Variables associated with recovery from acute exacerbations of chronic bronchitis and chronic...

11
Respiratory Medicine (2005) 99, 955965 Variables associated with recovery from acute exacerbations of chronic bronchitis and chronic obstructive pulmonary disease $ Marc Miravitlles a, , Carles Llor b , Karlos Naberan c , Josep Marı ´a Cots b , Jesu´sMolina d , for the EFEMAP study group a Pneumology Department, Institut Clı ´nic del To `rax (IDIBAPS), Red Respira FIS-ISCIII-RTIC-03/11, Hospital Clı ´nic, Villarroel 170, 08036 Barcelona, Catalonia, Spain b Infectious Diseases Group of the Catalan Society of Family Medicine, Spain c Pneumology Group of the Catalan Society of Family Medicine, Spain d Member of the Respiratory Group in Primary Care (GRAP), Spain Received 31 August 2004 Summary Objective: To identify risk factors for late recovery and failure after ambulatory treatment of exacerbations of chronic bronchitis (CB) and chronic obstructive pulmonary disease (COPD). Methods: Observational, non-randomised study of risk factors carried out in 2001 and 2002 in Primary Care practices. Patients aged 40 or older diagnosed with an exacerbation of CB or COPD of probable bacterial etiology were included in the study and followed up for 10 days. Patients were treated with amoxicillin plus clavulanic acid (co-amoxiclav) 500125 mg tds for 10 days, clarithromycin 500 mg bd for 10 days or moxifloxacin 400 mg od for 5 days. Results: Two hundred and fifty-two general practitioners participated, registering 1147 valid patients. The rate of failure at day 10 was 15.1% without significant differences among the antibiotic treatments. Median time to recovery was 5 days. Factors significantly associated with late recovery (45 days) on multivariate analysis were: use of long-term oxygen (OR ¼ 1.96; 95%CI ¼ 1.352.85); use of short-acting beta-2 agonists (OR ¼ 1.51; 1.171.92). The use of moxifloxacin had a ‘‘protective’’ effect against late recovery compared to co-amoxiclav (OR ¼ 0.34; 0.260.45) and clarithromycin (OR ¼ 0.41; 0.312.85). Factors associated with therapeutic failure were: previous hospitalisation (OR ¼ 1.61; 1.082.42); and 2 or more exacerbations ARTICLE IN PRESS KEYWORDS Chronic bronchitis; COPD; Exacerbations; Treatment failure 0954-6111/$ - see front matter & 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.rmed.2005.01.013 $ The list of investigators participating in the EFEMAP study is included in the appendix. The EFEMAP study was funded by Bayer Spain. Corresponding author. Tel./fax: +34 932275549. E-mail address: [email protected] (M. Miravitlles).

Transcript of Variables associated with recovery from acute exacerbations of chronic bronchitis and chronic...

ARTICLE IN PRESS

Respiratory Medicine (2005) 99, 955–965

KEYWORDChronic brCOPD;ExacerbatiTreatment

0954-6111/$ - sdoi:10.1016/j.r

$The list ofSpain.�CorrespondiE-mail addr

Variables associated with recovery from acuteexacerbations of chronic bronchitis and chronicobstructive pulmonary disease$

Marc Miravitllesa,�, Carles Llorb, Karlos Naberanc, Josep Marıa Cotsb,Jesus Molinad, for the EFEMAP study group

aPneumology Department, Institut Clınic del Torax (IDIBAPS), Red Respira FIS-ISCIII-RTIC-03/11,Hospital Clınic, Villarroel 170, 08036 Barcelona, Catalonia, SpainbInfectious Diseases Group of the Catalan Society of Family Medicine, SpaincPneumology Group of the Catalan Society of Family Medicine, SpaindMember of the Respiratory Group in Primary Care (GRAP), Spain

Received 31 August 2004

Sonchitis;

ons;failure

ee front matter & 2005med.2005.01.013

investigators participat

ng author. Tel./fax: +3ess: [email protected]

SummaryObjective: To identify risk factors for late recovery and failure after ambulatorytreatment of exacerbations of chronic bronchitis (CB) and chronic obstructivepulmonary disease (COPD).Methods: Observational, non-randomised study of risk factors carried out in 2001and 2002 in Primary Care practices. Patients aged 40 or older diagnosed with anexacerbation of CB or COPD of probable bacterial etiology were included in the studyand followed up for 10 days. Patients were treated with amoxicillin plus clavulanicacid (co-amoxiclav) 500–125mg tds for 10 days, clarithromycin 500mg bd for 10 daysor moxifloxacin 400mg od for 5 days.Results: Two hundred and fifty-two general practitioners participated, registering1147 valid patients. The rate of failure at day 10 was 15.1% without significantdifferences among the antibiotic treatments. Median time to recovery was 5 days.Factors significantly associated with late recovery (45 days) on multivariate analysiswere: use of long-term oxygen (OR ¼ 1.96; 95%CI ¼ 1.35–2.85); use of short-actingbeta-2 agonists (OR ¼ 1.51; 1.17–1.92). The use of moxifloxacin had a ‘‘protective’’effect against late recovery compared to co-amoxiclav (OR ¼ 0.34; 0.26–0.45) andclarithromycin (OR ¼ 0.41; 0.31–2.85). Factors associated with therapeutic failurewere: previous hospitalisation (OR ¼ 1.61; 1.08–2.42); and 2 or more exacerbations

Elsevier Ltd. All rights reserved.

ing in the EFEMAP study is included in the appendix. The EFEMAP study was funded by Bayer

4 93 2275549.s (M. Miravitlles).

