COPD is a common disease, associated with high. Acute Exacerbation of COPD* Factors Associated With Poor Treatment Outcome

Acute Exacerbation of COPD* Factors Associated With Poor Treatment Outcome Naresh A. Dewan, MBBS, FCCP; Salem Rafique, MD; Badar Kanwar, MD; Hemant Satpathy, MD; Kay Ryschon, MS; Glenn S. Tillotson, MS; and Michael S. Niederman, MD, FCCP Objectives: To determine the effect of age, severity of lung disease, severity and frequency of exacerbation, steroid use, choice of an antibiotic, and the presence of comorbidity on the outcome of treatment for an acute exacerbation of COPD. Design: A retrospective chart analysis over 24 months. Setting: A university Veterans Affairs medical center. Patients: Outpatients with COPD who were treated with an antibiotic over a period of 24 months for an acute exacerbation of COPD. Methods: Severity of an acute exacerbation of COPD was defined using the criteria of Anthonisen et al: increased dyspnea, increased sputum volume, and increased sputum purulence. Severity of lung disease was stratified based on FEV 1 percent predicted using American Thoracic Society guidelines (stage I, FEV 1 > 50%; stage II, FEV 1 35 to 49%; stage III, FEV 1 < 35%). Treatment outcome was judged successful when the patient had no return visit in 4 weeks for a respiratory problem. Failure was defined as a return visit for persistent respiratory symptoms that required a change of an antibiotic in < 4 weeks. Results: One-hundred seven patients with COPD (mean age SD, 66.9 9.5 years) experienced 232 exacerbations over 24 months. First-line antibiotics (trimethoprim-sulfamethoxazole, ampicillin/amoxicillin, and erythromycin) were used to treat 78% of all exacerbations. Treatment failure was noted in 12.1% of first exacerbations and 14.7% of all exacerbations, with more than half the failures requiring hospitalization. Host factors that were independently associated with treatment failure included the following: FEV 1 < 35% (46.4% vs 22.4%; p 0.047), use of home oxygen (60.7% vs 15.6%; p < 0.0001), frequency of exacerbation (3.8 2.0 vs 1.6 0.91; p < 0.001), history of previous pneumonia (64.3% vs 35.1 p < 0.007), history of sinusitis (28.6% vs 8.8%; p < 0.009) and use of maintenance steroids (32.1% vs 15.2% p 0.052). Using stepwise logistic regression analysis to identify the top independent variables, the use of home oxygen (p 0.0002) and frequency of exacerbation (p < 0.0001) correctly classified failures in 83.3% of the patients. Surprisingly, age, the choice of an antibiotic, and the presence of any one or more comorbidity did not affect the treatment outcome. Conclusion: The results of our study suggest that patient host factors and not antibiotic choice may determine treatment outcome. Prospective studies in appropriately stratified patients are needed to validate these findings. (CHEST 2000; 117:662 671) Key words: acute exacerbation; chronic obstructive pulmonary disease; outcome factors Abbreviation: CHF congestive heart failure COPD is a common disease, associated with high morbidity and mortality, and serves as the fourth leading cause of death in the United States. 1 Morbidity from this disease is also high, with an estimated 14-million office visits and 500,000 hospitalizations each year. 2 The vast majority of office visits *From the Department of Medicine, Division of Pulmonary and Critical Care (Drs. Dewan, Rafique, Kanwar, and Satpathy) and Division of Cardiology (Ms. Ryschon), Creighton University and Veterans Affairs Medical Center, Omaha, NE; Pulmonary and Critical Care Medicine (Dr. Niederman), Winthrop University Hospital, Mineola, NY; and Bayer Pharmaceutical Division (Mr. Tillotson), West Haven, CT. Presented at the International Assembly of American College of Chest Physicians, Toronto, Canada, November 10, 1998. Dr. Dewan is the recipient of a research grant from Bayer Pharmaceuticals, Inc. and serves on their speaker s bureau. Manuscript received March 8, 1999; revision accepted October 26, 1999. Correspondence to: Naresh A. Dewan, MBBS, FCCP, Associate Professor, Division of Pulmonary and Critical Care, Department of Internal Medicine, Creighton University, St. Joseph Hospital, 601 North 30th St, Omaha, NE 68131; e-mail: ndewan@ creighton.edu 662 Clinical Investigations

are presumed to be secondary to acute exacerbation of COPD. Costs are estimated to be in excess of $2 billion/year. The goals of treatment in acute exacerbation of COPD are the prompt resolution of symptoms and minimization of failure rate. In a randomized placebo-controlled trial, the use of antibiotics in acute exacerbation of COPD had a higher success rate with clinical improvement and a lower failure rate when compared to placebo. 3 The antibiotics appeared to benefit patients who had moderate to severe exacerbation as judged by the presence of at least two of the following: increased sputum volume, increased sputum purulence, and increased dyspnea. A recent meta-analysis of nine randomized, placebo-controlled studies demonstrated a small but statistically significant benefit with the use of antibiotics as compared to placebo in acute exacerbation of COPD. 4 Ball and colleagues 5 attempted to assess the clinical parameters that predicted treatment outcome in patients with acute exacerbation of chronic bronchitis. A total of 471 patients were enrolled at the offices of 100 general practitioners. This study demonstrated that the presence of coexisting cardiopulmonary disease and more than four exacerbations in the prior 12 months had a 75% sensitivity in predicting a return to the office with a chest problem within the next 4 weeks. The severity of clinical symptoms and physical findings, however, did not affect the treatment outcome. Furthermore, this study did not report the presence of other comorbid medical conditions, choice of antibiotics, and severity of underlying lung disease as judged by the pulmonary function tests. The Canadian guidelines on management of chronic bronchitis propose that in patients 65 years of age, or in patients with four or more exacerbations in the prior 12 months, or FEV 1 50% of predicted and the presence of comorbid medical conditions, may be classified as patients with complicated chronic bronchitis. 