Comparison of First-Line With Second- Line Antibiotics for Acute Exacerbations of Chronic Bronchitis* A Metaanalysis of Randomized Controlled Trials

Transcription

1 CHEST Comparison of First-Line With Second- Line Antibiotics for Acute Exacerbations of Chronic Bronchitis* A Metaanalysis of Randomized Controlled Trials Original Research George Dimopoulos, MD, FCCP; Ilias I. Siempos, MD; Ioanna P. Korbila, MD; Katerina G. Manta, MD; and Matthew E. Falagas, MD, MSc, DSc COPD Background: Although acute exacerbations of chronic bronchitis (AECBs) are common, there has been no metaanalysis that focused on the optimum regimen. Methods: To evaluate the comparative effectiveness and safety of first-line antimicrobial agents (ie, amoxicillin, ampicillin, pivampicillin, trimethoprim/sulfamethoxazole, and doxycycline) and second-line antimicrobial agents (ie, amoxicillin/clavulanic acid, macrolides, second-generation or third-generation cephalosporins, and quinolones) for the treatment of patients with AECB, in an era of increasing antimicrobial resistance among the microbes responsible for AECB, we performed a metaanalysis of randomized controlled trials (RCTs) retrieved through searches of the PubMed and the Cochrane databases. Results: Twelve RCTs were included in the metaanalysis. First-line antibiotics were associated with lower treatment success compared to second-line antibiotics in the clinically evaluable patients (odds ratio [OR], 0.51; 95% confidence interval [CI], 0.34 to 0.75). There were no differences among the compared regimens regarding mortality (OR, 0.64; 95% CI, 0.25 to 1.66) or treatment success (OR, 0.56; 95% CI, 0.22 to 1.43) in microbiologically evaluable patients, or adverse effects in general (OR, 0.75; 95% CI, 0.39 to 1.45) or diarrhea in particular (OR, 1.58; 95% CI, 0.74 to 3.35). Conclusions: Compared to first-line antibiotics, second-line antibiotics are more effective, but not less safe, when administered to patients with AECB. The available data did not allow for stratified analyses according to the presence of risk factors for poor outcome, such as increased age, impaired lung function, airway obstruction, and frequency of exacerbations; this fact should be taken into consideration when interpreting the findings of this metaanalysis. (CHEST 2007; 132: ) Key words: amoxicillin/clavulanic acid; COPD; macrolides; penicillins; quinolones; Streptococcus pneumoniae Abbreviations: AECB acute exacerbation of chronic bronchitis; CB chronic bronchitis; CE clinically evaluable; CI confidence interval; FEM fixed-effects model; ITT intention to treat; ME microbiologically evaluable; OR odds ratio; RCT randomized controlled trial; REM random-effects model; SMX sulfamethoxazole; TMP trimethoprim *From the Critical Care Department (Dr. Dimopoulos), Attikon University Hospital, Athens, Greece; and Alfa Institute of Biomedical Sciences (Drs. Siempos, Korbila, Manta, and Falagas), Athens, Greece. The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Manuscript received January 24, 2007; revision accepted April 16, Patients with chronic bronchitis (CB) consume a large number of antimicrobial agents for the management of exacerbations of their disease. 1,2 An antimicrobial agent administered for the treatment of acute exacerbations of CB (AECBs) should have significant in vitro activity against the pathogens more frequently implicated in AECBs (ie, Haemophilus influenzae, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: Matthew E. Falagas, MD, MSc, DSc, Alfa Institute of Biomedical Sciences (AIBS), 9 Neapoleos St, Marousi, Greece; m.falagas@aibs.gr DOI: /chest CHEST / 132 / 2/ AUGUST,

2 Streptococcus pneumoniae, and Moraxella catarrhalis), good penetration into sputum and bronchial mucosa, as well as minimal adverse effects. 3 5 Adherence to therapy and cost-effectiveness 6,7 should also be taken into consideration. Traditionally, amoxicillin, ampicillin, doxycycline, or trimethoprim (TMP)/sulfamethoxazole (SMX) have been considered for the treatment of patients with AECBs. However, given the increasing resistance of S pneumoniae and H influenzae to these older antimicrobial agents, the authors of the Canadian guidelines for the management of AECBs 8 stated that therapy with selected second-generation or third-generation cephalosporins or macrolides may be preferable for this patient population. In addition, in light of the recognition that relatively resistant bacterial species, including P aeruginosa, may be the etiologic agents of AECBs, especially among patients with significant impairment of lung function, 9 some investigators 10,11 have advocated the administration of even more potent broad-spectrum antibiotics (eg, quinolones) for the treatment of, at least, this subgroup of CB patients. As the accumulated evidence regarding the emerging antimicrobial resistance