Clinical Policy: Critical Issues in the Management of ...Pneumonia can be divided into 4 categories...
Transcript of Clinical Policy: Critical Issues in the Management of ...Pneumonia can be divided into 4 categories...
INFECTIOUS DISEASE/CLINICAL POLICY
Clinical Policy: Critical Issues in the Management of AdultPatients Presenting to the Emergency Department With
Community-Acquired Pneumonia
From the American College of Emergency Physicians Clinical Policies Subcommittee (Writing Committee) on Critical Issuesin the Evaluation and Management of Adult Patients Presenting to the Emergency Department With Community-AcquiredPneumonia:
Devorah J. Nazarian, MD, ChairOrin L. Eddy, MDThomas W. Lukens, MD, PhDScott D. Weingart, MDWyatt W. Decker, MD
Members of the American College of Emergency Physicians Clinical Policies Committee (Oversight Committee):
Wyatt W. Decker, MD (Co-Chair 2006-2007, Chair 2007-2009)
Andy S. Jagoda, MD (Chair 2003-2006, Co-Chair 2006-2007)
Deborah B. Diercks, MDBarry M. Diner, MD (Methodologist)Jonathan A. Edlow, MDFrancis M. Fesmire, MDJohn T. Finnell, II, MD, MSc (Liaison for Emergency
Medical Informatics Section 2004-2006)Steven A. Godwin, MDSigrid A. Hahn, MDJohn M. Howell, MDJ. Stephen Huff, MDEric J. Lavonas, MDThomas W. Lukens, MD, PhDDonna L. Mason, RN, MS, CEN (ENA Representative
2004-2006)Edward Melnick, MD (EMRA Representative 2007-2008)Anthony M. Napoli, MD (EMRA Representative 2004-2006)
704 Annals of Emergency Medicine
Devorah Nazarian, MDAnnMarie Papa, RN, MSN, CEN, FAEN (ENA
Representative 2007-2009)Jim Richmann, RN, BS, MA(c), CEN (ENA Representative
2006-2007)Scott M. Silvers, MDEdward P. Sloan, MD, MPHMolly E. W. Thiessen, MD (EMRA Representative 2006-
2008)Robert L. Wears, MD, MS (Methodologist)Stephen J. Wolf, MDCherri D. Hobgood, MD (Board Liaison 2004-2006)David C. Seaberg, MD, CPE (Board Liaison 2006-2009)Rhonda R. Whitson, RHIA, Staff Liaison, Clinical Policies
Committee and Subcommittees
Approved by the ACEP Board of Directors, June 23, 2009
Supported by the Emergency Nurses Association, July21, 2009
Policy statements and clinical policies are the official policies of the American College of EmergencyPhysicians and, as such, are not subject to the same peer review process as articles appearing in the printjournal. Policy statements and clinical policies of ACEP do not necessarily reflect the policies and beliefsof Annals of Emergency Medicine and its editors.
0196-0644/$-see front matterCopyright © 2009 by the American College of Emergency Physicians.doi:10.1016/j.annemergmed.2009.07.002
Volume , . : November
Clinical Policy
[Ann Emerg Med. 2009;54:704-731.]
ABSTRACTThis clinical policy from the American College of Emergency
Physicians focuses on critical issues concerning the managementof adult patients presenting to the emergency department (ED)with community-acquired pneumonia. It is an update of the2001 clinical policy for the management and risk stratificationof adult patients presenting to the ED with community-acquired pneumonia. A subcommittee reviewed the currentliterature to derive evidence-based recommendations to helpanswer the following questions: (1) Are routine blood culturesindicated in patients admitted with community-acquiredpneumonia? (2) In adult patients with community-acquiredpneumonia without severe sepsis, is there a benefit in mortalityor morbidity from the administration of antibiotics within aspecific time course? The evidence was graded andrecommendations were given based on the strength of evidence.
INTRODUCTIONCommunity-acquired pneumonia (CAP) is a major health
problem in the United States. CAP is the seventh leading cause ofdeath in the United States, with 1.7 million hospital admissions peryear.1,2 The annual economic costs of CAP-related hospitalizationshave been estimated at $9 billion.3 Pneumonia carries an age-adjusted mortality rate up to 22%.1 Despite clinical advances,pneumonia mortality rates have not decreased significantly sincepenicillin became routinely available.4
Pneumonia can be divided into 4 categories based on the site ofacquisition of illness: CAP, hospital-acquired pneumonia (HAP),ventilator-associated pneumonia (VAP), and health care-associatedpneumonia (HCAP).5 CAP has recently been defined as an acutepulmonary infection in a patient who is not hospitalized or living ina long-term care facility 14 or more days before presentation anddoes not meet the criteria for HCAP.5 HAP is defined as a newinfection occurring 48 hours or longer after hospital admission.VAP is defined as pneumonia occurring 48 to 72 hours afterendotracheal intubation. HCAP encompasses many patientspreviously defined as having CAP. HCAP is defined as infectionoccurring within 90 days of a 2-day or longer hospitalization; in anursing home or long-term care residence; within 30 days ofreceiving intravenous antibacterial therapy, chemotherapy, orwound care or after a hospital or hemodialysis clinic visit; or in anypatient in contact with a multidrug-resistant pathogen.6 Anemerging body of evidence suggests that patients with HCAP moreclosely resemble patients with HAP and may require HAP-liketreatments.6-8
Given the significance of CAP, improving pneumonia care hasbecome a recent focus of many organizations such as The JointCommission and the Centers for Medicare & Medicaid Services(CMS). There are a number of core measures for patients admittedwith the diagnosis of pneumonia. Core measures that evaluate theemergency department (ED) care of CAP patients include blood
culture collection prior to first antibiotic administration (when EDVolume , . : November
blood cultures are drawn), administration of initial antibioticswithin 6 hours of ED arrival (previously within 4 hours), andappropriate antibiotic selection.9
To comply with antibiotic quality measures and CMS andprivate payer pay for performance programs, some EDs havemoved toward treating possible CAP patients with antibioticsbefore the diagnosis is confirmed.10 In this age of increasingantibiotic resistance, this may have negative consequences inexcess of any putative benefit. Kanwar et al11 studied 2 cohortsof patients with the ED diagnosis of CAP, before and after theimplementation of antibiotic timing guidelines. To achieve anincrease in the number of patients with time to first antibioticdose less than 4 hours, an additional 17% of patients wereunnecessarily treated with antibiotics. Khalil et al12 performed aretrospective analysis of factors associated with the eventualdiagnosis of CAP in patients presenting to the ED. Of 1,948patients who presented with respiratory complaints, only 198eventually were diagnosed with CAP. If half of the patients inthis study received empiric antibiotics, at least 40% of thepatients would have received antibiotics unnecessarily,potentially increasing antibiotic resistance in the community. Inan online questionnaire, Pines et al10 found that 37% ofacademic EDs administer antibiotics before obtaining chestradiograph. In a retrospective chart review of patients admittedwith pneumonia, 22% of the patients presented in a mannerthat can result in delayed antibiotics delivery as a result ofdiagnostic uncertainty.13 The most recent iteration of the CMSguidelines includes provisions for diagnostic uncertainty whenassessing time to first antibiotic dose. With the current EDcrowding crisis, the feasibility of rapid antibiotic administrationcan be difficult.14-16
The disposition of patients with pneumonia is a majordecision for emergency physicians, with impact on patientoutcome. Prognostic tools such as the Pneumonia SeverityIndex (PSI) and severity-of-illness indexes such as the CURBand CURB-65 scores have been validated in several studies andcan be used to aid in admission decisions.17,18 The PSI stratifiespatients into 5 categories on the basis of mortality risk. It hasbeen suggested that patients in groups I and II be treated asoutpatients, those in group III be treated in an observation unitor with a short hospitalization, and those patients who fall intogroups IV and V be admitted for treatment.19 CURB-65 is aneasy-to-use severity-of-illness score that uses the followingfactors as indicators of increased mortality: Confusion, Urea,Respiratory rate, low Blood pressure, and age 65 or greater. Limet al20 suggested that patients with a CURB-65 score of 2 betreated as inpatients; those with a score of 3 or greater will oftenrequire an ICU.* These prognostic tools do not take intoaccount the psychosocial factors and other comorbidities that
*Confusion based on specific mental test or disorientation to person,place, or time, Urea �7 mmol/L (20 mg/dL), Respiratory Rate �30breaths/min, Blood pressure systolic �90 mm Hg or diastolic �60 mm
Hg, and age �65 years.Annals of Emergency Medicine 705
Clinical Policy
may also play a role in the emergency physician’s determinationof the best site of treatment for patients with CAP.
Most patients admitted for CAP are first cared for in theED.21 This clinical policy critically evaluates the availableevidence about 2 often controversial critical issues in the care ofpatients admitted with the diagnosis of CAP.11,13,22-25 Thefocused critical questions addressed in this policy include thefollowing:1. Are routine blood cultures indicated in patients admitted
with CAP?2. In adult patients with CAP without severe sepsis, is there a
benefit in mortality and morbidity from the administrationof antibiotics within a specific time course?
METHODOLOGYThis clinical policy was created after careful review and
critical analysis of the medical literature. Multiple searches ofMEDLINE, MEDLINE In-Process, and the Cochrane databasewere performed. Specific key words/phrases used in the searchesare identified under each critical question. All searches werelimited to English-language sources, human studies, and adults.Additional articles were reviewed from the bibliography ofarticles cited and from published textbooks and review articles.Subcommittee members supplied articles from their own files,and more recent articles identified during the process were alsoincluded.
The reasons for developing clinical policies in emergencymedicine and the approaches used in their development havebeen enumerated.26 This policy is a product of the AmericanCollege of Emergency Physicians (ACEP) clinical policydevelopment process, including expert review, and is based onthe existing literature; when literature was not available,consensus of emergency physicians was used. Expert reviewcomments were received from individual emergency physiciansand from individual members of the American College of ChestPhysicians, the American College of Physicians, the InfectiousDiseases Society of America, the Institute for Clinical SystemsImprovement, the Society for Academic Emergency Medicine,and ACEP’s Section on Critical Care Medicine. Their responseswere used to further refine and enhance this policy; however,their responses do not imply endorsement of this clinical policy.Clinical policies are scheduled for revision every 3 years;however, interim reviews are conducted when technology or thepractice environment changes significantly.
All articles used in the formulation of this clinical policy weregraded by at least 2 subcommittee members for strength ofevidence and classified by the subcommittee members into 3classes of evidence on the basis of the design of the study, withdesign 1 representing the strongest evidence and design 3representing the weakest evidence for therapeutic, diagnostic,and prognostic clinical reports, respectively (Appendix A).Articles were then graded on 6 dimensions thought to be mostrelevant to the development of a clinical guideline: blindedversus nonblinded outcome assessment, blinded or randomized
allocation, direct or indirect outcome measures (reliability and706 Annals of Emergency Medicine
validity), biases (eg, selection, detection, transfer), externalvalidity (ie, generalizability), and sufficient sample size. Articlesreceived a final grade (Class I, II, III) on the basis of apredetermined formula, taking into account design and qualityof study (Appendix B). Articles with fatal flaws were given an“X” grade and not used in formulating recommendations in thispolicy. Evidence grading was done with respect to the specificdata being extracted and the specific critical question beingreviewed. Thus, the level of evidence for any one study may varyaccording to the question, and it is possible for a single article toreceive different levels of grading as different critical questionsare answered. Question-specific level of evidence grading may befound in the Evidentiary Table included at the end of thispolicy.
Clinical findings and strength of recommendations regardingpatient management were then made according to the followingcriteria:
Level A recommendations. Generally accepted principles forpatient management that reflect a high degree of clinicalcertainty (ie, based on strength of evidence Class I oroverwhelming evidence from strength of evidence Class IIstudies that directly address all of the issues).
Level B recommendations. Recommendations for patientmanagement that may identify a particular strategy or range ofmanagement strategies that reflect moderate clinical certainty(ie, based on strength of evidence Class II studies that directlyaddress the issue, decision analysis that directly addresses theissue, or strong consensus of strength of evidence Class IIIstudies).
Level C recommendations. Other strategies for patientmanagement that are based on preliminary, inconclusive, orconflicting evidence, or in the absence of any publishedliterature, based on panel consensus.
There are certain circumstances in which therecommendations stemming from a body of evidence shouldnot be rated as highly as the individual studies on which theyare based. Factors such as heterogeneity of results, uncertaintyabout effect magnitude and consequences, strength of priorbeliefs, and publication bias, among others, might lead to such adowngrading of recommendations.
This policy is not intended to be a complete manual on theevaluation and management of adult patients with CAP but rathera focused examination of critical issues that have particularrelevance to the current practice of emergency medicine.
It is the goal of the Clinical Policies Committee to providean evidence-based recommendation when the medical literatureprovides enough quality information to answer a criticalquestion. When the medical literature does not contain enoughquality information to answer a critical question, the membersof the Clinical Policies Committee believe that it is equallyimportant to alert emergency physicians to this fact.
Recommendations offered in this policy are not intended torepresent the only diagnostic and management options that theemergency physician should consider. ACEP clearly recognizes
Volume , . : November
Clinical Policy
the importance of the individual physician’s judgment. Rather,this guideline defines for the physician those strategies for whichmedical literature exists to provide support for answers to thecrucial questions addressed in this policy.
Scope of Application. This guideline is intended forphysicians working in hospital-based EDs.
Inclusion Criteria. This guideline is intended for patients18 years of age or older with signs and symptoms of CAP andradiographic evidence of pneumonia.
Exclusion Criteria. This guideline is not intended forpatients who are pregnant, or immunocompromised (includingpatients with HIV/AIDS, organ transplant, or recipients ofcorticosteroids, antineoplastic therapy, or otherimmunosuppressive agents), or have been hospitalized withinthe last 30 days.
CRITICAL QUESTIONS1. Are routine blood cultures indicated in patients
admitted with CAP?
Patient Management RecommendationsLevel A recommendations. None specified.Level B recommendations. Do not routinely obtain blood
cultures in patients admitted with CAP.Level C recommendations. Consider obtaining blood
cultures in higher-risk patients admitted with CAP (eg, severedisease, immunocompromise, significant comorbidities, or otherrisk factors for infection with resistant organisms).
Key words/phrases for literature searches: pneumonia,community-acquired pneumonia, blood cultures, microbiology,bacteremia, utility of blood cultures, timeline 1996 – May 20,2009.