6 They suggest that treating this group of patients with newer more broad-spectrum antibiotics, as opposed to the usual first-line antibiotics, may improve treatment outcome. Data from a retrospective analysis from our University Center demonstrated that the use of newer antibiotics, when compared to the usual firstline antibiotics in the treatment of acute exacerbation of chronic bronchitis, reduced both the hospitalization and failure rate. 7 Another recent study demonstrated that patients treated with amoxicillin had higher recurrence rates of COPD exacerbations than patients who received other therapies. 8 These studies suggest that the choice of an antibiotic is likely to affect treatment outcome. The goal of this study was to determine the effect of age, severity, and frequency of exacerbation, comorbid medical conditions, severity of underlying lung disease, and the choice of an antibiotic on the outcome of treatment in acute exacerbation of COPD. Materials and Methods Study Design and Patient Selection This study was a retrospective collection of data over 24 months in outpatients with COPD who were treated with an antibiotic for an acute exacerbation of COPD. Patients were initially screened from the pharmacy records for an ongoing antibiotic study in acute exacerbation of COPD. Patients who were either receiving an inhaled or an oral bronchodilator, or both, and had received an antibiotic prescription were identified from the pharmacy records. Between January 1995 and June 1997, a total of 228 patients met this screening criteria. Patients were excluded from further analysis if they did not meet the definition of COPD according to the American Thoracic Society guidelines, 9 or if they had a diagnosis of pneumonia, active pulmonary tuberculosis, cystic fibrosis, bronchiectasis, lung cancer, or AIDS. Patients were also excluded if they had received an antibiotic for reasons other than acute exacerbation of COPD or did not receive all their care at our Veterans Affairs Medical Center. The majority of the 121 patients excluded were due to an antibiotic prescription for reasons other than an acute exacerbation of COPD. The remaining 107 patients who met the definition of COPD and had received an antibiotic for acute exacerbation of COPD were eligible for this study. Each patient was retrospectively followed for a period of 24 months from the time of the first exacerbation. The ongoing antibiotic study was approved by our institutional review board, but no approval was required for this retrospective analysis. Clinical Data The following demographic and clinical data were collected for every patient. Demographic data included age, sex, smoking history, history of previous pneumonia, history of sinusitis, vaccination status, pulmonary function tests, use of home oxygen, comorbid medical conditions, and the use of concomitant steroids. History of any prior pneumonia and sinusitis was recorded from the patient problem list. Data on the duration of COPD 10 years could not be reliably obtained in more than half the patients and was thus excluded from analysis. Severity of lung disease was classified based on the degree of airflow obstruction using the American Thoracic Society criteria for FEV 1 as a percent of predicted. 1 Stage I is mild airflow obstruction with FEV 1 50% predicted; stage II is moderate airflow obstruction with FEV 1 35 to 49% predicted; and stage III is severe airflow obstruction with FEV 1 35% predicted. The best postbronchodilator FEV 1 within 5 years was recorded. The majority of the pulmonary function tests were done within 24 months, and these studies were available in 104 patients. The following comorbid medical conditions were recorded: (1) congestive heart failure (CHF), (2) hypertension, (3) coronary artery disease, (4) diabetes mellitus, (5) chronic liver disease, (6) history of alcoholism, and (7) chronic renal failure. Most of these diagnoses were recorded from the patient problem list maintained in the charts. Patients with chronic liver failure had abnormal liver enzymes, but most patients with history of chronic CHEST / 117 / 3/ MARCH, 2000 663

