Impact of Hospital Formularies on Fluoroquinolone Prescribing in Emergency Departments

May 1, 2007
Sherrie L. Aspinall, PharmD, MSc

,
Joshua P. Metlay, MD, PhD

,
Judith H. Maselli, MSPH

,
Ralph Gonzales, MD, MSPH

Volume 13, Issue 5

Objective: To examine factors associated withfluoroquinolone prescribing among adults receivingantibiotics for acute respiratory tract infections(ARIs) in emergency departments.

Study Design: Cross-sectional.

International Classification of Diseases, Ninth

Revision

Methods:We analyzed data from 8 Departmentof Veterans Affairs medical centers and7 nonfederal US hospitals. At each hospital,we randomly sampled 200 ARI visits withdischarge diagnoses for nonspecificupper respiratory infections, acute bronchitis,pharyngitis, sinusitis, and pneumonia betweenNovember 1, 2003, and February 29, 2004.Patient and provider factors associated witheach visit were extracted from medical records.System characteristics were obtained by surveyingpharmacy directors. Multivariable logisticregression was used to evaluate independentpredictors of fluoroquinolone prescribing.

P

Results: Fluoroquinolones accounted for 14% ofthese prescriptions. At hospitals with at least1 unrestricted fluoroquinolone on formulary(n = 12), the average fluoroquinolone prescriptionrate was 17%, compared with a 6% prescriptionrate at hospitals where fluoroquinolone accesswas restricted by the hospital formulary (n = 3)(< .0001). Factors associated with increasedfluoroquinolone prescription rates were hospitaladmission (odds ratio [OR] = 1.8; 95% confidenceinterval [CI] = 1.1, 3.1) and the diagnoses of acutebronchitis (OR = 2.3; 95% CI = 1.3, 4.2), acuteexacerbations of chronic bronchitis (OR = 2.6;95% CI = 1.2, 5.6), and pneumonia (OR = 6.4;95% CI = 3.3, 12.4). Restricted hospital statuswas associated with decreased fluoroquinolonesaccounted for 14% of the antibiotic prescriptions.

Conclusion: Hospital formulary policies representa potentially important target for influencing outpatientdrug prescribing in emergency departments.

(Am J Manag Care. 2007;13:241-248)

Guidelines state that first-line use of respiratory fluoroquinolones shouldbe restricted to indications for which their therapeutic superiority has beendemonstrated, because of the emergence and spread of fluoroquinoloneresistantbacteria.

  • Respiratory fluoroquinolones are prescribed for many indications thatare commonly nonbacterial such as nonspecific upper respiratory tractinfections and acute bronchitis.
  • Hospitals with 1 or more unrestricted fluoroquinolones on formularyhad significantly higher rates of fluoroquinolone prescribiing in emergencydepartments than hospitals where fluoroquinolone access was restricted.
  • Hospital formulary restrictions can affect patterns of outpatient antibioticprescribing, which then can influence patterns of resistance in thecommunity.

Antibiotic treatment of adults with acute respiratory tractinfections (ARIs) in the outpatient setting has declined duringthe past 10 years, particularly for infections with a likelyviral etiology (ie, nasopharyngitis, acute bronchitis, nonspecificupper respiratory tract infections). Between 1995 and 2001, theproportion of office visits for these conditions that were treated withantibiotics declined from 60% to 49%. Unfortunately, during this sametime period the proportion of prescribed antibiotics that were categorizedas broad spectrum increased from 41% to almost 77%.1

