Abstract
Objective
To describe antibiotic prescribing patterns in acute exacerbations of chronic obstructive pulmonary disease (AECOPD) managed in emergency departments (EDs), assess the appropriateness of antibiotic selection according to Spanish guideline recommendations, and analyse its association with 30-day clinical outcomes.
Methods
Analysis derived from the multicentre retrospective EPOC-URG CAM registry (19 EDs in the Community of Madrid, 2022–2024), which included episodes of AECOPD. Antibiotic appropriateness was defined a priori according to national guideline recommendations. The primary outcome was the occurrence of an adverse event within 30 days, defined as recurrence of the exacerbation, ED revisit, hospitalisation, or death. The primary analysis was restricted to patients discharged directly from the ED with oral antibiotic treatment. Relative risks (RRs) were estimated using Poisson regression.
Results
A total of 4,136 episodes were analysed (mean age 74.5 ± 11.1 years; 68.3% male; 45.8% with a Charlson index ≥3). In the ED, 2,626 patients (63.5%) received antibiotics, with levofloxacin being the most frequently prescribed agent both in the ED (32.6%) and at discharge (39.3%). Patients’ clinical profiles were very similar regardless of the antibiotic class prescribed at discharge. Of the 1,736 (42%) patients discharged directly from the ED, 987 (56.9%) received antibiotic treatment, and the prescription was considered inappropriate in 158 (16.0%). The incidence of 30-day adverse events was higher among patients receiving inappropriate antibiotic treatment (20.9% vs 13.3%; p=0.019). In the multivariable analysis, appropriate antibiotic treatment was the only modifiable factor independently associated with a lower risk of an adverse event (adjusted RR 0.59; 95% CI 0.41–0.85; p=0.004), whereas previous exacerbations (adjusted RR 1.80; 95% CI 1.15–2.83; p=0.010) and moderate exacerbation (adjusted RR 1.52; 95% CI 1.10–2.11; p=0.012) were associated with a higher risk.
Conclusions
Antibiotic prescribing for AECOPD managed in the ED was characterised by a predominance of fluoroquinolones and limited individualisation according to patients’ clinical profiles. Among patients discharged directly from the ED, antibiotic selection in accordance with national recommendations was independently associated with a reduced risk of a 30-day adverse event.
Keywords: Chronic obstructive pulmonary disease, Exacerbation, Antibiotic therapy, Treatment appropriateness, Emergency departments, Antimicrobial stewardship
Abstract
Objetivo
Describir el patrón de prescripción antibiótica en la agudización de la enfermedad pulmonar obstructiva crónica (AEPOC) atendida en los servicios de urgencias hospitalarios (SUH), evaluar la adecuación de la selección antibiótica a las recomendaciones de las guías españolas y analizar su asociación con la evolución clínica a 30 días.
Método
Análisis derivado del registro multicéntrico retrospectivo EPOC-URG CAM (19 SUH de la Comunidad de Madrid, 2022-2024), que incluyó episodios de AEPOC. La adecuación antibiótica se definió a priori conforme a las recomendaciones de las guías nacionales. La variable principal de resultado fue la aparición de un evento adverso a los 30 días, definido como recurrencia de la agudización, revisita a urgencias, hospitalización o muerte. El análisis principal se restringió a los pacientes dados de alta directamente desde urgencias con tratamiento antibiótico oral. Se estimaron riesgos relativos (RR) mediante regresión de Poisson.
Resultados
Se analizaron 4.136 episodios (edad media de 74,5 ± 11,1 años; 68,3% varones; 45,8% con índice de Charlson ≥3). En urgencias, 2.626 pacientes (63,5%) recibieron antibiótico, siendo levofloxacino el más prescrito, tanto en urgencias (32,6%) como al alta (39,3%). El perfil clínico de los pacientes fue muy similar con independencia de la familia antibiótica prescrita al alta. De los 1.736 (42%) pacientes dados de alta directamente desde urgencias, 987 (56,9%) recibieron tratamiento antibiótico y, en 158 (16,0%), la prescripción se consideró inadecuada. La incidencia de eventos adversos a 30 días fue superior en los pacientes con tratamiento antibiótico inadecuado (20,9% frente a 13,3%; p=0,019). En el análisis multivariable, la adecuación del tratamiento antibiótico fue el único factor modificable asociado de forma independiente con un menor riesgo de evento adverso (RRa 0,59; IC95% 0,41-0,85; p=0,004), mientras que las agudizaciones previas (RRa 1,80; IC95% 1,15-2,83; p=0,010) y la agudización moderada (RRa 1,52; IC95% 1,10-2,11; p=0,012) se asociaron con un mayor riesgo.