alcoholism had normal liver enzymes. All patients with chronic renal failure had renal insufficiency with serum creatinine 2.5 mg/dl. There was no patient on hemodialysis or peritoneal dialysis. The use of maintenance steroids was defined as an oral dosage of prednisone of at least 5 mg/d for a period of 3 consecutive months. Booster steroid was defined as a short course of prednisone during an acute exacerbation of COPD. An average dose of 300 mg of prednisone was used for booster steroid. The average cumulative dosage of prednisone (milligrams per day) was calculated by adding the maintenance steroid dose plus the total dose for booster steroid divided by the number of days over 2 years from the time of first exacerbation. Acute exacerbation of COPD was defined as the presence of any one of the following three symptoms: (1) increased cough and sputum volume, (2) increased sputum purulence, or (3) increased dyspnea. In addition, patients may have had one or more symptoms of fever, malaise, fatigue, and chest congestion. Severity of an acute exacerbation of COPD was defined as type 1 when patients had all three symptoms, type 2 when patients had any two out of three symptoms, and type 3 when patients had any one out of three symptoms. 3 The number of exacerbations that required antibiotic treatment over 24 months were recorded for every patient. All patients were followed for 24 months after the first infective exacerbation. Subsequent exacerbation was defined as an exacerbation that occurred 4 weeks after a successful treatment. Treatment of a failure was not considered as a new exacerbation, ie, need for further antibiotics for 7 to 14 days after commencement of initial therapy. The following data were recorded for each exacerbation: (1) clinical signs and symptoms, (2) chest radiograph, (3) sputum for Gram s stain and cultures, (4) arterial blood gases, (5) antibiotics, and (6) use of booster steroids. Decisions for ordering the chest radiograph, sputum culture, and arterial blood gases were made by the individual treating physician. These tests were done sparingly and generally reflected the outpatient treatment practice at our center. Patients who had history of sinusitis were judged to be experiencing an acute exacerbation of COPD rather than an acute episode of sinusitis, but this distinction can be difficult sometimes. We feel reasonably comfortable that patients who did not have a chest radiograph were unlikely to have had pneumonia, since none of the patients who failed and required admission had an infiltrate at the time of hospitalization. The treatment outcome response for each exacerbation was recorded as follows. Patients were considered a treatment success if they had no return visit in 4 weeks for respiratory problem. Patients were considered a treatment failure if they had a return visit in 4 weeks with persistent respiratory symptoms that required a change of an antibiotic or hospitalization with change of an antibiotic. No patient had a respiratory problem that did not require a change in antibiotic or a return visit for a nonrespiratory problem. Data Analysis Demographic and baseline clinical characteristics were summarized as a mean SD for continuous variables and as a percentage of the group for categorical or dichotomous variables. Univariate testing of independent factors that were associated with exacerbation success or failure was done using the 2 statistic for dichotomous variables and Student s t test for continuous variables. The correlation between clinical variables was evaluated using the Pearson s correlation coefficient. A p value 0.05 was considered significant. Due to exclusion of a large number of patients, the study was powered at 50% to detect a significant difference at the 5% level. Logistic regression analysis was used to identify the subset of independent variables that were associated with exacerbation failure. Both forward stepwise selection and backward stepwise elimination methods were used for model development. Logistic regression analysis used in this fashion, in this group of patients, was considered to be more efficient than normal discriminant and multivariate analysis. 10,11 The likelihood-ratio test was used to determine which variable should be added or removed from the model. Results One hundred seven patients with COPD experienced a total of 232 acute exacerbations over a period of 24 months. Demographic and baseline clinical characteristics are listed in Table 1. The age (mean SD) was 66.9 9.5 years. Ninety-five patients (88.8%) were either current smokers or exsmokers. Sixty-four patients (61.5%) had moderate (stage II) to severe (stage III) airflow obstruction. Twenty-nine patients (27.1%) were using home oxygen, 21 patients (19.6%) were on maintenance steroids, 45 patients (42.1%) had a history of previous pneumonia, and 15 patients (14%) had history of sinusitis. A majority of the patients (80.4%) had received influenza vaccination, while only half the patients (48.6%) had documentation for pneumococcal vaccination. The incidence of comorbidity varied significantly: CHF (29.9%), coronary artery disease (51.5%), hypertension (65.4%), diabetes (17.8%), chronic liver disease (2.8%), history of alcoholism Table 1 Demographic and Baseline Characteristics* Characteristics Data, n 107 Age, yr 66.9 9.5 Sex, male 107 (100) Smoking history Nonsmoker 12 (11.2) Current/ex-smoker 95 (88.8) Severity of lung disease FEV 1 (% predicted, n 104) 47.1 18.3 FEV 1 50% 40 (38.5) FEV 1 35 49% 34 (32.7) FEV 1 35% 30 (28.8) Home oxygen 29 (27.1) Maintenance steroids 21 (19.6) History of previous pneumonia 45 (42.1) History of sinusitis 15 (14) Influenza vaccination 86 (80.4) Pneumococcal vaccination 52 (48.6) Comorbidity CHF 32 (29.9) Coronary artery disease 55 (51.4) Hypertension 70 (65.4) Diabetes 19 (17.8) Liver disease 3 (2.8) History of alcoholism 48 (44.9) Chronic renal failure 14 (13.1) *Data are presented as No. (%) or mean SD. 664 Clinical Investigations