The following have been identified as CMS core measures forpatients admitted with CAP: (1) the collection of blood culturesprior to antibiotic administration, when ED blood cultures aredrawn; (2) blood cultures performed within 24 hours prior to or24 hours after hospital arrival for patients who were transferredor admitted to the ICU within 24 hours of presentation to thehospital.9 The 2007 American Thoracic Society and InfectiousDiseases Society of America guidelines for the management ofpatients with CAP recommended pretreatment blood culturesfor those patients hospitalized with the following conditions:cavitary infiltrates, leukopenia, active alcohol abuse, chronicsevere liver disease, asplenia, positive test result forpneumococcal urinary antigen, pleural effusion, or thoseadmitted to the ICU. Blood cultures are optional for thosewithout the specifically listed conditions.27
Ideally, blood cultures identify a pathogen and its susceptibility,allowing antibiotic therapy to be customized for each patient.However, blood cultures are infrequently positive, and bloodculture results do not often lead to change in management. Avariety of Class II and III studies have reported the incidence ofpositive culture results in patients admitted with CAP. The yield
reported ranges from 0% in a series of 74 patients with nonsevereVolume , . : November
CAP without significant comorbidities28 to 33% in 146 ICUpatients with CAP from Reunion Island.29 Typically, the range is1% to 16%.30-41
A number of Class II and III studies have investigated theimpact of blood cultures on antibiotic management in CAPpatients. Antibiotic therapy was changed based on blood cultureresults in 0% to 5% of patients cultured.31-33,38,39,42-44 Changein patient condition (either improvement or deterioration) wasmore likely to prompt antibiotic modification than results ofblood cultures.33,44,45 Few changes were made for coverage ofresistant organisms identified by blood cultures. The Class IIstudy by Campbell et al31 found that only 0.4% of bloodcultures drawn yielded an organism resistant to recommendedempiric antibiotics. Similarly, the Class II study by Kennedy etal39 noted 4 of 414 cultures drawn (1%) yielded resistantorganisms, resulting in 2 patients having their initial treatmentchanged (2 others had coverage altered to more effectiveantibiotics before culture results were known). One Class IIstudy45 and multiple Class III studies reporting changes inempiric therapy based on blood culture results demonstratesimilar findings. These studies, ranging in size from 86 to 517patients, reported organisms resistant to empiric therapy in 0%to 2.7% of patients that were cultured.32,33,38,42-46
There are few data about blood culture performance in CAPpatients and association with outcomes such as mortality, timeto clinical stability, and length of stay. In a Class II multicenterstudy, Dedier et al47 retrospectively examined 1,062 patientswith a primary admission diagnosis of pneumonia. They foundno difference in mortality or length of stay between patientswho had blood cultures and those who did not have bloodcultures before receiving antibiotics and no difference inmortality or length of stay between patients who had bloodcultures and those who did not have blood cultures within 24hours of admission. In the frequently cited Class III study byMeehan et al,48 investigators retrospectively examined a nationalstudy set of 1,343 Medicare patients with a discharge diagnosisof pneumonia. The authors concluded that blood culturecollection within 24 hours was associated with lower 30-daymortality; however, the odds ratio (OR) was 0.9, with aconfidence interval (CI) of 0.81 to 1.0 and a nonsignificant Pvalue of 0.07. This same study examined collection of bloodcultures before or after antibiotic administration and found nosignificant association with lower mortality if patients had bloodcultures collected before receiving antibiotics.
Blood culture results may be misleading and may causeunintended consequences. False-positive or contaminatedspecimens are common, and in some studies, rates of false-positive blood cultures approach those of true-positive.32,33,39-40,42
Treatment based on preliminary false-positive blood cultureresults may lead to unnecessary antibiotic coverage andincreased length of stay, pending final identification of theorganism. Metersky et al40 retrospectively analyzed 13,043Medicare patients with CAP and found 7% with true-positive
blood cultures and 5% false-positive blood cultures. PatientsAnnals of Emergency Medicine 707
Clinical Policy
with contaminated blood cultures had an average length of stayof 1 day longer than those who did not have contaminatedblood cultures (P�0.01). False-positive blood cultures are alsocostly. Bates et al49 reported that total hospital charges were$4,000 greater for patients with contaminated blood culturescompared with those with negative blood cultures.
Data suggest that blood cultures are more likely to provideresults leading to a change in management in select patients.Liver disease, hypotension, hypothermia or fever, tachycardia,uremia, hyponatremia, and leucopenia or leukocytosis have beenidentified as independent predictors of bacteremia.40
Immunocompromised patients and patients from nursinghomes or other long-term care facilities are more likely to haveunusual or resistant pathogens identified by bloodcultures.34,39,50 Patients with severe pneumonia may also benefitfrom blood culture tests.29,51 In a prospective Class III study of209 patients, Waterer and Wunderink38 found that bloodculture results led to change in antibiotics only in patients withPSI class IV and V disease, whereas patients in PSI class I to IIIhad no antibiotic changes based on blood culture results.
In summary, the routine use of blood cultures in all patientsadmitted with CAP has a low yield and rarely leads to change inmanagement or outcome for patients admitted with CAP. False-positive blood culture results may complicate the course for patientsadmitted with CAP. Therefore, blood cultures should be tailored tothe individual patient. Patients with severe pneumonia, who areimmunocompromised or have other significant comorbidities, maybenefit from having blood cultures drawn. Because antibioticadministration before blood culture testing decreases blood cultureyield, when blood cultures are necessary, they should be drawnbefore antibiotic administration.37,40,41
2. In adult patients with CAP without severe sepsis, isthere a benefit in mortality or morbidity from theadministration of antibiotics within a specific time course?
Patient Management RecommendationsLevel A recommendations. None specified.Level B recommendations. There is insufficient evidence to
establish a benefit in mortality or morbidity from antibioticsadministered in less than 4, 6, or 8 hours from ED arrival.
Level C recommendations. Administer antibiotics as soon asfeasible once the diagnosis of CAP is established; there isinsufficient evidence to establish a benefit in morbidity or mortalityfrom antibiotics administered within any specific time course.
Key words/phrases for literature searches: pneumonia,community-acquired pneumonia, time to treatment, rapidantibiotic delivery, morbidity, mortality, outcomes, length ofstay, quality of care, timeline 1988 – May 20, 2009.
The timely administration of antibiotics to infected patientsis good emergency medical practice. Before giving antibiotics, areasonable assurance of the diagnosis is essential to avoidmistreatment, medication overuse, and increased antibiotic
13,22,52
resistance.708 Annals of Emergency Medicine
In the most recent consensus guidelines on the managementof CAP in adults, the Infectious Diseases Society of Americaand the American Thoracic Society agreed that there is a paucityof data to support a specific time recommendation for theadministration of antibiotics in ED patients with CAP.27 Theirrecommendation states: for patients admitted through the ED, thefirst antibiotic dose should be administered while [the patient is]still in the ED.†
Four-Hour CutoffIn a frequently cited article, Houck et al53 analyzed whether
the time to first antibiotic dose might be associated withreductions in mortality and morbidity. In a retrospectivemulticenter, Class III study, Houck et al53 examined the chartsof 13,771 Medicare patients with a primary or secondaryInternational Classification of Diseases, Ninth Revision (ICD-9) diagnosis of pneumonia, who had not received out-of-hospital antibiotics. The patients analyzed were older than 65years, had not received out-of-hospital antibiotics, and hadradiographic evidence of pneumonia in the ED. This studyshowed an association between antibiotics administered within4 hours and a decreased 30-day mortality, with an OR of 0.85(95% CI 0.76 to 0.95). There was also a significant associationwith reduction of inhospital mortality and reduction of lengthof stay exceeding the 5-day median.
This study’s limitations include the following: more patientsin the group with time to first antibiotic dose less than 4 hoursreceived appropriate antibiotics, though this was included inmultivariate analysis.53 There was a post hoc determination ofthe 4-hour cutoff. Any of the cutoff times from 3 to 8 hourswere associated with similar 30-day mortality. The researcherschose the 4-hour cutoff, even though adjusted ORs of the 4-and 8-hour cutoffs were identical. They attempted to controlfor confounders through the performance of multivariateanalysis. Although the study controlled for many possibleconfounders, the possibility of missing others potentially biasesthe results, which may account for the fact that despite themultivariate analysis, patients who received antibiotics between0 and 2 hours did not have any significant mortality reduction.
Early administration of antibiotics is reliant on the earlydiagnosis of pneumonia. Patients whose disease is more difficult todiagnose because of atypical presentations may receive theirantibiotics later. If any of the factors that lead to the delayeddiagnosis are also associated with mortality, then the link betweenearly antibiotic administration and mortality may be spurious.Waterer et al54 examined these factors in a prospective Class IIstudy. The researchers performed an observational study of time tofirst antibiotic dose in patients older than 18 years and diagnosedwith CAP during their hospitalization. In univariate analysis, thisstudy confirmed the aforementioned association between time tofirst antibiotic dose less than 4 hours and mortality. However,when the data were examined for factors that can cause a delayed
†Infectious Diseases Society of America/American Thoracic Society
grading: moderate recommendation, level III evidence.Volume , . : November
Clinical Policy
diagnosis of pneumonia, 3 factors emerged: altered mental status,the absence of hypoxia, and the absence of fever. When reanalyzedcontrolling for these factors, all of the mortality benefit associatedwith time to first antibiotic dose less than 4 hours disappeared.Altered mental status and the absence of fever remained associatedwith increased mortality after the multivariate analysis. This study’sresults indicate that for patients presenting with CAP and alteredmental status or the inability to mount a febrile response, it may bemore difficult to rapidly diagnose pneumonia, and they may be athigher risk for death.54 The study by Houck et al53 did notspecifically control for altered mental status or the presence of feverin the multivariate analysis.
In a prospective, observational Class II study, Silber et al55
examined the differences in time to clinical stability‡ in 409patients based on their door-to-antibiotic time. Three cohortswere analyzed: antibiotics in less than 4 hours, antibiotics in 4to 8 hours, and antibiotics in greater than 8 hours. There wereno statistically significant differences in time to clinical stabilitybetween the groups.
In another Class II study, Marrie and Wu56 implemented aCAP pathway for non-ICU patients at 6 Canadian hospitals.They prospectively analyzed the effects of time to first antibioticdose on inhospital mortality. Of the 3,043 patients included inanalysis, the mortality rate for time to first antibiotic dose lessthan 4 hours was 9.2% and the rate for time to first antibioticdose greater than 4 hours was 8.6%. If patients who receivedantibiotics before their arrival at the ED were removed (as in thestudy by Houck et al53), the mortality rate for time to firstantibiotic dose less than 4 hours was 8.3% and the mortalityrate for time to first antibiotic dose greater than 4 hours was8.1%, a nonsignificant difference.
Battleman et al57 performed a Class III, multicenter,retrospective analysis of 609 patients with a chart-codingdiagnosis of pneumonia. They examined the associationbetween time to first antibiotic dose and prolonged length ofstay (prolonged length of stay was defined as �9 days). They foundan association between shorter time to first antibiotic dose andfewer patients with prolonged length of stay. This finding was alsoobserved in patients who received their antibiotics in the ED ratherthan on the floor. This study excluded patients who died, and theactual data analysis of prolonged length of stay was not provided.Potential factors that may lead to a delayed diagnosis were notincluded in the analysis.
Six-Hour CutoffNo research has specifically examined a 6-hour cutoff for the
time to first antibiotic dose. This time period was part of thedata of the study by Houck et al53 mentioned above. This cutoffhad a significant association with reduced mortality (adjusted
‡Time to clinical stability is a composite measure of the first 24-hourperiod during which the patient has all of the following: systolic bloodpressure �90 mm Hg, pulse rate �100 beats/min, respiratory rate �24breaths/min, temperature �101°F, O2 saturation �90, and the ability to
eat.Volume , . : November
OR 0.84; 95% CI 0.73 to 0.95); but the conclusions are limitedby all of the same factors present in the 4-hour cutoff.
Beyond 6 HoursAn 8-hour cutoff for time to first antibiotic dose has been
analyzed in a number of studies. A large, multicenter, retrospective,Class III study by Meehan et al48 demonstrated an associationbetween antibiotic administration within 8 hours of ED arrival andmortality (adjusted OR 0.85; 95% CI 0.75 to 0.96). This studyshares the same methodology as the analysis by Houck et al,53 andits conclusions are limited by many of the same issues. Patientswere included based on claims data, which may have led toselection bias. Confounding factors such as altered mental status,the absence of fever, and other clinical factors hindering diagnosiswere not included in the multivariate analysis.
The study by Marrie and Wu56 mentioned above alsoincluded data on time to first antibiotic dose less than 8 hourscompared with greater than 8 hours. There was no significantmortality difference between these 2 groups. Even when patientswho received antibiotics before arrival at the hospital wereremoved from the cohorts, no significant mortality benefitemerged for early antibiotic administration.§
Dedier et al47 retrospectively studied 1,062 CAP patientsfrom 38 hospitals. This Class III study examined the effect oftime to first antibiotic dose less than 8 hours on inpatientmortality, length of stay, and time to clinical stability. Therewere no significant associations with rapid antibioticadministration in any of these measures. There is insufficientevidence to establish a specific cutoff time for antibioticsadministration in patients who are diagnosed with CAP in theED. In the noncritically ill patient, it is prudent to administerantibiotics as soon as possible after a definitive diagnosis ismade.
Relevant industry relationships of subcommitteemembers: There were no relevant industry relationshipsdisclosed by the subcommittee members.
Relevant industry relationships are those relationshipswith companies associated with products or services thatsignificantly impact the specific aspect of disease addressedin the critical question.
Earn CME Credit: Continuing Medical Education is available forthis article at: http://www.ACEP.EMedHome.com.
REFERENCES1. Arias E, Smith BL. Deaths: preliminary data for 2001. Natl Vital
Stat Rep. 2003;51:1-44.2. Centers for Medicare and Medicaid Services. Medicare and
Medicaid statistical supplement, 1995. Healthcare Finance RevStat. Suppl 1995; 1–388.
3. Halm EA, Tierstein AS. Clinical practice. Management of community-acquired pneumonia. N Engl J Med. 2002;347:2039-2045.
§P values of 0.81 were calculated from the study data with the SPSS 14.0
statistical package (SPSS, Inc, Chicago, IL).Annals of Emergency Medicine 709
Clinical Policy
4. Feikin DR, Schuchat A, Kolczak M, et al. Mortality from invasivepneumococcal pneumonia in the era of antibiotic resistance,1995-1997. Am J Public Health. 2000;90:223-229.
5. Talan PA, DeBleuix P, Kollef MH. A new paradigm in emergencymedicine: healthcare-associated pneumonia. Clinical Courier.2007;25:1-16.
6. American Thoracic Society, Infectious Diseases Society ofAmerica. Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associatedpneumonia. Am J Respir Crit Care Med. 2005;171:388-416.
7. Kollef MH, Shorr A, Tabak YP, et al. Epidemiology and outcomesof health-care-associated pneumonia: results from a large USdatabase of culture-positive pneumonia. Chest. 2005;128:3854-3862.