Conclusiones
La prescripción antibiótica en la AEPOC atendida en urgencias muestra un predominio de fluoroquinolonas y una escasa individualización según el perfil clínico del paciente. En los pacientes dados de alta directamente desde urgencias, la adecuación de la selección antibiótica a las recomendaciones nacionales se asoció de forma independiente con una reducción del riesgo de evento adverso a 30 días.
Palabras clave: Enfermedad pulmonar obstructiva crónica, Agudización, Antibioterapia, Adecuación terapéutica, Servicios de urgencias, Optimización del uso de antimicrobianos
Introduction
Chronic obstructive pulmonary disease (COPD) is the third leading cause of death worldwide and the fourth in Spain, and its exacerbations are among the most frequent reasons for consultation in hospital emergency departments (EDs) [1,2]. An exacerbation is not an incidental event in the course of the disease but rather marks a prognostic turning point. Between 15% and 40% of deaths in patients with COPD are related to an exacerbation, and the eight weeks following discharge are associated with a high risk of treatment failure, relapse, and readmission [2,3]. Identifying the patients at greatest risk of reconsultation has therefore become a priority for Spanish EDs [4–6].
Bacterial infection of the tracheobronchial tree is the most frequent cause of exacerbation, implicated in 50-70% of episodes [7]. The remainder are due to non-infectious causes, including air pollution, whose role as a trigger has been documented even in low-pollution areas [8]. The antibiotic management of these patients, however, involves two clearly distinct clinical decisions: whether antibiotic treatment is indicated and, if so, which antibiotic to select. Whereas the first decision has been extensively studied since the description of the Anthonisen criteria and later with the incorporation of biomarkers and severity scores [7,9], the second has received far less attention, particularly in the ED setting.
The main Spanish guidelines and consensus documents agree on recommending amoxicillin–clavulanate and cefditoren as first-line options for most mild and moderate exacerbations, reserving fluoroquinolones for patients with suspected Pseudomonas aeruginosa infection or when β-lactams cannot be used [2,7,10–12]. Likewise, they advise against the use of cefuroxime, cefixime, and macrolide monotherapy owing to their microbiological or pharmacokinetic limitations [2,10–12]. These recommendations are consistent with the warnings issued by the European Medicines Agency (EMA) and the Food and Drug Administration (FDA), which restrict the use of fluoroquinolones when effective alternatives exist, given their safety profile [13,14].
Despite these recommendations, little is known about how antibiotics are selected in routine ED practice. More importantly, there are scarcely any data evaluating whether the appropriateness of that selection has consequences for the patient’s clinical course. This question is particularly relevant because the choice of antibiotic is one of the few potentially modifiable decisions during emergency care, in contrast to other non-modifiable prognostic determinants such as age, comorbidity, or the exacerbator phenotype.
The aim of this study was to describe the pattern of antibiotic prescribing in COPD exacerbations managed in EDs, to determine whether antibiotic selection reflects the patient’s clinical profile or is independent of it, and to analyse whether the appropriateness of antibiotic selection according to Spanish guideline recommendations is associated with 30-day clinical outcomes.
Methods
Design and setting
This was an observational analytical study derived from the EPOC-URG CAM registry, a multicentre retrospective cohort promoted by the Respiratory Pathology Working Group of SEMES, in which 20 EDs in the Community of Madrid participated. AECOPD episodes managed between 1 January 2022 and 31 December 2024 in patients with spirometry-confirmed COPD were included.
