitalized for nearly 16 days, compared to only 7 days in patients hospitalized but not receiving MV.7 Partially reflecting this difference, the costs of care for SCAP requiring MV are $30,000 greater than the costs related to a hospitalization for uncomplicated pneumonia. Moreover, among the elderly, who are at high risk for CAP, only 7% require MV but account for nearly half of all expenditures by Medicare for the treatment of CAP.7 If one estimates that a day of MV results in an additional $3,000 to $5,000 in health-care costs, guideline compliance can yield a net savings of from $9,000 to $15,000 per case. Coupled with the fact that guideline compliance is a process-of-care variable that is certainly amenable to changes in physician behavior, these estimates should serve as an impetus to hospitals, professional societies, and third-party payers to stress efforts to promote adherence with official recommendations for antibiotic use in SCAP. Furthermore, projects to educate clinicians about antibiotic guidelines and to reward guideline compliance will likely prove cost-effective in light of the potential return on this investment. As protocols for antibiotic selection in other disease states have been shown to improve compliance with guidelines, their potential in SCAP to result in financial savings is considerable.920
Our methodologic approach has several strengths that merit comment. First, we were able to control for multiple possible confounders ranging from process of care variables to both chronic comorbidities and severity of illness. Second, the data were generated from a multi-institutional undertaking. Third, the sample size allowed us to examine the interaction among multiple potential covariates.
There are several important limitations, however. First, the approach we employed was technically retrospective and hence prone to several types of bias. However, rather than being purely retrospective, the data were generated prospectively as part of a large registry. Second, only one investigator determined guideline compliance. Third, since the study was conducted in Spain, the results may have limited generalizablility to other nations. There seems no reason, in principle, why variation in health-care practice would differentially affect the importance of guideline compliance on resource utilization. Fourth, many factors affect the duration of MV, such as the sedation practice and the use of protocols for liberation from MV.2223 We unfortunately lacked information as to why an individual patient remained on MV, so we cannot comment as to whether this factor was potentially modifiable or if it was directly related to the initial reason for admission. Since critical care practice was not standardized across the multiple participating ICUs, it is likely that these issues may confound our findings. It may be that guideline compliance may represent a marker for use of other evidence-based interventions in the ICU that have been shown to effectively shorten the duration of MV. The rate of guideline compliance itself, in short, may simply be an epiphenomenon. Although possible we believe this less likely because we did not detect center-specific effects in our multivariate model, and we actively sought out this potential by including a variable to capture how frequently particular institutions followed the IDSA recommendations. Finally, there may have been other center-specific affects that we could not address. The wide range in the rate of guideline adherence across participating centers underscores this point. As we note above, institutions with particularly low rates of guideline adherence contributed few total patients to the final cohort. Furthermore, there were no statistical differences in the process of care variables, which likely capture institution-specific practice styles, between patients receiving IDSA-compliant regimens and those whose antibiotics varied from these recommendations. As with any concern regarding other unmeasured factors that may affect duration of MV, we acknowledge this concern, but we certainly saw no evidence that enhanced guideline compliance led to worse morbidity even though it concomitantly improves survival. Often in critical care, mortality is reduced at the expense of morbidity. This does not seem to be the case with guideline compliance for antibiotic prescribing in SCAP.
In summary, failure to follow formal antibiotic-prescribing guidelines in SCAP adversely impacts morbidity and resource use. More precisely, guideline nonadherence adds approximately 3 days to the duration of MV among eventual survivors and in turn likely adds appreciably to the costs of this disease.
Appendix
CAPUCI Study Investigators
J. Solé, Dr. Negrin Hospital, Gran Canaria; J. Blanquer, Clinic Hospital, Valencia; J. Jiménez, Virgen Del Rocio Hospital, Sevilla; V. De La Torre, Virgen De La Victoria Hospital, Malaga; Jm. Sirvent, Josep Trueta Hospital, Girona; M. Bodí, Joan Xxiii Hospital, Tarragona; J. Almirall, Mataró Hospital, Mataró (Barcelona); A. Doblas, Juan Ramon Jimenez Hospital, Huelva; Jr. Badía, Clinic Hospital, Barcelona; F. García, General Hospital, Albacete; A. Mendia, Nuestra Señora De Aranzazu Hospital, San Sebastian; R. Jordá, Son Dureta Hospital, Palma De Mallorca; F. Bobillo, Clinico Hospital, Valladolid; J. Vallés, Hospital Parc Tauli, Sabadell (Barcelona); Mj. Broch, Sagunto Hospital, Valencia; N. Carrasco, Princesa Hospital, Madrid; Ma. Herranz, Rio Hortega Hospital, Valladolid; F. Alvarez Lerma, Del Mar Hospital, Barcelona; E. Mesalles, Trias I Pujol Hospital, Badalona (Barcelona); B. Alvarez, General Hospital, Alicante; Jc. Robles, Reina Sofia Hospital, Córdoba; E. Maraví, Virgen Del Camino Hospital, Pamplona; F. Barcenilla, Arnau De Vilanova Hospital, Lleida; Ma. Blasco, Peset Aleixandre Hospital, Valencia; G. Masdeu, Verge De La Cinta Hospital, Tortosa (Tarragona); Mj. López Pueyo, General De Yagüe Hospital, Burgos; A. Margarit, Virgen Meritxell Hospital, Andorra; J. Fierro, Poniente Hospital, Almeria; F. Renedo, Leon Hospital, Leon; A. Lores, Bellvitge Hospital, Barcelona; R. Alonso, General De Asturias Hospital, Oviedo; M. J. Huertos, Puerto Real Hospital, Cadiz; Mj. López Cambra, General Hospital, Segovia.
Footnotes
Abbreviations: ABG = arterial blood gas; APACHE = acute physiology and chronic health evaluation; ATS = American Thoracic Society; CAP = community-acquired pneumonia; CI = confidence interval; ISDA = Infectious Disease Society of America; LOS = length of stay; MV = mechanical ventilation; SCAP = severe community-acquired pneumonia
This study was supported in part by Instituto de Salud Carlos III (RTIC03/11), FISS PI04/1500, and 2005/SGR/920.
Received for publication December 13, 2005. Accepted for publication March 8, 2006.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