among the microbes responsible for AECBs have forced a reexamination of antimicrobial choices, we sought to evaluate the potential superiority, if there is one, of the second-line antimicrobial agents compared to the first-line antimicrobial agents regarding effectiveness and safety for the treatment of patients with AECBs. Literature Search Materials and Methods To identify the randomized controlled trials (RCTs) that were appropriate for inclusion in the metaanalysis, we conducted a systematic search of the PubMed and the Cochrane Central Register of Controlled Trials. We used the following search terms: acute exacerbations ; COPD ; and chronic bronchitis combined with amoxicillin, clavulanic acid, ampicillin, sulfamethoxazole, trimethoprim, doxycycline, quinolones, macrolides, Haemophilus influenzae, Moraxella catarrhalis, and Streptococcus pneumoniae. The reference lists of all retrieved articles were reviewed as well. The search was limited to articles written in English, French, and German. Study Selection and Data Extraction Two reviewers (IPK and KGM) independently searched the aforementioned databases to identify potentially eligible trials. To be included in our analysis, a trial had to be an RCT, and to focus on the comparison of the effectiveness and/or toxicity between a first-line antibiotic and a second-line (advanced) antibiotic. From each of the selected studies, the following data were extracted: year of publication; study design; patient population; dosage and duration of antimicrobial treatment as well as treatment success in the intention-to-treat (ITT), the clinically evaluable (CE), and the microbiologically evaluable (ME) patients; all-cause mortality; and toxicity (ie, any adverse effect, diarrhea, and the number of patients who were withdrawn from the RCTs due to drug-related adverse effects). In addition, the methodological quality of the RCTs included in the analysis was assessed according to a modified Jadad score. 12 In detail, the existence of randomization, blinding, and information on withdrawals in each RCT was examined, and the appropriateness of randomization and blinding, if present, was evaluated. One point was awarded for the presence of each of the first three criteria, whereas the last two criteria could take the values of 1 (inappropriate), 0 (no data), and 1 (appropriate). Thus, the maximum score for a study was 5, and a score 3 points denoted a good-quality RCT. Definitions First-Line and Second-Line Antibiotics: Amoxicillin, ampicillin, pivampicillin, TMP/SMX, and doxycycline were considered to be first-line antibiotics for the management of patients with AECB. On the other hand, amoxicillin/clavulanic acid, macrolides (ie, roxithromycin, clarithromycin, and azithromycin), second-generation or third-generation cephalosporins (ie, cefaclor), and quinolones were considered to be advanced or second-line antibiotics for this indication according to published guidelines. 8,13 CB and AECB: The diagnosis of CB in all RCTs included in this metaanalysis was based on the history of cough and expectoration on most days during a period of at least 3 consecutive months for 2 consecutive years, as the authors stated. We classified AECBs according to the criteria of Anthonisen et al, 14 based on the information provided by the investigators. The exacerbation classified as type I by Anthonisen et al 14 met all three of the following criteria: increases in amount of sputum; purulence of sputum; and dyspnea. Type II met two of the above three criteria, and type III met only one criterion. 14 Outcomes of This Metaanalysis: Treatment success (defined as remission of all baseline symptoms of acute infection [clinical cure] or amelioration of symptoms without their complete disappearance [improvement]) in both ITT and CE patients and adverse effects probably or possibly related to study antibiotics were considered as outcome measures for this metaanalysis. In addition, all-cause mortality in the ITT population during the study period, the number of patients who experienced diarrhea or were withdrawn from the RCTs due to drug-related adverse effects, treatment success in ME patients (ie, the absence of a baseline pathogen or the absence of adequate culturable material from a patient exhibiting clinical cure or improvement), and pathogen eradication (documented or presumed) of H influenzae, M catarrhalis, and S pneumoniae isolates were also regarded as outcomes for this metaanalysis. Statistical Analysis Statistical analyses were performed using a statistical software package (S-PLUS, version 6.1; Insightful Corp; Seattle, WA). Using a 2 test, we assessed the heterogeneity among RCTs. Publication bias was assessed by the funnel plot method using the test of Egger et al. 15 Pooled odds ratios (ORs) and 95% confidence intervals (CIs) for all primary and secondary outcomes were calculated by using both the Mantel-Haenszel fixed effects model (FEM) 16 and the DerSimonian-Laird 17 random-effects models (REMs). For all analyses, results from the FEM are presented only when there was no statistically significant heterogeneity among RCTs; otherwise, results from the REM are presented. Selected Studies Results This metaanalysis was performed in accordance with the QUOROM statement. 18 Figure 1 displays 448 Original Research

