Expanded Analysis of 20 Pneumococcal Serotypes Associated With Radiographically Confirmed Community-acquired Pneumonia in Hospitalized US Adults.

BACKGROUND: Streptococcus pneumoniae is a causative agent of community-acquired pneumonia (CAP). The 13-valent pneumococcal conjugate vaccine (PCV13) has significantly decreased the burden of PCV13-serotype pneumococcal disease; however, disease from nonvaccine serotypes remains substantial. A recent study documented the persistence of PCV13 serotypes among US adults hospitalized with radiographically confirmed CAP. The current analysis used a recently developed urinary antigen detection (UAD) assay (UAD2) to extend these results to additional serotypes included in an investigational PCV20 vaccine. METHODS: This prospective study enrolled adults aged ≥18 years hospitalized with radiographically confirmed CAP between October 2013 and September 2016. Presence of S pneumoniae was determined by blood and respiratory sample culture, BinaxNOW urine testing, and UAD. In addition to Quellung on cultured isolates when available, serotypes were identified from urine specimens using UAD1 for PCV13 serotypes and UAD2 for 7 PCV20-unique serotypes (8, 10A, 11A, 12F, 15B, 22F, and 33F) and 4 additional serotypes (2, 9N, 17F, and 20). RESULTS: Among 12 055 subjects with radiographically confirmed CAP, 1482 were positive for S pneumoniae. PCV13- and PCV20-unique serotypes were associated with 37.7% (n = 559) and 27.0% (n = 400) of cases, respectively; 288 subjects were exclusively diagnosed as positive for S pneumoniae by UAD2. Demographic and clinical disease characteristics were similar between subjects with CAP caused by PCV13 and PCV20-unique serotypes. CONCLUSIONS: The current analysis using UAD2 identified a sizeable proportion of hospitalized adult CAP associated with PCV20-unique serotypes. PCV20 may therefore address the burden of CAP caused by the additional serotypes present in the vaccine.

Higher clinical success in patients with ventilator-associated pneumonia due to methicillin-resistant Staphylococcus aureus treated with linezolid compared with vancomycin: results from the IMPACT-HAP study.

INTRODUCTION: Controversy exists regarding optimal treatment for ventilator-associated pneumonia (VAP) due to methicillin-resistant Staphylococcus aureus (MRSA). The primary objective of this study was to compare clinical success of linezolid versus vancomycin for the treatment of patients with MRSA VAP. METHODS: This was a multicenter, retrospective, observational study of patients with VAP (defined according to Centers for Disease Control and Prevention criteria) due to MRSA who were treated with linezolid or vancomycin. MRSA VAP was considered when MRSA was isolated from a tracheal aspirate or bronchoalveolar lavage. Clinical success was evaluated by assessing improvement or resolution of signs and symptoms of VAP by day 14. After matching on confounding factors, logistic regression models were used to determine if an association existed between treatment arm and clinical success. RESULTS: A total of 188 patients were evaluated (101 treated with linezolid and 87 with vancomycin). The mean ± standard deviation Acute Physiology and Chronic Health Evaluation (APACHE) II score was 21 ± 11 for linezolid- and 19 ± 9 for vancomycin-treated patients (P = 0.041). Clinical success occurred in 85% of linezolid-treated patients compared with 69% of vancomycin-treated patients (P = 0.009). After adjusting for confounding factors, linezolid-treated patients were 24% more likely to experience clinical success than vancomycin-treated patients (P = 0.018). CONCLUSIONS: This study adds to the evidence indicating that patients with MRSA VAP who are treated with linezolid are more likely to respond favorably compared with patients treated with vancomycin.

Analysis of pathogen and host factors related to clinical outcomes in patients with hospital-acquired pneumonia due to methicillin-resistant Staphylococcus aureus.