(44.9%), and chronic renal failure (13.1%). The presence of any one or more comorbid conditions also varied significantly: no comorbidity (13.1%), one comorbidity (15%), two comorbidities (33%), three comorbidities (19%), four comorbidities (15%), and five comorbidities (5%). The distribution and outcome of exacerbations among all patients is shown in Figure 1. Seventynine (73%) patients had a total of 126 exacerbations with no failures, while 28 patients (27%) had 106 exacerbations that resulted in 34 failures. Of these 28 patients with failures, 23 patients had one failure, 4 patients had two failures, and 1 patient had three failures. Approximately 82% of the patients had less than four exacerbations over 24 months, while only 18% of the patients had four or more exacerbations. The frequency distribution for all exacerbations in 107 patients is shown in Figure 2. The clinical and laboratory data for all exacerbations is listed in Table 2. Based on the criteria of Anthonisen et al, 190 exacerbations (82%) were moderate to severe in intensity (type 1 and type 2). A majority of the exacerbations (81%) were associated with abnormal auscultation sounds (wheezing, rhonchi, and decreased breath sounds), while very few exacerbations had significant tachypnea (1%), tachycardia (18%), and elevated temperature (15%). Sputum for culture was obtained prior to antibiotics in 42 exacerbations (18%), and a majority of the cultures demonstrated normal flora. A chest radiograph was done in 101 exacerbations (44%), arterial blood gases in 41 exacerbations (18%), and booster steroids were used in 51 exacerbations (22%). First-line antibiotics (amoxicillin/ampicillin, trimethoprim-sulfamethoxazole, and erythromycin) were used in 182 exacerbations (78%). Figure 1. Distribution of all patients with successful exacerbations and failures. Seventy-nine patients had no failures, while 28 patients had 1 failures, with a total of 34 failed exacerbations. Treatment Outcome: An analysis of the treatment outcome for all exacerbations demonstrated a successful outcome in 198 exacerbations (85.3%), while 34 exacerbations (14.7%) failed to respond within 4 weeks of treatment. The mean time of failure (mean SD) was 16.7 8.3 days. The failure rate for the first exacerbation in these 107 patients was 12.1% and very similar to the outcome of all exacerbations. Of these 34 failures, 16 exacerbations (48%) in 10 patients required a change in antibiotic, while 18 exacerbations (52%) in 18 patients required hospitalization. Correlation of Initial Antibiotic Choice With Treatment Outcome: The frequency of antibiotic use and success rate respectively for different antibiotics in all 232 exacerbations were as follows: trimethoprim-sulfamethoxazole, 50.4% and 86.3%; ampicillin/amoxicillin, 19.4% and 84.4%; erythromycin, 9.5% and 81.8%; amoxicillin-clavulanic acid, 8.6% and 80%; azithromycin, 4.5% and 91%; first generation cephalosporins, 3.5% and 75%; and other antibiotics (ciprofloxacin, cefixime, cefuroxime, and doxycycline), 3.8% and 100%. Overall, first-line antibiotics were used in 182 exacerbations (78%), while the newer second-line antibiotics were used in only 50 exacerbations (22%). Although, a small subgroup of patients who were treated with the newer antibiotics had a 100% success rate, the numbers were too small for any meaningful correlation. Overall, the choice of an antibiotic did not affect the treatment outcome. Analysis of 28 patients who experienced 34 failures demonstrated that the use of first-line antibiotics in the failure group was 76.5% (26/34) and was similar when compared to the successful group 78% (56/72; p 0.653). The success rate with first-line antibiotics in this group of 28 patients was 68.2% (56/82) and similar to the success rate with second-line antibiotics 63.6% (14/22). Similarly, analysis of 30 patients with FEV 1 35% demonstrated that the use of first-line antibiotics was similar in the failure group (64.7%; 11/17) as compared to the successful group (72.1%; 44/61; p 0.463). The success rate with first-line antibiotics in this group of 30 patients was 80% (44/55) and similar to the success rate with second-line antibiotics 72.7% (16/22). A subgroup analysis of seven patients who had 4 exacerbations over 24 months and the use of home oxygen demonstrated similar use of the first-line antibiotics in the failure group (85%; 10/12) as compared to the successful group (74%; 20/27; p 0.526). The success rate with first-line antibiotics in this subgroup of seven patients was 66.7% (20/30) and similar to the success rate with second-line antibiotics (77.7%; 7/9). Surprisingly, the choice of antibiotic treatment did not affect the treatment outcome in any of these high-risk groups. CHEST / 117 / 3/ MARCH, 2000 665

Figure 2. Frequency distribution of exacerbations in 107 patients. Approximately 82% of the patients had less than four exacerbations, while 18% of the patients had four or more exacerbations. Table 2 Clinical and Laboratory Data for Exacerbations* Variables Data, n 232 Severity of exacerbation Type 1 85 (37) Type 2 105 (45) Type 3 42 (18) Total 232 Clinical findings Respiratory rate (30 beats/ 3 (1) min Temperature 99 F 33 (15) Heart rate 100 beats/min 41 (18) Abnormal auscultation sounds 189 (81) Sputum culture 42 (18) Normal flora 30 (13) H influenzae 4 (2) H parainfluenzae 4 (2) M catarrhalis 1 (0.5) S aureus 2 (1) P aeruginosa 1 (0.5) Chest radiograph 101 (44) Booster steroids 51 (22) First-line antibiotics 182 (78) Arterial blood gases 41 (18) *Data are presented as No or No. (%). Treatment Outcome Factors: The treatment outcome factors for all exacerbations are listed in Table 3. Factors that were independently associated with treatment failure included the following: FEV 1 35% (46.4% vs 22.4%; p 0.047), use of home oxygen (60.7% vs 15.6%; p 0.0001); frequency of exacerbation over 24 months (3.8 2.0 vs 1.6 0.9; p 0.001), history of previous pneumonia (64.3% vs 35.1%; p 0.007), history of sinusitis (28.6% vs 8.8%; p 0.009), and the use of maintenance steroids (32.1% vs 15.2%; p 0.052). The average cumulative steroid dosage (milligrams per day) over 24 months was also significantly higher in the group of patients who had at least one failure as compared to patients with no failure (12.1 8.2 vs 5.1 3.7 mg; p 0.017). Surprisingly, age and the presence of any one, two, or more than two comorbidities did not affect the treatment outcome. A subgroup analysis of the 18 patients who were hospitalized demonstrated a significantly lower FEV 1 35% (67% vs 10%; p 0.006) and a more frequent use of home oxygen (78% vs 30%; p 0.020) when compared to the 10 patients who failed but did not require hospitalization. Pathogens recovered from sputum cultures in these hospitalized patients included Moraxella catarrhalis, 1 Haemophilus influenzae, 1 and Pseudomonas aeruginosa. 