3 Potentially relevant articles identified in PubMed and the Cochrane central register of randomized controlled trials (RCTs) (n=177) RCTs retrieved for further evaluation (n=124) RCTs regarding the comparison of different antibiotics used in AECB (n=97) 12 RCTs included in the present meta-analysis Articles excluded because they: were review articles (n=5) or provided duplicated data (n=4) or were not randomized trials (n=33) were written in Japanese (n=1) or Spanish (n=3) or Norwegian (n=1) or Romanian (n=1) or Swedish (n=1) or Russian (n=1) compared antibiotics with aminophylline (n=1) or triamphenicol (n=1) or supristol (n=1) RCTs that did not meet inclusion criteria because: regarded cost/effectiveness issues (n=1) regarded the prevention of AECB (n=5) regarded respiratory tract infections other than AECB (n=5) were placebo-controlled RCTs (n=5) regarded the comparison of the same antibiotic in different dosages or mode of administration (n=9) included paediatric patients (n=1) or were cross-over studies (n=1) RCTs excluded because they: regarded antibiotics not used in AECB (n=35) regarded the comparison between 2 penicillins (n=10) or between TMP/SMX and cotrifamole (n=1) regarded the comparison between a firstgeneration cephalosporin and a penicillin (n=6),or between penicillins and TMP/SMX (n=5), between tetracyclines and penicillins (n=2) or between teracyclines and TMP/SMX (n=1) regarded the comparison between amoxicillin/clavulanic acid and quinolones (n=5), amoxicillin/clavulanic acid and macrolides (n=8 )or between amoxicillin/ clavulanic acid and second- or thirdgeneration cephalosporins (n=12) Abbreviations: AECB: acute exacerbation of chronic bronchitis, TMP/SMX: trimethoprim/sulfamethoxazole Figure 1. Flow diagram. the process of identifying the relevant articles. Of the 177 articles that were initially retrieved, 80 were excluded due to reasons presented in Figure 1. In addition, 35 RCTs were omitted because they regarded the comparison of antibiotics that are not commonly used today or were withdrawn from the market for the treatment of AECBs. Also, 25 RCTs were excluded because they compared first-line antibiotics, and another 25 RCTs were excluded because they compared advanced antibiotics. At last, 12 RCTs that compared pivampicillin with amoxicillin/clavulanic acid (n 1), 28 TMP/SMX with cefaclor (n 1), 30 penicillins with macrolides (n 5), 19,23 26 quinolones (n 2), 20,21 or cephalo- CHEST / 132 / 2/ AUGUST,

4 sporins (n 2), 27,29 and doxycycline with roxithromycin (n 1) 22 were eligible for evaluation. The main characteristics of the 12 selected RCTs, which enrolled 2,261 adult patients, are presented in Table 1. The mean quality score of these RCTs was 3.3 (range, 1 to 5); 10 of them 19 22,24 26,28 30 had a quality score of 3. Nine RCTs 20 22,24 27,29,30 were double-blinded, and three RCTs 19,23,28 were singleblinded. Most RCTs included a mixed population of both inpatients and outpatients. Information on the use of systemic corticosteroids before AECBs was given in three trials, 21,22,26 while demographic information (ie, mean age and sex) related to the patients included in the eligible RCTs is depicted in Table 1. Treatment Success and Mortality Data regarding treatment success in ITT patients were provided in only 2 RCTs 24,30 of the 12 eligible RCTs, as shown in Table 2. Data on treatment success for the CE patients were available for all 12 selected RCTs First-line antibiotics were less effective than second-line antibiotics (treatment success, 1,145 CE patients; OR, 0.51; 95% CI, 0.34 to 0.75; FEM). The ORs for the treatment success of the compared antibiotics for the CE patients in the individual RCTs, as well as the pooled ORs, are presented in Figure 2, top left, A. The result did not substantially change when 10 high-quality RCTs 19 22,24 26,28 30 were put in a sensitivity analysis (997 CE patients; OR, 0.46; 95% CI, 0.30 to 0.70; FEM) as well as after the exclusion of the trial that provided the larger number of patients included in this metaanalysis 21 (864 CE patients; OR, 0.53; 95% CI, 0.33 to 0.84; FEM). This was also the case for the subgroup analyses that included only the double-blinded RCTs (1,465 CE patients; OR, 0.57; 95% CI, 0.37 to 0.89; FEM [data from nine trials 20 22,24 27,29,30 ]), only the single-blinded RCTs (796 CE patients; OR, 0.36; 95% CI, 0.16 to 0.81 [data