ARTICLE IN PRESS

M. Miravitlles et al.956

the previous year (OR ¼ 1.51; 1.04–2.17); criteria of CB had a protective effectagainst failure (OR ¼ 0.53; 0.35–0.79).Conclusions: There are readily identifiable risk factors for ambulatory treatmentfailure of exacerbations of CB and COPD. In addition, long-term oxygen therapy andshort-acting beta-2 agonists are associated with late recovery, and the use ofmoxifloxacin compared with co-amoxiclav and clarithromycin is associated withfaster recovery of symptoms.& 2005 Elsevier Ltd. All rights reserved.

Introduction

Chronic pulmonary disease, particularly chronicbronchitis (CB) and chronic obstructive pulmonarydisease (COPD), constitutes one of the principaldemands on health care in primary care. Accordingto recent studies, it is estimated that in Spainup to 8–10% of the population over 40 yearsof age may have COPD1,2 and, in men over 65years of age, this figure may rise to 20%.2 Thecourse of both CB and COPD can be affected bycrises of worsening of the respiratory symptoms,which are known as exacerbations. According toreported findings, a high percentage of theseexacerbations, between 50% and 70%, are causedby bacteria.3

Studies performed in the community have shownthat patients with COPD have a mean of twoexacerbations per year, 90% of which are treatedwith antibiotics.4 However, the failure rate of anexacerbation after ambulatory treatment may beas high as 20–25%.5–7 Some recent guidelines havepointed out that the ability to identify patients athigh risk for therapeutic failure should help toguide decisions concerning hospital admissionand follow-up appointments.8 Therefore, studiesaimed at identifying risk factors for treatmentfailure in real life conditions are required. Inaddition, the speed of recovery of symptoms ofexacerbation is important in patients with chronicbronchial disease. Milder patients may return towork or normal activities earlier and more severepatients may avoid respiratory failure if recovery ofsymptoms is achieved in a short period of time.Knowing the time course of symptoms and the riskfactors for late recovery may help physicians toschedule follow-up visits and interpret the out-comes of therapy. In a previous analysis of dataderived from this cohort, we evaluated the clinicalevolution of exacerbations under different anti-biotic regimens.9 The aim of the present study wasto identify which variables of the patients andtreatment are significantly associated with anincreased risk of treatment failure or late recoveryof symptoms.

Method

Patients

Patients elegible for participation in the EFEMAPstudy (Estudio FarmacoEconomico de Moxifloxacinoen las Agudizaciones de la EPOC) had a clinicaldiagnosis of exacerbation of CB and/or COPD ofprobable bacterial aetiology and were aged 40years or older. Diagnosis of an exacerbation ofprobable bacterial aetiology was established whenthe patient presented with a type I or type IIexacerbation as described below. Classification ofexacerbations followed the criteria described byAnthonisen et al.10 increase in the usual level ofdyspnoea, increase in sputum volume, and/orincrease in the purulence of the sputum. Exacer-bations presenting one of these symptoms wereclassified as type III, those with two symptoms astype II and those with all three symptoms as typeI.10 Patients with type III exacerbations were notincluded in the study, since their probability ofbacterial infection is very low and usually do notrequire antibiotic therapy.

To establish the diagnosis of CB, the patientshould refer a productive cough for at least 3months a year for 2 consecutive years; for thediagnosis of COPD, the observation of a non-reversible airflow obstruction was required, char-acterised on forced spirometry by an FEV1 o80% ofthe theoretical value and an FEV1/FVC ratio ofo70% in a stable phase.11 Patients with bronchialasthma, cystic fibrosis, bronchiectasis, malignancy,pneumonia and those patients who fulfilled criteriafor hospitalisation were excluded.

Study design

Full details of the study design have been publishedelsewhere.9,12 This was an observational study of apopulation of patients with exacerbated CB orCOPD being followed in primary care centresdistributed throughout Spain.

ARTICLE IN PRESS

Variables associated with acute exacerbations of CB and COPD 957

After diagnosis of the exacerbation, the partici-pating investigators were free to prescribe any ofthe three following options of antibiotic treatment:association of co-amoxiclav 500mg/8 h for 10 days,moxifloxacin 400mg/day for 5 days, or clarithro-mycin 500mg/12 h for 10 days, in accordance withthe current guidelines.13 A medical appointmentwas made for all the patients 10 days after the firstvisit, when they were asked about the number ofdays which had passed from starting treatmentuntil resolution of the symptoms of exacerbation.

The first outcome of the study was time to theresolution of symptoms. Fast recovery was definedas the resolution of symptoms in days below themedian duration of symptoms for the whole group.Treatment failure was defined as failure to returnto baseline before 10 days or the requirement ofnew treatments or medications during the first 10days for persistence or aggravation of symptoms.

This study was notified to the Spanish HealthAuthorities and complied with Spanish legislationon observational studies. The medication pre-scribed was that recommended in the currentSpanish guidelines13 and all data in the databasewere coded to maintain confidentiality.

Data collection

All the data generated were collected on a form inelectronic format designed specifically for thestudy. This form included information on demo-graphic variables, respiratory risk factors, co-morbidity, number of previous exacerbations de-fined as episodes of aggravation of respiratorysymptoms requiring treatment with antibiotics,corticosteroids or both and which were documen-ted in clinical records, characteristics of thepatient’s respiratory disease, degree of dyspnoea(0, no dyspnoea; I, dyspnoea when climbing twofloors; II, dyspnoea when climbing one floor; III,dyspnoea when walking on flat ground; IV, restingdyspnoea), usual medication, characteristics of theexacerbation and treatment administered. Theform was loaded into a portable personal handheldcomputer (PCA), Hewlett Packard Jornada 545(Colour Pocket PC), using a 32 bit processor, 16MBof RAM and 16MB of ROM. Data transmission to aunified database was carried out on-line by tele-phone connection via a CompactModem modem forPocket PCs at 56 K and 3.3 V. To ensure confidenti-ality, the data transferred contained no informa-tion which could identify the participatingindividuals. The information was included in asingle database located on the server and adminis-tered by the centre for data recording and analysis

(Biomedical Systems Group, S.A. Barcelona). Apilot study was performed to test the correctfunctioning of the system and the results of thisstudy were published recently.12

Statistical analysis

The association of categorical independent vari-ables with failures was assessed by the chi-squaretest, and the significance of continuous variableswas assessed with the Student’s t-test and Wilcox-on’s Rank Sum Tests. A regression model wasconstructed to identify variables independentlyand significantly associated with treatment failure.The model was constructed using a randomlyselected subsample of 90% of the subjects inclu-ded—developmental model. The model obtainedwas tested with data derived from the remaining10% of the total population—validation sample.The developmental and validation samples werecreated by assigning each patient a random numberbetween 0 and 1. Patients with a random number of0.90 or lower formed the developmental sampleand the remaining patients formed the validationsample.