Methicillin-resistant Staphylococcus aureus (MRSA) is a major cause of nosocomial pneumonia. To characterize pathogen-derived and host-related factors in intensive care unit (ICU) patients with MRSA pneumonia, we evaluated the Improving Medicine through Pathway Assessment of Critical Therapy in Hospital-Acquired Pneumonia (IMPACT-HAP) database. We performed multivariate regression analyses of 28-day mortality and clinical response using univariate analysis variables at a P level of <0.25. In isolates from 251 patients, the most common molecular characteristics were USA100 (55.0%) and USA300 (23.9%), SCCmec types II (64.1%) and IV (33.1%), and agr I (36.7%) and II (61.8%). Panton-Valentine leukocidin (PVL) was present in 21.9%, and vancomycin heteroresistance was present in 15.9%. Mortality occurred in 37.1% of patients; factors in the univariate analysis were age, APACHE II score, AIDS, cardiac disease, vascular disease, diabetes, SCCmec type II, PVL negativity, and higher vancomycin MIC (all P values were <0.05). In multivariate analysis, independent predictors were APACHE II score (odds ratio [OR], 1.090; 95% confidence interval [CI], 1.041 to 1.141; P < 0.001) and age (OR, 1.024; 95% CI, 1.003 to 1.046; P = 0.02). Clinical failure occurred in 36.8% of 201 evaluable patients; the only independent predictor was APACHE II score (OR, 1.082; 95% CI, 1.031 to 1.136; P = 0.002). In summary, APACHE II score (mortality, clinical failure) and age (mortality) were the only independent predictors, which is consistent with severity of illness in ICU patients with MRSA pneumonia. Interestingly, our univariate findings suggest that both pathogen and host factors influence outcomes. As the epidemiology of MRSA pneumonia continues to evolve, both pathogen- and host-related factors should be considered when describing epidemiological trends and outcomes of therapeutic interventions.

Surgical site infection prophylaxis strategies for cardiothoracic surgery: a decision-analytic model.

BACKGROUND: Methicillin-resistant Staphylococcus aureus (MRSA) is a common cause of invasive surgical site infection (SSI) in the USA. Antimicrobial prophylaxis for SSI typically includes a cephalosporin. Vancomycin is used to provide MRSA coverage, but the timing of administration is challenging. Linezolid is an attractive agent for SSI prophylaxis, particularly for the prevention of SSI due to MRSA. METHODS: We developed a decision-analytic model to evaluate linezolid use for cardiothoracic SSI prophylaxis. A theoretical cohort of 10,000 cardiothoracic surgery patients was followed through 2 stages: (1) occurrence of SSI, and (2) mortality after SSI. All patients were administered cefuroxime, vancomycin, or linezolid between 1 and 180 min prior to surgical incision. SSIs were categorized into 3 pathogen categories: (1) methicillin-susceptible Gram-positive, (2) methicillin-resistant Gram-positive, and (3) other organisms. The most effective strategy resulted in the fewest SSIs. Assumptions for antibiotic effectiveness, impact of administration time, and pathogens were based on the published literature. RESULTS: Compared with cefuroxime, there was a 1% increase in the total number of SSIs in the linezolid group (mean SSI increase = 7), while there was a 12% increase in the vancomycin group (mean SSI increase = 86). Linezolid prophylaxis resulted in fewer SSIs due to methicillin-resistant Gram-positive infections (n = 108) compared with cefuroxime (n = 200, 46% reduction in the linezolid group) and vancomycin (n = 119, 9% reduction in the linezolid group). CONCLUSIONS: This simulation indicates that linezolid may offer benefits for SSI prophylaxis over existing prophylactic agents, particularly for the prevention of SSI due to Gram-positive methicillin-resistant pathogens.

Frequency of stool specimen collection and testing for Clostridioides difficile of hospitalized adults and long-term care facility residents with new-onset diarrhea in Louisville, Kentucky.

OBJECTIVES: This study aimed to determine the stool specimen collection and Clostridioides difficile (C. difficile) testing frequency from inpatients and long-term care facility (LTCF) residents with new-onset diarrhea. METHODS: A cross-sectional study was conducted in all wards of 9 adult hospitals (3532 beds) and 14 LTCFs (1205 beds) in Louisville, Kentucky to identify new-onset diarrhea (≥3 loose stools in the past 24 h and not present in the preceding 24 h) among Louisville adults via electronic medical record review, nurse interviews, and patient interviews during a 1-2 week observation period in 2018-2019. RESULTS: Among Louisville-resident inpatients, 167 patients with 9731 inpatient-days had new-onset diarrhea (1.7/100 inpatient-days). Stool specimens were collected from 32% (53/167); 12 (23%) specimens were laboratory-confirmed for C. difficile infection (CDI) (12.3 cases/10,000 inpatient-days). Among LTCF residents, 63 with 10,402 LTCF resident-days had new-onset diarrhea (0.6/100 LTCF resident-days). Stool specimens were collected from 32% (20/63); 9 (45%) specimens were laboratory-confirmed for CDI (8.6 cases/10,000 LTCF resident-days). CONCLUSIONS: New-onset diarrhea was common among inpatients and LTCF residents. Only one-third of patients with new-onset diarrhea had a stool specimen collected and tested for C. difficile-indicative of a potential CDI underdiagnosis-although, further studies are needed to confirm the extent of CDI underdiagnosis.