from three trials 19,23,28 ]), or the RCTs published after 1991 (1,287 CE patients; OR, 0.46; 95% CI, 0.29 to 0.73 [data from six trials ]), but not for the analysis of the RCTs published before 1991 (974 CE patients; OR, 0.80; 95% CI, 0.38 to 1.69 [data from six trials ]). The analysis of the data on all-cause mortality during the study period showed no differences among the regimens compared (1,392 ITT patients; OR, 0.64; 95% CI, 0.25 to 1.66; FEM [data from five trials 21,26 28,30 ]) [Fig 2, top right, B]. Adverse Effects Regarding the drug-related adverse events (1,619 ITT patients; OR, 0.75; 95% CI, 0.39 to 1.45; REM [data from eight trials 19 23,25,26,30 ] [Fig 2, bottom left, C]) and study withdrawal due to drug-related adverse effects (1,166 ITT patients; OR, 0.61; 95% CI, 0.35 to 1.06; FEM [data from five trials 19 22,30 ]), there were no differences among the studied regimens. Neither were there any differences among patients receiving first-line or second-line antibiotics concerning diarrhea (554 ITT patients; OR, 1.58; 95% CI, 0.74 to 3.35; FEM [data from five trials 19,23,25,27,30 ]). Adverse effects other than diarrhea commonly caused by first-line agents were nausea (four trials 19,21 23 ), abdominal pain, vomiting, disorders of the CNS, and skin rashes (in one trial each). On the other hand, second-line agents were frequently associated with nausea (four trials 20,21,23,25 ), abdominal pain (two trials 19,22 ), vomiting, insomnia, and dizziness (in one trial each). Microbiological Outcomes Table 3 depicts the microbiological outcomes of nine RCTs 19 23,25 27,30 that provided relevant data. There were no differences among the groups compared with regard to treatment success in the ME populations (ie, in the proportion of patients from whom the causative pathogen was not isolated after treatment among those patients from whom this pathogen was initially isolated; 638 ME patients; OR, 0.56; 95% CI, 0.22 to 1.43; REM) [Fig 2, bottom right, D]. Data on the eradication rates of the three most common pathogens isolated at baseline (ie, H influenzae, M catarrhalis, and S pneumoniae) were reported in seven of the eligible RCTs. 20,21,23 26,30 Treating patients with first-line antibiotics as opposed to second-line antibiotics was not associated with lower eradication rates of H influenzae (241 isolates; OR, 0.62; 95% CI, 0.29 to 1.29; FEM), M catarrhalis (98 isolates; OR, 0.33; 95% CI, 0.07 to 1.53; FEM), or S pneumoniae (137 isolates; OR, 1.86; 95% CI, 0.67 to 5.20; FEM). Subgroup Analyses We analyzed treatment success in CE patients in trials comparing penicillins (ie, amoxicillin, ampicillin, and pivampicillin) to second-line antibiotics (ie, macrolides, quinolones, amoxicillin/clavulanic acid, and cephalosporins); penicillins were less effective than comparators (943 CE patients; OR, 0.46; 95% CI, 0.30 to 0.72; FEM [data from nine trials 19 21,23 27,29 ]). Penicillins also proved to be associated with lower treatment success in the ME population (610 ME patients; OR, 0.35; 95% CI, 0.20 to 0.60; FEM [data from seven trials 19 21,23,25 27 ]). With regard to mortality 21,26 27 as well as the drug-related adverse events in general, 19 21,23,25 26 and diarrhea in particular, 19,23,25,27 and withdrawals due to adverse events, no differences were found among the groups compared in this subgroup analysis (data not shown). The treatment of patients with penicillins was not associated 450 Original Research

5 Table 1 Main Characteristics of RCTs Included in the Metaanalysis* Study/Year/Country Study Design Study Population Mean Age, yr/ Proportion of Women Among Enrolled Patients, % Regimen (Duration of Treatment) Use of Systemic Corticosteroid 1 2 Before AECB Enrolled Patients, No. ITT Patients, No. Study Quality Score Schouenbourg et al 19 /2000/ Denmark Chodosh 20 /1993/ United States Ulmer 21 /1993/ Germany De Vlieger et al 22 / 1992/ Belgium Guay and Craft 23 / 1992/United States Mertens et al 24 /1992/ the Netherlands Aldons 25 /1991/ Australia Bachand 26 /1991/ United States Trigg et al 27 /1991/ United Kingdom McGhie et al 28 /1986/ United Kingdom Law et al 29 /1983/ United Kingdom Anderson et al 30 / 1981/United Kingdom MC, SB, RCT Outpatients 18 yr old with CB and an Anthonisen type II or III AECB MC, DB, RCT Hospitalized patients and outpatients 18 yr old with CB and an Anthonisen type I AECB MC, DB, RCT Hospitalized patients and outpatients 18 yr old with CB and an Anthonisen type II AECB MC, DB, RCT Patients 18 yr old with CB and an Anthonisen type II AECB MC, SB, RCT Outpatients 18 yr old with CB and an Anthonisen type II or III AECB DB, RCT Hospitalized patients and outpatients (adults) with CB and an Anthonisen type I, II AECB MC, DB, RCT Outpatients 18 yr old with CB and an Anthonisen type I, II or III AECB MC, DB, RCT Outpatients 18 yr old with