We wanted our model to accurately reflect thefailure experience of the patient sample whilecontaining the minimum number of variablesnecessary to calibrate and discriminate well inthe developmental and validation samples. Onlyclearly definable and reliably obtained terms wereincluded and the use of laboratory values, radi-ological examinations and measurements, whichcannot be performed as part of routine patientcare, were avoided. Candidate variables includedin the model were: age, sex, smoking habits (activesmoker vs. non- or ex-smoker), years of evolutionof the disease, presence or absence of chronicbronchitis, and co-morbidity coded as ‘‘0 ¼ no,1 ¼ yes’’ for any of the following: ischemic heartdisease, diabetes mellitus or hypertension, chosenby presenting a prevalence of 5% or greater of thestudy population, severity of the underlying diseasequantified by the degree of dyspnoea, number ofexacerbations the previous year, severity of theexacerbation based on classification as types I or II,and treatment with oral steroids, inhaled short-acting beta-2 agonists, inhaled long-acting beta-2agonists, anticholinergics, long-term oxygen ther-apy, inhaled corticosteroids, and antibiotics (moxi-floxacin, clarithromycin or co-amoxiclav). Lungfunction parameters were not included in themodel since only 54% of the population hadspirometric evaluation performed.

ARTICLE IN PRESS

M. Miravitlles et al.958

Another model was constructed for identifyingvariables independently and significantly associatedwith the speed of recovery. For this model, thepopulation was divided into two groups, the groupwith a fast resolution of symptoms (equal to or lessthan 5 days) and the group with slow resolution ofsymptoms (6 days or longer). A cut-off point of 5days was chosen, since 5 days was the median timeto recovery for the whole group. The sameindependent variables were included in the model.

Variables were eligible for entry into a multiplelogistic regression model if they were significantlyassociated with failures at a P-value of less than0.10 and at least 5% of the population exhibitedthat factor. Correlations among the regressioncoefficients were used to screen for multicollinear-ity.14 Absolute values of r greater than 0.7 wereconsidered significant. The k statistic was used toassess inter-rater reliability of variables. Estimatedcoefficients and their SE’s were calculated using themethod of maximum likelihood. Variables wereeliminated from the model one at a time based onlikelihood ratio tests.

When all statistically non-significant ðP40:05Þvariables had been eliminated from the multi-variate model, calibration was assessed using theHosmer–Lemeshow goodness-of-fit test.15 This testevaluates the degree of correspondence between amodel’s estimated probability of failure and theactual failure rate of patients over groups spanningthe entire range of probabilities.

Discrimination was assessed using the area underthe receiver operating characteristic (ROC) curve16

to evaluate how well the model-distinguishedpatients who failed from patients who did not.The statistic represents, for all possible pairs ofpatients, the proportion in which the patient whofailed had a higher probability of relapse than thepatient who did not.

Modeling for fast recovery of exacerbation, i.e.in 5 days or less, followed the same process as themodeling for failure.

All the data were analysed using the SAS 6.04statistical package (SAS Institute, Cary, NC, USA).

Results

Study population

A total of 252 primary care physicians participatedin the study. Information was received on 1656patients, representing a mean of 6.6 patients perinvestigator. Two hundred patients were excludedfrom the study as they did not fulfil the inclusion

criteria, they fulfilled one or more of the exclusioncriteria or their follow-up data were incomplete. Inaddition, 309 exacerbations were type III and wereexcluded since they did not fulfill the criteria forprobable bacterial etiology. This left 1147 patientsvalid for the final analysis. A total of 630 patients(55%) had a diagnosis of COPD confirmed byspirometry.

Most of the patients were men (81%) and themean age of the population was of 68.7 years (SD:9.4). Of the patients with spirometry, the meanFEV1 was of 51.4% (SD: 14%), signifying a moderatealteration of pulmonary function. The mean num-ber of exacerbations during the previous year was2.4 (SD: 2.2) and only 4.8% of the patients had noexacerbation during the previous year.

Globally, 25.6% of the exacerbations were classi-fied as type I and 74.4% as type II. Patients with atype I exacerbation had a significantly lower FVCðP ¼ 0:012Þ and a tendency towards a lower FEV1. Inaddition, patients with type I exacerbations hadmore frequent exacerbations in the past, with77.5% experiencing two or more, compared with70.5% of the patients who presented a type IIexacerbation ðP ¼ 0:022Þ (Table 1).

Treatment and course of the exacerbations

The antibiotic treatment administered was co-amoxiclav in 358 cases (31%), moxifloxacin in 482(41%) and clarithromycin in 324 (28%). No signifi-cant differences were found in the characteristics ofthe patients and the severity of exacerbationsamong the treatment groups (Table 2). Oral corti-costeroids were prescribed to 24.6% of exacerba-tions without differences among groups. The clinicalcharacteristics of the patients and the comparion ofthe different variables between patients in whomthe exacerbation resolved and those with treatmentfailure are presented in Table 3.