Streptococcus pneumoniae

Pseudomonas aeruginosa

S pneumoniae

Fluoroquinolone antibiotics, which are typically classified as broadspectrum, were the most commonly prescribed class of antibiotics foradult outpatients with any type of infection in 2002.2 Increased use of fluoroquinolonesis a concern because it promotes the emergence and spreadof fluoroquinolone-resistant bacteria, including ,Enterobacteriaceae, and .3-7 In the era of multidrugresistance among species, guidelines emphasize theneed to restrict first-line use of respiratory fluoroquinolones (RFQs) toindications in which therapeutic superiority has been demonstrated,regardless of the setting.8,9 Antimicrobial management programs, includingprior authorization and formulary restrictions, have been promoted asa means of improving antibiotic selection among inpatients, but littledata are available on potential interventions to modify antibiotic selectionin emergency departments (EDs).10 System factors with policy implicationsare of particular interest (eg, exposure to pharmaceuticalmarketing representatives, antibiotics available in the ED, formulary policies),although how these characteristics compare with patient andprovider factors in influencing RFQ prescribing in patients with ARIs isunknown. Once the issues are more clearly identified, evidence-basedinterventions to increase the appropriate use of these antibiotics could beimplemented. Therefore, the purpose of this study was to examine thepatient, provider, and hospital factors associated with RFQ prescribingamong adults receiving antibiotics for ARIs in EDs.

METHODS

Study Design/Setting

This study uses data from the baseline year of the ImprovingAntibiotic Use in Acute CareTreatment (IMPAACT) project. TheIMPAACT project includes a clusterrandomized,controlled trial designedto evaluate the impact of a multidimensional educational intervention on antibiotic treatmentof ARIs in EDs. The project involves 8 Department ofVeterans Affairs (VA) hospitals and 8 nonfederal acutecare hospitals. Details of the hospital sampling strategywere previously described.11 In brief, we surveyed all 135major VA medical centers and all US nonfederal hospitalswith accredited ED residency programs and/or prior participationin a national ED clinical trial network to identify hospitalsthat were willing to participate in an ED-based antibioticquality improvement program. Among all interested hospitals,we sampled pairs of VA and nonfederal hospitals, stratifyingby US region (ie, Northeast, South, Midwest, andWest), and identified 8 pairs in major metropolitan regions. Inthe first year, 1 of the nonfederal hospitals dropped out priorto data collection, leaving 15 eligible hospitals for this study.The patient and provider characteristics of these hospitalswere described previously.12

In the first year of the IMPAACT project, random samplesof 50 ARI visits per month between November 1, 2003, andFebruary 29, 2004 (ie, 200 visits per hospital), were identifiedin the 15 EDs to establish a baseline rate of antibiotic prescribing.Sample sizes were based on power calculations for theIMPAACT project's intervention trial. No interventions wereconducted during this baseline year. This sampling ensuredequal representation of visits throughout the entire respiratoryvirus season at each site. The institutional review boardsat each hospital approved the study prior to initiating datacollection.

Patient Eligibility

International Classification of Diseases, Ninth Revision

ICD-9

ICD-9

ICD-9

All adult (age >18 years) visits were sampled based on the()discharge diagnoses. The ARI visits selected had principalcodes corresponding to a list of the most common ARIdiagnoses, including nonspecific upper respiratory tract infections(URIs), acute bronchitis, pharyngitis, otitis media,sinusitis, and pneumonia (ie, all codes between 460and 491 [including subcodes], 786.2, and 034). We purposelychose a broad list of codes to identify all potential visits ofinterest.

Data Collection

Patient and Provider Factors.

ICD-9

Research coordinatorsfrom each hospital received training on appropriate data collectionmethods. At each hospital, the research coordinatorreviewed the medical records for the selected visits and collectedthe following clinical characteristics: demographics,smoking status, arrival time, discharge time, comorbidities(eg, chronic obstructive pulmonary disease, ischemic heartdisease, diabetes, stroke), vital signs, chief complaint (eg,cough, sore throat, hemoptysis), diagnostic tests (eg, chestradiograph, rapid strep test, blood culture), provider-assigneddiagnosis, antibiotic prescriptions, and disposition (eg, home,hospital, nursing home). Although ARI visits were sampledaccording to the primary discharge diagnosis inadministrative databases, all analyses were conducted usingthe physician-assigned diagnoses as written in the assessmentand plan for each encounter. Provider type for the primaryclinician documenting the encounter (ie, attending physician,resident/intern/fellow, nurse practitioner, physicianassistant) was collected based on documentation in the medicalnotes.