8. Hiramatsu K, Niederman MS. Health-care-associated pneumonia:a new therapeutic paradigm. Chest. 2005;128:3784-3787.
9. The Joint Commission. Specifications manual for national hospitalquality measures. Available at: http://www.jointcommission.org/PerformanceMeasurement/PerformanceMeasurement/Current�NHQM�Manual.htm. Accessed May 15, 2009.
10. Pines JM, Hollander JE, Lee H, et al. Emergency departmentoperational changes in response to pay-for-performance andantibiotic timing in pneumonia. Acad Emerg Med. 2007;14:545-548.
11. Kanwar M, Brar N, Khatib R, et al. Misdiagnosis of community-acquired pneumonia and inappropriate utilization of antibiotics:side effects of the 4-h antibiotic administration rule. Chest.2007;131:1865-1869.
12. Khalil A, Kelen G, Rothman RE. A simple screening tool foridentification of community-acquired pneumonia in an inner cityemergency department. Emerg Med J. 2007;24:336-338.
13. Metersky ML, Sweeney TA, Getzow MB, et al. Antibiotic timingand diagnostic uncertainty in Medicare patients with pneumonia.Is it reasonable to expect all patients to receive antibiotics within4 hours? Chest. 2006;130:16-21.
14. Fee C, Weber EJ, Maak CA, et al. Effect of emergency departmentcrowding on time to antibiotics in patients admitted withcommunity-acquired pneumonia. Ann Emerg Med. 2007;50:501-509.
15. Pines JM, Localio AR, Hollander JE, et al. The impact ofemergency department crowding measures on time to antibioticsfor patients with community-acquired pneumonia. Ann EmergMed. 2007;50:510-516.
16. Pines JM, Hollander JE, Localio AR, et al. The associationbetween emergency department crowding and hospitalperformance on antibiotic timing for pneumonia and percutaneousintervention for myocardial infarction. Acad Emerg Med. 2006;13:873-878.
17. Capelastegui A, Espana PP, Quintana JM, et al. Validation of apredictive rule for the management of community-acquiredpneumonia. Eur Respir J. 2006;27:151-157.
18. Aujesky D, Auble TE, Yealy DM, et al. Prospective comparison ofthree validated prediction rules for prognosis in community-acquired pneumonia. Am J Med. 2005;118:384-392.
19. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identifylow-risk patients with community-acquired pneumonia. N EnglJ Med. 1997;336:243-250.
20. Lim WS, van der Eerden MM, Laing R, et al. Defining communityacquired pneumonia severity on presentation to hospital: aninternational derivation and validation study. Thorax. 2003;58:377-382.
21. Yealy DM, Auble TE, Stone RA, et al. The emergency departmentcommunity-acquired pneumonia trial: methodology of a quality
improvement intervention. Ann Emerg Med. 2004;43:770-782.710 Annals of Emergency Medicine
22. Walls RM, Resnick J. The Joint Commission on Accreditation ofHealthcare Organizations and Center for Medicare and MedicaidServices community-acquired pneumonia initiative: what wentwrong? Ann Emerg Med. 2005;46:409-411.
23. Fee C, Sharpe BA, Nguy M, et al. JCAHO/CMS core measures forcommunity-acquired pneumonia. Ann Emerg Med. 2006;47:505-506.
24. Moran GJ, Abrahamian FM. Blood cultures for community-acquiredpneumonia: can we hit the target without a shotgun? Ann EmergMed. 2005;46:407-408.
25. Rothman RE, Quianzon CCL, Kelen GD. Narrowing in on JCAHOrecommendations for community-acquired pneumonia. AcadEmerg Med. 2006;13:983-985.
26. Schriger DL, Cantrill SV, Greene CS. The origins, benefits, harms,and implications of emergency medicine clinical policies. AnnEmerg Med. 1993;22:597-602.
27. Mandell LA, Wunderink RG, Anzueto A, et al. Infectious DiseasesSociety of America/American Thoracic Society consensusguidelines on the management of community-acquired pneumoniain adults. Clin Infect Dis. 2007;44:S27-S72.
28. Theerthakarai R, El-Halees W, Ismail M, et al. Nonvalue of theinitial microbiological studies in the management of nonseverecommunity-acquired pneumonia. Chest. 2001;119:181-184.
29. Paganin F, Lilienthal, F, Bourdin A, et al. Severe community-acquired pneumonia: assessment of microbial aetiology asmortality factor. Eur Respir J. 2004;24:779-785.
30. Beovic B, Bonac, B, Kese D, et al. Aetiology and clinicalpresentation of mild community-acquired bacterial pneumonia.Eur J Clin Microbiol Infect Dis. 2003;22:584-591.
31. Campbell SG, Marrie TJ, Anstey R, et al. The contribution of bloodcultures to the clinical management of adult patients admitted tothe hospital with community-acquired pneumonia: a prospectiveobservational study. Chest. 2003;123:1142-1150.
32. Chalasani NP, Valdecanas MA, Gopal AK, et al. Clinical utility ofblood cultures in adult patients with community-acquiredpneumonia without defined underlying risks. Chest. 1995;108:932-936.
33. Corbo J, Friedman B, Bijur P, et al. Limited usefulness of initialblood cultures in community acquired pneumonia. Emerg Med J.2004;21:446-448.
34. El-Solh AA, Sikka P, Ramadan F, et al. Etiology of severepneumonia in the very elderly. Am J Respir Crit Care Med. 2001;163:645-651.
35. Ewig S, Bauer T, Hasper E, et al. Value of routine microbialinvestigation in community-acquired pneumonia treated in atertiary care center. Respiration. 1996;63:164-169.
36. Fine MJ, Stone RA, Singer DE, et al. Processes and outcomes ofcare for patients with community-acquired pneumonia: resultsfrom the Pneumonia Patient Outcomes Research Team (PORT)cohort study. Arch Intern Med. 1999;159:970-980.
37. Glerant JC, Hellmuth D, Schmit JL, et al. Utility of blood culturesin community-acquired pneumonia requiring hospitalization:influence of antibiotic treatment before admission. Respir Med.1999;93:208-212.
38. Waterer GW, Wunderink RG. The influence of the severity ofcommunity-acquired pneumonia on the usefulness of bloodcultures. Respir Med. 2001;95:78-82.
39. Kennedy M, Bates DW, Wright SB, et al. Do emergencydepartment blood cultures change practice in patients withpneumonia? Ann Emerg Med. 2005;46:393-400.
40. Metersky ML, Ma A, Bratzler DW, et al. Predicting bacteremia inpatients with community-acquired pneumonia. Am J Respir CritCare Med. 2004;169:342-347.
41. van der Eerden MM, Vlaspolder F, de Graaff CS, et al. Value of
intensive diagnostic microbiological investigation in low- and high-Volume , . : November
Clinical Policy
risk patients with community-acquired pneumonia. Eur J ClinMicrobiol Infect Dis. 2005;24:241-249.
42. Ramanujam P, Rathlev NK. Blood cultures do not changemanagement in hospitalized patients with community-acquiredpneumonia. Acad Emerg Med. 2006;13:740-745.
43. Socan M, Marinic-Fiser N, Kraigher A, et al. Microbial aetiology ofcommunity-acquired pneumonia in hospitalized patients. EurJ Clin Microbiol Infect Dis. 1999;18:777-782.
44. Woodhead MA, Arrowsmith J, Chamberlain-Webber R, et al. Thevalue of routine microbial investigation in community-acquiredpneumonia. Respir Med. 1991;85:313-317.
45. Sanyal S, Smith PR, Saha AC, et al. Initial microbiologic studiesdid not affect outcome in adults hospitalized with community-acquired pneumonia. Am J Respir Crit Care Med. 1999;160:346-348.
46. Waterer GW, Jennings SG, Wunderink RG. The impact of bloodcultures on antibiotic therapy in pneumococcal pneumonia.Chest. 1999;116:1278-1281.
47. Dedier J, Singer DE, Chang Y, et al. Processes of care, illnessseverity, and outcomes in the management of community-acquired pneumonia at academic hospitals. Arch Intern Med.2001;161:2099-2104.
48. Meehan TP, Fine MJ, Krumholz HM, et al. Quality of care,process, and outcomes in elderly patients with pneumonia. JAMA.1997;278:2080-2084.
49. Bates DW, Goldman L, Lee TH. Contaminant blood cultures andresource utilization. The true consequences of false-positiveresults. JAMA. 1991;265:365-369.
Volume , . : November
50. Lujan M, Gallego M, Fontanals D, et al. Prospective observationalstudy of bacteremic pneumococcal pneumonia: effect of discordanttherapy on mortality. Crit Care Med. 2004;32:625-631.
51. Moine P, Vercken JB, Chevret S, et al. Severe community-acquired pneumonia. Etiology, epidemiology, and prognosisfactors. French Study Group for Community-Acquired Pneumoniain the Intensive Care Unit. Chest. 1994;105:1487-1495.
52. Houck PM. Antibiotics and pneumonia: is timing everything or justa cause of more problems? Chest. 2006;130:1-3.
53. Houck PM, Bratzler DW, Nsa W, et al. Timing of antibioticadministration and outcomes for Medicare patients hospitalizedwith community-acquired pneumonia. Arch Intern Med. 2004;164:637-644.
54. Waterer GW, Kessler LA, Wunderink RG. Delayed administrationof antibiotics and atypical presentation in community-acquiredpneumonia. Chest. 2006;130:11-15.
55. Silber SH, Garrett C, Singh R, et al. Early administration ofantibiotics does not shorten time to clinical stability in patientswith moderate-to-severe community-acquired pneumonia. Chest.2003;124:1798-1804.
56. Marrie TJ, Wu L. Factors influencing in-hospital mortality incommunity-acquired pneumonia: a prospective study ofpatients not initially admitted to the ICU. Chest. 2005;127:1260-1270.
57. Battleman DS, Callahan M, Thaler HT. Rapid antibiotic deliveryand appropriate antibiotic selection reduce length of hospital stayof patients with community-acquired pneumonia: link betweenquality of care and resource utilization. Arch Intern Med. 2002;162:682-688.
Annals of Emergency Medicine 711
Evi
dent
iary
Tab
le.
Stud
y Y
ear
Des
ign
Inte
rven
tion(
s)/T
est(
s)/M
odal
ityO
utco
me
Mea
sure
/Cri
teri
on
Stan
dard
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Thee
rthak
arai
et
al28
2001
Pr
ospe
ctiv
e ob
serv
atio
nal
stud
y; 1
ho
spita
l in
Pate
rson
, NJ
Enro
lled
cons
ecut
ive
patie
nts w
ith
the
diag
nosi
s of C
AP
to a
sses
s the
va
lue
of th
e in
itial
mic
robi
olog
ical
st
udie
s, co
nsis
ting
of sp
utum
Gra
m’s
st
ain,
sput
um c
ultu
re, a
nd b
lood
cu
lture
, in
the
etio
logi
c di
agno
sis o
f C
AP
with
out c
omor
bidi
ty
212
patie
nts
scre
ened
,74
patie
nts
incl
uded
; age
s 22-
64
y; a
ll pa
tient
s had
: sp
utum
Gra
m’s
stai
n (a
ll m
ixed
flor
a),
sput
um c
ultu
re (4
pa
thog
ens 5
%),
bl
ood
cultu
res (
all
nega
tive)
No
posi
tive
bloo
d cu
lture
re
sults
in th
is lo
w-r
isk
popu
latio
n w
ith
nons
ever
e C
AP;
all
patie
nts h
ad im
prov
ed
sym
ptom
s by
48 h
and
be
cam
e af
ebril
e in
96
h;
no p
atie
nt re
quire
d a
chan
ge in
em
piric
an
tibio
tic c
over
age
inst
itute
d at
adm
issi
on
Smal
l sam
ple
size
; un
usua
lly lo
w y
ield
on
cultu
res;
no
base
line
patie
nt
com
paris
ons;
stud
y in
clud
ed
only
thos
e pa
tient
s abl
e to
pr
oduc
e va
lid sp
utum
sa
mpl
e, th
ey c
ould
diff
er
from
all
patie
nts w
ith C
AP;
po
ssib
le se
lect
ion
bias
; 21
(28%
) did
not
mee
t ATS
gu
idel
ine
crite
ria fo
r ad
mis
sion
; mul
tiple
ex
clus
ion
crite
ria,
esse
ntia
lly e
limin
atin
g al
l hi
gh-r
isk,
eld
erly
, and
sick
pa
tient
s
III
Paga
nin
et
al29
2004
Pr
ospe
ctiv
e ob
serv
atio
nal
stud
y 19
95-
2004
; dat
a fr
om 1
hos
pita
l on
a F
renc
h is
land
in th
e In
dian
Oce
an
Con
secu
tive
patie
nts a
dmitt
ed fr
om
the
ED to
ICU
for C
AP
from
9/1
995-
12/2
000;
stud
y ob
ject
ive:
to a
sses
s th
e et
iolo
gy a
nd p
rogn
ostic
fact
ors o
f C
AP
patie
nts a
dmitt
ed to
the
ICU
; ex
clus
ion
crite
ria: s
ever
e im
mun
osup
pres
sion
146
patie
nts,
34
excl
uded
as t
hey
did
not m
eet d
efin
ition
of
CA
P; 1
12 to
tal
incl
uded
; 94
(84%
) m
ale,
70
(62.
5%)
alco
holic
, 48
(43%
) di
ed; 5
5 pa
tient
s PSI
I-
II-I
II; 5
7 pa
tient
s PS
I IV
-V; a
ll ha
d at
le
ast 1
blo
od c
ultu
re;
37 (3
3%) p
ositi
ve
bloo
d cu
lture
; 23
S pn
eum
onia
e, 9
K
lebs
iella
, 2 c
ases
of
resi
stan
t S
pneu
mon
iae
Blo
od c
ultu
re m
ore
likel
y to
be
posi
tive
in si
cker
pa
tient
s, an
d po
sitiv
e bl
ood
cultu
re w
as a
n in
depe
nden
t ris
k fa
ctor
fo
r dea
th in
sick
er
patie
nts w
ith C
AP
(rel
ativ
e ris
k 2.
7; C
I 0.8
-8.
9; P
=0.0
002)
, als
o se
ptic
shoc
k, h
igh
SAPS
II
scor
e an
d in
fect
ion
with
Kle
bsie
lla
Stud
y se
tting
and
pop
ulat
ion
(Fre
nch
isla
nd in
the
Indi
an
Oce
an),
mos
tly m
ale,
m
ostly
alc
ohol
ic; n
ot
gene
raliz
able
, sel
ectio
n bi
as; l
ow le
vel o
f ant
ibio
tic
resi
stan
ce; 5
5 pa
tient
s PSI
I-II
-III
(why
wer
e th
ese
in th
e IC
U?)