Data were collected retrospectively through a review of the electronic medical record, using a common case report from hosted in REDCap. Exacerbation severity was classified according to the Rome proposal [15]. One centre that contributed only 11 episodes and had no information on antibiotic prescribing was excluded from this analysis. A further 72 episodes for which no antibiotic prescription had been recorded were also excluded, leaving an analytical cohort of 4,136 episodes from 19 EDs.
The STROBE recommendations for observational studies were followed. No a priori sample size was calculated, as all available episodes meeting the inclusion criteria during the study period were included.
Variables
Demographic data, baseline status (Charlson index, Barthel index), comorbidities, recent disease history (exacerbations and ED visits in the previous 12 months, hospital admission in the previous 3 months, previous antibiotic exposure), clinical and laboratory variables of the index episode, including temperature and C-reactive protein (CRP), and the antibiotic treatment prescribed in the ED and at discharge, including the specific agent(s), were recorded.
Definition of appropriateness
Antibiotic appropriateness was defined a priori based on the oral regimen prescribed at discharge, in accordance with the Spanish guidelines and consensus documents in force during the study period [2,7,10–12]. Levofloxacin, moxifloxacin, cefditoren, and amoxicillin–clavulanate were considered appropriate. Amoxicillin monotherapy, cefuroxime, cefixime, ciprofloxacin, and azithromycin monotherapy were considered inappropriate. Azithromycin combined with cefditoren, amoxicillin–clavulanate, levofloxacin, or moxifloxacin was considered appropriate, as it represents a strategy to cover intracellular pathogens using an active beta-lactam or quinolone backbone; its combination with ciprofloxacin was considered inappropriate. Regimens recorded under the “other” category were classified individually by expert judgement according to their activity against Streptococcus pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis. Coverage of all three key pathogens was required for a regimen to be classified as appropriate, given that the patients were discharged directly from the ED and treatment was prescribed empirically, without a microbiological diagnosis of the episode. A discharge regimen classified as inappropriate was considered so regardless of the antibiotic administered previously in the ED.
Analysis populations
Patients who received antibiotics in the ED and at discharge, or at discharge only, were considered treated. Patients who received a single dose of an antibiotic in the ED and were discharged directly without a subsequent oral regimen were considered untreated, as an isolated dose does not constitute a therapeutic course.
The primary analysis was restricted to patients discharged directly from the ED with an oral antibiotic regimen.
Outcome variable
The outcome variable, prespecified in the registry protocol and common to the analyses derived from it, was a 30-day composite adverse event following the index episode, defined by the occurrence of any of the following: exacerbation recurrence, ED revisit, hospitalisation, or all-cause death during follow-up after discharge. A breakdown of the components of the composite is also presented.
Statistical analysis
Categorical variables are described as absolute and relative frequencies and compared using the chi-squared test. Continuous variables are described as the mean (standard deviation, SD) or median (interquartile range, IQR) according to their distribution, and compared using the Kruskal–Wallis or Mann–Whitney U tests. The association between appropriateness and 30–day adverse events was assessed using modified Poisson regression with robust variance. Models were adjusted for age, sex, Charlson index ≥3, Barthel index <90, exacerbation severity according to Rome, previous exacerbations, previous ED visits, admission in the previous 3 months, previous antibiotic exposure, comorbidities, and antibiotic administration in the ED. Results are expressed as relative risk (RR) with its 95% confidence interval (95% CI).
Values of p<0.05 or a 95% CI for the RR excluding 1 was considered statistically significant. All statistical analyses were performed using IBM® SPSS Statistics version 20 (IBM Corp., Armonk, NY, USA) and Stata® version 16.1 (StataCorp LLC, College Station, TX, USA).
Ethical considerations
The study complied with local and international ethical standards, including the Declaration of Helsinki, for the use of patient data, which were coded to ensure confidentiality. The Ethics Committee of the reference centre (Hospital Clínico Universitario de Valladolid) approved the study (PI-25-501-C). As this was a retrospective study, the Ethics Committee waived the requirement for informed consent.