Hospital Factors.

Starting in August 2004, a singleresearch coordinator from 1 of the EDs involved in the studycontacted the pharmacy director at each of the IMPAACTsites to conduct a brief survey of hospital factors that couldinfluence antibiotic selection. The pharmacy director wasinstructed to refer the coordinator to someone else in thepharmacy department if that person was more familiar withantibiotic use at the facility (eg, a clinical pharmacist). Wereviewed a list of 25 antibiotics commonly used for the treatmentof ARIs and asked the respondent to indicate the availabilityof each at the hospital. Response options anddefinitions included (1) on the formulary and unrestricted (ie,available without any prescribing restrictions); (2) on the formularybut restricted access (eg, available for use by the infectiousdisease service or in the treatment of pneumonia); (3)nonformulary (ie, not usually available, but would be obtainedfor an individual patient with appropriate justification); and(4) unavailable (ie, not available and will not be obtained;provider must prescribe another antibiotic). In addition, wecollected information on the presence of published guidelinesfor antibiotics that are available on a restricted or nonformularybasis, the names of antibiotics stocked in the ED,availability of samples in the ED, availability of a hospitalgenerateddrug susceptibility report for common pathogens(ie, an antibiogram), and existence of policies regarding pharmaceuticalsales representatives (ie, access to providers, presenceof industry-sponsored educational programs, presence ofindustry-provided meals in conjunction with programs ormeetings). The surveys were completed prior to analysis ofdata on antibiotic prescription rates. A copy of the surveyinstrument is available on request.

Data Analysis

For this study, our primary analysis focused on identifying factorsassociated with prescribing an RFQ (ie, levofloxacin, gatifloxacin,moxifloxacin) among the ARI visits that involved anantibiotic prescription. Analyses werelimited to these visits to evaluate antibioticselection rather than the decisionof whether or not to prescribe an antibiotic.If the patient was taking antibioticsprior to the visit or if the patientwas immunosuppressed (eg, active cancer,HIV/AIDS), the visit was excludedfrom the study. We also excluded visitswith a primary discharge diagnosis ofotitis media because there were no visitsfor otitis media where an RFQ was prescribedacross all sites.

P

P

We used ?2 tests and Mantel-Haenszel ?2 tests for trend to check fordifferences in RFQ prescribing bypatient, provider, and hospital characteristics.For hospital-level factors, formularystatus was dichotomized asfollows: hospitals with at least 1 RFQavailable on an unrestricted basis wereclassified as having "hospital-unrestricted"RFQs, and those hospitalswith RFQs available only on a restrictedor nonformulary basis were consideredto have "hospital-restricted"RFQs. The categories of restricted andnonformulary were combined becausehospitals had prescribing guidelines inplace regardless of formulary status (ie,formulary restricted vs nonformulary).Although individual RFQs may havebeen unavailable at some of the hospitals,all had at least 1 RFQ available foruse under restricted or nonformularystatus. Therefore, the category of notavailable for an individual RFQ did notapply at the hospital level. All analyseswere performed using the 9th version ofthe SAS statistical software program(SAS Institute, Inc, Cary, NC). Foradjusted analyses, backward stepwiselogistic regression was used to evaluate independent predictorsof RFQ prescribing, and generalized estimating equations(PROC GENMOD) were used to control for clustering of RFQprescriptions by unique provider. All patient-, physician-, andhospital-level factors with the potential to influence antibioticselection were included in the regression model if they wereassociated with RFQ prescribing at < .2 in unadjusted analyses.13 Those that remained significant at < .05 were retainedin the model.