III
Clinical Policy
712 Anna
ls of Emergency Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/Mod
ality
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Beo
vic
et a
l30
2003
Pr
ospe
ctiv
e;
mul
ticen
ter i
n Sl
oven
ia
Con
secu
tive
patie
nts w
ith C
AP
pres
entin
g to
7 st
udy
cent
ers l
ooki
ng
at e
tiolo
gy a
nd c
linic
al p
ictu
re o
f m
ild C
AP;
stud
y pa
tient
s wer
e bo
th
inpa
tient
and
out
patie
nts
116
patie
nts e
nrol
led,
11
3 in
clud
ed in
stud
y 10
9 ha
d co
mpl
ete
data
; 96/
109
(88%
) w
ere
PSI I
or I
I; 1
patie
nt h
ad a
pos
itive
bl
ood
cultu
re (S
pn
eum
onia
e);
etio
logy
est
ablis
hed
in 6
8 (6
2.4%
), 17
ty
pica
l, 42
aty
pica
l, 9
mix
ed
Aty
pica
l pat
hoge
ns p
lay
an im
porta
nt ro
le in
mild
C
AP;
ther
e w
as a
su
bsta
ntia
l sim
ilarit
y in
th
e cl
inic
al p
rese
ntat
ion
of p
neum
onia
cau
sed
by
diff
eren
t age
nts;
blo
od
cultu
res a
re v
ery
rare
ly
posi
tive
in m
ild C
AP
treat
ed w
ith o
ral
antib
iotic
s
Trea
tmen
t with
ora
l age
nts
was
incl
usio
n cr
iteria
; po
tent
ial s
elec
tion
bias
; ve
ry sm
all n
umbe
r of
patie
nts g
iven
that
en
rollm
ent i
nclu
ded
7 st
udy
cent
ers;
stud
y pa
tient
s wer
e bo
th in
patie
nt a
nd
outp
atie
nts;
inve
stig
ator
s do
not r
epor
t how
man
y w
ere
inpa
tient
s
III
Clinical Policy
Volume , . : November
Annals of Emergency Medicine 713Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/Mod
ality
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Cam
pbel
l et
al31
2003
Pr
ospe
ctiv
e;
mul
ticen
ter,
19 c
ente
rs in
C
anad
a
Patie
nts a
dmitt
ed w
ith C
AP
eith
er
rece
ivin
g ca
re p
er c
linic
al g
uide
line
or c
onve
ntio
nal m
anag
emen
t;
clin
ical
use
fuln
ess o
f blo
od c
ultu
re in
th
e m
anag
emen
t of p
atie
nts
hosp
italiz
ed w
ith C
AP
2,80
4 pa
tient
s en
rolle
d, 1
,061
ex
clud
ed; 7
16
inte
rven
tion
arm
; 1,
027
conv
entio
nal
arm
; 760
(74%
) bl
ood
cultu
res d
raw
n,
43 (5
.7%
) “s
igni
fican
t” p
ositi
ve
bloo
d cu
lture
; 3
patie
nts (
0.4%
, 3/
760)
cha
nged
to
broa
der s
pect
rum
as
indi
cate
d by
blo
od
cultu
re, 1
MSS
A, 1
PR
SP, 1
MR
SA;
6 ch
ange
d to
bro
ader
sp
ectru
m n
ot
indi
cate
d by
blo
od
cultu
re; 1
2 ch
ange
d to
nar
row
er/c
heap
er
as in
dica
ted
by b
lood
cu
lture
; 2 c
hang
ed to
na
rrow
er/c
heap
er n
ot
indi
cate
d by
blo
od
cultu
re; 1
7 co
ntin
ued
empi
ric th
erap
y de
spite
blo
od c
ultu
re
indi
catio
n to
step
do
wn;
blo
od c
ultu
re
resu
lts d
id n
ot
corr
elat
e w
ith P
SI
Ther
e w
as a
2%
cha
nce
(15/
760)
of h
avin
g a
chan
ge o
f the
rapy
di
rect
ed b
y bl
ood
cultu
re
resu
lts; i
n on
ly 0
.4%
was
th
is c
hang
e lik
ely
to h
ave
impr
oved
the
outc
ome
for
the
patie
nt; t
hose
with
po
sitiv
e bl
ood
cultu
re h
ad
a 39
% c
hanc
e of
hav
ing
ther
apy
chan
ged
due
to
bloo
d cu
lture
resu
lts, a
nd
a 42
% c
hanc
e of
hav
ing
ther
apy
cont
inue
d no
t in
dica
ted
by b
lood
cul
ture
re
sults
; rou
tine
bloo
d cu
lture
s rar
ely
cont
ribut
e si
gnifi
cant
ly to
the
clin
ical
man
agem
ent o
f C
AP
Dat
a pu
lled
from
stud
y on
us
e of
clin
ical
pat
hway
for
man
agin
g C
AP
— li
mits
in
tern
al v
alid
ity; l
arge
nu
mbe
r of p
atie
nts e
xclu
ded
— p
oten
tial s
elec
tion
bias
; in
terv
entio
n ar
m p
atie
nts
may
be
less
like
ly to
step
do
wn
or c
hang
e dr
ugs
beca
use
drug
is su
pplie
d;
in
terv
entio
n pa
tient
s mor
e lik
ely
to h
ave
bloo
d cu
lture
dr
awn
(58%
vs.
33%
);
limits
val
idity
; bas
elin
e ch
arac
teris
tics o
f pat
ient
s no
t com
pare
d; se
lect
ion
bias
; fal
se-p
ositi
ve
cont
amin
ants
not
cou
nted
or
disc
usse
d
II
Clinical Policy
714 Annals of
Emergency Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/Mod
ality
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Cha
lasa
ni e
t al
32
1995
R
etro
spec
tive;
si
ngle
in
stitu
tion
Cha
rt re
view
of a
dults
hos
pita
lized
w
ith C
AP
to d
eter
min
e th
e cl
inic
al
utili
ty o
f obt
aini
ng ro
utin
e bl
ood
cultu
re b
efor
e th
e ad
min
istra
tion
of
antib
iotic
s in
certa
in n
on-
imm
unos
uppr
esse
d pa
tient
s
1,25
0 pa
tient
s id
entif
ied
with
di
scha
rge
diag
nosi
s of
CA
P, 5
17 p
atie
nts
met
stud
y cr
iteria
; 6.
6% (3
4) tr
ue-
posi
tive
bloo
d cu
lture
, 4.8
% (2
5)
cont
amin
ated
blo
od
cultu
re; 5
6 pa
tient
s ha
d an
tibio
tics
chan
ged:
42
patie
nts
with
neg
ativ
e bl
ood
cultu
re a
nd 1
4 pa
tient
s with
pos
itive
bl
ood
cultu
re; 1
.4%
(7
of 5
17 p
atie
nts)
ha
d an
tibio
tic c
hang
e as
a re
sult
of b
lood
cu
lture
resu
lts, 6
na
rrow
ed, a
nd 1
br
oade
ned
to c
over
H
influ
enza
e
Blo
od c
ultu
res h
ave
limite
d cl
inic
al u
tility
and
qu
estio
nabl
e co
st-
effe
ctiv
enes
s; n
o pe
nici
llin
resi
stan
ce
note
d; ra
te o
f tru
e-
posi
tive
bloo
d cu
lture
si
mila
r to
rate
of
cont
amin
ated
blo
od
cultu
re
Ret
rosp
ectiv
e de
sign
; pa
tient
s ide
ntifi
ed b
y di
scha
rge
diag
nosi
s;
sele
ctio
n bi
as; l
ow ra
te o
f an
tibio
tic re
sist
ance
co
mpa
red
to c
urre
nt 2
007
rate
s; c
onta
min
ant
dete
rmin
ed b
y tre
atin
g ph
ysic
ian;
reas
on fo
r an
tibio
tic c
hang
e in
ferr
ed
for t
he c
hart,
not
nec
essa
rily
docu
men
ted
III
Clinical Policy
Volume , . : November
Annals of Emergency Medicine 715Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Cor
bo e
t al33
2004
R
etro
spec
tive;
si
ngle
in
stitu
tion
in
the
Bro
nx
In E
D p
atie
nts h
ospi
taliz
ed w
ith
CA
P, th
e hy
poth
esis
that
the
prop
ortio
n of
fals
e-po
sitiv
e bl
ood
cultu
res w
ould
exc
eed
the
prop
ortio
n of
true
pos
itive
s was
te
sted
; a se
cond
ary
aim
was
to
quan
tify
the
freq
uenc
y w
ith
whi
ch a
ntib
iotic
ther
apy
was
ch
ange
d ba
sed
on b
lood
cul
ture
re
sults
821
patie
nts a
dmitt
ed, 3
55
had
bloo
d cu
lture
s;20
%
posi
tive
bloo
d cu
lture
s (7
0/35
5), 3
3 tru
e-po
sitiv
e (9
%) a
nd 3
7 fa
lse-
posi
tive
(10%
); 23
8 pa
tient
s had
ch
ange
in a
ntib
iotic
s; 2
5 tru
e-po
sitiv
e ch
ange
d an
tibio
tics:
10
due
to
bloo
d cu
lture
s, 10
due
to
clin
ical
impr
ovem
ent,
1 du
e to
wor
seni
ng, 4
for
othe
r rea
sons
; 26
fals
e po
sitiv
e ch
ange
d an
tibio
tics:
6 d
ue to
blo
od
cultu
res,
187/
285
with
ne
gativ
e bl
ood
cultu
res
chan
ged
antib
iotic
s with
2
chan
ges d
ue to
blo
od
cultu
re re
sults
; ove
rall,
18
patie
nts (
5%) h
ad
antib
iotic
cha
nge
attri
bute
d to
blo
od
cultu
re: 1
0 tru
e-po
sitiv
e w
ith a
ntib
iotic
cha
nge
(7
narr
owed
, 3 b
road
ened
[n
ot b
ecau
se re
sist
ant])
, 6
fals
e po
sitiv
e w
ith
antib
iotic
cha
nge,
2 tr
ue-
nega
tive
with
ant
ibio
tic
chan
ge; 1
51 (4
3%) h
ad
antib
iotic
s cha
nged
due
to
clin
ical
impr
ovem
ent
and
23 (6
%) w
ith
antib
iotic
s cha
nged
due
to
clin
ical
det
erio
ratio
n
Rat
e of
con
tam
inat
ed
bloo
d cu
lture
s equ
aled
ra
te o
f tru
e-po
sitiv
e bl
ood
cultu
res;
clin
ical
co
nditi
on is
use
d m
uch
mor
e fr
eque
ntly
than
bl
ood
cultu
re to
cha
nge
antib
iotic
s; n
o or
gani
sm
was
iden
tifie
d by
blo
od
cultu
re th
at w
as re
sist
ant
to a
ntib
iotic
regi
men
or
igin
ally
cho
sen
Ret
rosp
ectiv
e de
sign
; un
derly
ing
cond
ition
s sta
ted
to b
e si
mila
r in
grou
ps b
ut
no ta
ble
prov
ided
; aut
hors
co
mm
ent t
hat l
engt
h of
stay
is
incr
ease
d w
hen
antib
iotic
co
vera
ge is
err
oneo
usly
br
oade
ned
to c
over
fals
e-
posi
tive
bloo
d cu
lture
re
sults
but
no
data
giv
en;
no d
ata
on m
orta
lity,
leng
th
of st
ay;P
SI n
ot re
porte
d
III
Clinical Policy
716 Annals o
f Emergency Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
El-S
olh
et
al34
2001
Pr
ospe
ctiv
e co
hort;
2
univ
ersi
ty
hosp
itals
in
New
Yor
k st
ate
Elde
rly p
atie
nts w
ith C
AP
adm
itted
to a
n IC
U w
hile
re
ceiv
ing
mec
hani
cal v
entil
atio
n st
udie
d to
det
erm
ine
the
prev
alen
ce o
f res
pira
tory
pa
thog
ens a
nd th
e ef
fect
of
com
orbi
dity
and
func
tiona
l st
atus
on
the
mic
robi
al e
tiolo
gy
of se
vere
pne
umon
ia in
the
very
el
derly
; nur
sing
hom
e, a
s wel
l as
CA
P pa
tient
s, in
clud
ed
136
patie
nts e
ligib
le, 1
04
patie
nts e
nrol
led,
57
from
ho
me,
47
from
nur
sing
ho
me;
in c
omm
unity
pa
tient
s the
mos
t com
mon
pa
thog
en w
as S
pn
eum
onia
e, le
gion
ella
; in
nur
sing
hom
e pa
tient
s th
e m
ost c
omm
on
path
ogen
was
S a
ureu
s(M
SSA
11,
MR
SA 3
); m
orta
lity
of 5
4.8%
not
di
ffer
ent b
etw
een
com
mun
ity v
s nur
sing
ho
me
patie
nts;
m
orta
lity
sign
ifica
ntly
hi
gher
in th
ose
who
re
ceiv
ed in
adeq
uate
an
timic
robi
al th
erap
y (3
9% v
s 4%
, P=0
.007
)
93 b
lood
cul
ture
s, 15
po
sitiv
e (1
6%),
mor
e po
sitiv
e fr
om n
ursi
ng
hom
e th
an h
ome
(10
vs 5
) bu
t not
stat
istic
ally
si
gnifi
cant
; eld
erly
nu
rsin
g ho
me
patie
nts
requ
iring
mec
hani
cal
vent
ilatio
n ar
e at
risk
for
path
ogen
s tha
t are
di
ffer
ent f
rom
the
usua
l C
AP
and
thos
e pa
thog
ens
are
pote
ntia
lly d
rug
resi
stan
t
Few
dat
a on
blo
od c
ultu
res;
ve
ry sp
ecifi
c, se
lect
po
pula
tion,
not
ge
nera
lizab
le; p
hysi
cian
ca
re n
ot st
anda
rdiz
ed
III
Ewig
et a
l3519
96
Ret
rosp
ectiv
e;
1 ho
spita
l in
Ger
man
y
CA
P pa
tient
s ref
erre
d to
a
terti
ary
care
cen
ter s
tudi
ed to
de
term
ine
the
diag
nost
ic y
ield
of
mic
robi
olog
ical
inve
stig
atio
ns
and
thei
r val
ue in
dire
ctin
g an
tibio
tic th
erap
y; re
latio
nshi
p be
twee
n m
icro
bial
resu
lts a
nd
asso
ciat
ion
with
pre
treat
men
t, se
verit
y of
dis
ease
, and
cha
nge
in a
ntib
iotic
s
93 e
piso
des i
n 92
pat
ient
s, 32
ICU
pat
ient
s 22
tran
sfer
s in
from
an
othe
r ins
titut
ion;
20
die
d; 7
4% (6
9) tr
eate
d w
ith a
t lea
st 1
ant
ibio
tic
befo
re a
dmis
sion
; 50
blo
od c
ultu
res d
one,
w
ith 7
pos
itive
(14%
); 52
sero
logy
with
12
defin
itive
pat
hoge
ns;
25 b
ronc
hosc
opy
with
1
defin
itive
pat
hoge
n;
56 sp
utum
cul
ture
—
excl
uded
to id
entif
y de
finiti
ve p
atho
gen
Res
ults
of m
icro
bial
in
vest
igat
ion
led
to
antib
iotic
cha
nge
in 9
ca
ses;
blo
od c
ultu
re
resu
lts le
d to
ant
ibio
tic
chan
ge in
0 c
ases
; de
finiti
ve p
atho
gen
iden
tifie
d in
8/3
2 (2
5%)
seve
re a
nd 1
1/51
(22%
) no
nsev
ere
CA
P; se
verit
y di
d no
t cor
rela
te w
ith
abili
ty to
iden
tify
path
ogen
—th
ey d
id n
ot
spec
ifica
lly a
ddre
ss b
lood
cu
lture
and
seve
rity
Smal
l stu
dy p
opul
atio
n gi
ven
data
from
8 y
; al
thou
gh n
o ba
selin
e pa
tient
ta
ble,
repo
rted
mix
is
atyp
ical
(mal
e:fe
mal
e 62
:30)
, als
o lo
ts o
f tra
nsfe
rs
in, m
uch
pote
ntia
l bia
s, ca
nnot
gen
eral
ize
to E
D
popu
latio
n; P
SI n
ot re
porte
d
III
Clinical Policy
Volume
, . : November Annals of Emergency Medicine 717Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/Mod
ality
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Fine
et a
l36
1999
Pr
ospe
ctiv
e ob
serv
atio
nal;
m
ultic
ente
r
Am
bula
tory
and
hos
pita
lized
CA
P pa
tient
s stu
died
for p
roce
ss-o
f-ca
re
bloo
d cu
lture
, oth
er la
bora
tory
and
m
icro
biol
ogic
test
ing,
leng
th o
f sta
y,
adm
it to
ICU
, mor
talit
y, ti
me
to
retu
rn to
usu
al a
ctiv
ities
12,5
00 p
oten
tial
case
s of C
AP
scre
ened
; 3,9
64
pote
ntia
l par
ticip
ants
; 2,
287
(57.