COPD, asthma, emphysema, bronchiectasis, and AECB DB, RCT Hospitalized patients 46 yr old with COPD and an Anthonisen type II AECB SB, RCT Patients 18 yr old with CB and AECB DB, RCT Hospitalized patients and outpatients 55 yr old with CB and an Anthonisen type II AECB DB, RCT Hospitalized patients 48 yr old with CB and AECB 61 vs 59/49 vs 49 Pivampicillin, 700 mg q12h (10 d) 65 vs 65/23 vs 41 Amoxicillin, 500 mg 63 vs 63/34 vs 35 Amoxicillin, 500 mg 58 vs 62/48 vs 37 Doxycycline, 200 mg q24h (7 14 d) 54 vs 58/46 vs 42 Ampicillin, 500 mg q6h (up to 14 d) 58 vs 65/56 vs 79 Amoxicillin, 500 mg q8h (5 d) 58 vs 60/43 vs 45 Ampicillin, 250 mg q6h (7 14 d) 56 vs 55/42 vs 41 Ampicillin, 250 mg q6h (10 d) 71 vs 68/79 vs 30 Amoxicillin, 500 mg 49 vs 52/50 vs 43 Pivampicillin, 250 mg, plus pivmecillinam, 200 mg q12h (10 d) 66 vs 65/20 vs 15 Amoxicillin, 500 mg q8h (10 d) 66 vs 64/16 vs 25 TMP, 80 mg/smx, 400 mg q24h (7 d) Azithromycin, 500 mg q24h (3 d) Fleroxacin, 400 mg q24h (7 d) Fleroxacin, 400 mg q24h (7 d) Roxithromycin, 300 mg q24h (7 14 d) Clarithromycin, 500 mg q12h (maximum of 14 d) Azithromycin, 500 mgq24h(3d) Clarithromycin, 250 mg q12h (7 14 d) Clarithromycin, 250 mg q12h (10 d) Cefaclor, 500 mg Amoxicillin, 250 mg, plus clavulanic acid, 125 mg q8h (10 d) Cefaclor, 500 mg q8h (10 d) Cefaclor, 500 mg NA vs NA vs Permitted vs Permitted vs 41 4 NA vs 53 1 NA vs 25 4 NA vs 60 5 Permitted vs NA vs 26 2 NA vs NA vs 40 3 NA vs 20 3 *All study antibiotics were administered orally. The use of antimicrobial agents other than those studied was not allowed. MC multicenter; SB single-blind; DB double-blind; NA not applicable. Values are reported for patients receiving first-line antimicrobial agents vs patients receiving second-line antimicrobial agents. According to a modified Jadad score. The data did not allow us to classify AECBs according to the Anthonisen criteria. CHEST / 132 / 2/ AUGUST,

6 Table 2 Outcome Data From the Selected RCTs for the Metaanalysis* 1 Treatment Success Adverse Effects ITT at TOCV CE at TOCV All-Cause Mortality Total Withdrawn Patients Diarrhea Study/Year NA 79/92 (86) vs 124/133 (93) NA 22/98 (22) vs 20/138 (14) 5/98 (5) vs 3/138 (2) 11/98 (11) vs 7/138 (5) Schouenbourg et al 19 / 2000 Chodosh 20 /1993 NA 22/29 (76) vs 21/22 (95) NA 14/92 (15) vs 42/102 (41) 5/92 (5) vs 13/102 (13) NA Ulmer 21 /1993 NA 111/136 (81) vs 131/145 (89) 3/310 (1) vs 3/313 (1) 27/310 (9) vs 61/313 (19) 7/310 (2) vs 20/313 (6) NA De Vlieger et al 22 /1992 NA 20/25 (80) vs 30/37 (81) NA 7/33 (21) vs 4/41 (10) 3/33 (9) vs 0/41 (0) NA Guay and Craft 23 /1992 NA 46/47 (98) vs 53/53 (100) NA 10/50 (20) vs 8/53 (15) NA 3/50 (6) vs 3/53 (6) Mertens et al 24 / /25 (80) vs 24/25 (96) 20/25 (80) vs 24/25 (96) NA NA NA NA Aldons 25 /1991 NA 21/23 (91) vs 27/28 (96) NA 1/65 (2) vs 7/60 (12) 1/65 (2) vs NA 0/65 (0) vs 0/60 (0) Bachand 26 /1991 NA 31/34 (91) vs 28/29 (97) 1/114 (1) vs 1/111 (1) 36/114 (32) vs 34/111 (31) NA NA Trigg et al 27 /1991 NA 13/25 (52) vs 11/23 (48) 2/25 (8) vs 4/26 (15) NA NA 1/25 (4) vs 2/26 (8) McGhie et al 28 /1986 NA 51/55 (93) vs 49/49 (100) 0/230 (0) vs 2/224 (1) NA NA NA Law et al 29 /1983 NA 33/36 (92) vs 36/38 (95) NA NA NA NA Anderson et al 30 / /19 (79) vs 15/20 (75) 15/17 (88) vs 15/19 (80) 1/19 (5) vs 1/20 (5) 0/19 (0) vs 0/20 (0) 0/19 (0) vs 0/20 (0) 0/19 (0) vs 0/20 (0) *Values are given as the No. of patients treated/total No. of patients in the group (%) for patients receiving first-line antimicrobial agents vs those receiving second-line antimicrobial agents. TOCV test-of-cure visit (1 to 7 days after the end of treatment). See Table 1 for abbreviation not used in the text. with lower eradication rates of H influenzae, M catarrhalis, or S pneumoniae 20 21,23 26 (data not shown). In another subgroup analysis of trials comparing penicillins to macrolides, treatment success in CE patients was accomplished less commonly by penicillins (489 CE patients; OR, 0.36; 95% CI, 0.17 to 0.75; FEM [data from five trials 19,23 26 ]). In this subset of patients, there were no differences regarding any adverse effects, 19,23,25 26 or diarrhea specifically, 19,23,25 treatment success in ME patients, 19,23,25 26 or eradication rates of H influenzae, M catarrhalis, or S pneumoniae (data not shown). Finally, two subgroup analyses were conducted by including the six RCTs that enrolled hospitalized patients 20,21,24,27,29,30 and the four RCTs that enrolled only ambulatory patients. 