RESULTS

ICD-9

Between November 1, 2003, and February 29, 2004, atotal of 2483 eligible visits with a discharge diagnosis of anARI were reviewed across the 15 study sites. (Only 4.7% ofthe charts identified by codes did not have a physician-assigned diagnosis of an ARI.) Fewer than 3000 visitswere analyzed because 16 patients were excluded due toimmunosuppression, 101 were excluded because they hadbeen taking antibiotics prior to the visit, and 12 hospitals hadfewer than 200 visits of interest in the time frame specified.Among the 2483 visits, 1646 visits (66%) had a documentedprescription for an antibiotic. Respiratory fluoroquinolonesaccounted for 14% (237/1646) of the total antibiotic prescriptions,ranging from 4% to 24% among the 15 individualhospitals. The other antibiotics prescribed includedmacrolides (39%), penicillins (32%), tetracyclines (6%),cephalosporins (4%), trimethoprim/sulfamethoxazole (3%),clindamycin (1%), and others (1%).

P

P

P

P

There were significant differences in the proportion ofARI visits associated with an RFQ prescription in relation topatient, provider, and hospital characteristics (Table 1).Among those visits where an antibioticwas prescribed, a higher percentageof patients in the age categories of45 to 64 years (39%) and > 65 years(37%) received RFQs compared withpatients aged 18 to 44 years (24%; < .001). In addition, a greater proportionof patients who had a dischargediagnosis of acute bronchitis (29%) orpneumonia (43%; < .0001) receivedprescriptions for RFQs compared withpatients who had other dischargediagnoses. In regards to provider characteristics,the proportion of ARI visitswhere an RFQ was prescribed waslower when the patient was evaluatedby a nurse practitioner or physicianassistant versus house staff or anattending physician (4% vs 20%[house staff] and 74% [attendingphysician]; < .001). The percentageof ARI visits where an RFQ was prescribedalso was lower when theseantibiotics were only available on thehospital formulary on a restrictedbasis (9% vs 91%; < .0001).

Table 2 contains the results of thesurvey regarding hospital policies andprocedures that could influenceantibiotic selection by hospital type,as well as the RFQ prescription ratesfor the 15 sites. For the VA medical centers, 7 of 8 sites hadat least 1 hospital-unrestricted RFQ (1 hospital had 2 unrestrictedRFQs). Of the 7 non-VA hospitals, 5 had at least 1hospital-unrestricted RFQ. Of 13 hospitals with data available,9 stocked an RFQ in the ED. However, only 1 of the 15sites had any patient drug samples available. Virtually allpatients who are evaluated in the ED at a VA hospital havetheir prescriptions filled at the on-site, outpatient pharmacy.At 5 of 6 non-VA hospitals, an outpatient pharmacy wasavailable for use by patients seen in the ED, and 4 of the 5pharmacies followed the hospital formulary. However,patients are not obligated to use these outpatient pharmacies.Regularly updated hospital antibiograms were distributed at13 of the hospitals. Provider detailing by pharmaceutical salesrepresentatives required scheduled appointments at all but 1hospital, where sales representatives could still conduct"drop-in" visits. Finally, 9 hospitals permitted industry-sponsoredcontinuing education programs, and 6 permitted salesrepresentatives to bring meals to those programs and/or othermeetings.