Results
Cohort characteristics
A total of 4,136 COPD exacerbation episodes from 19 EDs were analysed (Figure 1). The population had a mean age of 74.5 (SD 11.1) years; 2,825 (68.3%) were male, 1,896 (45.8%) had a Charlson index ≥3, and 1,052 (25.4%) had a Barthel index <90 (Table 1). Almost four in ten patients, 1,519 (36.7%), had two or more exacerbations in the previous 12 months; 918 (22.2%) had attended the ED four or more times in the past year; 1,664 (40.2%) had received a course of antibiotics in the previous three months; and 1,278 (30.9%) had received more than three courses in the past year. According to the Rome criteria, 1,787 (43.2%) had a mild episode, 1,883 (45.5%) a moderate episode, and 456 (11.0%) a severe one.
Figure 1. Study flow diagram.
*Disposition was not recorded in 5 episodes.
Table 1.
Baseline characteristics and clinical outcomes of the cohort (N = 4,136)
| Variable | n (%) or mean (SD) |
|---|---|
| Demographic data | |
| Age, years | 74.5 (11.1) |
| Male sex | 2,825 (68.3) |
|
| |
| Baseline status | |
| Charlson index ≥3 | 1,896 (45.8) |
| Barthel index <90 | 1,052 (25.4) |
|
| |
| Comorbidities | |
| Ischaemic heart disease | 685 (16.6) |
| Heart failure | 954 (23.1) |
| Diabetes mellitus | 1,107 (26.8) |
| Chronic kidney disease | 649 (15.7) |
| Atrial fibrillation | 861 (20.8) |
|
| |
| Recent history | |
| ≥2 exacerbations <12 months | 1,519 (36.7) |
| ≥4 ED visits <12 months | 918 (22.2) |
| Hospital admission <3 months prior | 1,197 (28.9) |
| Antibiotic in previous 3 months | 1,664 (40.2) |
| >3 antibiotic courses in past year | 1,278 (30.9) |
|
| |
| Index episode | |
| Rome severity: mild | 1,787 (43.2) |
| Rome severity: moderate | 1,883 (45.5) |
| Rome severity: severe | 456 (11.0) |
| CRP, median (IQR), mg/L | 19.5 (5.0–66.2) |
|
| |
| Outcomes | |
| Antibiotic in the ED | 2,626 (63.5) |
| Hospital admission | 2,395 (57.9) |
| In-hospital mortality | 219 (9.1) |
|
| |
| 30-day adverse event | |
| Exacerbation recurrence | 431 (10.4) |
| ED revisit | 333 (8.1) |
| Hospitalisation | 255 (6.2) |
| 30-day death | 152 (3.7) |
| Any | 772 (18.7) |
SD: standard deviation; CRP: C-reactive protein; IQR: interquartile range.
The Rome severity classification was available for 4,126 episodes; in 10 it was not recorded. Severity percentages are calculated over the whole cohort (N=4,136). In-hospital mortality is calculated over admitted patients (n=2,395); all other percentages are over the whole cohort.
Of all the patients managed, 2,395 (57.9%) were admitted. Mortality during the index admission was 9.1% (219 of the 2,395 admitted patients). During the 30-day follow-up period, 772 patients (18.7%) experienced some component of the prespecified composite adverse event: exacerbation recurrence (431; 10.4%), ED revisit (333; 8.1%), hospitalization (255; 6.2%), and death (152; 3.7%). Mortality during the index admission (219 patients) and the death component of the 30-day event (152 patients) correspond to different definitions (death during the index hospitalisation versus death recorded during follow-up of the episode) and do not represent nested quantities.
Antibiotic prescribing pattern
During ED care, 2,626 (63.5%) of the 4,136 patients managed received antibiotics. Among the 1,736 patients discharged directly from the ED, 987 (56.9%) did so with oral antibiotic therapy.
Levofloxacin was the most frequently prescribed antibiotic, both in the ED and at discharge (Table 2). In the ED, it was administered to 855 patients (32.6% of those treated), followed by amoxicillin-clavulanate (681; 25.9%), ceftriaxone (662; 25.2%), and azithromycin (373; 14.2%). In the 987 patients discharged directly from the ED with an oral antibiotic regimen, levofloxacin was prescribed to 388 patients (39.3%), amoxicillin-clavulanate to 309 (31.3%), and cefditoren to 116 (11.8%). Among the options advised against by the national guidelines, azithromycin monotherapy was prescribed to 82 (8.3%) patients, cefixime to 25 (2.5%), amoxicillin to 15 (1.5%), cefuroxime to 12 (1.2%), and ciprofloxacin to 11 (1.1%). The median duration of the regimen was 7 days (IQR 5-7), and 5.2% of regimens exceeded 7 days.