1 There was no significant difference in the treatment outcome between the first-line and the second-line antibiotics in the group of patients who were admitted, and also when compared to the patients who were not admitted. The risk for at least one failure appeared to increase progressively as the number of exacerbations increased, and the chance for failure was 100% for patients who had more than four exacerbations over the 24-month period (Fig 3). The failure rate for type 1 exacerbation was significantly greater than the failure rate for type 3 exacerbation (22% vs 7.1%; p 0.037), but not significantly different than the 666 Clinical Investigations

Table 3 Comparison of Patient Characteristics Between Failure and Success Groups* Characteristics Failure, n 28 Success, n 79 p Value Age, yr 64.6 10.7 67.7 9.0 0.148 Severity of lung disease FEV 1 (% predicted) 41.8 20.5 49.0 17.1 0.071 FEV 1 50% 21 (75.0) 43/76 (56.5) 0.087 FEV 1 50% (stage I) 7 (25) 33/76 (43.4) 0.087 FEV 1 35 49% (stage II) 8 (28.6) 26/76 (34.2) 0.587 FEV 1 35% (stage III) 13 (46.4) 17/76 (22.4) 0.047 Home oxygen 17 (60.7) 12/77 (15.6) 0.0001 Frequency of exacerbation over 24 mo 3.8 2.0 1.6 0.9 0.001 Maintenance steroids 9 (32.1) 12/79 (15.2) 0.052 Average cumulative steroid dosage over 24 mo, mg/d 12.1 8.2 5.1 3.75 0.017 Previous pneumonia 18 (64.3) 27/77 (35.1) 0.007 History of sinusitis 8 (28.6) 7/79 (8.8) 0.009 Influenza vaccination 24 (85.7) 62/79 (78.5) 0.408 Pneumococcal vaccination 17 (60.7) 35/79 (44.3) 0.135 Comorbidity CHF 9 (32.1) 23/77 (29.9) 0.823 Coronary artery disease 15 (53.6) 40/76 (52.6) 0.932 Hypertension 18 (64.3) 52/78 (66.7) 0.819 Diabetes 4 (14.3) 15/78 (19.2) 0.558 Liver disease 2 (7.1) 1/78 (1.3) 0.109 History of alcoholism 15 (53.6) 33/78 (42.3) 0.304 Chronic renal failure 2 (7.1) 12/78 (15.4) 0.269 *Data are presented as No. (%) or mean SD. Numbers 79 represent missing values. Figure 3. Failure risk stratified by the frequency of exacerbations in 107 patients. The percent of patients with at least one failure appeared to increase as the number of exacerbations increased. Patients with more than four exacerbations had a 100% chance of having at least one failure. failure rate for type 2 exacerbation (22% vs 12.4%; p 0.081). There was also good correlation between the severity of lung disease and the treatment outcome. The percent of exacerbations that resulted in failures progressively increased as the severity of airflow obstruction increased. Stage III patients with an FEV 1 35% had significantly greater percentage of failures than successful exacerbations (46.4% vs 22.4%; p 0.047; Fig 4). This analysis used one 2 3 2 table that compared the failures and suc- CHEST / 117 / 3/ MARCH, 2000 667

Figure 4. Treatment outcome and distribution of patients by severity of lung disease. The percent of patients in the group with at least one failure (n 28) increased as the severity of lung disease worsened. In stage III patients with FEV 1 35% there were more exacerbation failures than successes (46.4% vs 22.4%; p 0.047). *Significant difference in failure rate between stage III and stage I patients (see text for details). cesses between all stages of COPD (stage I, II, and III). On further analysis, using 2 2 2 tables, the failure rate between stage III patients (FEV 1 35%) was significantly greater than the failure rate for stage I patients (FEV 1 50%; 65% vs 35%; p 0.018), but not with stage II patients (FEV 1 35 to 49%; 61% vs 38%; p 0.092) or between stage II and stage I patients (53% vs 47%; p 0.520). Among the top six independent variables, there was significant correlation between home oxygen and severity of airflow obstruction (r 0.437; p 0.001), frequency of exacerbation (r 0.235; p 0.016) and use of maintenance steroids (r 0.020; p 0.0026), maintenance steroids and severity of airflow obstruction (r 0.276; p 0.0005), and frequency of exacerbation (r 0.245; p 0.011). There was no correlation between history of sinusitis and the other five independent variables. Using stepwise logistic regression analysis for the top six independent variables that were associated with exacerbation failure, a model was constructed to predict the likelihood for failure. Variables in the final model were use of home oxygen (p 0.0002) and frequency of exacerbation (p 0.0001). The overall model 2 had a significance level of p 0.0001 and correctly classified 83.3% of the patients. The odds of failure for patients on home oxygen progressively increased with greater frequency of exacerbations (Table 4). Table 4 Odds of Failure in Relation to Home Oxygen Therapy and Number of Exacerbations Over 24 Months* Variables Discussion Odds of Failure Home oxygen and one exacerbation 0.311 Home oxygen and two exacerbations 1.008 Home oxygen and three exacerbations 3.274 Home oxygen and four exacerbations 10.627 Home oxygen and five exacerbations 34.707 *Odds of failure calculated from a logistic model, which included home oxygen and frequency of exacerbation as significant predictors. The overall 2 was significant at p 0.0001. The outcome of any given treatment is generally dependent on the severity of illness, host factors, and the specific treatment given. 