P

P

P

Table 3 shows the results of the multivariable regressionanalyses to determine which patient-, provider-, and hospitallevelfactors were independently associated with RFQ prescriptions.All models controlled for clustering ofpatient-level observations within provider, and there was noadditional clustering within hospital. Factors that were eligiblefor model entry included those that were significant at< .2 in the univariable analyses (ie, age, sex, chronic obstructivepulmonary disease, diagnosis, admission to hospital,provider type, hospital type, formulary status, pharmaceuticalindustry access, RFQ stocked in ED). Patient characteristicsthat remained significantly (< .05) associated with increasedRFQ prescribing were admission to the hospital and dischargediagnoses of acute bronchitis, acute exacerbation of chronicbronchitis (AECB), and pneumonia (compared with a nonspecificURI diagnosis). Hospital formulary status remainedindependently associated with RFQ prescription rates, wherethe availability of RFQs only on a restricted basis was associatedwith a lower rate of RFQ prescribing in outpatients (adjustedodds ratio = 0.2; 95% confidence interval = 0.1, 0.4). Therewas no significant interaction between hospital type (VA vsnon-VA) and hospital formulary status (= .69). No otherpatient, provider-, or hospital-level characteristics wereindependently associated with RFQ prescribing.

DISCUSSION

Clostridium difficile

We found that although overall levels of RFQ prescribingfor ARIs in hospital EDs were relatively low (14% of allantibiotics prescribed for ARIs), RFQs were prescribed formany ARIs that are commonly nonbacterial, including nonspecificURIs and acute bronchitis. This result is consistentwith other recent studies noting inappropriate broad-spectrumantibiotic use in the primary care and ED settings for avariety of diagnoses, including ARIs.1,2,14-16 We also foundthat the diagnoses of pneumonia and AECB were independentlyassociated with RFQ prescribing. The use of RFQs forpneumonia would be considered appropriate based on publishedguidelines,9,17,18 and some studies promote the use ofRFQs for the treatment of AECB in select patients.19,20However, some organizations and experts recommend notusing RFQs as first-line therapy for AECB to reduce overuseof these drugs, slow the development of resistance, and preservetheir antibacterial activity.21 Fluoroquinolone resistanceis emerging among gram-positive and gram-negativebacteria,3,22,23 and has been linked with recent outbreaks of astrain that is associated with high morbidityand mortality.24,25

community

More important, our results seem to indicate that hospitalpolicies, namely hospital formulary restrictions, are associatedwith patterns of RFQ prescribing for non-VA patients who arelikely to have their antibiotics filled in pharmacies.This is compatible with the results of a recent survey of physicians,clinical pharmacists, and formulary committee membersthat found formulary status, presence of prescribingrestrictions, and drug safety and effectiveness were all highlyinfluential in the prescribing decision process.26 Health maintenanceorganizations (HMOs) and pharmacy benefits managershave successfully used formularies as a means to affectambulatory drug prescribing for some time.27 Depending onthe criteria, formulary restrictions may negatively influenceprescribing in some conditions (eg, pneumonia) and positivelyinfluence it in others (eg, acute bronchitis). In communityacquiredpneumonia, inadequate antibiotic therapy has beenassociated with poor outcomes.28,29 It is important to recognizethe risks and benefits of antimicrobial drug restriction policies,particularly when the policies are focused at the drug level buthave implications across a variety of different diagnoses.

Policies and practices that influence physician prescribinghave been characterized as "direct" and "indirect."30,31Examples of direct methods include limiting access to specificdrugs through formulary restrictions and preapproval requirements.Indirect methods include educational programs frompharmaceutical sales representatives or professional colleaguesand distribution of clinical practice guidelines.31 Within theVA, the impact of the formulary is direct because most EDpatients have their prescriptions filled on-site. Therefore, theycannot obtain restricted or nonformulary medications unlessprespecified criteria are met or appropriate justification isgiven. These restrictions are feasible because alternativeantibiotics that cover the organisms associated with ARIs inmost patients often are available without restrictions. Innon-VA settings, the impact of the hospital formulary onambulatory prescribing is indirect because it does not directlyaffect the antibiotic that can be filled at an outside (community)pharmacy. However, in academically affiliated EDs,such as those in our study, house staff may rotate through theED and inpatient settings, and some ED physicians may periodicallyserve as inpatient attending physicians. Therefore,it is likely that their prescribing decisions in the ED arebeing influenced by their prior interactions with the hospitalformulary.