A total of 173 (15.1%) patients had treatmentfailure and they were patients with a more severebaseline pulmonary disease, demonstrated by anincreased degree of baseline dyspnoea ðP ¼ 0:002Þ:They also had more frequent previous admissionsfor an exacerbation ðP ¼ 0:004Þ and a non-signifi-cant trend towards more frequent exacerbations inthe past year ðP ¼ 0:07Þ: In contrast, in patientswho failed symptoms of CB were less frequent (71%vs. 82%, Po0:001Þ:

Regarding treatment, failure was significantlymore frequent in patients receiving inhaledcorticosteroids (71.6% vs. 61.1%; P ¼ 0:009Þ: Nosignificant differences were found in failure ratesamong the different antibiotic options ðP ¼ 0:19Þ:

ARTICLE IN PRESS

Table 1 Demographic and clinical characteristics of patients with CB and/or COPD and type I or IIexacerbations.

Variables Type I (n ¼ 294) Type II (n ¼ 853) P-value

Sex, men 80.3% 81.2% 0.71Age, years 68.5 (9.69) 68.8 (9.3) 0.73Active smoker 21.8% 23.5% 0.55Pack-years 36.4 (22.7) 37.3 (23.9) 0.80Years of evolution 12.6 (8.2) 12.4 (7.8) 0.89Comorbidity 56.1% 51.9% 0.20CB 82.3% 80.3% 0.45

Baseline dyspneaDegree 0 or 1 49.5% 47.0% 0.71Degree 2 34.8% 37.5%Degree 3 or 4 15.7% 15.5%

Lung function ðn ¼ 630ÞFVC (ml) 2469 (893) 2624 (879) 0.012FVC (%) 67.2 (15.3) 68.2 (16.6) 0.43FEV1 (ml) 1474 (632) 1525 (643) 0.29FEV1 (%) 51.9 (14.2) 51.2 (13.9) 0.48

Exacerbations previous year0 or 1 22.5% 29.5% 0.0222 or more 77.5% 70.5%Admission previous year 0.3 (0.7) 0.3 (0.7) 0.43

Antibiotic prescribedMoxifloxacin 42.5% 40.8% 0.85Co-amoxiclav 30.6% 30.9%Clarithromycin 26.8% 28.2%

Failure rate 16.4% 14.2% 0.35

Variables associated with acute exacerbations of CB and COPD 959

Median time to resolution of symptoms was 5days and 659 (57%) patients recovered before or at5 days. Factors significantly associated with fasterrecovery were a younger age ðP ¼ 0:016Þ andlower number of previous admissions ðP ¼ 0:044Þ(Table 4). Patients treated with short-acting beta-2agonists had a higher rate of late recovery ðP ¼

0:008Þ; as did patients receiving long-term oxygentherapy ðP ¼ 0:001Þ: A 70.3% of patients treatedwith moxifloxacin recovered in 5 days or lesscompared with 44.4% with amoxi-clav or 49.7%with clarithromycin ðPo0:0001Þ:

Results of the multivariate models for failureand fast recovery

Table 5 shows the results of the multivariateanalysis of factors independently and significantlyassociated with failure. Hospitalisation for anexacerbation the previous year was significantlyassociated with failure. Similarly, frequent exacer-

bations the previous year were also significantlyassociated with failure. In contrast, patients withcriteria of CB had a reduced risk of failure(OR ¼ 0.53; 95% CI ¼ 0.35–0.79). The model waswell calibrated, as demonstrated by a P value of0.99 in the Hosmer–Lemeshow test, and showed afair discrimination (c statistic or area under theROC curve 0.62).

In contrast to the previous model, only variablesrelated to treatment were significantly associatedwith slow recovery. The use of long-term oxygenwas associated with a prolonged duration ofsymptoms, as well as the use of short-acting beta-2 agonists. The use of moxifloxacin compared toamoxi-clav was a ‘‘protective’’ factor for slowrecovery (OR ¼ 0.34; 95% CI ¼ 0.26–0.45). Similarresults were observed comparing moxifloxacin withclarithromycin (OR ¼ 0.41; 95% CI ¼ 0.31–0.55),(Table 5). The model also showed a good calibration(Hosmer-Lemeshow test, P ¼ 0:68) and fair discri-mination (c statistic or area under the ROC curve0.73).

ARTICLE IN PRESS

Table 2 Demographics of study patients by treatment group ðn ¼ 1164Þ:

All Moxilfoxacinðn ¼ 482Þ

Co-amoxiclavðn ¼ 358Þ

Clarithromycinðn ¼ 324Þ

P-value

Patient’s characteristicsAge (years) 68.7 (9.4) 69.2 (10.1) 68.6 (10) 68.5 (9.9) 0.82Males (%) 81 82 80 80 0.87Years of evolution of CB/COPD 13 (8) 13 (9) 11 (7) 13 (9) 0.67More than 2 comorbidities (%) 27 25 25 28 0.43Smokers and exsmokers (%) 81 81 80 78 0.41Pack-years 43 (28) 43 (28) 43 (26) 44 (25) 0.67Pulmonary function testsðn ¼ 630Þ

2497 (872) 2532 (820) 2506 (843) 2572 (833) 0.52

FVC (ml) 68 (16) 68 (17) 67 (16) 68 (16) 0.94FVC (%) 1497 (653) 1490 (644) 1488 (639) 1501 (630) 0.59FEV1 (ml) 51 (14) 51 (16) 51 (13) 51 (14) 0.95FEV1 (%) 58 (13) 57 (14) 59 (15) 58 (14) 0.43FEV1/FVCDuring year prior to enrollment,number of exacerbations

2.4 (2.2) 2.5 (1.6) 2.5 (1.5) 2.3 (1.8) 0.28

Characteristics of exacerbationsAnthonisen type I (%) 26.6 27.8 26.2 25.6 0.77Treatment with oral steroids (%) 24.6 26.5 23 23.5 0.45

Data expressed as mean (standard deviation) unless otherwise specified. P-values with respect to the results of the comparisonbetween the three groups, Kruskal–Wallis test and chi-squared test when appropriate.