Table 2.
Antibiotics prescribed in the ED and at discharge
| Antibiotic | In the ED (n = 2,626) N (%) | At discharge (n = 987) N (%) |
|---|---|---|
| Levofloxacin | 855 (32.6) | 388 (39.3) |
|
| ||
| Amoxicillin-clavulanate | 681 (25.9) | 309 (31.3) |
|
| ||
| Ceftriaxone | 662 (25.2) | – |
|
| ||
| Cefditoren | – | 116 (11.8) |
|
| ||
| Azithromycin | 373 (14.2) | 104 (10.5) |
| — as monotherapy | 61 (2.3) | 82 (8.3) |
|
| ||
| Piperacillin-tazobactam | 228 (8.7) | – |
|
| ||
| Meropenem | 116 (4.4) | – |
|
| ||
| Cefixime | – | 25 (2.5) |
|
| ||
| Amoxicillin | 11 (0.4) | 15 (1.5) |
|
| ||
| Moxifloxacin | 25 (1.0) | 14 (1.4) |
|
| ||
| Cefuroxime | – | 12 (1.2) |
|
| ||
| Ciprofloxacin | 40 (1.5) | 11 (1.1) |
|
| ||
| Other | 210 (8.0) | 22 (2.2) |
A single patient may receive more than one agent, so the sum of the percentages in each column may exceed 100%. The “- as monotherapy” row is a subcategory of azithromycin, not an independent category.
Prescribing decision and antibiotic selection
Patients treated with antibiotics were significantly older (mean age 75.1 vs 72.9 years; p<0.001), had a higher comorbidity burden (mean Charlson index 3.1 vs 2.5; p<0.001), had more severe exacerbations (12.5% vs 7.0% severe exacerbations; p<0.001), and had higher inflammatory biomarkers (median CRP 30.5 vs 6.0 mg/L, p<0.001; fever ≥38 °C in 7.4% vs 1.9%, p<0.001). Among patients who received antibiotic treatment during their ED stay, 1,976 (75.2%) were ultimately admitted. Conversely, 742 (28.3%) of the patients treated with antibiotics had neither fever nor elevated CRP.
Not receiving antibiotics was not associated with the 30-day composite adverse event after multivariable adjustment, either in the overall cohort (adjusted RR 1.09; 95% CI 0.86-1.37; p=0.48) or in non-admitted patients (adjusted RR 1.08; 95% CI 0.83-1.41; p=0.56).
Furthermore, we identified no clinically relevant differences in patient characteristics according to the antibiotic class prescribed. When comparing the therapeutic groups defined at discharge [amoxicillin–clavulanate (n=309), cefditoren (n=115), levofloxacin (n=388), moxifloxacin (n=13), and other therapies (n=162)], none of the 15 variables analysed showed significant differences (Table 3): age, sex, Charlson index, Barthel index, CRP (median), exacerbation severity (Rome), previous exacerbations, previous ED visits, hospital admission in the previous three months, and recent antibiotic exposure (course in the previous 3 months or more than three courses in the past year).
Table 3.