The results of our study should be interpreted with severallimitations in mind. Although we included many factors thatcan influence antibiotic selection in the ED, other patient,provider, and hospital characteristics not measured in our studycould have influenced RFQ prescribing decisions. For example,Steinman and colleagues found that lack of health insurance,black race, and membership in an HMO were associated with alower rate of broad-spectrum antibiotic prescribing for ARIs inoffice practices.15 Antibiotic selection should not be affected bya lack of insurance within the VA; however, this may have beena factor in non-VA hospitals, where we observed an overalllower prescribing rate for RFQs compared with the rate seen atVA sites. Patient factors such as a b-lactam allergy orrecent antibiotic use could also have entered into thedecision to prescribe a fluoroquinolone; therefore, useof an RFQ may have been appropriate in some ofthese situations. In addition, the data collected for thisstudy did not allow us to examine certain patientsubgroups for whom an RFQ could be consideredappropriate (eg, those with severe AECB). In thisregard, it is possible that some of the increased use ofRFQs in VA versus non-VA settings was appropriateand reflected a greater severity of illness and diseaseburden among patients being seen for respiratory tractinfections at VA sites. Providers may have decided touse a more broad-spectrum antibiotic, such as an RFQ,because they were concerned about the ability of certain outpatientsto return for care if they did not improve after treatmentwith a narrow-spectrum agent, or because they were concernedabout the potential for antibiotic-resistant infections. Althoughwe considered hospital-level characteristics such as marketingefforts by the pharmaceutical industry, these were hospital-widepolicies, and we do not know the details of programs or mealsthat occurred in the ED. In addition, the study may have beenunderpowered to detect an association between some hospitallevelvariables (eg, pharmaceutical industry access, stockingRFQs in the ED) and RFQ prescribing because there were only15 sites in the study. The study may have been underpowered todetect associations between provider type and RFQ prescribingas well. Other studies of antibiotic prescribing for ARIs havehighlighted differences by provider type; however, the resultshave not been consistent.32-34We believe that admission to thehospital, diagnosis, and formulary status remain strong factors ininfluencing antibiotic selection and may limit the impact ofsome of these other variables. Finally, our results may not begeneralizable to small, community, or nonteaching hospitalsand hospitals that are unwilling to participate in an antibioticquality improvement program.

CONCLUSION

In conclusion, hospital formulary policies appear to representa potentially important target for influencing drug prescribingpractices in ED settings. It is important for local policymakers to consider the impact of hospital formulary restrictionson patterns of outpatient antibiotic prescribing, which thencan influence patterns of resistance in the community.

Author Affiliations:

From the Center for Health Equity Research andPromotion, VA Pittsburgh Healthcare System, and University of PittsburghSchool of Pharmacy, Pittsburgh, Pa (SLA); the Center for Health EquityResearch and Promotion, Philadelphia VA Medical Center, and University ofPennsylvania School of Medicine, Philadelphia, Pa (JPM); and the Division ofGeneral Internal Medicine, Department of Medicine, University of California,San Francisco, Calif (JHM, RG).

Funding Sources:

Financial support was provided by joint grants from theHealth Services Research and Development (HSR&D) Service of the Departmentof Veterans Affairs (AVA-03-239) and the Agency for HealthcareResearch and Quality (R01-HS13915) under the Translating Research intoPractice program. Neither organization had a role in the design and conduct ofthe study; collection, management, analysis, and interpretation of the data;and preparation, review, or approval of the manuscript. Dr. Aspinall was a VAHSR&D Research Career Development awardee at the time of the study.Presented in part at the 43rd Annual Meeting of the Infectious DiseasesSociety of America, October 7, 2005, San Francisco, Calif.

Correspondence Author:

Sherrie L. Aspinall, PharmD, MSc, VAPittsburgh Healthcare System, Center for Health Equity Research andPromotion (151C-U), University Dr C, Pittsburgh, PA 15240. E-mail:sherrie.aspinall@va.gov.