Severity of disease and clinical outcomes in patients with hospital-acquired pneumonia due to methicillin-resistant Staphylococcus aureus strains not influenced by the presence of the Panton-Valentine leukocidin gene.

BACKGROUND: Patients with community-acquired pneumonia (CAP) infected with methicillin-resistant Staphylococcus aureus (MRSA) strains carrying the Panton-Valentine leukocidin (PVL) gene have severe clinical presentation and poor clinical outcomes. Antibiotics that suppress toxin production have been suggested for the management of these patients. The objective of this study was to compare the severity of disease and clinical outcomes of patients with hospital-acquired pneumonia/ventilator-associated pneumonia (HAP/VAP) infected with MRSA carrying the PVL gene with those patients infected with MRSA strains that do not carry the PVL gene. METHODS: This was a multicenter observational study of patients with HAP and VAP. MRSA isolates were subjected to genetic analysis to define the presence of the PVL gene, the USA type and the staphylococcal cassette chromosome mec type. Severity of disease was evaluated with the Acute Physiology and Chronic Health Evaluation II (APACHE II) score. The primary clinical outcome was mortality at hospital discharge. RESULTS: A total of 109 cases of MRSA HAP/VAP were evaluated. The incidence of PVL(+) MRSA was 27%. APACHE II score at diagnosis of HAP/VAP was 21 ± 8 for PVL(+) MRSA and 20 ± 6 for PVL(-) MRSA (P = .67). Mortality was 10% (3/29) for patients with PVL(+) MRSA versus 10% (8/80) for patients with PVL(-) MRSA (P > .99). CONCLUSIONS: In patients with HAP or VAP due to MRSA, severity of disease and clinical outcomes are not influenced by the presence of the PVL gene. Therapeutic strategies directed to block PVL exotoxin may not impact outcomes in these patients.

Development and implementation of a performance improvement project in adult intensive care units: overview of the Improving Medicine Through Pathway Assessment of Critical Therapy in Hospital-Acquired Pneumonia (IMPACT-HAP) study.

INTRODUCTION: In 2005 the American Thoracic Society and Infectious Diseases Society of America (ATS/IDSA) published guidelines for managing hospital-acquired pneumonia (HAP), ventilator-associated pneumonia (VAP), and healthcare-associated pneumonia (HCAP). Although recommendations were evidence based, collective guidelines had not been validated in clinical practice and did not provide specific tools for local implementation. We initiated a performance improvement project designated Improving Medicine Through Pathway Assessment of Critical Therapy in Hospital-Acquired Pneumonia (IMPACT-HAP) at four academic centers in the United States. Our objectives were to develop and implement the project, and to assess compliance with quality indicators in adults admitted to intensive care units (ICUs) with HAP, VAP, or HCAP. METHODS: The project was conducted in three phases over 18 consecutive months beginning 1 February 2006: 1) a three-month planning period for literature review to create the consensus pathway for managing nosocomial pneumonia in these ICUs, a data collection form, quality performance indicators, and internet-based repository; 2) a six-month implementation period for customizing ATS/IDSA guidelines into center-specific guidelines via educational forums; and 3) a nine-month post-implementation period for continuing education and data collection. Data from the first two phases were combined (pre-implementation period) and compared with data from the post-implementation period. RESULTS: We developed a consensus pathway based on ATS/IDSA guidelines and customized it at the local level to accommodate formulary and microbiologic considerations. We implemented multimodal educational activities to teach ICU staff about the guidelines and continued education throughout post-implementation. We registered 432 patients (pre- vs post-implementation, 274 vs 158). Diagnostic criteria for nosocomial pneumonia were more likely to be met during post-implementation (247/257 (96.1%) vs 150/151 (99.3%); P = 0.06). Similarly, empiric antibiotics were more likely to be compliant with ATS/IDSA guidelines during post-implementation (79/257 (30.7%) vs 66/151 (43.7%); P = 0.01), an effect that was sustained over quarterly intervals (P = 0.0008). Between-period differences in compliance with obtaining cultures and use of de-escalation were not statistically significant. CONCLUSIONS: Developing a multi-center performance improvement project to operationalize ATS/IDSA guidelines for HAP, VAP, and HCAP is feasible with local consensus pathway directives for implementation and with quality indicators for monitoring compliance with guidelines.