7%)
patie
nts e
nrol
led,
944
ou
tpat
ient
s, 1,
343
inpa
tient
s; 8
.5%
(77)
of
out
patie
nts h
ad
bloo
d cu
lture
bef
ore
antib
iotic
s, 2.
6% (2
) w
ere
posi
tive;
71.
2%
(951
) inp
atie
nts h
ad
bloo
d cu
lture
bef
ore
antib
iotic
s, 82
(8.6
%)
wer
e po
sitiv
e
Mos
t pat
ient
s with
pn
eum
onia
hav
e pn
eum
onia
of u
nkno
wn
etio
logy
, neg
ativ
e bl
ood
cultu
re; S
pne
umon
iae
and
H
influ
enza
e m
ost c
omm
on
path
ogen
s ide
ntifi
ed; b
lood
cu
lture
rec
omm
ende
d de
spite
low
yie
ld b
ecau
se
of th
e pr
ogno
stic
im
porta
nce
of b
acte
rem
ia
and
the
pote
ntia
l to
dire
ct
ther
apy
agai
nst a
spec
ific
path
ogen
Larg
e nu
mbe
r of e
ligib
le
patie
nts n
ot e
nrol
led;
en
rolle
d pa
tient
s wer
e yo
unge
r, m
ore
likel
y to
be
whi
te, m
ore
likel
y to
be
low
ris
k fo
r mor
talit
y; fe
w
outp
atie
nts h
ad b
lood
cu
lture
don
e; st
udy
did
not
dire
ctly
ass
ess t
he e
ffec
t of
rout
ine
mic
robi
olog
ic
test
ing
on m
edic
al o
utco
mes
III
Gle
rant
et
al37
1999
Pr
ospe
ctiv
e ob
serv
atio
nal;
1
hosp
ital i
n Fr
ance
Patie
nts h
ospi
taliz
ed fo
r mod
erat
e C
AP
(non
-IC
U) t
o co
mpa
re th
e ut
ility
and
cos
t ben
efits
of b
lood
cu
lture
in p
atie
nts w
ho h
ad o
r had
no
t rec
eive
d an
tibio
tic th
erap
y be
fore
ad
mis
sion
53 p
atie
nts;
all
had
bloo
d cu
lture
s;
30 n
o pr
evio
us
antib
iotic
, 23
had
prev
ious
ant
ibio
tic;
30 w
ithou
t pre
viou
s an
tibio
tics h
ad 7
4 bl
ood
cultu
res d
raw
n,
8 po
sitiv
e in
5
patie
nts;
23
with
pr
evio
us a
ntib
iotic
s ha
d 62
blo
od c
ultu
res
draw
n, 0
true
- po
sitiv
e, 2
co
ntam
inan
ts;
bact
erem
ia in
pat
ient
s w
ithou
t pre
viou
s an
tibio
tic 5
/30
vs
with
ant
ibio
tic 0
/23
P<0.
05; a
ll is
olat
ed
orga
nism
s wer
e su
scep
tible
to a
nti-
biot
ic in
itial
ly c
hose
n
Ther
e is
redu
ced
clin
ical
ut
ility
and
cos
t ben
efit
of
bloo
d cu
lture
s in
patie
nts
hosp
italiz
ed fo
r mod
erat
e C
AP
who
hav
e re
ceiv
ed a
n an
tibio
tic tr
eatm
ent b
efor
e
adm
issi
on; b
lood
cul
ture
s no
t lik
ely
to b
e po
sitiv
e in
m
oder
atel
y ill
hos
pita
lized
pa
tient
s pre
viou
sly
treat
ed
with
ant
ibio
tics
Aut
hors
do
not s
tate
how
m
any
CA
P pa
tient
s wer
e m
isse
d or
not
enr
olle
d;
smal
l stu
dy p
opul
atio
n;
auth
ors s
tate
coe
xist
ing
illne
sses
sim
ilar i
n pr
etre
ated
and
not
pre
treat
ed
grou
ps; h
owev
er, n
o ta
ble
or
stat
istic
s pro
vide
d to
show
ba
selin
e ch
arac
teris
tics o
f th
e 2
grou
ps; P
SI n
ot
repo
rted
III
Clinical Policy
718 Ann
als of Emergency Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Wat
erer
and
W
unde
rink38
2001
Pr
ospe
ctiv
e co
hort;
1
hosp
ital i
n M
emph
is
Pros
pect
ivel
y st
udie
d th
e yi
eld
and
effe
ct o
f blo
od
cultu
re in
pat
ient
s adm
itted
w
ith C
AP;
stud
ied
the
rela
tions
hip
betw
een
bloo
d cu
lture
yie
ld a
nd
corr
elat
ion
with
PSI
, as
wel
l as w
heth
er b
lood
cu
lture
resu
lts le
d to
a
chan
ge in
man
agem
ent;
hypo
thes
ized
that
blo
od
cultu
re w
ould
onl
y ha
ve a
si
gnifi
cant
eff
ect o
n pa
tient
m
anag
emen
t in
patie
nts i
n PS
I gra
des I
V a
nd V
; in
clud
ed o
nly
if su
bjec
ts
had
2 bl
ood
cultu
res b
efor
e
any
antib
iotic
; exc
lusi
on
crite
ria: n
onam
bula
tory
nu
rsin
g ho
me
patie
nts,
had
chem
othe
rapy
in p
ast 3
0 da
ys, h
ad p
revi
ous
hosp
italiz
atio
n in
pas
t 30
days
, AID
S,
imm
unos
uppr
essa
nt
ther
apy
Hig
her P
SI c
orre
late
d w
ith
high
er y
ield
from
blo
od c
ultu
re
P=0.
02
PSI #
+blo
od c
ultu
re
I – 1
(5.3
%)
II –
6 (1
0.2%
) II
I – 4
(10.
3%)
IV –
10
(26.
7%)
V –
8 (1
3.9%
); ch
ange
in m
anag
emen
t bas
ed
on b
lood
cul
ture
resu
lts; n
o di
ffer
ence
in m
orta
lity
in
patie
nts w
ith e
mpi
ric a
ntib
iotic
ch
ange
(16%
) vs t
hose
with
ch
ange
bas
ed o
n m
icro
biol
ogic
al re
sults
(25%
) (s
igni
fican
ce n
ot re
porte
d);
20 S
pne
umon
ia is
olat
ed,
3 ha
d M
IC>2
for p
enic
illin
, 11
resi
stan
t to
eryt
hrom
ycin
209
subj
ects
; all
had
bloo
d cu
lture
s;22
(10.
5%) d
ied,
38
(18.
2%) p
ositi
ve b
lood
cu
lture
, 9
(4%
) con
tam
inan
ts,
29 (1
3.9%
) tru
e-po
sitiv
e bl
ood
cultu
re,
12/2
9 ha
d ch
ange
in
man
agem
ent b
ased
on
bloo
d cu
lture
resu
lts: i
n 7
antib
iotic
ther
apy
was
in
tens
ified
, cha
nged
in 1
pa
tient
, and
dec
reas
ed in
5
patie
nts;
for P
SI I-
III,
11/1
17 h
ad p
ositi
ve b
lood
cu
lture
, 0 h
ad c
hang
e in
m
anag
emen
t bas
ed o
n bl
ood
cultu
re; f
or P
SI IV
-V
, 18/
92 h
ad p
ositi
ve
bloo
d cu
lture
, 12
had
chan
ge in
man
agem
ent
base
d on
blo
od c
ultu
re;
bloo
d cu
lture
isol
ate
resi
stan
t to
empi
ric
antib
iotic
in 1
cas
e; b
lood
cu
lture
resu
lts le
d to
a
chan
ge in
man
agem
ent
only
in si
cker
pat
ient
s with
PS
I IV
-V
.
Pros
pect
ive
coho
rt, n
ot c
lear
th
at th
is w
as c
onse
cutiv
e pa
tient
s; o
nly
incl
uded
pa
tient
s who
had
2 b
lood
cu
lture
s bef
ore
antib
iotic
s;
auth
ors d
o no
t rep
ort h
ow
man
y to
tal p
atie
nts w
ith
CA
P w
ere
adm
itted
and
did
no
t hav
e bl
ood
cultu
re; a
lso
auth
ors d
o no
t rep
ort
num
ber o
f pat
ient
s with
C
AP
not e
nrol
led;
pot
entia
l se
lect
ion
bias
; con
clus
ions
ab
out p
atie
nts w
ith p
ositi
ve
bloo
d cu
lture
are
lim
ited
by
the
smal
l num
ber o
f the
se
patie
nts,
n=29
; the
1 p
atie
nt
with
a b
lood
cul
ture
sh
owin
g a
resi
stan
t or
gani
sm le
adin
g to
a
chan
ge in
ant
ibio
tic d
ied
III
Clinical Policy
Volume , . :
November Annals of Emergency Medicine 719Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Ken
nedy
et
al39
2005
Pr
ospe
ctiv
e ob
serv
atio
nal;
1
hosp
ital i
n B
osto
n
Patie
nts a
dmitt
ed w
ith
bloo
d cu
lture
don
e an
d C
AP
diag
nose
d, a
nd th
e re
latio
nshi
p be
twee
n bl
ood
cultu
re re
sults
and
cha
nge
in e
mpi
ric th
erap
y in
blo
od
cultu
re–p
ositi
ve p
atie
nts
3,76
2 ED
pat
ient
s had
blo
od
cultu
res,
414
patie
nts
diag
nose
d w
ith p
neum
onia
; 7%
(29)
blo
od c
ultu
res t
rue-
po
sitiv
e, 3
60 b
lood
cul
ture
s ne
gativ
e; 6
% b
lood
cul
ture
s co
nsid
ered
con
tam
inat
ed;
3 pa
tient
s die
d be
fore
blo
od
cultu
re re
sults
; 15
patie
nts h
ad
ther
apy
alte
red
by b
lood
cu
lture
resu
lts: 1
1 na
rrow
ed, 4
br
oade
ned;
in 1
1 pa
tient
s the
th
erap
y un
chan
ged,
and
of
thes
e, 8
cou
ld h
ave
been
na
rrow
ed; 4
pat
ient
s had
blo
od
cultu
res p
ositi
ve fo
r org
anis
m
resi
stan
t to
empi
ric th
erap
y; 2
ha
d th
erap
y ch
ange
d to
bet
ter
antib
iotic
bef
ore
bloo
d cu
lture
re
sults
(bas
ed o
n cl
inic
al
cond
ition
); al
l 4 o
f the
se
patie
nts h
ad ri
sk fa
ctor
s for
re
sist
ant o
rgan
ism
s: 3
nur
sing
ho
me
resi
dent
s and
1 a
lcoh
olic
w
ith m
ultip
le c
omor
bidi
ties;
30 o
rgan
ism
s ide
ntifi
ed in
29
patie
nts;
12/
30 n
onsu
scep
tible
to
at l
east
1 a
ntib
iotic
; 9/3
0 no
nsus
cept
ible
to a
gent
s in
mor
e th
an 1
ant
ibio
tic c
lass
Blo
od c
ultu
res a
re lo
w
yiel
d an
d in
freq
uent
ly
chan
ge m
anag
emen
t; 3.
