infection not clinically resolved, clinic visit or call, development of resistance,

COMPARATIVE RESEARCH Barry Patel, H. Ed Perez, Thomas A. Stacy, Ramin PourJarZib OBJECTIVE: To determine appropriateness of outpatient antibiotic use for common community-acquired infections and to compare the therapeutic outcomes of various antibiotics. DESIGN: Retrospective chart review using a standardized data collection form. PATIENTS: 209 members of a mixed-model health maintenance organization in Tampa, Florida, who had received antibiotics during January 1996. The study population proportionately represented all patients who were prescribed antibiotics, so that severity of illness did not affect results. MAIN OUTCOME MEASURES: This study gauged the appropriateness of empiric therapy based on patient characteristics, infection site, and laboratory/culture results. The outcome measures included appropriate empiric therapy based on patient characteristics and site of infection, as well as laboratory test and microbiological culture results. Antibiotic therapy was categorized as successful if one or more of the following criteria applied: absence of symptoms post therapy, no clinical signs of infection, and/or no clinic visit or call. Unsuccessful therapy resulted in the presence of symptoms post therapy, infection not clinically resolved, clinic visit or call, development of resistance, change in drug therapy, emergency room visit, and/or hospitalization. RESULTS: Antibiotic therapy was judged appropriate in 67% of patients and successful in 75% of patients. Of 56 patients with unsuccessful therapy cours- es, none visited an emergency room or was hospitalized. Infection was confirmed in approximately 10% of patients; 50% of patients initially received a broad-spectrum antibiotic. Antibiotic success rates were 85% for amoxicillin/ clavulanate, 79.5% for amoxicillin, 69.9% for first-generation s, and 47.4% for second-generation s. CONCLUSION: Treatment algorithms are needed to provide recommendations for empilic first-line antibiotic therapy of specific infectious diseases. Use of the quality improvement process could reduce the incidence of antibiotic resis- tance and pharmacy costs. KEYWORDS: Amoxicillin, Cephalosporins, Community-acquired infection, Managed care, Outcomes assessment J Managed Care Pharm 1999: 122-26... AUTHORS BARRY PATEL, PHARM.D., is President oj Total Therapcutic Management, Inc., Tampa, FL; H. ED PEREZ, PHARM.D., is CEO oj Total Therapeutic Management, Inc; THOMAS A. STAO: PHARM.O., is Director of Operations, Total Therapeutic Management, Inc.; RAMIN POURFARZlB, PH.D., is Adjunct projessol; Mercer University, Atlanta, GA. AUTHOR CORRESPONDENCE: Barry Patel, Pha/'m.D., Total Therapeutic Management, Inc., 125 Townpark Olive, Suite 360, Kennesaw, GA 30144. ACKNOWLEDGEMENT: This study was supported by unrestlicted educational g,'ants fmm Pfizer Phanl1aceuticals, Blistol-Myers Squibb Pharmaceuticals, and SmithKline Beecham Pharmaccuticals. Copyright(Þ 1999 Academy of Managed Care Pharmacy, Inc. All lights reserved. Community-acquired respiratory tract and urinary tract infections account for a significant number of physi- cian office visits and substantial antibiotic use in managed care environments. Because antimicrobial agents represent a major portion of the pharmacy budget in most managed care organizations, their inappropriate use can increase costs substantially. I In addition, inappropriate antibiotic use can induce the development of resistant microorganisms or pseudo-. membranous colitis diarrhea, resulting in negative patient outcomes.' Therefore, managed care organizations should imple- ment quality improvement programs for antibiotic use. Implementing a continuous quality improvement plan involves defining the problem through a formal outcomes evaluation process and then sequentially developing appropriate care plans.3 Outpatient use of antibiotics, including the penicillins, s, macrolides, and quinolones, primarily focuses on empiric treatment of symptoms related to a community-acquired infection. Empiric antibiotic therapy is defined as treatment initiated before microbiologic confirmation of infection.4 In selecting empiric therapy, physicians should consider previous antibiotic use, the most common pathogens for the site of infection, and the risks of treatment.s In practice, however, a physician's reasons for selecting a par- ticular empiric antibiotic regimen can vary \Nidely, based on his or her clinical experience and on the promotional efforts of 122 Journal of Managed Care Pharmacy jmcp March/April 1999 VoI.S, No.2