M. Miravitlles et al.960

Discussion

The results of the present study indicate that theuse of long-term oxygen therapy and inhaled short-acting beta-2 agonists are significantly associatedwith a prolonged time to recovery from exacerba-tions of CB/COPD. Although no significant differ-ences were observed in the rate of treatmentfailure with the different antibiotics prescribed inthe study, the use of moxifloxacin was associatedwith a faster recovery compared to both, amoxi-clav and clarithromycin. The treatment failure rateat day 10 was 15.1% and patients who had a hospitaladmission or two or more exacerbations theprevious year were at increased risk of failure. Incontrast, patients with criteria of CB were lesslikely to fail after treatment of an exacerbationcompared with patients with emphysema.

In a previous study using data derived from thiscohort, we observed that patients treated withmoxifloxacin had a higher rate of cure at days 3 and5 after initiating treatment for the exacerbation.However, cure rates at day 10 did not significantlydiffer for the different antibiotic options.9 In thepresent study, we extended our analysis to identifyrisk factors for slow recovery among the character-istics of the patients, the severity of the exacer-bation and the treatment prescribed for the

baseline bronchial disease and the exacerbations.It is important for the patients to recover as quicklyas possible from their respiratory symptoms, thusavoiding repeated visits to the doctor, symptomaticmedication, and even the use of a further course ofantibiotics or substitution of the antibiotic when alack of efficacy is suspected. Furthermore, thereare major economic implications, since earlyrecovery facilitates earlier return to work or usualactivities. Comparative results should be inter-preted with caution, since this study is neitherexperimental nor randomised and is, therefore,subject to a possible selection bias in favour of onetreatment or another. Nonetheless, the participa-tion of 252 investigators and the inclusion of morethan one thousand patients makes it unlikely forthe same bias to be systematically repeated in asignificant number of cases. Furthermore, the threepatient groups analysed were perfectly balanced inall their demographic, clinical and functionalcharacteristics, as well as in the severity of theexacerbations.9 Another argument in favour of thesimilarity of the three groups is that the cure ratesachieved at 10 days did not significantly differ,making selection bias very unlikely.

The prolonged time to recovery in patientsreceiving long-term oxygen therapy is notsurprising, since this is a well-known marker of

ARTICLE IN PRESS

Table 3 Demographic and clinical characteristics of patients with CB and/or COPD and type I or II exacerbationsaccording to the outcome of the exacerbation.

Variables All (1147) Cure (974) Relapse (173) P-value

Sex, men 81.0% 80.8% 82.1% 0.69Age, years 68.7 (9.3) 68.8 (9.2) 69.4 (9.5) 0.50Active smoker 23.5% 22% 23.7% 0.61Pack-years 36.9 (23.7) 37.2 (23.7) 36 (24) 0.79Years of evolution 12.4 (7.8) 12.4 (8) 12.5 (6.5) 0.85Comorbidity 54.2% 53.5% 58.1% 0.27CB 80.5% 82.1% 71.1% o0.001

Baseline dyspnoeaDegree 0 or 1 47.4% 88.1% 11.9% 0.002Degree 2 37.2% 85.8% 14.2%Degree 3 or 4 15.4% 77% 23%

Lung function ðn ¼ 630ÞFVC (ml) 2573 (785) 2590 (864) 2557 (889) 0.23FVC (%) 68.3 (15.7) 69.1 (16.3) 68.4 (17.2) 0.46FEV1 (ml) 1511 (618) 1521 (593) 1480 (747) 0.11FEV1 (%) 52.0 (13.3) 52.2 (13.5) 49.4 (13) 0.54

Exacerbations previous year0 or 1 27.8% 88.2% 11.8% 0.0762 or more 72.2% 84% 16%Admission previous year 0.31 (0.6) 0.28 (0.67) 0.49 (0.85) 0.004

TreatmentShort-acting beta-2 67.3% 66.3% 71.2% 0.21Long-acting beta-2 69.5% 69.2% 75.1% 0.12Ipratropium bromide 61.0% 59.8% 65.9% 0.13Inhaled steroids 62.5% 61.1% 71.6% 0.009Long-term oxygen 11.2% 10.5% 15.5% 0.057

Treatment of exacerbationMoxifloxacin 41.3% 87.1% 12.9% 0.19Co-amoxiclav 30.9% 83.9% 16.1%Clarithromycin 27.8% 82.8% 17.2%Oral steroids 24.6% 24% 26.5% 0.41

Variables associated with acute exacerbations of CB and COPD 961

severity.6,17 However, the association between theuse of short-acting beta-2 agonists and longerduration of symptoms is a new finding. It can behypothesized that the use of short-acting beta-2agonists for rapid relief of dyspnoea by somepatients may substitute the use of more activedrugs such as oral corticosteroids or long-actingbronchodilators. This effect has been demon-strated in asthma. Patients using short-actingbeta-2 agonists instead of inhaled or oral corticos-teroids during an exacerbation have a poorerprognosis.18

In the present study, only exacerbations with ahigh probability of bacterial aetiology, defined byfulfilling the criteria for types I or II exacerbations10

were included. There were no significant differ-ences in the outcomes for both types of exacerba-

tions. Only the number of previous exacerbationswas different in both groups, with patients with atype I exacerbation being more likely to have hadmore frequent exacerbations in the past. This is inagreement with the study by Donaldson et al.19

that showed that as the number of exacerbationsincrease, their severity also increases.

The observed failure rate of 15.1% at day 10 is inaccordance with other studies performed in thecommunity. In a previous study by our group, afailure rate of 21% at 30 days was demonstrated in apopulation of 2414 exacerbated patients.7 In thislatter study, the presence of cardiac comorbidity,an increased severity of the baseline respiratorydisease and the increased number of previousexacerbations were risk factors for therapeuticfailure.7 Other published studies have reported

ARTICLE IN PRESS

Table 4 Demographic and clinical characteristics of patients with CB and/or COPD and type I or II exacerbationsaccording to the time to recovery from the symptoms of exacerbation.