6%
of a
ll pa
tient
s had
blo
od
cultu
re; i
n bl
ood
cultu
re
posi
tive
patie
nts,
bloo
d cu
lture
lead
s to
chan
ge in
m
anag
emen
t in
52%
(1
5/29
); 10
0 bl
ood
cultu
res
wou
ld h
ave
to b
e do
ne in
C
AP
patie
nts t
o id
entif
y 1
patie
nt w
ith a
resi
stan
t or
gani
sm; a
ll pa
tient
s with
bl
ood
cultu
res p
ositi
ve fo
r re
sist
ant p
atho
gens
had
ris
k fa
ctor
s for
resi
stan
t or
gani
sms:
3 n
ursi
ng h
ome
resi
dent
s and
1 a
lcoh
olic
w
ith m
ultip
le
com
orbi
ditie
s; ra
te o
f tru
e-po
sitiv
e bl
ood
cultu
res
sim
ilar t
o ra
te o
f co
ntam
inat
ed b
lood
cu
lture
s
Ana
lysi
s of b
lood
cul
ture
- po
sitiv
e pa
tient
s as a
gro
up
is p
robl
emat
ic b
ecau
se th
ere
are
only
29
patie
nts;
lo
w ra
te o
f pen
icill
in
resi
stan
ce (2
0%);
ob
tain
ing
bloo
d cu
lture
was
pa
rt of
stud
y in
clus
ion
crite
ria; m
ay o
vere
stim
ate
bloo
d cu
lture
yie
ld;
stud
y di
d no
t inc
lude
pa
tient
s with
CA
P w
ho d
id
not h
ave
a bl
ood
cultu
re
done
: sel
ectio
n bi
as
II
Clinical Policy
720 Annals of Emerg
ency Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Met
ersk
y et
al
4020
04
Ret
rosp
ectiv
e;
mul
ticen
ter
natio
nal s
tudy
fr
om M
edic
are
clai
ms d
atab
ase
Rev
iew
of M
edic
are
Nat
iona
l Pne
umon
ia
Proj
ect/C
MS
QI p
rogr
am
data
base
to d
eter
min
e pr
edic
tors
of b
acte
rem
ia;
deci
sion
tool
mad
e an
d va
lidat
ed;
deriv
atio
n of
rule
: 4/
1998
-3/1
999;
va
lidat
ion
of ru
le:
7/20
00-3
/200
1
Der
ivat
ion
stud
y; 3
9,24
2 ca
ses
of p
neum
onia
,16,
327
excl
uded
—
no
bloo
d cu
lture
; 5,1
80
excl
uded
bas
ed o
n cr
iteria
and
4,
692
excl
uded
for m
issi
ng
data
; 13,
043
case
s rev
iew
ed:
7% (8
86) b
acte
rem
ia;
5% (6
43) c
onta
min
ated
blo
od
cultu
res;
mul
tivar
iate
ana
lysi
s sh
owed
incr
ease
d le
ngth
of
stay
due
to fa
lse-
posi
tive
bloo
d cu
lture
resu
lts; u
se o
f an
tibio
tics b
efor
e bl
ood
cultu
re
was
neg
ativ
ely
asso
ciat
ed w
ith
bact
erem
ia; i
ndep
ende
nt
pred
icto
rs o
f bac
tere
mia
: liv
er d
isea
se, s
ysto
lic B
P <9
0 m
m H
g; te
mpe
ratu
re <
35° o
r >4
0° C
; pul
se >
125
bea
ts/m
in;
bloo
d ur
ea n
itrog
en >
30
mg/
dL; s
odiu
m <
130
mm
ol/L
W
BC
<5,
000/
mm
3 or
>20,
000/
mm
3 ; age
, res
pira
tory
co
mpr
omis
e no
t ass
ocia
ted
with
bac
tere
mia
; val
idat
ion
stud
y: 1
2,77
1 pa
tient
s,7%
(9
54) b
acte
rem
ic; 8
49
bact
erem
ic p
atie
nts w
ould
be
iden
tifie
d by
dec
isio
n to
ol, 1
05
mis
sed;
583
(5%
) con
tam
inan
ts
Patie
nts w
ith c
onta
min
ated
bl
ood
cultu
res h
ad lo
nger
le
ngth
of s
tay
than
thos
e w
ho d
id n
ot P
<0.0
1;us
e of
an
tibio
tics b
efor
e bl
ood
cultu
re w
as n
egat
ivel
y as
soci
ated
with
bac
tere
mia
;de
cisi
on to
ol id
entif
ied
88%
-89%
of p
atie
nts w
ith
bact
erem
ia w
hile
redu
cing
38
% o
f blo
od c
ultu
res
done
;20%
mor
talit
y am
ong
patie
nts w
ith
bact
erem
ia w
ould
hav
e be
en m
isse
d by
dec
isio
n ru
le; P
SI n
ot si
gnifi
cant
ly
asso
ciat
ed w
ith b
acte
rem
ia
Patie
nts i
dent
ified
from
cl
aim
s dat
a w
ith
retro
spec
tive
revi
ew,
po
tent
ial s
elec
tion
bias
; pa
tient
s age
>65
y, p
oten
tial
for b
ias;
not
gen
eral
izab
le;
tool
is b
ette
r at d
etec
ting
pneu
moc
occa
l bac
tere
mia
th
an o
ther
pat
hoge
ns; o
nly
dete
cted
65%
of n
on-
pneu
moc
occa
l St
rept
ococ
cus s
p;
a pr
oble
m b
ecau
se o
ne g
oal
of b
lood
cul
ture
is to
id
entif
y un
usua
l org
anis
ms;
st
udy
not d
esig
ned
to
anal
yze
outc
ome;
rule
not
te
sted
pro
spec
tivel
y
II fo
r bl
ood
cultu
re
yiel
d;
III f
or
othe
r co
nclu
-si
ons
Clinical Policy
Volume , . : N
ovember Annals of Emergency Medicine 721Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
van
der
Eerd
en e
t al
41
2005
Pr
ospe
ctiv
e ob
serv
atio
nal;
1 la
rge
hosp
ital
in th
e N
ethe
rland
s
Eval
uate
d th
e di
agno
stic
yi
eld
of d
iffer
ent
mic
robi
olog
ical
test
s in
hosp
italiz
ed p
atie
nts w
ith
CA
P
262
patie
nts,
158
(60%
), pa
tient
s with
iden
tifie
d pa
thog
en, 4
0 (1
5%) p
ositi
ve
bloo
d cu
lture
s; n
o pe
nici
llin
or
mac
rolid
e re
sist
ant S
pn
eum
onia
e id
entif
ied;
pr
etre
atm
ent w
ith a
ntib
iotic
s le
d to
low
er b
lood
cul
ture
yi
eld:
5/6
6 (8
%) v
s 35/
188
(19%
), P=
0.03
; com
bina
tion
sput
um e
xam
inat
ion
with
G
ram
’s st
ain,
cul
ture
, and
pn
eum
ococ
cal a
ntig
en sh
owed
th
e hi
ghes
t dia
gnos
tic y
ield
(4
9%),
follo
wed
by
urin
ary
PCA
test
(20%
), fo
llow
ed b
y bl
ood
cultu
re (1
6%);
no
corr
elat
ion
betw
een
bloo
d cu
lture
yie
ld a
nd d
isea
se
seve
rity/
PSI
Inve
stig
atio
n of
sput
um
with
Gra
m’s
stai
n; c
ultu
re
and
pneu
moc
occa
l ant
igen
pr
ovid
ed th
e la
rges
t yie
ld
in d
eter
min
ing
the
etio
logy
of C
AP;
pr
etre
atm
ent w
ith
antib
iotic
s dec
reas
es b
lood
cu
lture
yie
ld
Tota
l num
ber o
f pat
ient
s ho
spita
lized
for p
neum
onia
an
d ho
w m
any
patie
nts w
ere
not e
nrol
led
and
not
repo
rted
— p
oten
tial
sele
ctio
n bi
as; s
ome
base
line
char
acte
ristic
s gi
ven
but n
o ta
ble
for
com
paris
on; l
ow a
ntib
iotic
re
sist
ance
rate
; not
ge
nera
lizab
le; n
o co
mm
ent
on e
ffec
t of b
lood
cu
lture
/mic
robi
olog
ic re
sults
on
mor
talit
y or
leng
th o
f st
ay, o
r cha
nge
in a
ntib
iotic
s
II fo
r bl
ood
cultu
re
yiel
d
Ram
anuj
am
and
Rat
hlev
42
2006
R
etro
spec
tive
obse
rvat
iona
l; si
ngle
hos
pita
l
Patie
nts a
dmitt
ed fr
om E
D
with
dia
gnos
is o
f CA
P in
w
hich
blo
od c
ultu
res w
ere
draw
n be
fore
ant
ibio
tics;
in
clud
ed IC
U p
atie
nts,
excl
uded
im
mun
osup
pres
sed,
re
cent
ly h
ospi
taliz
ed a
nd
nurs
ing
hom
e pa
tient
s;
all p
atie
nts w
ere
treat
ed
with
eith
er
ceftr
iaxo
ne+a
zith
rom
ycin
or
levo
floxa
cin
Num
ber o
f pos
itive
blo
od
cultu
res a
nd c
hang
es in
an
tibio
tics d
ue to
blo
od c
ultu
re
resu
lts; r
ecov
ery
of re
sist
ant
orga
nism
s and
if e
mpi
ric
antib
iotic
s are
suff
icie
nt fo
r pa
tient
s with
CA
P
532
ED p
atie
nts
hosp
italiz
ed w
ith C
AP;
28
9 pa
tient
s enr
olle
d;13
(4
.5%
) pat
ient
s had
true
- po
sitiv
e bl
ood
cultu
res,
13
had
fals
e-po
sitiv
e bl
ood
cultu
res;
org
anis
ms
isol
ated
wer
e se
nsiti
ve to
em
piric
ant
ibio
tics i
n al
l ca
ses;
no
patie
nt re
quire
d an
ant
ibio
tic c
hang
e du
e to
re
sist
ance
; 4 p
atie
nts h
ad
chan
ge in
ant
ibio
tics d
ue to
de
terio
ratio
n of
clin
ical
st
atus
Ret
rosp
ectiv
e de
sign
; sm
all s
tudy
pop
ulat
ion;
m
any
CA
P pa
tient
s did
not
ha
ve a
blo
od c
ultu
re —
po
ssib
le se
lect
ion
bias
; sm
all n
umbe
r of p
ositi
ve
bloo
d cu
lture
s with
no
resi
stan
t org
anis
ms,
diff
icul
t to
say
whe
ther
em
piric
an
tibio
tics a
re a
lway
s ap
prop
riate
III
Clinical Policy
722 Annals of
Emergency Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Soca
n et
al43
19
99
Pros
pect
ive;
1
hosp
ital i
n Sl
oven
ia
Adu
lt pa
tient
s >15
y o
f age
ad
mitt
ed w
ith p
neum
onia
(in
clud
ed n
ursi
ng h
ome
patie
nts)
to d
eter
min
e th
e m
icro
bial
etio
logy
of
pneu
mon
ia in
adu
lt pa
tient
s
211
patie
nts,
195
had
bloo
d cu
lture
s with
blo
od c
ultu
res
posi
tive
in 2
3 (1
2%);
em
piric
ther
apy
chan
ged
beca
use
of b
lood
cul
ture
resu
lts
in 2
(1%
of a
ll bl
ood
cultu
res
or 9
% o
f pos
itive
blo
od
cultu
res)
pat
ient
s
Blo
od c
ultu
re re
sults
do
not o
ften
lead
to c
hang
e in
th
erap
y in
this
setti
ng
Tota
l num
ber o
f pat
ient
s ho
spita
lized
for p
neum
onia
an
d nu
mbe
r not
enr
olle
d no
t re
porte
d; p
oten
tial s
elec
tion
bias
; unu
sual
ly lo
w ra
te o
f pn
eum
ococ
cal p
neum
onia
5.
7%, a
nd lo
w ra
te o
f an
tibio
tic re
sist
ance
; lim
its
gene
raliz
abili
ty; o
ne th
ird o
f pa
tient
s wer
e ta
king
an
tibio
tic b
efor
e ad
mis
sion
to
hos
pita
l
III
Woo
dhea
d et
al44
1991
Pr
ospe
ctiv
e;
2 B
ritis
h ho
spita
ls
How
mic
robi
olog
ical
in
vest
igat
ions
are
use
d in
an
uns
elec
ted
grou
p of
ad
ult p
atie
nts w
ith C
AP,
an
d ev
alua
te th
e us
eful
ness
of
the
resu
lts o
btai
ned
in
chan
ging
ant
ibio
tic
regi
men
; con
secu
tive
adul
ts
adm
itted
with
CA
P;
patie
nts i
dent
ified
pr
ospe
ctiv
ely,
cha
rts
revi
ewed
retro
spec
tivel
y;
excl
uded
: pat
ient
s ad
mitt
ed to
ger
iatri
c w
ard,
co
mm
unic
able
dis
ease
uni
t, m
alig
nanc
y,
imm
unos
uppr
essi
on
Cha
nge
in a
ntib
iotic
ther
apy
due
to m
icro
biol
ogic
al
iden
tific
atio
n of
pat
hoge
ns;
antib
iotic
cha
nges
occ
urre
d in
: 33
(31%
) pat
ient
s tot
al;
13/2
8 (4
6%) o
f pat
ient
s with
pa
thog
en id
entif
ied
(by
any
met
hod)
; 20/
78 (2
6%) o
f pa
tient
s with
out p
atho
gen
iden
tifie
d; 9
(8%
) pat
ient
s had
ch
ange
bec
ause
of r
esul
ts o
f m
icro
biol
ogic
al te
sts;
18
(17%
) ha
d ch
ange
bec
ause
of c
linic
al
cond
ition
122
patie
nts i
dent
ified
, 106
in
clud
ed; 2
8 (2
6%) h
ad
caus
ativ
e pa
thog
en
iden
tifie
d; 8
6 (8
1%) h
ad
bloo
d cu
lture
don
e, 9
(1
0%) p
ositi
ve; 4
(4%
) had
ch
ange
bec
ause
of b
lood
cu
lture
resu
lts; 2
(2%
) had
co
vera
ge b
road
ened
be
caus
e of
blo
od c
ultu
re
resu
lts; b
lood
cul
ture
s are
in
freq
uent
ly p
ositi
ve a
nd
rare
ly c
hang
e m
anag
emen
t
No
info
rmat
ion/
repo
rting
on
antib
iotic
resi
stan
ce,
ther
efor
e un
sure
whe
ther
st
udy
has e
xter
nal v
alid
ity;
olde
r dat
a fr
om B
ritai
n,
limits
gen
eral
izab
ility
; ab
solu
te n
umbe
r of p
atie
nts
with
ant
ibio
tic c
hang
es is
lo
w, d
iffic
ult t
o m
ake
conc
lusi
ons b
ased
on
33
patie
nts
III
Clinical Policy
Volume , . : No
vember Annals of Emergency Medicine 723Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Sany
al e
t al
45
1999
R
etro
spec
tive;
si
ngle
hos
pita
l
Rev
iew
of a
ll ad
ult p
atie
nts
with
CA
P di
scha
rged
in
1996
trea
ted
by 1
993
ATS
gu
idel
ines
to d
eter
min
e w
heth
er re
sults
of
mic
robi
olog
ic st
udie
s led
to
chan
ge in
ant
ibio
tics i
n pa
tient
s who
fail
to re
spon
d to
initi
al a
ntib
iotic
s (n
onre
spon
ders
); co
mpa
red
patie
nts w
ith se
vere
and
no
nsev
ere
CA
P
184
patie
nts,
94.6
% (1
74) h
ad
bloo
d cu
lture
s, 11
% (1
9/17
4)
bloo
d cu
lture
s pos
itive
; 116
ha
d sp
utum
ana
lysi
s, 34
%
(40/
116)
pos
itive
; no
diff
eren
ce in
rate
of p
ositi
ve
bloo
d cu
lture
bet
wee
n se
vere
C
AP
and
nons
ever
e C
AP
(11%
fo
r eac
h); 1
4% (2
5/18
4) d
id
not r
espo
nd to
initi
al
antib
iotic
s; 6
non
seve
re C
AP,
no
ne h
ad p
ositi
ve b
lood
cu
lture
, cha
nges
in a
ntib
iotic
s m
ade
empi
rical
ly; 1
9 se
vere
C
AP:
4 d
ied
<72
h, 1
3 ha
d po
sitiv
e m
icro
biol
ogic
stud
ies,
1 ha
d an
tibio
tic c
hang
e ba
sed
on b
lood
cul
ture
(gre
w
MR
SA);
11 p
atie
nts h
ad
mic
robi
olog
ic st
udie
s sen
sitiv
e to
initi
al a
ntib
iotic
s, bu
t an
tibio
tics w
ere
chan
ged
empi
rical
ly b
ecau
se o
f clin
ical
de
terio
ratio
n; p
atie
nts w
ith
bact
erem
ia h
ad g
reat
er
mor
talit
y th
an n
onba
cter
emic
pa
tient
s (21
% v
s 6.5
%,
P<0.