3,12,13 Host factors that may affect the treatment outcome in acute exacerbation of COPD include age, severity of underlying lung disease, and the presence of comorbidity. 5,6 The results from our study in this group of COPD patients demonstrate that patients who had severe underlying lung disease and had greater frequency and severity of exacerbation were more likely to fail. Surprisingly, age, the presence of comorbidity, and the choice of an antibiotic did not affect the treatment outcome. 668 Clinical Investigations

Prior studies have demonstrated that the mortality in COPD patients increases sharply beyond age 65 years. 14 Because of this concern for high mortality, the Canadian guidelines have classified patients older than age 65 years as having complicated chronic bronchitis and a greater likelihood for poor outcome. 6 However, we could not find any study in the literature that demonstrated clear evidence of poor outcome in patients who were older than 65 years. In our study, the mean age for the group of patients who experienced at least one failure was 65 years and was slightly less but not significantly different than the group of patients with no failures. This suggests that age, in and by itself, may not be an independent variable that may affect the treatment outcome. However, since the age range and the study sample size was limited, it is possible that age, as a predictor of treatment outcome, may not have been adequately tested in this study. The presence of coexisting cardiopulmonary disease has been identified as a risk factor, and patients with this disease were twice as likely to return to their physicians with recurrence of respiratory symptoms than patients without. 5 The Canadian guidelines have included other comorbid medical conditions such as CHF, diabetes mellitus, chronic renal failure, and liver disease as risk factors that may adversely affect the treatment outcome. 6 In this study, despite the high prevalence of several comorbid medical conditions including CHF, the presence of any one, two, or more than two comorbidities did not affect the treatment outcome. The presence of severe underlying lung disease as judged by FEV 1 35% was associated with poor outcome in our study. The Canadian guidelines have classified patients with chronic bronchitis and an FEV 1 50% of predicted as patients with complicated chronic bronchitis. 5 Duration of COPD 10 years and increased severity of chronic bronchitis have also been shown to be risk factors for poor outcome with greater chance of hospitalization. 15 More recent data in the literature demonstrate that patients with severe obstructive lung disease are more likely to be infected with Gram-negative pathogens during acute and severe exacerbation of chronic bronchitis. 16,17 These data suggest that patients with severe lung disease when treated with the usual first-line antibiotics for an acute exacerbation could be more likely to fail and to have a poor outcome. Other host factors that were associated with poor outcome in our study were the use of home oxygen, use of maintenance steroids, history of previous pneumonia, and history of sinusitis. Statistical analysis demonstrated good correlation between the first three factors and the severity of lung disease as judged by FEV 1 35%. Furthermore, the use of logistic regression analysis demonstrated that among all these variables, the use of home oxygen and frequency of exacerbation correctly classified failures in a majority of the patients. The results of our study thus support the rationale that patients with severe lung disease are more likely to require home oxygen, and maintenance steroids and are at greater risk for frequent infections with poor treatment outcome. This suggests that host factors may have a greater impact on treatment outcome than the choice of treatment. The frequency of acute exacerbation was also associated with poor outcome in our study. Ball and colleagues 5 demonstrated that patients with more than four exacerbations in the prior 12 months were twice as likely to fail when compared to patients with less than three exacerbations. The Canadian guidelines categorize patients with chronic bronchitis who have more than four exacerbations in the prior 12 months as having complicated chronic bronchitis. 6 The findings from our study also demonstrate that patients with more frequent exacerbations were more likely to fail than patients with less-frequent exacerbations. In fact, the highest odds of failure were in the group of patients who were receiving home oxygen and had more than four exacerbations over 24 months. However, the patients with frequent exacerbations (four or more) were 20% of our study subjects, while the majority of the patients had infrequent exacerbations (less than four exacerbations over 24 months). The usual first-line antibiotics are recommended for patients with infrequent exacerbations, while newer antibiotics are suggested for patients with frequent exacerbations. 