Author Disclosure.

The authors (SLA, JPM, JHM, RG) report no relationshipor financial interest with any entity that would pose a conflict ofinterest with the subject matter discussed in this manuscript.

Authorship Information.

Concept and design and analysis and interpretationof data (SLA, JPM, JHM, RG); acquisition of data (SLA, JPM); draftingmanuscript and critical revision of the manuscript for importantintellectual content (SLA, JPM, RG); statistical analysis (JPM, JHM); provisionof study materials or patients (RG); obtaining funding (JPM, RG); administration,technical, or logistic support and supervision (JPM).

J Gen Intern

Med.

1. Roumie CL, Halasa NB, Grijalva CG, et al.Trends in antibiotic prescribingfor adults in the United States—1995 to 2002. 2005;20:697-702.

Am J Med.

2. Linder JA, Huang ES, Steinman MA, et al. Fluoroquinolone prescribingin the United States: 1995 to 2002. 2005;118:259-268.

Streptococcus pneumoniae

Clin Infect Dis.

3. Ho PL, Tse WS,Tsang KWT, et al. Risk factors for acquisition of levofloxacin-resistant : a case-control study.2001;32:701-707.

Streptococcus pneumoniae

Clin

Infect Dis.

4. Fuller JD, Low DE. A review of infectiontreatment failures associated with fluoroquinolone resistance. 2005;41:118-121.

Clin Infect Dis.

5. Zervos MJ, Hershberger E, Nicolau DP, et al. Relationship betweenfluoroquinolone use and changes in susceptibility to fluoroquinolonesof selected pathogens in 10 United States teaching hospitals, 1991-2000. 2003;37:1643-1648.

Clin Infect Dis.

6. Lautenbach E, Strom BL, Nachamkin I, et al. Longitudinal trends influoroquinolone resistance among Enterobacteriaceae isolates frominpatients and outpatients, 1989-2000: differences in the emergenceand epidemiology of resistance across organisms. 2004;38:655-662.

Escherichia coli

Klebsiella

pneumoniae

Arch Intern Med.

7. Lautenbach E, Fishman NO, Bilker WB, et al. Risk factors for fluoroquinoloneresistance in nosocomial and infections. 2002;162:2469-2477.

Streptococcus pneumoniae

Arch Intern Med.

8. Heffelfinger JD, Dowell SF, Jorgensen JH, et al. Management ofcommunity-acquired pneumonia in the era of pneumococcal resistance.A report from the Drug-Resistant Therapeutic Working Group. 2000;160:1399-1408.

Am J Respir Crit Care Med.

9. Niederman MS, Mandell LA, Anzueto A, et al. Guidelines for themanagement of adults with community-acquired pneumonia. Diagnosis,assessment of severity, antimicrobial therapy, and prevention.2001;163:1730-1754.

Am J Infect Control.

10. Fishman N. Antimicrobial stewardship. 2006;34(5 suppl 1):S55-S63.

Acad Emerg Med.

11. Metlay JP, Camargo CA Jr, Bos K, Gonzales R. Assessing the suitabilityof intervention sites for quality improvement studies in emergencydepartments. 2005;12:667-670.

Acad Emerg

Med.

12. Gonzales R, Camargo CA Jr, MacKenzie T, et al. Antibiotic treatmentof acute respiratory infections in acute care settings. 2006;13:288-294.

Applied Logistic Regression.

13. Hosmer DW, Lemeshow S. 2nd ed.New York, NY: John Wiley & Sons, Inc; 2000.

Ann Intern Med.

14. Steinman MA, Gonzales R, Linder JA, Landefeld CS. Changing useof antibiotics in community-based outpatient practice. 2003;138:525-533.

JAMA.

15. Steinman MA, Landefeld CS, Gonzales R. Predictors of broad-spectrumantibiotic prescribing for acute respiratory tract infections in adultprimary care. 2003;289:719-725.