pharmaceutical companies'" Significant potential exists for the overutilization of broad-spectrum antibiotics as empiric thera- py. Therefore, overall outcomes need to be assessed through a formal evaluation. The present evaluation assessed clinical and economic outcomes related to antibiotic therapy of community-acquired infections in a managed care setting. This type of evaluation can serve as the first phase in an overall management program for infectious disease outcomes. It helped determine overall cost effectiveness of physician prescribing habits and identified potential quality and cost improvements. METHODS The study was a retrospective examination of the medical charts of patients who received antibiotics within a one-month study period at a mixed-model health maintenance organiza- tion (HMO). The strength of this approach was its ability to allow the HMO to measure drug utilization without altering physician prescribing behavior. This examination used a computerized database to randomly select from the insured population patients who had received an antibiotic during the month of January 1996. Randomization was formal to ensure that the study population was representative of the overall population that received antibiotics, that severity of illness did not affect results, and to obtain similar numbers of patients in each antibiotic drug class. To conduct the randomization, the pharmacy director provided a computer file that contained antibiotic utilization data for the month of January 1996. Clinical pharmacists collected data from patient medical records using a data collection tool designed to capture all information needed to evaluate outcomes follov:,ing antibiotic use. Major data collection variables included patient history and characteristics (age, renal function, history of antibiotic exposure, allergies); clinical risk factors for infection (comorbid diseases, surgical interventions, immunocompetency, placement of catheter or chronic intravenous line, hospitaliza- tion); signs and symptoms of infection (chief complaint or reason for office visit, physical signs and symptoms); diagnosis; site of infection; clinical indicators of infection (complete blood count, urinalysis, microbiology/culture results); and drug treatment regimen (drug, dose, dosing schedule, length of therapy, dosage adjustment for renal function). The most common documented infection site was the respiratory tract, occurring in 167 patients (80.0%), followed by skin and soft tissue in 17 patients (8.1 %), and urinary tract in three patients 0.4%); 20 (9.6%) patients did not have a docu- mented diagnosis (see Table 1). Of the 167 patients with respiratory tract infections, 8.4% were smokers, and 7.8% had some type of chronic obstructive pulmonary disease (COPD). Ten of the 20 patients without a documented diagnosis had symptoms of cough and congestion, suggesting a probable respiratory tract infection, while the remaining 10 patients had no chief complaint or symptoms. Not surprisingly, the most common diagnosis in pediatric patients was otitis media. Severity of infection was evaluated by the number of symptoms experienced by patients. A majority of patients (73%) had two or more symptoms (see Table 2). Since the predominant complaint was respiratory tract infection, the symptoms most commonly reported were sore throat, cough, congestion, and cold symptoms; the most common symptom for pediatric patients was ear pain. Fever, usually a hallmark of infection, was documented in only 40 patients 09.1 %). Laboratory testing (complete blood count, urinalysis) was performed in only nine cases, all for respiratory tract infections. Physicians obtained cultures in only 12 of the 209 patients; four (two respiratory tract, two skin and soft tissue) had no growth, and eight (five respiratory tract, three skin and soft tissue) had growth of gram-positive organisms. Organisms cultured from patients with respiratory tract infections included Group A, Group C, and non beta hemolytic Streptococci. The primary organism cultured from patients with skin and soft tissue infections was Staphylococcus aureus. Antibiotic Table 1. Diagnosis of Antibiotic Prescriptions for Community-Acquired Infections by Gender and Physician Specialty Number (%) of Patients Respiratory Tract Urinary Tract Skin and Soft Tissue I I Other Infection No Diagnosis TOlal (n=i67) (n=3) (n= 17) (n=2) (n=20) (n=209) Female (65.5) III (66.5) 1 (333) 8 (47.1) 2 (100) 15 (750) 137 Male (345) 56 (335) 2 (66 7) 9 (52.9) 0(0) 5 (25.0) 72 Family Practice (26.8) 41 (246) o (0) 8 (47.1) 1 (500) 6 (500) 56 Internal Medicine (39.7) 64 (38.3) 3 (100) 6 (353) 1 (500) 9(45.0) 83 Pediatrics (33.0) 62 (371) 0(0) 3 (17.6) 0(0) 4 (200) 69 Rheumatology 0(0) 0(0) 0(0) 0(0) 1 (05) 1 (05) Vat5, No.2 MarchJApril1999 ]MCP Journal of Managed Care Pharmacy 123