18 The severity of an acute exacerbation as judged by the number of symptoms is likely to have impact on the treatment outcome. Anthonisen et al 3 demonstrated that patients with severe exacerbation who experienced all three symptoms of increased dyspnea, sputum volume, and sputum purulence (type 1) were more likely to fail than patients with less severe exacerbations. Other workers 5 found no correlation between the severity of symptoms and the treatment outcome. The results of our study are similar to the study by Anthonisen et al, 3 with the failure rate being significantly higher in patients with type 1 exacerbation when compared to type 3 exacerbation. The choice of an antibiotic in the treatment of acute exacerbation can affect the treatment outcome. Studies done in the early 1980s showed no differences in the outcome among the usual first-line antibiotics that included amoxicillin/ampicillin, trimethoprim-sulfamethoxazole, and erythromycin. With the changing bacterial etiology 17,19 and resistance patterns notably for H influenzae, M catarrhalis, CHEST / 117 / 3/ MARCH, 2000 669

and Streptococcus pneumoniae, 20,21.22 it would not be surprising to expect some antibiotics to do better than others. The use of ciprofloxacin in a prospective 1-year study demonstrated improved clinical outcome in patients with moderate to severe chronic bronchitis and at least four acute exacerbations of chronic bronchitis in the previous year when compared to the usual antibiotic treatment. 15 One recent study demonstrated that the use of amoxicillin in the treatment of acute exacerbation had significantly higher recurrence rate (65%) for COPD exacerbations than other agents. 8 Another study conducted at a university center demonstrated a significantly higher failure (19%) and hospitalization rate (18%) in patients who were treated with the usual first-line antibiotics as compared to the newer antibiotics. 7 In this study, first-line antibiotics were used in 78% of the patients with a successful outcome in a majority of the patients (85%), and no differences in treatment outcome were observed among the different antibiotics. However, since the newer antibiotics were used only in a small group of patients (22%), we cannot conclude that all antibiotics were similar in efficacy and that no antibiotic therapy would have lead to improved outcome. We can only speculate that the use of newer antibiotics in our study did not show any differences in the treatment outcome because of their limited use, and lumping of all second-line agents. The results of our study also suggest that patient factors and not antibiotic therapy may determine treatment outcome. Using other end points besides failure and admission for treatment outcome, a prospective study comparing first-line vs newer antibiotics in appropriately stratified COPD patients may help to resolve this issue. The results of our study also demonstrate that 12.1% of the first exacerbations and 14.7% of all acute exacerbations of COPD failed to respond to antibiotics within 4 weeks of treatment. This failure rate was slightly higher than failure rate of 10% reported by Anthonisen et al, 3 but similar to the results of more recent studies. 5,7 Of greater clinical significance, more than half the exacerbations that resulted in failures required hospitalization. There are several limitations to our study. First, the screening process from the pharmacy records used for this study only allowed inclusion of patients who were treated with antibiotics for an acute exacerbation of COPD. It is possible that some patients may have had acute exacerbations that were not treated with antibiotics and their outcome is not known. Second, the number of patients who had sputum cultures and chest radiographs was limited. We believe that this reflected our clinical practice in an outpatient setting. Evidence suggests that approximately half of the exacerbations involve bacterial infection, while one third of the exacerbations are caused by viral infections. 23 However, it is difficult to distinguish bacterial infections from viral infections and other noninfectious causes of exacerbations, and most patients are treated empirically with an antibiotic. In our study, patients were treated based on the clinician s impression that these patients were experiencing an acute exacerbation of COPD that required an antibiotic. It is conceivable that some of the exacerbations could have been nonbacterial in origin and antibiotics may not have made a difference. It is also possible that a few patients may have had acute sinusitis and they would have benefited from the antibiotics. Third, the incidence of recovery of pathogens in the sputum culture was low. This could be related to a variety of factors to include sputum collection, sputum transport, and handling in the laboratory. Fourth, the study had fewer patients than we had anticipated and was not sufficiently powered. Fifth, the retrospective design of this study did not allow for monitoring of compliance with the treatment and in determining the precise end points for success or failure. Patients who did not show up for subsequent follow-up in the next 4 weeks were considered to be presumed successes. It is possible that a small percent of patients could have gone for treatment elsewhere and we may have missed a few treatment failures; however, we selected a group of patients who lived within the vicinity of the medical center and had received all their care at this hospital. We had access to all their outpatient and inpatient records including pharmacy records and it is unlikely that we would have missed a large number of failures. Finally, the results of our study, which was done in a population of veterans, may not be applicable to the general population. In