Variables Fast recovery (o5days) ðn ¼ 659Þ

Slow recovery (45days) ðn ¼ 505Þ

P-value

Sex, men 81.9% 79.8% 0.36Age, years 68 (9.8) 69.6 (8.7) 0.016Active smoker 24.6% 21.4% 0.21Pack-years 36.7 (24.1) 37.2 (23.2) 0.67Years of evolution 12.1 (7.8) 12.8 (7.9) 0.13Comorbidity 54.6% 54.5% 0.96CB 80.9% 80.4% 0.84

Baseline dyspneaDegree 0 or 1 58.4% 41.6% 0.09Degree 2 52.2% 47.8%Degree 3 or 4 50.9% 49.1%

Lung function ðn ¼ 630ÞFVC (ml) 2581 (891) 2570 (881) 0.84FVC (%) 67.5 (16.2) 68.6 (16) 0.12FEV1 (ml) 1516 (653) 1499 (620) 0.96FEV1 (%) 50.6 (14.1) 52.4 (13.8) 0.053

Exacerbations previous year0 or 1 60.9% 39.1% 0.062 or more 54.8% 45.2%Admission previous year 0.28 (0.7) 0.36 (0.61) 0.044

TreatmentShort-acting beta-2 64.1% 71.4% 0.008Long-acting beta-2 68.6% 70.6% 0.44Ipratropium bromide 59.4% 63.0% 0.22Inhaled steroids 61.6% 63.6% 0.47Long-term oxygen 8.5% 14.4% 0.001

Anthonisen’s typeType I 55.1% 44.9% 0.58Type II 57.0% 43.0%

Treatment of exacerbationMoxifloxacin 70.3% 29.7% o.0001Co-amoxiclav 44.4% 55.6%Clarithromycin 49.7% 50.3%Oral steroids 19.7% 29.8% 0.18

M. Miravitlles et al.962

failure rates between 10% and 26%.5–7,10 Our resultsconsistently show that the increased number ofprevious exacerbations and admissions was asso-ciated with higher risk of failure. In addition, thepresence of symptoms of CB was associated with adecreased risk of failure. It is possible that patientswith chronic mucus hypersecretion are more proneto develop exacerbations that may be milder,20 incontrast with patients with emphysema who maydevelop respiratory failure more frequently duringthe course of an exacerbation. Other risk factorsfor failure proposed, such as older age and moresevere impairment in lung function were notobserved in the present study. A plausible explana-

tion is the relative uniformity of the populationsample due to its selection from primary carecenters, which prevents the inclusion of significantnumbers of elderly patients. In addition, lungfunction could not be included in the model,because this information was lacking in a significantnumber of individuals.

In summary, ambulatory treatment of exacerba-tions fails in 15.1% of patients with exacerbationsof CB/COPD. Patients with the emphysematousform of the respiratory disease and those withfrequent exacerbations or admissions are at in-creased risk of failure. The mean duration ofsymptoms of exacerbation is 5 days. Patients

ARTICLE IN PRESS

Table 5 Variables in the model for failure andlate recovery.

Variables EstimatedadjustedOR

95% CI

(a) Variables in the model for failure�

Previous hospitalization 1.61 1.08–2.42Two or moreexacerbations previousyear

1.51 1.04–2.17

CB 0.53 0.35–0.79

(b) Variables in the model for late recovery (longerthan 5 days)y

Long-term oxygen 1.96 1.35–2.85Short-acting beta-2agonists

1.51 1.17–1.92

Moxifloxacin vs. amoxi/clav

0.34 0.26–0.45

Moxifloxacin vs.clarithromycin

0.41 0.31–0.55

�Hosmer–Lemeshow goodness-of-fit test, P ¼ 0:99: ROCarea under the curve 0.62.

yHosmer–Lemeshow goodness-of-fit test, P ¼ 0:68: ROCarea under the curve 0.73.

Variables associated with acute exacerbations of CB and COPD 963

receiving long-term oxygen and short-actingbeta-2 agonists presented an increased risk ofprolonged exacerbations. Treatment withmoxifloxacin was associated with shorter durationof symptoms compared to co-amoxiclav andclarithromycin.

Appendix

Investigators participating in the EFEMAP study:Angel Abad Revilla, Inigo Aguirre Arratibel, Ma

Dolores Aicart Bort, Rafael Alario Segura, JesusAlbert Alvarez, Guillermo Alfaro Alonso, FelixAliaga Celna, Rafael Alonso Matia, Juan CarlosAlvarez Vicente, Javier Amiana Ruiz, Javier AntonOrtega, Felix Aparicio Ortigosa, Jose Jesus AparisiRomero, Jon Aranguren Castro, Antxon Arbulu Arin,Salvador Arias Moreno, Javier Arnal Crespo, Guil-lermo Arranz Gallardo, Fernando Artal Moneva,Francisco Atienza Martın, Lisardo Baena Camus,Juan Antonio Balda Soria, Juan Julio BallesterosOrtega, Ricardo Barcina Pajares, Juan Jose BarquinSolera, Jose Carlos Bastida Calvo, Gerard BernatLujan, Jorge Bertola Lomba, Pilar Blanco Prieto,Gemma Borque Gonzalo, Julia Boveda Fontan, JoseLuis Bravo Corrales, A. Gerardo Bravo Infante,