19,23,25,26 Again, the effectiveness of old antimicrobial agents (540 CE patients; OR, 0.55; 95% CI, 0.33 to 0.90; FEM [data from six trials 20,21,24,27,29,30 ]) as opposed to advanced antimicrobial agents (439 CE patients; OR, 0.41; 95% CI, 0.19 to 0.90; REM [data from four trials 19,23,25,26 ]) for the treatment of CE patients with AECBs was found to be lower in both of these analyses. Discussion The results of this metaanalysis suggest that the administration of first-line antimicrobial agents to CE patients with AECBs was associated with lower treatment success compared to the administration of second-line agents. However, no differences were found among the compared regimens with regard to mortality, microbiological outcomes (ie, treatment success in ME patients and pathogen eradication of most common pathogens in patients with AECBs), and safety (ie, drug-related adverse effects in general, diarrhea, or study withdrawals due to adverse effects). The observed clinical advantage of the second-line antibiotics should be very carefully interpreted in light of issues related to the design of the RCTs included in this metaanalysis, concerning mainly the characteristics of the patients who were included in the RCTs. Specifically, the selected RCTs did not provide adequate clinical information on whether the enrolled AECB patients had or did not have risk factors for poor outcome (ie, age 65 years, FEV 1 50% predicted at baseline, more than three AECBs in the previous year, or comorbid illness [especially cardiac disease]). 8 In patients with the above characteristics, aggressive therapy with second-line antimicrobial agents may be more appropri- 452 Original Research

7 Figure 2. ORs of the outcomes of individual RCTs comparing first-line and second-line antibiotics for the treatment of patients with AECBs and the pooled analysis. Top left, A: treatment success in CE patients. Top right, B: all-cause mortality. Bottom left, C: total adverse effects. Bottom right, D: treatment success in ME patients. Vertical line no difference point between the two regimens; square OR (the size of each square denotes the proportion of information given by each trial); diamond pooled ORs for all RCTs; horizontal lines 95% CI. ate than therapy with first-line agents. 31,32 Note that at least half of the selected RCTs 20,21,24,27,29,30 enrolled hospitalized patients along with or without outpatients. It is more likely that inpatients experience a higher degree of pulmonary impairment and comorbidities, and, thus, their management requires the administration of advanced antibiotics. It is interesting that the superiority of the advanced antimicrobial agents over the older agents regarding treatment success in CE patients with AECBs was maintained in the various subanalyses that we performed. In detail, advanced antibiotics were found to have higher treatment success than comparators even after the exclusion of the RCT 21 that provided the largest number of patients included in this metaanalysis. This was also the case in the sensitivity analysis of RCTs with high methodological quality scores 19 22,24 26,28 30 as well as in the subgroup analyses that included only the double-blinded RCTs 20 22,24 27,29,30 or the RCTs published after It is noteworthy that in the analysis of the RCTs published before 1991, no difference was found between advanced regimens and old regimens regarding effectiveness, a finding that is presumably explained by the lack of resistance of pathogens toward both advanced and old antibiotics in that time period. One might postulate that the findings of the present metaanalysis seem to contravene the attempt to preserve new antibiotics and to prevent further development of resistance in an era of increasing incidence of multi-drug-resistant bacteria. However, the scope of our contribution was only to estimate the differences among the administered antimicrobial agents for a group of patients who need antimicrobial therapy. Besides, treatment directed toward resistant pathogens with adequately effective antimicrobial agents is presumably expected to prevent a lack of response to therapy and, therefore, to avoid prolonged hospitalization and repeated courses of antibiotics. 1,33 Finally, although not examined in this study, it seems plausible that the emergence of resistance may be successfully limited by avoidance of the administration of antibiotics in patients with Anthonisen type III (mild) AECBs CHEST / 132 / 2/ AUGUST,

8 Table 3 Microbiological Outcomes From the Selected RCTs for the Metaanalysis* Study/Year Treatment Success (Microbiological Evaluation) Pathogen Eradication H influenzae M catarrhalis S pneumoniae Schouenbourg et al 19 / 32/37 (86) vs 49/54 (91) NA NA NA 2000 Chodosh 20 / /30 (60) vs 21/22 (95) 7/11 (64) vs 10/10 (100) 5/5 (100) vs 6/6 (100) 0/1 (0) vs 1/1 (100) Ulmer 21 / /138 (83) vs 142/148 (96) 31/38 (82) vs 59/59 (100) 20/22 (91) vs 19/19 (100) 31/35 (89) vs 15/19 (79) De Vlieger et al 22 /1992 6/7 (86) vs 5/5 (100) NA NA NA Guay and Craft 23 / /12 (100) vs 23/23 (100) 10/10 (100) vs 10/10 (100) 11/11 (100) vs 11/11 (100) 20/20 (100) vs 19/19 (100) Mertens et al 24 /1992 NA 7/11 (64) vs 8/12 (67) 2/2 (100) vs 6/6 (100) 6/7 (86) vs 3/4 (75) Aldons 25 / /24 (100) vs 