summary, this study demonstrates that patients who had severe underlying lung disease and experienced frequent and severe exacerbations were more likely to have poor treatment outcome. Patients with severe airflow obstruction were more likely to require home oxygen, use maintenance steroids, and were at greater risk for frequent infections. Age, the presence of the comorbidity, and the choice of an antibiotic did not affect the treatment outcome. These findings support the concept that patient factors and not antibiotic therapy may determine treatment outcome. In conclusion, the results of our study suggest that stratification of patients correlates with treatment outcome. The data from our study can be used to design future studies, to determine if specific patient subsets would benefit from specific types of antibiotics, which may lead to improved treatment outcomes. In addition, the selective use of newer antibiotics based on stratification would also limit the emergence of drug-resistant organism. More prospective studies using other end points beside failure 670 Clinical Investigations

and admission are needed to compare first-line antibiotics vs newer antibiotics in appropriately stratified patients. ACKNOWLEDGMENT: The authors greatly appreciate the assistance of Mike Caldwell, RRT, and Chris Harvey, RRT, for retrieving charts and data entry, and Mona Coburn and Cindy Hanneman for secretarial assistance. References 1 Celli BR, Snider GL, Heffner J, et al. Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1995; 152(suppl): 77 120 2 Morbidity and mortality: chartbook on cardiovascular, lung, and blood diseases. Rockville, MD: National Heart Lung and Blood Institute, National Institutes of Health; May 1996; Publication No. 96 50 3 Anthonisen NR, Manfreda J, Warren CPW, et al. Antibiotic therapy in exacerbations of chronic obstructive pulmonary disease. Ann Intern Med 1987; 106:196 204 4 Saint S, Bent S, Vittinghoff E, et al. Antibiotics in chronic obstructive pulmonary disease exacerbations: a meta-analysis. JAMA 1995; 273:957 960 5 Ball P, Harris JM, Lowson D, et al. Acute Infective exacerbation of chronic bronchitis. Q J Med 1995; 88:61 68 6 Balter MS, Hyland RH, Low DE, et al. Recommendations on the management of chronic bronchitis: a practical guide for Canadian physicians. Can Med Assoc J 1994; 151(suppl 10):5 23 7 Destache CJ, Dewan NA, O Donohue WJ, Jr, et al. Clinical and economic considerations in acute exacerbation of chronic bronchitis. J Antimicrob Chemother 1999; 43(Suppl A):107 113 8 Adams S, Mel J, Anzueto A. Effects of antibiotics on the recurrence rates of chronic obstructive pulmonary disease exacerbations [abstract]. Chest 1997; 112:22S 9 American Thoracic Society. Chronic bronchitis, asthma, and pulmonary emphysema: a statement by the Committee on Diagnostic Standards for Non-tuberculosis Respiratory Diseases. Am Rev Respir Dis 1962; 1962:85:762 768 10 Efron B. The efficiency of logistic regression compared to normal discriminant analysis. J Am Stat Assoc 1975; 70:892 898 11 Afifi AA, Clark V. Computer-aided multivariate analysis. Belmont, CA: Lifetime Learning Publications, 1984 12 Chodosh S. Treatment of chronic bronchitis: state of the art. Am J Med 1991; 91(6A):87S 92S 13 Niederman MS. Acute exacerbations of chronic bronchitis: the role of infection and the selection of appropriate therapy. Pulm Crit Care Update 1996; 11:1 7 14 Higgins M. Epidemiology of pulmonary disease. In: Casaburi R, Petty TL, eds. Principles and practice of pulmonary rehabilitation. Philadelphia, PA: WB Saunders; 1993; 10 17 15 Grossman RG, Mukherjee J, Vaughn D, et al. A 1-year community-based health economic study of ciprofloxacin vs usual antibiotic treatment in acute exacerbations of chronic bronchitis. Chest 1998; 113:131 141 16 Eller J, Ede A, Schaberg T, et al. Acute infective exacerbation of chronic bronchitis: relation between bacteriologic etiology and lung function. Chest 1998; 113:1542 1548 17 Soler N, Torres A, Ewig S, et al. Bronchial microbial patterns in severe exacerbations of chronic obstructive pulmonary disease (COPD) requiring mechanical ventilation. Am J Respir Crit Care Med 1998; 157:1698 1705 18 Ball P, Make B. Acute exacerbations of chronic bronchitis: an international comparison. Chest 1998; 113(suppl):199S 204S 19 Leeper KV, Jones AM, Tillotson G. The changing bacterial etiology of chronic obstructive pulmonary disease [abstract]. Chest 1997; 112(suppl):21S 20 Doern GV, Brueggeman AB, Pierce G, et al. Antibiotic resistance among clinical isolates of Haemophilus Influenzae in the United States in 1994 and 1995 and detection of B-lactamase-positive strains resist to amoxicillin-clavulanate: results of a national multicenter surveillance study. Antimicrob Agents Chemother 1997; 41:292 297 21 Doern GV, Brueggeman AB, Pierce G, et al. Prevalence of antimicrobial resistance among 723 outpatient clinical isolates Moraxella catarrhalis in the United States in 1994 and 1995: results of a 30-center national surveillance study. Antimicrob Agents Chemother 1996; 40:2884 2886 22 Doern GV, Brueggeman A, Holley Jr PH, et al. Antimicrobial resistance of Streptococcus pneumoniae recovered from outpatients in the United States during the winter months of 1994 to 1995: results of a 30-center national surveillance study. Antimicrob Agents Chemother 1996; 40: 1208 1213 23 Smith CB, Golden CA, Kanner RE, et al. Association of viral and Mycoplasma Pneumoniae Infections with Acute Respiratory Illness in Patients with Chronic Obstructive Pulmonary Disease. Am Rev Respir Dis 1980; 121:225 232 CHEST / 117 / 3/ MARCH, 2000 671