Patient-Reported Consequences of Community-Acquired Pneumonia in Patients with Chronic Obstructive Pulmonary Disease.

Community acquired pneumonia (CAP) carries high morbidity, mortality, and economic burden, which is even higher in adults diagnosed with chronic obstructive pulmonary disease (COPD). While several studies have assessed the clinical burden and mortality risk of CAP and COPD, very few studies focus on CAP burden from a COPD patient perspective. Individuals recently diagnosed with CAP and with pre-existing COPD were recruited through the COPD Foundation. The CAP Burden of Illness Questionnaire (CAP-BIQ), a content validated questionnaire assessing CAP symptomatology, duration of symptoms and CAP impact on work, activities and family, was administered at baseline and at 30-days follow-up. Of the 490 participants recruited, 481 had data sufficient for analysis. The prevalence of respiratory-related symptoms was very high (>90%) at the time of diagnosis with other generalized symptoms such as fatigue, trouble sleeping, headaches and confusion present in more than 60% of participants. Mean duration of symptoms varied from approximately 2 weeks for headaches and fever to more than a month for fatigue, wheezing, dyspnea, and cough. Employed participants missed an average of 21 days of work and those not employed missed 36 days of usual activities. Over 84% required help from family, friends or care givers. CAP is a serious and burdensome condition for people with COPD, a condition that can impair activities for weeks, frequently requires care from family or friends, and includes lingering symptoms. The patient-reported impact of CAP reported in this study underscores the need for prevention strategies in this population.

Pneumococcal epidemiology among us adults hospitalized for community-acquired pneumonia.

BACKGROUND: Few studies have measured the burden of adult pneumococcal disease after the introduction of 13-valent pneumococcal conjugate vaccine (PCV13) into the US infant vaccination schedule. Further, most data regarding pneumococcal serotypes are derived from invasive pneumococcal disease (IPD), which represents only a fraction of all adult pneumococcal disease burden. Understanding which pneumococcal serotypes cause pneumonia in adults is critical for informing current immunization policy. The objective of this study was to measure the proportion of radiographically-confirmed (CXR+) community-acquired pneumonia (CAP) caused by PCV13 serotypes in hospitalized US adults. METHODS: This observational, prospective surveillance study recruited hospitalized adults aged ≥18 years from 21 acute care hospitals across 10 geographically-dispersed cities in the United States between October 2013 and September 2016. Clinical and demographic data were collected during hospitalization. Vital status was ascertained 30 days after enrollment. Pneumococcal serotypes were detected via culture from the respiratory tract and normally-sterile sites (including blood and pleural fluid). Additionally, a novel, Luminex-based serotype-specific urinary antigen detection (UAD) assay was used to detect serotypes included in PCV13. RESULTS: Of 15,572 enrolled participants, 12,055 eligible patients with CXR+CAP were included in the final analysis population. Mean age was 64.1 years and 52.7% were aged ≥65 years. Common comorbidities included chronic obstructive pulmonary disease (43.0%) and diabetes mellitus (28.6%). PCV13 serotypes were detected in 552/12,055 (4.6%) of all patients and 265/6347 (4.2%) of those aged ≥65 years. Among patients aged 18-64 years PCV13 serotypes were detected in 3.8-5.3% of patients depending on their risk status. CONCLUSIONS: After implementation of a pneumococcal conjugate vaccination program in US children, and despite the herd protection observed in US adults, a persistent burden of PCV13-type CAP remains in this population.

Understanding why low-risk patients accept vaccines: a socio-behavioral approach.