. Table 2. Number of Reported Symptoms by Type of Infection Number of Symptoms Respiratory Tract Urinary Tract Skin and Soft Tissue Other Infection No Diagnosis Total 0 1 1 2 0 10 14 (6.7%) 1 29 2 4 1 6 42 (20.1%) 2 58 0 8 0 4 70 (33.5%) 3 46 0 3 0 0 49 (23.4) 4 31 0 0 1 0 32 (15.3%) 5 2 0 0 0 0 2 (0.9%) Total 167 3 17 2 20 209 (100%) therapy was initiated in the three patients with urinary tract infections based on symptoms only. The proportions of antibiotic prescriptions by class in patients evaluated, as shown in Table 3, are similar to those for antibiotic utilization in the overall plan. Based on the approved indications for the antibiotics prescribed, antibiotic use was appropriate in most patients (see Table 3). However, if the criterion for appropriate therapy is confirmation of an infection, nearly all courses of therapy were inappropriate. In addition, second-line broad-spectrum antibiotics (amoxi- cillin/clavulanate, s, extended-spectrum macrolides, and quinolones) generally were prescribed for patients who did not have a history of failure or allergy to first-line agents. Of the 110 patients who received broad-spectrum antibiotics, only 31 had a prior history of failure with the first-line, narrow-spectrum agents amoxicillin, ampicillin, doxycycline, trimethoprim/sulfamethoxazole, and erythromycin. Figure 1 illustrates use of second-line antibiotics by history of failure or allergy to first-line agents. The majority of patients (82.3%, 186/209 patients) did not report a history of allergy to antibiotics. Penicillins (penicillin, 14 patients; amox- icillin, eight patients) and sulfonamides (13 patients) were Table 3. Antibiotic Prescriptions by Type of Infection Number (%) of Prescriptions Respiratory Urinary Skin and Other No Tract Tract Soft Tissue Infection Diagnosis Total Amoxicillin 67 (31.6) 0(0) 0(0) 0(0) 6 (2.8) 73 (34.4) AmoxicilliniClavulame 18 (85) 0(0) 1 (0.5) 0(0) 1 (0.5) 20 (9.4) Cephalosporins First generation 13 (6.1) 0(0) 9 (4.2) 0(0) 1 (0.5) 23 (l0.8) Second generation 19 (9.0) 0(0) 0(0) 0(0) 0(0) 19 (90) Third generation 6 (2.8) 0(0) 0(0) 0(0) 2 (0.9) 8 (3.8) Macrolides Erythromycin 18 (8.5) 0(0) 0(0) 0(0) 2 (0.9) 20 (9.4) Extended spectrum 20 (94) 0(0) 2 (0.9) 0(0) 7 (3.3) 29 (l3.7) Quinolones 1 (0.5) 3 (14) 2 (0.9) 0(0) 0(0) 6 (2.8) Tetracyclines 3(14) 0(0) 4 (1.9) 0(0) 0(0) 7 (3.3) Otheri' 3 (l4) 0(0) 1 (0.5) 2 (0.9) 1 (0.5) 7 (3.3) Total * 168 (79.2) 3 (1.4) 19 (9.0) 2 (0.9) 20 (9.4) 212 (100) *Threc patients received combination therapy. tother includes trimethoprim/sulfamethoxazole (3 patients), penicillin (2 patients), ampicillin (J patient), erythromycinlsulfamethoxazole (J patient). 124 Journal of Managed Care Pharmacy JMl.p MarchiApril1999 Vol.5, NO.2

Figure 1. Use of Second-line Antibiotics Amoxicill in -clavu lana te Extended -spectrum macrolides I st -generation 2nd-generation 3rd-generation TOTAL 35.5 (7/20 most commonly associated with previous drug allergy; some patients had a history of allergy to multiple antibiotics. Patients allergic to penicillin-type drugs were treated appropriately with either a macrolide or a sulfonamide. In general, length of drug therapy was appropriate, corresponding to that recommended for the diagnosis. A notable exception occurred with clarithromycin, which frequently was prescribed for 10 days when a seven-day regimen would have been adequate. Azithromycin was prescribed for an average of five days. Overall, antibiotic therapy was judged successful in 75% of patients. Therapy success rates by antibiotic class are shown in Figure 2. Most patients with unsuccessful therapy (33 of 56 patients) were switched to other antibiotics because of symptoms post therapy; none required an emergency room visit or hospitalization. The low success rate with extended-spectrum macrolide therapy (72.4%) is largely due to clarithromycin; success rates for clarithromycin and azithromycin were 64.7% 01117 patients) and 81 % (9111 patients), respectively.. Amoxicillin ~iii Erythromycin o 10 20 Percent of patients Figure 2. Therapy Success Rates by Antibiotic Class AmoxiciIlin Amoxicillinclavulanate I st -generation 2nd -generation 3rd-generation Erythromycin Extended -spectrum marcolides Quinolones Tetracyclines Other Overall I 30 75.0% (159/212) o 25 50 75 100 Percent of successful courses 40 DISCUSSION Diagnosis and treatment of community-acquired infections should be evaluated in all patient populations, as they represent a substantial number of physician visits annually. This study identified patterns of care in patients treated by physicians in an actual outpatient managed care setting. Unlike clinical trials, the study environment was uncontrolled, and therefore, evaluations of the various treatment regimens reflect decisions made by prescribers in the actual clinical environment. Findings from the medication outcomes assessment can be used to educate health care professionals and develop treatment