05)
Blo
od c
ultu
re c
hang
ed
man
agem
ent i
n 1
patie
nt,
0.5%
(1/1
74) o
f all
bloo
d cu
lture
s or 5
% (1
/19)
of
posi
tive
bloo
d cu
lture
s;
antib
iotic
s cha
nged
em
piric
ally
mor
e fr
eque
ntly
than
per
resu
lts
of m
icro
biol
ogic
stud
ies
(85%
vs 1
5%; n
o P
valu
e re
porte
d); i
n no
nres
pond
ers
ther
e w
as n
o di
ffer
ence
in
mor
talit
y be
twee
n th
ose
in
who
m a
ntib
iotic
s wer
e ch
ange
d em
piric
ally
and
th
ose
with
mic
robi
olog
ic
stud
y-gu
ided
cha
nges
Diff
icul
t to
com
e to
co
nclu
sion
abo
ut
nonr
espo
nder
s bec
ause
ac
tual
num
ber o
f no
nres
pond
ers (
25) i
s low
; re
trosp
ectiv
e de
sign
; pa
tient
s ide
ntifi
ed b
y di
scha
rge
diag
nosi
s;
low
leve
l of a
ntib
iotic
re
sist
ance
; all
S pn
eum
onia
e is
olat
ed b
y bl
ood
cultu
res
wer
e su
scep
tible
to
peni
cilli
n; a
ll pa
tient
s for
w
hom
mic
robi
olog
ic st
udie
s ch
ange
d m
anag
emen
t cam
e fr
om lo
ng-te
rm c
are
faci
lity
II fo
r bl
ood
cultu
re
yiel
d
Clinical Policy
724 Annals of Emerge
ncy Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Wat
erer
et
al46
19
99
Ret
rosp
ectiv
e;
1 ho
spita
l in
Mem
phis
To d
eter
min
e ho
w o
ften
phys
icia
ns c
hang
e m
anag
emen
t bas
ed o
n bl
ood
cultu
re re
sults
po
sitiv
e fo
r S p
neum
onia
; in
clud
ed p
atie
nts a
dmitt
ed
with
dia
gnos
is o
f CA
P,
bloo
d cu
lture
dra
wn
befo
re
antib
iotic
s and
at l
east
1
posi
tive
bloo
d cu
lture
for S
pn
eum
onia
; re
trosp
ectiv
e ch
art r
evie
w p
erfo
rmed
1,80
5 pa
tient
s with
CA
P;
118
patie
nts w
ith p
ositi
ve
bloo
d cu
lture
for S
pne
umon
ia;
105
char
ts a
vaila
ble;
74
patie
nts w
ith C
AP
and
bloo
d cu
lture
pos
itive
for S
pn
eum
onia
e in
clud
ed in
stud
y;
15 is
olat
es w
ere
peni
cilli
n re
sist
ant,
4 “h
igh
grad
e” (o
nly
one
with
MIC
=4);
4 is
olat
es
wer
e ce
phal
ospo
rin re
sist
ant,
1 hi
gh g
rade
; 51
patie
nts w
ithou
t pe
nici
llin
alle
rgy
grew
S
pneu
mon
iae
susc
eptib
le to
pe
nici
llin;
ant
ibio
tics w
ere
chan
ged
to p
enic
illin
in o
nly
11 o
f the
se (2
1.6%
)
Blo
od c
ultu
re c
hang
ed
man
agem
ent i
n 31
(42%
of
posi
tive
bloo
d cu
lture
); an
tibio
tic c
hang
ed in
2
(2.7
%) p
atie
nts b
ecau
se o
f re
sist
ance
; no
corr
elat
ion
betw
een
dise
ase
seve
rity
and
bloo
d cu
lture
po
sitiv
ity; p
hysi
cian
s ofte
n do
not
nar
row
ther
apy
as
indi
cate
d by
blo
od c
ultu
re
resu
lts
Ret
rosp
ectiv
e de
sign
; se
lect
pop
ulat
ion—
stud
y lo
oks a
t adm
itted
CA
P pa
tient
s with
blo
od c
ultu
res
posi
tive
for S
pne
umon
iae
only
; res
ista
nce
rate
low
and
le
vel o
f res
ista
nce
low
co
mpa
red
with
200
7 ra
tes o
f re
sist
ance
; aut
hors
do
not
repo
rt ho
w m
any
patie
nts
with
CA
P ha
d bl
ood
cultu
res d
one;
ther
efor
e ca
nnot
cal
cula
te b
lood
cu
lture
yie
ld; f
urth
erm
ore,
ca
nnot
cal
cula
te th
e ov
eral
l ut
ility
of b
lood
cul
ture
(of
the
tota
l num
ber o
f blo
od
cultu
res d
one,
wha
t pe
rcen
tage
led
to a
cha
nge
in m
anag
emen
t?)
III
Ded
ier e
t al
47
2001
R
etro
spec
tive
char
t rev
iew
; m
ultic
ente
r; 38
Uni
ted
Stat
es a
cade
mic
ho
spita
ls
CA
P pa
tient
s stu
died
to
dete
rmin
e re
latio
nshi
p be
twee
n pr
ompt
ac
hiev
emen
t of p
roce
ss o
f ca
re m
arke
rs (b
lood
cul
ture
w
ithin
24
hour
s of a
dmit,
bl
ood
cultu
re b
efor
e an
tibio
tic, a
ntib
iotic
with
in
8 h
of h
ospi
tal a
rriv
al,
oxyg
enat
ion
mea
sure
men
t w
ithin
24
h) a
nd o
utco
mes
(r
each
ing
clin
ical
stab
ility
w
ithin
48
h of
hos
pita
l ad
mis
sion
, dec
reas
ed
leng
th o
f sta
y an
d in
patie
nt
deat
hs)
1,45
7 pa
tient
s, 1,
062
elig
ible
; 89
% a
dmitt
ed th
roug
h ED
; 76
.2%
had
ant
ibio
tics w
ithin
8
h; 8
2.5%
blo
od c
ultu
res b
y 24
h;
72.
3% h
ad b
lood
cul
ture
s be
fore
ant
ibio
tics;
94.
5% h
ad
oxyg
en m
easu
red
by 2
4 h;
in
crea
sed
seve
rity
of il
lnes
s w
as a
ssoc
iate
d w
ith b
lood
cu
lture
per
form
ance
(P=0
.009
) an
d sh
orte
r tim
e to
ant
ibio
tics
(P=0
.04)
No
impr
ovem
ent i
n de
ath,
le
ngth
of s
tay
for p
atie
nts
with
blo
od c
ultu
re b
efor
e
antib
iotic
s or p
atie
nts w
ith
bloo
d cu
lture
s with
in 2
4 h;
no
con
sist
ent r
elat
ions
hip
betw
een
proc
ess-
of-c
are
mar
ker a
chie
vem
ent a
nd
impr
ovem
ent i
n th
e cl
inic
al
outc
omes
Ret
rosp
ectiv
e de
sign
; pa
tient
s ide
ntifi
ed b
y di
scha
rge
diag
nosi
s;
sele
ctio
n bi
as; m
edia
n nu
mbe
r of p
atie
nts f
rom
ea
ch h
ospi
tal 2
8, w
hich
se
ems l
ow; l
arge
num
ber o
f pa
tient
s exc
lude
d; h
igh
num
ber o
f low
-ris
k pa
tient
s in
the
stud
y po
pula
tion
(29%
PSI
I-II
); da
ta n
ot
give
n ex
plic
itly
for P
SI IV
-V
pat
ient
s; n
o pr
open
sity
m
atch
ing
perf
orm
ed d
espi
te
low
rate
of o
utco
me
II fo
r bl
ood
cultu
re;
III f
or
anti-
biot
ics
Clinical Policy
Volume ,
. : November Annals of Emergency Medicine 725Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Mee
han
et
al48
1997
R
etro
spec
tive;
m
ultic
ente
r na
tiona
l stu
dy
from
Med
icar
e cl
aim
s dat
abas
e
Rev
iew
of c
laim
s dat
a fr
om
Med
icar
e na
tiona
l cla
ims
hist
ory
file
and
patie
nt
char
ts to
ass
ess q
ualit
y of
ca
re fo
r Med
icar
e pa
tient
s ho
spita
lized
with
pn
eum
onia
and
to
dete
rmin
e w
heth
er p
roce
ss-
of-c
are
perf
orm
ance
is
asso
ciat
ed w
ith lo
wer
30-
day
mor
talit
y; 4
pro
cess
es
of c
are
inve
stig
ated
: bl
ood
cultu
res b
efor
e
antib
iotic
s, bl
ood
cultu
res
with
in 2
4 h,
tim
e to
an
tibio
tic a
dmin
istra
tion,
an
d ox
ygen
atio
n as
sess
men
t with
in 2
4 h
500
pote
ntia
l cas
es w
ere
sele
cted
rand
omly
from
eac
h st
ate,
DC
, and
Pue
rto R
ico;
26
,000
pot
entia
l cas
es, 1
4,06
9 ag
greg
ate
stud
y se
t;2,
500
subs
et o
f sam
pled
cas
es
crea
ted,
exc
lusi
on c
riter
ia
appl
ied
to c
reat
e 1,
343
natio
nal
stud
y se
t; m
ean
age
79.4
y;
23.4
% fr
om n
ursi
ng h
omes
; 58
.2%
had
at l
east
1
com
orbi
dity
; inh
ospi
tal
mor
talit
y 10
.3%
; 30-
day
mor
talit
y 15
.3%
; bl
ood
cultu
re c
olle
ctio
n w
ithin
24
h o
f adm
issi
on a
ssoc
iate
d w
ith lo
wer
30-
day
mor
talit
y:
OR
0.9
(95%
CI 0
.81-
1.0)
, P=
0.07
; blo
od c
ultu
re
colle
ctio
n be
fore
ant
ibio
tic
adm
inis
tratio
n w
as n
ot
sign
ifica
ntly
ass
ocia
ted
with
hi
gher
or l
ower
mor
talit
y O
R
0.92
(95%
CI 0
.82-
1.2)
, P=
0.10
Adm
inis
terin
g an
tibio
tics
with
in 8
h o
f hos
pita
l ar
rival
and
col
lect
ing
bloo
d cu
lture
s with
in 2
4 ho
urs
wer
e as
soci
ated
with
im
prov
ed su
rviv
al
Ret
rosp
ectiv
e re
view
of
clai
ms d
ata;
pot
entia
l se
lect
ion
bias
; stu
dy
popu
latio
n ol
der,
ofte
n fr
om
nurs
ing
hom
e, o
ften
with
co
mor
bidi
ties —
pat
ient
s m
ore
likel
y to
hav
e bl
ood
cultu
res a
nyw
ay; K
appa
for
abst
ract
ors a
s low
as 0
.48
for r
ecen
t che
mot
hera
py,
0.52
for m
enta
l sta
tus;
K
appa
for
blo
od c
ultu
re
0.83
with
in 2
4 h;
stud
y po
pula
tion
olde
r and
sick
er
than
gen
eral
ED
pat
ient
s;
conc
lusi
on th
at b
lood
cu
lture
don
e w
ithin
24
h is
as
soci
ated
with
redu
ced
mor
talit
y co
mes
from
dat
a w
ith P
=0.0
7, C
I inc
lude
s 1;
stat
istic
ally
sign
ifica
nt?