BACKGROUND: Vaccines are one of the most important public health interventions. Understanding factors associated with vaccine acceptance is critical. The objectives of this study were to evaluate the impact of the three constructs of the Theory of Planned Behavior (TPB) on the intention to be vaccinated among healthy individuals being seen for pre-travel care, and to evaluate if behavioral intention was associated with vaccine acceptance. METHODS: We surveyed individuals seeking vaccination at the University of Louisville Vaccine and International Health and Travel Clinic. Linear and two stage least squares regression models were used to define the associations between constructs of the TPB and the intention to be vaccinated, as well as the association between the intention to be vaccinated and vaccine acceptance. RESULTS: A total of 183 individuals were included in the analysis. None of the constructs of the TPB were associated with intention to be vaccinated. Behavioral intention was not associated with vaccination acceptance. CONCLUSIONS: This study suggests that the TPB does not predict the intention to get vaccinated among individuals attending our Vaccine and International Health and Travel Clinic. It will be critical to define better predictors of vaccine uptake in healthy, low-risk individuals to increase vaccine acceptance.

Absence of gender-based differences in outcome of patients with hospital-acquired pneumonia.

BACKGROUND: The objective of this analysis was to evaluate the association between gender and clinical outcomes in intensive care unit (ICU) patients with hospital-acquired pneumonia (HAP) since data thus far are controversial. METHODS: Data from a convenience sample of ICU patients with HAP, including ventilator-associated and health care-associated pneumonia, were retrospectively collected from four academic institutions (Improving Medicine through Pathway Assessment of Critical Therapy in Hospital-Acquired Pneumonia [IMPACT-HAP] study). Outcomes included 28-day mortality, clinical failure at day 14, hospital and ICU length of stay (LOS), and duration of mechanical ventilation. We compared baseline characteristics and performed multivariate analysis to identify factors independently associated with mortality. RESULTS: Among 416 patients, 271 were men and 145 were women. Women were older (62.4±16.9 vs. 55.7±16.5 years, p<0.001) and more critically ill, with Acute Physiology and Chronic Health Evaluation (APACHE) II scores of 21 vs. 19 (p=0.004). Day-28 mortality was 30% for women and 24% for men (p=0.25). Increased 28-day mortality was associated with severity of illness, age, ventilator-associated pneumonia, vascular disease, and hospital LOS prior to pneumonia diagnosis. No significant differences were found in the distribution of bacteria pathogens or in clinical failure rates (36% vs. 31%) between genders. Duration in days of mechanical ventilation, ICU LOS and hospital LOS after the diagnosis of pneumonia were not significantly different between men and women. Analyzing data for women based on presumed pre- or postmenopausal status (age breakpoint of 50 years), showed an increased in ICU LOS (15 vs. 25 days; p=0.0026) and hospital LOS (22 vs. 30 days; p=0.05) for women ≤50 years. No differences were noted in 28-day mortality (24.3% vs. 13.1%; p=0.18) in women ≤50 years of age. CONCLUSIONS: In ICU patients with pneumonia, female gender was not associated with worse outcomes or increased resource utilization compared to male gender. Further studies are needed to evaluate menopausal status and outcomes in women with pneumonia.

Emergence of methicillin-resistant Staphylococcus aureus USA300 genotype as a major cause of late-onset nosocomial pneumonia in intensive care patients in the USA.

OBJECTIVE: To compare demographic and clinical characteristics, and methicillin-resistant Staphylococcus aureus (MRSA) strain characteristics, in patients with early-onset (EO) and late-onset (LO) MRSA nosocomial pneumonia. METHODS: This was a retrospective analysis of data from a multicenter observational study of nosocomial pneumonia patients admitted between November 2008 and July 2010. Laboratory analyses performed on MRSA isolates included confirmation of antimicrobial susceptibility and heteroresistance to vancomycin, USA typing, staphylococcal cassette chromosome (SCC) mec typing, and detection of Panton-Valentine leukocidin (PVL) genes. RESULTS: We identified 134 patients; 42 (31%) had EO MRSA pneumonia and 92 (69%) had LO MRSA pneumonia. The patients in the LO group were more likely to have risk factors for multidrug-resistant pathogens (98% vs. 76%, p<0.001). The MRSA USA300 strain was found with equal frequency in the EO and LO groups. Likewise, both groups had similar frequencies of isolates exhibiting PVL and SCCmec type IV. CONCLUSIONS: Our findings provide further evidence of the continued migration of community-associated MRSA into the healthcare setting in the USA. MRSA USA300 genotype has emerged as a significant cause of LO nosocomial pneumonia in intensive care units. Appropriate anti-MRSA antimicrobial therapy should be considered for both EO and LO hospital-acquired pneumonia and ventilator-associated pneumonia.

Incidence of nephrotoxicity and association with vancomycin use in intensive care unit patients with pneumonia: retrospective analysis of the IMPACT-HAP Database.