guidelines to COrrect inappropriate drug usage as part of a continuous quality improvement program for managing community-acquired infections. Proper documentation is a very important component in evaluating any disease management program. In the present assessment, 20 patients (9.6%) did not have a documented diagnosis, although most had documented symptoms suggestive of an infectious disease process. Although patients who smoke or have chronic obstructive pulmonary disease tend to have a higher incidence of and more severe respiratory tract infections, these groups accounted for only a small percentage of patients with respiratory tract infections. Confirmation of infection may be even more important in evaluating the appropriateness of drug therapy. Generally, antibiotic therapy is prescribed empirically based on an evalu- ation of symptoms; however, infection should be confirmed with laboratory tests and culture results whenever possible. In the present study, physicians ordered laboratory testing (complete blood count, urinalysis) for about 4% of patients (9/209 patients), and cultures for only 6% of patients (2/209). Although it is well documented that sputum cultures are not '- r I Vo1.5. No.2 March/April1999 jmcp Journal of Managed Care Pharmacy 125

totally reliable in confirming respiratory tract infections/ the 4% culture rate (7/167 patients) in the present study is extremely low. Improvements in diagnostic testing for patients with suspected infection can be a cost-effective step for managed care organizations, especially when more expensive second- and third-line antibiotics are prescribed. The randomized procedure assured that the drugs utilized in the study population were indicative of those prescribed in the overall plan. Amoxicillin/clavulanate, s, extended-spectrum macrolides, and quinolones are considered second-line agents that should be reserved for patients who have failed first-line therapy with narrow spectrum antibiotics. There is growing concern that using broad-spectrum, second- line antibiotics rather than narrow-spectrum antibiotics may cause common pathogens to develop resistance to these drugs'" The approximate 3.5:1 ratio of amoxicillin to amoxicillin/clavulanate prescriptions in this evaluation suggests that the latter drug is being prescribed appropriately. In contrast, the ratio of prescriptions for extended-spectrum macrolides versus traditional erythromycin was 1.5: 1. In the present eval- uation, eight of 20 patients who received amoxicillin/clavulanate and eight of 29 patients receiving extended-spectrum macrolides had failed first-line therapy. Interestingly, the therapy success rate with erythromycin was far superior to that of the broader-spectrum macrolides. The appropriateness of drug therapy was determined based on history of drug allergies, diagnosis, culture results, and history of failure to first-line therapy. Overall, 66.5% of prescriptions were appropriate. The majority of inappropriate prescriptions were for extended-spectrum macrolides. Education and quality improvement efforts to encourage physicians to select first-line antibiotic therapy could improve overall antibiotic utilization. Based on the results of this study, we conclude that treat- ment algorithms that provide recommendations for appropriate antibiotic choices, particularly the use of first-line drugs, need to be developed for specific infections. As part of a quality improvement process, these algorithms could reduce the incidence of antibiotic resistance and pharmacy costs. À Rcfuenccs I. SchentagJ, Ballow CH, Fritz AL, et al. Changes in antimicrobial agent usage resulting from interactions among clinical pharmacy, the infectious disease division, and the microbiology laboratory. Diagn Microbiol Infect Dis 1993; 16: 255-4. 2. Murray B. Problems and dilemmas of antimicrobial resistance. Pharmacotherapy 1992; 12(6 pt 2): 865-935. 3. Marr J, Moffet H, Kunin C. Guidehnes for improving the use of antimicrobial agents in' hospitals: a statement by the Infectious Diseases Society of America J Infect Dis 1988; 157: 869-76. 4. Evans 5, Classen D, Pestotnik 5, et al. Improving empiric antibiotic selec- tion using computer decision support. Arch Intern Med 1994; 154: 878-84. 5 Briceland L, Nightingale C, Quintilliani R, et al. Antibiotic streamlining from combination therapy to monotherapy utihzing an interdisciphnary approach. Arch Intern Med 1988; 148: 2019-22. 6. Yu V, Stoehr G, Starling R, Shogan J. Empiric antibiotic selection by physicians: evaluation of reasoning strategies. Am J Med Sci 1991; 301: 165-72. 7. Dunagan W, Woodward R, MedotT G, et al. Antimicrobial misuse in patients with positive blood cultures. Am J Med 1989; 87: 253-9. 126 Journal of Managed Care Pharmacy jmcp March/April 1999 Vo15, No.2