Javier Bris Pertinez, Francisco Brotons Munto,Enrique Caamano Romero, Ma Luz Caballero Hijon,Julian Cabrera Biosquece, Marcelino Calvino Cer-queiro, Antonio Calvo Valero, Fco Javier Cano Sanz,Emilia Capilla Ramirez, Antonio Capote Galindo,Manuel Carpintero Navarro, Emilio Carreno Iglesias,Ana Carvajal De La Torre, Juan Casado Martın,Josep Casajuana Brunet, Ma Jose Castany Fabregas,Jesus Castellano Alarcon, Manuel Castro Pazos,Juan Carlos Cayuela Poleto, Francisco JavierCanellas Pons, Francisco Cebrian Montolio, ManuelCercos Aparisi, Carlos Cervero Sobrevela, EnriqueCimas Hernando, Joaquim Collado Peiro, Jose LuisColomar Martınez, Felipe Cordero Mendez, EstherCordon Rodrıguez, Ma Angeles Coucheteaux Lour-ido, Francisco Crespo Martınez, Jose Ma CrespoPeleu, Silvia Crispi Cifuentes, Juan CuadradoGomez, Ignacio Cueto Bulnes, Mikhail DahdouhKuri, Josep Davins Miralles, Francisco De Alba,Paula De Blas, Jose Luis De Burgo, Jose Javier DeCastro, Juan Carlos De La Fuente Ibanez, Manuel DeLa Cueva Espinosa, Francisco Javier Del Corral,Lourdes Del Rio, Julio Del Valle, Elena DiezHuidonlon, Manuel Domınguez Rodrıguez, Jose Ma

Dura Altabella, Pancracio Eguren Lages, FernandoFacal Garcıa, Raimundo Fernandez Montero, JoseAntonio Fernandez Rodrıguez, Luis Antonio Fernan-dez Saura, Jose Ma Fernandez Villaverde, MarianoFerrer Marın, Carlos Fluixa Carrascosa, Lino Fontai-na Perez, Jose Galera Galera, Jose Damian GarcesRanz, Concepcion Garcıa Domingo, Raquel GarcıaLecina, Francisco Garcıa Mateos, Jose Luis GarcıaRodes, Fco Javier Garijo Cobo, Leopoldo GarroAzkarate, Endika Gerediaga Goikolea, Vicente GilFabra, Marta Giron Davina, Jose Antonio GodinezMarquez, Francisco Javier Gomez Torrado, JoaquimGonzalez Zaera, Patricia Gonzalez Loureiro, Ma

Isabel Gonzalez Saavedra, Susana Gonzalez Tejon,Ma Angeles Goterris Pinto, Lluıs Gracia Pardo, JuanGutierrez Cebollada, Adolfo Gutierrez Doblas,Carlos Gutierrez Maturana, Silvia Hernandez Ana-don, Enric Hernandez Huet, David Alejandro Her-nandez Vivancos, Rafael Hervas Samperio, MariaJose Hidalgo Fajardo, Pablo Huelin Martınez,Francisco Ibanez San Diego, Gloria Iglesias Bello,Ricardo Iglesias Losada, Joan Carles Iglesias Serra,Begona Iraola Sierra, Marıa Jose IribarrenOscoz, Jose Ignacio Jaen Dıaz, Santos JaimeSanchez, Eugenio Jimeno Saenz, Juan Luis JuezSenduilla, Manuel Lado Lopez, Alfonso LagunaMuguerza, Santiago Lancho Lancho, Jon LarizgoitiaLekue, Montserrat Latorre Casedas, Alberto RamonLeon Estella, Carlos Llor Vila, Enric Llopis Sorio,Jose L. Lopez Bocanegra, Rafael Lopez Garcıa, JoseLuıs Lopez Jimenez, Eloy Lopez Neira, AlbertoLosada Martınez, Mikel Luque Jimenez, Fco Javier

ARTICLE IN PRESS

M. Miravitlles et al.964

Manzaneda Lopez, Monserrat Manzano Gallego,Albert Marcos Estruch, Pedro Marino Campos,Miguel Angel Martı Martınez, Jose L. MartınCarmona, Jose Martın Clos, Javier MartınEsquillor, Miguel Angel Martın Rodrıguez, CarlosMartın Zamora, Xavier Martınez Artes, Jose ManuelMartınez Eizaguirre, Agustın Martınez Gonzalez,Juan Jose Martinez La Huerta, Arturo MartınezLarios, Leopoldo Martınez Roldan, Miguel AngelMartınez Solorzano, Carlos Martın, Esther MarzoJurico, Alvaro Mas Lacave, Pilar Mazas Vazquez,Luis Mejıas Perez, Marıa Dolores Mena Ruiz, JuanLuis Mendia Corostion, Vicens Miralles Belda, MiguelMolina De Heras, Jesus Molina Parıs, Ma JoseMonedero Mira, Carlos Montamarta Martınez, JavierMontoro Martı, Federico Montosa Martın, EnriqueMora Requena, Julio Moreno Banos, Javier MunozGutierrez, Karlos Naberan Tona, Javier NadalOltra, Fernando Navarro Tarın, Vicente NavarroTarın, B. Ramon Navarro Villa, Enrique NietoPol, Monserrat Oliveres Boadella, Rafael OrdovasCasaurran, Jacinto Ortiz Morinya, Ramon PachecoSarabia, Tomas Paga Bueso, Albert PagesArtunedo, Jesus Pages Corbera, Loreto PalacioElua, Domingo Pardo Albajez, MontserratPeraferrer Puigpelat, Jose Joaquim Perairo Fornan,Jose Antonio Pere Pedrol, Juan Manuel PerezGomez, Carlos Perez Perez, Benito Pinal Hermida,Enrique Piney Palacio, Eduardo Plaza Alarcon, JoseAntonio Plaza Hernandez, Francesc Pont Barrio,Manuel Prieto Estevez, Jose Miguel Puente Sota,Francisco Asıs Pulido Carretero, Angels QuesadaVazquez, Gloria Rabanaque Mallen, Jose VicenteRaga Casasus, Rosaura Reig Puig, EnriqueRevilla Pascual, Carlos Rodrigo Benito, TomasRodrıguez Camacho, Rafael Rodrıguez Fernandez,Florentino Rodrıguez Gimeno, Ma Francisca Rodrı-guez Hervas, Daniel Rodrıguez Lopez, ManuelRodrıguez Pineiro, Manuel Royo Del Rıo, Jose FelixRubio Fernandez, Jaime Ruiz Baixauli, FranciscoJose Saez Martınez, Alfonso Salgado Cachafeiro,Juan Salgado Saenz, Antonio Salvador Calvo, RamonSalvador Diego, Fernando Sanchez Dıez, ValentınSanchez Moran, Gonzalo Sans Domınguez, FernandoSapina Ontola, Jesus Seco Rubio, Pere SimonetAineto, Angel Soubriet Velasco, Jose Manuel SuarezSanchez, Julio Tarin Tamarit, Juan Pedro TobalimaAragues, Juan Toledano Galdeano, Matilde TorralbaLopez, Joaquim Torras Salles, Santiago TorreiraMonro, Juan Antonio Trigueros Carrero, VictorinoTurrado Turrado, Dolores Teresa Ungo Del Ganso,Pablo Vaamonde Garcıa, Jose Miguel ValeroPerez, Consuelo Vilades Laborda, Rosa VillafafilaFerrero, Juan Luis Villamor Bueno, FernandoVillasante Claudios, Ma Adela Vinregra Domınguez,Ines Yeguas Sanchez