28/29 (96) 17/17 (100) vs 19/20 (95) 1/1 (100) vs 6/6 (100) 6/6 (100) vs 2/2 (100) Bachand 26 / /34 (94) vs 25/29 (86) 18/21 (86) vs 7/11 (64) 1/2 (50) vs 6/7 (86) 7/7 (100) vs 11/11 (100) Trigg et al 27 /1991 5/15 (33) vs 9/15 (60) NA NA NA McGhie et al 28 /1986 NA NA NA NA Law et al 29 /1983 NA NA NA NA Anderson et al 30 /1981 8/8 (100) vs 4/8 (50) 5/5 (100) vs 4/6 (67) NA 3/3 (100) vs 0/2 (0) * See Table 1 for abbreviation not used in the text. Values are given as the No. of patients in whom the specified outcome occurred/total No. of patients in the group (%) for patients receiving first-line antimicrobial agents vs those receiving second-line antimicrobial agents. Values are given as the No. of isolates eradicated/total No. of isolates for each microbe (%) in patients receiving first-line antimicrobial agents vs those receiving second-line antimicrobial agents. This metaanalysis has substantial limitations. First, we acknowledge as a major limitation the fact that the eligible RCTs did not stratify patients according to the existence of the previously outlined risk factors for poor outcome (eg, age, degree of pulmonary impairment and airway obstruction, and frequency of exacerbations), and, moreover, some of them enrolled both inpatients and outpatients. However, no substantially different results were produced regarding treatment effectiveness in the subsets of RCTs that included hospitalized patients or enrolled only ambulatory patients. Second, we grouped different classes of antimicrobial agents (ie, amoxicillin, ampicillin, TMP/SMX, and doxycycline in the group of first-line agents, and amoxicillin/clavulanic acid, macrolides, cefaclor, and quinolones in the group of second-line agents) with different in vitro activity against the most common pathogens that cause AECBs and, thus, possibly with different effectiveness. However, TMP/SMX (which is not recommended any more for the management of patients with AECBs, according to the relevant European guidelines 13 ) and doxycycline were compared to a second-line antibiotic in only one RCT each. Besides, in a very recent metaanalysis 34 of RCTs comparing macrolides, quinolones, and amoxicillin/clavulanic acid for the treatment of patients with AECBs, no difference was revealed among these second-line agents regarding effectiveness. Third, the scarcity of specific data with regard to treatment success in ITT patients did not allow us to pool them; in fact, only two RCTs 24,30 provided such information. Fourth, we did not evaluate other interesting outcomes, such as AECB-free interval and time to recovery, due to lack of available relevant evidence. Fifth, no data were available on the eradication of pathogens other than S pneumoniae, H influenzae, and M catarrhalis, such as Pseudomonas aeruginosa, which may also be the cause of AECBs. Sixth, no information was provided regarding the emergence of Clostridium difficile acquisition in patients receiving antimicrobial agents for the treatment of AECBs. Seventh, the fact that the included RCTs were conducted in different time periods, in which significant changes in the pathogens and the patterns of resistance took place, may also impact the outcomes of this metaanalysis. In an attempt to address this issue, we conducted two subgroup analyses by including the trials published before and after In addition, one may support the idea that a metaanalytic approach other than the present one, comparing penicillins with all other antimicrobial agents for the management of patients with AECBs, would also be of value. Indeed, provided the -lactamase production in H influenzae and M catarrhalis, as well as the high level -lactam resistance in several strains of S pneumoniae, penicillins might not be expected to be as effective as other antibiotics for this purpose. Finally, a comparison of second-line and first-line antibiotics regarding cost-effectiveness was out of the scope of this metaanalysis. In conclusion, second-line antimicrobial agents appear to be more effective than first-line antimicrobial agents for the management of patients with AECBs. This finding provides further support to the 454 Original Research