Arch Intern Med.

16. Lautenbach E, Larosa LA, Kasbekar N, et al. Fluoroquinolone utilizationin the emergency departments of academic medical centers.2003;163:601-605.

Clin Infect Dis.

17. Bartlett JG, Dowell SF, Mandell LA, et al. Practice guidelines for themanagement of community-acquired pneumonia in adults. InfectiousDiseases Society of America. 2000;31:347-382.

Clin Infect Dis.

18. Mandell LA, Bartlett JG, Dowell SF, et al. Update of practice guidelinesfor the management of community-acquired pneumonia inimmunocompetent adults. 2003;37:1405-1433.

Am J Med.

19. Grossman RF, Rotschafer JC,Tan JS. Antimicrobial treatment oflower respiratory tract infections in the hospital setting. 2005;118(suppl 7A):29S-38S.

Am J Med.

20. Martinez FJ, Anzueto A. Appropriate outpatient treatment of acutebacterial exacerbations of chronic bronchitis. 2005;118(suppl 7A):39S-44S.

N Engl J Med.

21. Stoller JK. Acute exacerbations of chronic obstructive pulmonarydisease. 2002;346:988-994.

Escherichia coli

Klebsiella

pneumoniae

Arch Intern Med.

22. Lautenbach E, Fishman N, Bilker WB, et al. Risk factors for fluoroquinoloneresistance in nosocomial and infections. 2002;162:2469-2477.

Clin Infect Dis.

23. Zervos MJ, Hershberger E, Nicolau DP, et al. Relationship betweenfluoroquinolone use and changes in susceptibility to fluoroquinolonesof selected pathogens in 10 United States teaching hospitals, 1991-2000.2003;37:1643-1648.

Clostridium difficile. N Engl J Med.

24. McDonald LC, Killgore GE,Thompson A, et al. An epidemic, toxingene-variant strain of 2005;353:2433-2441.

Clostridium difficile

N Engl J Med.

25. Loo VG, Poirier L, Miller MA, et al. A predominantly clonal multiinstitutionaloutbreak of -associated diarrhea withhigh morbidity and mortality. 2005;353:2442-2449.

Ann Pharmacother.

26. Schumock GT,Walton SM, Park HY, et al. Factors that influenceprescribing decisions. 2004;38:557-562.

Am J Manag Care.

27. Carroll NV. How effectively do managed care organizations influenceprescribing and dispensing decisions? 2002;8:1041-1054.

Chest.

28. Houck PM, MacLehose RF, Niederman MS, Lowery JK. Empiricantibiotic therapy and mortality among Medicare pneumonia inpatientsin 10 western states: 1993, 1995, and 1997. 2001;119:1420-1426.

Am J Med.

29. Mortensen EM, Restrepo MI, Anzueto A, Pugh JA. Antibiotic therapyand 48-hour mortality for patients with pneumonia. 2006;119:859-864.

DICP.

30. Raisch DW. A model of methods for influencing prescribing, part I:a review of prescribing models, persuasion theories, and administrativeand educational methods. 1990;24:417-421.

DICP.

31. Raisch DW. A model of methods for influencing prescribing, part II:a review of educational methods, theories of human inference, anddelineation of the model. 1990;24:537-542.

Am J Med.

32. Roumie CL, Halasa NB, Edwards KM, et al. Differences in antibioticprescribing among physicians, residents, and nonphysician clinicians.2005;118:641-648.

Ann

Pharmacother.

33.Thorpe JM, Smith SR,Trygstad TK.Trends in emergency departmentantibiotic prescribing for acute respiratory tract infections. 2004;38:928-935.

J Am Acad Nurse Pract.

34. Ladd E.The use of antibiotics for viral upper respiratory tract infections:an analysis of nurse practitioner and physician prescribing practicesin ambulatory care, 1997-2001. 2005;17:416-424.