III
Clinical Policy
726 Annals of Emergency M
edicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Luja
n et
al50
20
04
Pros
pect
ive
obse
rvat
iona
l; 1
hosp
ital i
n B
arce
lona
Patie
nts a
ge >
18 y
ho
spita
lized
with
CA
P w
ith
bloo
d cu
lture
pos
itive
for S
pn
eum
onia
e to
eva
luat
e th
e ef
fect
of d
isco
rdan
t em
piric
al th
erap
y on
ou
tcom
e in
bac
tere
mic
pn
eum
ococ
cal C
AP;
ou
tcom
es e
xam
ined
in
clud
ed 2
8-da
y m
orta
lity,
us
e of
vas
oact
ive
med
icat
ions
, and
su
ppur
ativ
e co
mpl
icat
ions
100
cons
ecut
ive
patie
nts,
29
pneu
moc
occa
l iso
late
s sho
wed
so
me
resi
stan
ce to
pen
icill
in:
17 in
term
edia
te m
inim
um
inhi
bito
ry c
once
ntra
tions
(0
.12-
1 µg
/mL)
, 12
high
m
inim
um in
hibi
tory
co
ncen
tratio
ns (>
2 µg
/mL)
; 18
non
susc
eptib
le to
m
acro
lides
, 2 n
onsu
scep
tible
to
ceph
alos
porin
, 27
patie
nts
imm
unoc
ompr
omis
ed;
10 p
atie
nts h
ad d
isco
rdan
t th
erap
y, 5
0% (5
/10)
pat
ient
s w
ith d
isco
rdan
t the
rapy
die
d co
mpa
red
with
13/
90 (1
4%)
who
had
con
cord
ant t
hera
py;
estim
ated
exc
ess m
orta
lity
for
initi
al d
isco
rdan
t the
rapy
was
35
.6%
(95%
CI 3
.73-
67.4
); on
ly 3
of 9
pat
ient
s stil
l aliv
e w
hen
bloo
d cu
lture
de
mon
stra
ted
disc
orda
nt
ther
apy
actu
ally
had
ther
apy
chan
ged
to a
ppro
pria
te th
erap
y
Sign
ifica
nt a
ssoc
iatio
n be
twee
n di
scor
dant
ther
apy
and
high
er m
oral
ity in
ba
cter
emic
pat
ient
s with
pn
eum
ococ
cal C
AP;
nu
rsin
g ho
me
resi
denc
e an
d im
mun
ocom
prom
ised
pa
tient
s wer
e si
gnifi
cant
ly
asso
ciat
ed w
ith p
enic
illin
an
d m
acro
lide
resi
stan
ce
Dis
cord
ant p
ool i
nclu
ded
patie
nts w
ith in
term
edia
te
resi
stan
ce p
ossi
ble
skew
ing
resu
lts to
show
dis
cord
ant
ther
apy
caus
es le
ss h
arm
; ve
ry sm
all n
umbe
r of
patie
nts w
ith d
isco
rdan
t th
erap
y (1
0) le
ads t
o ve
ry
wid
e C
I; sp
ecifi
c gr
oup
of
patie
nts —
blo
od c
ultu
re
posi
tive
for S
pne
umon
ia;
6 pa
tient
s rec
eivi
ng
disc
orda
nt th
erap
y tre
ated
w
ith a
mox
icill
in-c
lavu
lana
te
as th
e in
itial
em
piric
an
tibio
tic, i
nclu
ding
2 w
ho
wer
e PS
I V; n
ot ty
pica
l of
empi
ric tr
eatm
ent f
or
hosp
italiz
ed p
atie
nts i
n th
e U
nite
d St
ates
; inc
lude
d im
mun
ocom
prom
ised
pa
tient
s — p
ossi
bly
bias
ing
to h
ighe
r mor
talit
y
III
Clinical Policy
Volume , . : Novemb
er Annals of Emergency Medicine 727Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Moi
ne e
t al
5119
94
Pros
pect
ive
obse
rvat
iona
l; m
ultic
ente
r; 15
Fr
ench
cen
ters
Con
secu
tive
patie
nts
hosp
italiz
ed w
ith se
vere
C
AP
in th
e IC
U to
de
term
ine
caus
ativ
e ag
ents
, th
e va
lue
of c
linic
al,
biol
ogic
, and
radi
olog
ic
feat
ures
in p
redi
ctin
g th
e et
iolo
gy, a
nd to
def
ine
prog
nost
ic fa
ctor
s in
patie
nts w
ith se
vere
CA
P
157
CA
P pa
tient
s, 25
ex
clud
ed; 1
32 st
udy
patie
nts:
98
mal
e, 3
4 fe
mal
e;
46 h
ad a
ntib
iotic
s bef
ore
ad
mis
sion
; 127
had
blo
od
cultu
res,
34 (2
7%) p
ositi
ve
bloo
d cu
lture
s, 22
S
pneu
mon
iae,
4 S
trep
toco
ccus
spec
ies,
1 Es
cher
ichi
a co
li, 5
K
lebs
iella
; 31
patie
nts h
ad
ther
apy
mod
ified
bas
ed o
n ba
cter
iolo
gic
resu
lts, 1
6 pa
tient
s with
uns
ucce
ssfu
l tre
atm
ent r
espo
nse
had
ther
apy
mod
ified
bas
ed o
n ba
cter
iolo
gic
resu
lts
(cha
nges
due
to b
lood
cul
ture
in
par
ticul
ar n
ot re
porte
d)
27%
bac
tere
mia
in th
is
popu
latio
n of
pat
ient
s with
se
vere
CA
P; b
lood
cul
ture
yi
eld
high
er in
sick
er
patie
nts;
bac
tere
mia
si
gnifi
cant
ly a
ssoc
iate
d w
ith d
eath
in th
is
popu
latio
n; d
eter
min
ing
the
etio
logy
did
not
im
prov
e su
rviv
al;
15/3
4 pa
tient
s with
po
sitiv
e bl
ood
cultu
re d
ied;
ba
cter
emia
sign
ifica
ntly
as
soci
ated
with
dea
th
(P=0
.004
)
Extre
me
mal
e pr
edom
inan
ce
98:3
4; a
lthou
gh 3
1 pa
tient
s ha
d th
erap
y m
odifi
ed b
ased
on
bac
terio
logi
c fin
ding
s, it
was
not
repo
rted
in h
ow
man
y bl
ood
cultu
re
spec
ifica
lly c
hang
ed
man
agem
ent
II
Hou
ck e
t al
5320
04
Mul
ticen
ter
retro
spec
tive
coho
rt
Enro
lled
18,2
09 p
atie
nts,
4,43
8 pa
tient
s exc
lude
d fo
r pr
etre
atm
ent a
ntib
iotic
s;
char
t rev
iew
of 1
3,77
1 pa
tient
s ≥6
5 y
with
ICD
-9
code
of p
neum
onia
from
m
ore
than
3,5
00 h
ospi
tals
w
ho d
id n
ot re
ceiv
e an
tibio
tics b
efor
e ar
rival
at
hosp
ital;
patie
nts g
athe
red
durin
g a
1-y
per
iod
base
d on
cla
ims d
ata
Inho
spita
l mor
talit
y, 3
0-da
y m
orta
lity,
and
leng
th o
f sta
y >
5 da
ys; a
s ass
ocia
ted
with
an
tibio
tic a
dmin
istra
tion
befo
re
or a
fter 4
h fr
om a
rriv
al
Afte
r per
form
ance
of
mul
tivar
iate
logi
stic
re
gres
sion
, ant
ibio
tic
adm
inis
tratio
n w
ithin
4 h
w
hen
com
pare
d to
>4
h yi
elde
d an
adj
uste
d O
R o
f 0.
85 (9
5% C
I 0.7
4-0.
98)
for i
nhos
pita
l mor
talit
y, a
n ad
just
ed O
R o
f 0.8
5 (9
5%
CI 0
.76-
0.95
) for
30-
day
mor
talit
y, a
nd a
n ad
just
ed
OR
of
0.9
(95%
CI 0
.83-
0.96
) for
leng
th o
f sta
y
grea
ter t
han
5 da
ys
4-h
cuto
ff w
as d
eter
min
ed
post
hoc
; 3- t
o 8-
h cu
toff
s ha
d ne
ar id
entic
al 3
0-da
y m
orta
lity
asso
ciat
ions
; th
ough
incl
uded
in th
e m
ultiv
aria
te a
naly
sis,
mor
e pa
tient
s in
the
antib
iotic
s <4
h gr
oup
rece
ived
ant
ibio
tic
regi
men
s dee
med
ap
prop
riate
; did
not
ana
lyze
fo
r alte
red
men
tal s
tatu
s;
enro
llmen
t bas
ed o
n cl
aim
s da
ta a
nd e
qual
num
bers
sa
mpl
ed p
er st
ate,
not
bas
ed
on st
ate
popu
latio
n; d
id n
ot
anal
yze
by in
divi
dual
ho
spita
l; ho
spita
ls th
at
diag
nose
mor
e ef
ficie
ntly
m
ay b
e as
soci
ated
with
be
tter o
vera
ll ca
re
III
Clinical Policy
728 Anna
ls of Emergency Medicine Volume , . : NovemberEvi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Wat
erer
et
al54
20
06
Sing
le c
ente
r pr
ospe
ctiv
e co
hort
451
patie
nts s
plit
into
an
tibio
tic b
efor
e or
afte
r 4-
h gr
oups
; abo
ut 5
0% o
f pa
tient
s in
each
coh
ort
Ant
ibio
tics b
efor
e or
afte
r 4 h
as
ass
ocia
ted
with
mor
talit
y;
also
look
ed a
t ass
ocia
tions
with
se
verit
y, se
ptic
shoc
k, h
ypox
ia,
and
decr
ease
d m
enta
l sta
tus
On
univ
aria
te a
naly
sis,
antib
iotic
s >4
h a
fter
arriv
al w
as a
ssoc
iate
d w
ith
incr
ease
d ris
k of
dea
th, b
ut
whe
n m
ultiv
aria
te a
naly
sis
perf
orm
ed, n
o st
atis
tical
ly
sign
ifica
nt in
crea
sed
risk
of d
eath
bas
ed o
n an
tibio
tic
time;
alte
red
men
tal s
tate
as
soci
ated
with
an
adju
sted
O
R 3
.33
(95%
CI 1
.28-
8.77
) and
abs
ence
of f
ever
w
as a
ssoc
iate
d w
ith
adju
sted
OR
2.5
5 (9
5% C
I 1.
02-6
.37)
for m
orta
lity
Sing
le c
ente
r; sm
all n
umbe
r of
mor
talit
ies i
n ag
e >6
5 y
popu
latio
n; n
o m
entio
n is
m
ade
abou
t whe
ther
any
pa
tient
s rec
eive
d ou
t-of-
hosp
ital a
ntib
iotic
s
II
Silb
er e
t al55
20
03
Pros
pect
ive
obse
rvat
iona
l co
hort
409
patie
nts >
21 y
(tho
ugh
mos
t >65
y) w
ith m
oder
ate
to se
vere
pne
umon
ia (b
ased
on
PO
RT
scor
e) w
ere
plac
ed in
to 3
gro
ups b
ased
on
thei
r tim
e fr
om a
rriv
al to
an
tibio
tics (
grou
p 1
rece
ived
ant
ibio
tics i
n <4
h,
grou
p 2
from
4 to
8 h
, gr
oup
3 in
>8
h)
Tim
e to
clin
ical
stab
ility
— a
co
mpo
site
mea
sure
of t
he fi
rst
24 h
per
iod
that
the
patie
nt h
as
all o
f the
follo
win
g: S
BP
≥90
mm
Hg,
pul
se ra
te ≤
100
beat
s/m
in, r
espi
rato
ry ra
te ≤
24
brea
ths/
min
, tem
pera
ture
≤1
01°F
, O2 S
at ≥
90, a
nd th
e ab
ility
to e
at
No
stat
istic
ally
sign
ifica
nt
diff
eren
ces b
etw
een
the
grou
ps in
tim
e to
clin
ical
st
abili
ty
Excl
uded
pat
ient
s who
re
ceiv
ed in
appr
opria
te
antib
iotic
s; e
xclu
ded
patie
nts w
ho n
ever
reac
hed
clin
ical
stab
ility
; mod
erat
e sa
mpl
e si
ze m
ay h
ave
mis
sed
diff
eren
ces
II
Mar
rie a
nd
Wu56
20
05
Mul
ticen
ter,
pros
pect
ive
obse
rvat
iona
l tri
al
3,04
3 pa
tient
s, m
ean
age
70
y; e
xclu
ded
patie
nts:
ad
mitt
ed to
the
ICU
from
th
e ED
, asp
iratio
n pn
eum
oniti
s (1s
t y o
nly)
, tu
berc
ulos
is, c
ystic
fibr
osis
, pr
egna
nt, o
r tak
ing
im
mun
osup
pres
sive
dr
ugs/
CD
4 <2
50
Impl
emen
ted
a ca
re p
athw
ay;
track
ed m
any
inte
rven
tions
and
pr
ogno
stic
fact
ors i
nclu
ding
an
tibio
tics b
efor
e or
afte
r 4 h
No
sign
ifica
nt d
iffer
ence
in
mor
talit
y w
ith a
4 o
r 8 h
cu
toff
Onl
y pe
rfor
med
uni
varia
te
anal
ysis
on
the
time
to
antib
iotic
and
mor
talit
y as
soci
atio
ns; l
ack
of
mul
tivar
iate
ana
lysi
s of
conf
ound
ing
fact
ors
decr
ease
s clin
ical
util
ity o
f th
ese
resu
lts
II
Clinical Policy
Volume , .
: November Annals of Emergency Medicine 729Evi
dent
iary
Tab
le (c
ontin
ued)
. St
udy
Yea
r D
esig
n In
terv
entio
n(s)
/Tes
t(s)
/M
odal
ity
Out
com
e M
easu
re/C
rite
rion
St
anda
rd
Res
ults
L
imita
tions
/Com
men
ts
Cla
ss
Bat
tlem
an e
t al
5720
02
Mul
ticen
ter
retro
spec
tive
coho
rt
609
patie
nts f
rom
7
hosp
itals
with
dia
gnos
is o
f pn
eum
onia
bas
ed o
n D
RG
co
ding
Prol
onge
d le
ngth
of s
tay
(d
efin
ed a
s ≥9
days
) as
asso
ciat
ed w
ith d
oor-
to-n
eedl
e tim
e an
d w
heth
er a
ntib
iotic
s w
ere
adm
inis
tere
d in
ED
or o
n flo
or
Dec
reas
ed n
umbe
r of
patie
nts w
ith p
rolo
nged
le
ngth
of s
tay
asso
ciat
ed
with
shor
ter d
oor-
to-n
eedl
e tim
es a
nd a
ntib
iotic
s ad
min
iste
red
in th
e ED
Excl
uded
mor
talit
ies f
rom
an
alys
is; d
ata
not s
how
n fo
r an
alys
is o
f doo
r-to
-nee
dle
time
III
ATS,
Am
eric
an T
hora
cic
Soci
ety;
BP,
blo
od p
ress
ure;
CAP
, com
mun
ity-a
cqui
red
pneu
mon
ia; C
I, co
nfid
ence
inte
rval
; CM
S, C
ente
rs fo
r Med
icar
e an
d M
edic
aid
Se
rvic
es; D
RG, D
iagn
osis
-Rel
ated
Gro
up; E
D, e
mer
genc
y de
partm
ent;
H, H
aem
ophi
lus;
h, h
our;
ICD
-9, I
nter
natio
nal C
lass
ifica
tion
of D
isea
ses,
Nin
th R
evis
ion;
IC
U, i
nten
sive
car
e un
it; M
IC, m
inim
um in
hibi
tory
con
cent
ratio
ns; M
RSA,
met
hici
llin-
resi
stan
t Sta
phyl
ococ
cus a
ureu
s; M
SSA,
met
hici
llin-
susc
eptib
le
Stap
hylo
cocc
us a
ureu
s; m
in, m
inut
e; O
2, ox
ygen
; OR,
odd
s rat
io; P
CA,
pne
umoc
occa
l ant
igen
; PO
RT, P
atie
nt O
utco
mes
Res
earc
h Te
am; P
RSP,
pen
icill
in
resi
stan
t Str
epto
cocc
us p
neum
onia
e; P
SI, p
neum
onia
seve
rity
inde
x; Q
I, qu
ality
impr
ovem
ent;
S,st
rept
ococ
cus;
SAP
S, si
mpl
ified
acu
te p
hysi
olog
y sc
ore;
Sat
, sa
tura
tion;
SBP
, sys
tolic
blo
od p
ress
ure;
vs,
vers
us; y
, yea
r; W
BC, w
hite
blo
od c
ell c
ount
.
Clinical Policy
730 Annals of Emergency Medicine
Volume , . : NovemberClinical Policy
Volume , . : November
Diagnosis‡
Prognosis§
cohort using a criterion standard Population prospective cohort
ve observational Retrospective cohortCase control
tonsensus, review)
Case seriesCase reportOther (eg, consensus, review)
lly.
Appendix A. Literature classification schema.*
Design/Class Therapy†
1 Randomized, controlled trial or meta-analysesof randomized trials
Prospective
2 Nonrandomized trial Retrospecti
3 Case seriesCase reportOther (eg, consensus, review)
Case seriesCase reporOther (eg, c
*Some designs (eg, surveys) will not fit this schema and should be assessed individua†Objective is to measure therapeutic efficacy comparing �2 interventions.‡Objective is to determine the sensitivity and specificity of diagnostic tests.§Objective is to predict outcome including mortality and morbidity.
Appendix B. Approach to downgrading strength of evidence.
Downgrading
Design/Class
1 2 3
None I II III1 level II III X2 levels III X XFatally flawed X X X
Annals of Emergency Medicine 731