9 suggestion by some experts that advanced antibiotics are preferable to the old antibiotics for this purpose when the administration of antimicrobial agents is recommended. References 1 Miravitlles M, Murio C, Guerrero T, et al. Pharmacoeconomic evaluation of acute exacerbations of chronic bronchitis and COPD. Chest 2002; 121: Niederman MS, McCombs JS, Unger AN, et al. Treatment cost of acute exacerbations of chronic bronchitis. Clin Ther 1999; 21: Sethi S, Evans N, Grant BJ, et al. New strains of bacteria and exacerbations of chronic obstructive pulmonary disease. N Engl J Med 2002; 347: Patel IS, Seemungal TAR, Wilks M, et al. Relationship between bacterial colonization and the frequency, character, and severity of COPD exacerbations. Thorax 2002; 57: Martinez FJ, Anzueto A. Appropriate outpatient treatment of acute bacterial exacerbations of chronic bronchitis. Am J Med 2005; 118(suppl):39S 44S 6 Adams SG, Anzueto A. Treating acute exacerbations of chronic bronchitis in the face of antibiotic resistance. Cleve Clin J Med 2000; 67: , Anzueto A. Treatment of acute exacerbations of chronic bronchitis: antibiotic therapy. Semin Respir Crit Care Med 2000; 21: Balter MS, La Forge J, Low DE, et al. Canadian guidelines for the management of acute exacerbations of chronic bronchitis. Can Respir J 2003; 10(suppl):3B 32B 9 Miravitlles M, Espinosa C, Fernández-Laso E, et al. Relationship between bacterial flora in sputum and functional impairment in patients with acute exacerbations of COPD. Chest 1999; 116: Sethi S. Moxifloxacin for the treatment of acute exacerbations of chronic obstructive pulmonary disease. Clin Infect Dis 2005; 41(suppl):S177 S Miravitlles M, Torres A. No more equivalence trials for the antibiotics in exacerbations of COPD, please. Chest 2004; 125: Moher D, Jones A, Cook DJ, et al. Does quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses? Lancet 1998; 352: Woodhead M, Blasi F, Ewig S, et al. Guidelines for the management of adult lower respiratory tract infections. Eur Respir J 2005; 26: Anthonisen NR, Manfreda J, Warren CP, et al. Antibiotic therapy in exacerbations of chronic obstructive pulmonary disease. Ann Intern Med 1987; 106: Egger M, Davey SG, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959; 22: DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: Moher D, Cook DJ, Eastwood S, et al. Improving the quality of reports of meta-analyses of randomized controlled trials: the QUOROM statement. Lancet 1999; 354: Schouenborg P, Gerdes N, Rasmussen HH, et al. Azithromycin versus Pivampicillin in the treatment of acute exacerbations of chronic bronchitis: a single-blind, double-dummy, multicenter study. J Int Med Res 2000; 28: Chodosh S. Efficacy of fleroxacin versus amoxicillin in acute exacerbations of chronic bronchitis. Am J Med 1993; 94(suppl):131S 135S 21 Ulmer W. Fleroxacin versus amoxicillin in the treatment of acute exacerbation of chronic bronchitis. Am J Med 1993; 94(suppl):136S 141S 22 De Vlieger A, Druart M, Puttemans M. Roxithromycin versus doxycycline in the treatment of acute exacerbations of chronic bronchitis. Diagn Microbiol Infect Dis 1992; 15(suppl):123S 127S 23 Guay DR, Craft JC. Comparative safety and efficacy of clarithromycin and ampicillin in the treatment of out-patients with acute bacterial exacerbation of chronic bronchitis. J Intern Med 1992; 231: Mertens JC, van Barneveld PW, Asin HR, et al. Double-blind randomised study comparing the efficacies and safeties of a short (3-day) course of azithromycin and a 5-day course of amoxicillin in patients with acute exacerbations of chronic bronchitis. Antimicrob Agents Chemother 1992; 36: Aldons PM. A comparison of clarithromycin with ampicillin in the treatment of outpatients with acute bacterial exacerbation of chronic bronchitis. J Antimicrob Chemother 1991; 27(suppl): Bachand RT Jr. Comparative study of clarithromycin and ampicillin in the treatment of patients with acute bacterial exacerbations of chronic bronchitis. J Antimicrob Chemother 1991; 27(suppl): Trigg CJ, Wilks M, Herdman MJ, et al. A double-blind comparison of the effects of cefaclor and amoxycillin on respiratory tract and oropharyngeal flora and clinical response in acute exacerbations of bronchitis. Respir Med 1991; 85: McGhie D, Kerfoot P, Talbot DJ. A comparative study of Miraxid (pivmecillinam plus pivampicillin) and Augmentin (amoxycillin plus clavulanic acid) in the treatment of lower respiratory tract infections in general practice. J Int Med Res 1986; 14: Law MR, Holt HA, Reeves DS, et al. Cefaclor and amoxycillin in the treatment of infective exacerbations of chronic bronchitis. J Antimicrob Chemother 1983; 11: Anderson G, Williams L, Pardoe T, et al. Co-trimoxazole versus cefaclor in acute on chronic bronchitis. J Antimicrob Chemother 1981; 8: Wilson R, Jones P, Schaberg T, et al. Antibiotic treatment and factors influencing short and long term outcomes of acute exacerbations of chronic bronchitis. Thorax 2006; 61: Wilson R, Allegra L, Huchon G, et al. Short-term and long-term outcomes of moxifloxacin compared to standard antibiotic treatment in acute exacerbations of chronic bronchitis. Chest 2004; 125: Sportel JH, Koeter GH, van Altena R, et al. Relation between beta-lactamase producing bacteria and patient characteristics in chronic obstructive pulmonary disease (COPD). Thorax 1995; 50: Siempos II, Dimopoulos G, Korbila IP, et al. Macrolides, quinolones, and amoxicillin/clavulanate for chronic bronchitis: a meta-analysis. Eur Respir J 2007; 29: CHEST / 132 / 2/ AUGUST,