References

1. Miravitlles M, Sobradillo V, Villasante C, et al. Estudioepidemiologico de la EPOC en Espana (IBERPOC): recluta-miento y trabajo de campo. Arch Bronconeumol1999;35:152–8.

2. Sobradillo Pena V, Miravitlles M, Gabriel R, et al. Geogra-phical variations in prevalence and underdiagnosis of COPD.Results of the IBERPOC multicentre epidemiological study.Chest 2000;118:981–9.

3. Miravitlles M. Epidemiology of chronic obstructive pulmon-ary disease exacerbations. Clin Pulm Med 2002;9:191–7.

4. Miravitlles M, Mayordomo C, Artes T, Sanchez-Agudo L,Nicolau F, Segu JL, The EOLO Group. Treatment of chronicobstructive pulmonary disease and its exacerbations inGeneral Practice. Respir Med 1999;93:173–9.

5. Adams SG, Melo J, Luther M, Anzueto A. Antibiotics areassociated with lower relapse rates in outpatients withacute exacerbations of COPD. Chest 2000;117:1345–52.

6. Dewan NA, Rafique S, Kanwar B, et al. Acute exacerbation ofCOPD. Factors associated with poor outcome. Chest2000;117:662–71.

7. Miravitlles M, Murio C, Guerrero T, On Behalf of the DAFNEStudy Group. Factors associated with relapse after ambula-tory treatment of acute exacerbations of chronic bronchitis.A prospective multicenter study in the community. EurRespir J 2001;17:928–33.

8. Snow V, Lascher S, Mottur-Pilson C. The evidence base formanagement of acute exacerbations of COPD. Chest2001;119:1185–9.

9. Miravitlles M, Llor C, Naberan K, Cots JM, Molina J, for theEFEMAP study group. The effect of various antimicrobialregimens on the clinical course of exacerbations of chronicbronchitis and chronic obstructive pulmonary disease inPrimary Care. Clin Drug Invest 2004;24:63–72.

10. Anthonisen NR, Manfreda J, Warren CPW, Hershfield ES,Harding GKM, Nelson NA. Antibiotic therapy in exacerba-tions of chronic obstructive pulmonary disease. Ann InternMed 1987;106:196–204.

11. American Thoracic Society. Standards for the diagnosis andcare of patients with chronic obstructive pulmonary disease.Am J Respir Crit Care Med 1995;152:S77–S120.

12. Miravitlles M, Llor C, Naberan K, Cots JM, on Behalf of theEFEMAP Study Group. Use of internet in a multicenter studyof chronic obstructive disease in primary care. Pilot phase ofthe EFEMAP study. Arch Bronconeumol 2002;38:427–30.

13. ALAT Work Group. Update to the Latin American ThoracicSociety (ALAT) recommendations on infectious exacerbationof COPD. Arch Bronconeumol 2004;40:315–25.

14. Glantz SA, Slinker BK. Primer of applied regression andanalysis of variance. New York, NY: McGraw-Hill Interna-tional Book Co; 1990. p. 512–68.

15. Hosmer DW, Lemeshow S. Applied logistic regression. NewYork, NY: Wiley; 1989.

16. Hanley JA, McNeil BJ. The meaning and use of the areaunder a receiver operating characteristic (ROC) curve.Radiology 1982;143:29–36.

17. Yohannes AM, Baldwin RC, Connolly M. Mortality predictorsin disabling chronic obstructive pulmonary disease in oldage. Age Ageing 2002;31:137–40.

18. Habashy D, Lam LT, Browne GJ. The administration of beta-2agonists for paediatric asthma and its adverse reaction inAustralian and New Zealand emergency departments: across-sectional survey. Eur J Emerg Med 2003;10:219–24.

19. Donaldson GC, Seemungal TAR, Patel IS, Lloyd-Owen SJ,Wilkinson TMA, Wedzicha JA. Longitudinal changes in the

ARTICLE IN PRESS

Variables associated with acute exacerbations of CB and COPD 965

nature, severity, and frequency of COPD exacerbations. EurRespir J 2003;22:931–6.

20. Miravitlles M, Guerrero T, Mayordomo C, Sanchez-Agudo L,Nicolau F, Segu JL, on Behalf of the EOLO Study Group.

Factors associated with increased risk of exacerbation andhospital admission in a cohort of ambulatory COPDpatients: a multiple logistic regression analysis. Respiration2000;67:495–501.