BACKGROUND: The 2005 guidelines from the American Thoracic Society and the Infectious Diseases Society of America recommend vancomycin trough levels of 15 to 20 mg/L for the therapy of hospital-acquired (HAP), ventilator-associated (VAP), and health care-associated (HCAP) pneumonia. OBJECTIVE: The goal of this article was to report the incidence of nephrotoxicity and associated risk factors in intensive care unit patients who received vancomycin for the treatment of HAP, VAP, and HCAP. METHODS: This was a retrospective analysis of data from a multicenter, observational study of pneumonia patients. Antibiotic-associated nephrotoxicity was defined as either an increase in serum creatinine ≥0.5 mg/dL or 50% above baseline, from initiation of vancomycin to 72 hours after completion of therapy. Univariate and multivariate logistic regression analyses were performed to identify risk factors for development of renal dysfunction. RESULTS: Of the 449 patients in the database, 240 received at least one dose of vancomycin and 188 had sufficient data for analysis. In these 188 patients, 63% were male. Mean (SD) age was 58.5 (17.2) years, and the mean Acute Physiology and Chronic Health Evaluation II score was 19.4 (6.4). Nephrotoxicity occurred in 29 of 188 (15.4%) vancomycin-treated patients. In multivariate analysis, initial vancomycin trough levels ≥15 mg/L (odds ratio [OR], 5.2 [95% CI, 1.9-13.9]; P = 0.001), concomitant aminoglycoside use (OR, 2.67 [95% CI, 1.09-6.54]; P = 0.03), and duration of vancomycin therapy (OR for each additional treatment day, 1.12 [95% CI, 1.02-1.23]; P = 0.02) were independently associated with nephrotoxicity. The incidence of nephrotoxicity increased as a function of the initial vancomycin trough level, rising from 7% at a trough <10 mg/L to 34% at >20 mg/L (P = 0.001). The mean time to nephrotoxicity decreased from 8.8 days at vancomycin trough levels <15 mg/L to 7.4 days at >20 mg/L (Kaplan-Meier analysis, P = 0.0003). CONCLUSIONS: Nephrotoxicity may be common among intensive care unit patients with pneumonia treated with broad-spectrum antibiotic therapy that includes vancomycin. The finding that an initial vancomycin trough level ≥15 mg/L may be an independent risk factor for nephrotoxicity highlights the need for additional studies to assess current recommendations for vancomycin dosing for ICU patients with pneumonia.

Implementation of guidelines for management of possible multidrug-resistant pneumonia in intensive care: an observational, multicentre cohort study.

BACKGROUND: The American Thoracic Society and Infectious Diseases Society of America provide guidelines for management of hospital-acquired, ventilator-associated, and health-care-associated pneumonias, consisting of empirical antibiotic regimens for patients at risk for multidrug-resistant pathogens. We aimed to improve compliance with these guidelines and assess outcomes. METHODS: We implemented a performance-improvement initiative in four academic medical centres in the USA with protocol-based education and prospective observation of outcomes. Patients were assessed for severity of illness and followed up until death, hospital discharge, or day 28. We included patients in intensive-care units who were at risk for multidrug-resistant pneumonia and were treated empirically. FINDINGS: 303 patients at risk for multidrug-resistant pneumonia were treated empirically, and prescribed treatment was guideline compliant in 129 patients and non-compliant in 174 patients. 44 (34%) patients died before 28 days in the compliance group and 35 (20%) died in the non-compliance group. Five patients in the compliance group and seven in the non-compliance group were lost to follow-up after day 14. Kaplan-Meier estimated survival to 28 days was 65% in the compliance group and 79% in the non-compliance group (p=0·0042). This difference persisted after adjustment for severity of illness. Median length of stay and duration of mechanical ventilation did not differ between groups. Compliance failures included non-use of dual treatment for Gram-negative pathogens in 154 patients and absence of meticillin-resistant Staphylococcus aureus coverage in 24 patients. For patients in whom pathogens were subsequently identified, empirical treatment was active in 79 (81%) of 97 of patients receiving compliant therapy compared with 109 (85%) of 128 of patients receiving non-compliant therapy. INTERPRETATION: Because adherence with empirical treatment was associated with increased mortality, we recommend a randomised trial be done before further implementation of these guidelines. FUNDING: Pfizer, US Medical.