Patient clinical profile according to the antibiotic prescribed at discharge
| Variable | Amox-clav (n=309) | Cefditoren (n=115) | Levofloxacin (n=388) | Moxifloxacin (n=13) | Other (n=162) | p value |
|---|---|---|---|---|---|---|
| Age, years, mean (SD) | 73.7 (11.0) | 71.3 (10.9) | 72.8 (10.5) | 73.4 (10.4) | 73.6 (11.7) | 0.269 |
| Male sex | 224 (72.7) | 80 (69.6) | 253 (65.2) | 7 (53.8) | 107 (66.0) | 0.187 |
| Charlson index, mean (SD) | 2.7 (2.0) | 2.5 (1.8) | 2.8 (2.5) | 3.4 (3.0) | 2.8 (2.0) | 0.686 |
| Charlson ≥3 | 127 (42.8) | 47 (41.2) | 158 (41.7) | 8 (61.5) | 71 (46.1) | 0.584 |
| Barthel index, mean (SD) | 90.5 (20.4) | 94.8 (13.6) | 92.7 (16.8) | 79.6 (31.7) | 92.2 (16.2) | 0.192 |
| Barthel <90 | 57 (20.7) | 13 (11.6) | 61 (17.4) | 4 (33.3) | 30 (21.1) | 0.132 |
| CRP, median (IQR), mg/L | 19.0 (5.1–53.6) | 16.2 (5.0–35.5) | 17.0 (5.3–53.0) | 14.0 (6.4–34.5) | 13.4 (3.8–45.6) | 0.325 |
| Moderate exacerbation (Rome) | 83 (26.9) | 27 (23.5) | 97 (25.1) | 4 (30.8) | 35 (21.6) | 0.747 |
| Severe exacerbation (Rome) | 8 (2.6) | 2 (1.7) | 12 (3.1) | 0 (0.0) | 4 (2.5) | 0.901 |
| ≥2 exacerbations in 12 months | 89 (28.8) | 26 (22.6) | 124 (32.0) | 3 (23.1) | 45 (27.8) | 0.371 |
| ≥4 ED visits in 12 months | 48 (15.6) | 21 (18.3) | 85 (21.9) | 3 (23.1) | 33 (20.4) | 0.314 |
| Hospital admission in previous 3 months | 42 (13.6) | 15 (13.0) | 79 (20.4) | 3 (23.1) | 30 (18.5) | 0.113 |
| Antibiotic course in previous 3 months | 107 (34.7) | 37 (32.2) | 139 (35.9) | 5 (38.5) | 65 (40.1) | 0.705 |
| >3 antibiotic courses <1 year | 63 (20.4) | 25 (21.7) | 114 (29.4) | 4 (30.8) | 46 (28.4) | 0.059 |
Amox-clav: amoxicillin–clavulanate; CRP: C-reactive protein; IQR: interquartile range.
Comparisons using the chi-squared test for categorical variables and the Kruskal–Wallis test for continuous variables. The five therapeutic groups are mutually exclusive and comprise all 987 patients; combinations of azithromycin on an active beta-lactam (n=14) were classified in the beta-lactam group. The “Other” group includes azithromycin (n=90), cefixime (n=24), amoxicillin (n=15), cefuroxime (n=12), ciprofloxacin (n=11), and other agents (n=19); within this group the agent subtotals are not mutually exclusive, as some patients received more than one of these agents, so they do not sum to the group size.
Clinical outcomes according to antibiotic selection
Of the 987 patients discharged directly from the ED with an oral antibiotic regimen, 158 (16.0%) received an inappropriate regimen. In 147 cases this was inappropriate monotherapy [azithromycin (n=82), cefixime (n=24), amoxicillin (n=12), cefuroxime (n=12), ciprofloxacin (n=9), and other agents without sufficient coverage of the key pathogens (n=8)], and in 11 cases, an inappropriate combination (including azithromycin in 7 and other combinations without adequate coverage in 4).
The 30-day adverse event occurred in 20.9% of patients with an inappropriate prescription versus 13.3% of those who received an appropriate regimen (p=0.019), representing an absolute risk difference of 7.6 percentage points (95% CI 0.9-14.4), equivalent to one additional 30-day adverse event for every 13 patients receiving an inappropriate regimen. In multivariable analysis, three variables were independently associated with the outcome (Table 4, Figure 2): previous exacerbations (≥2 in 12 months: adjusted RR 1.80; 95% CI 1.15-2.83; p=0.010), moderate exacerbation severity according to Rome (adjusted RR 1.52; 95% CI 1.10-2.11; p=0.012), and appropriateness of antibiotic treatment, the only factor independently associated with a lower risk (adjusted RR 0.59; 95% CI 0.41-0.85; p=0.004). Neither comorbidity, frailty, nor previous antibiotic exposure retained significance after adjustment. The severe exacerbation category was not associated with the event (adjusted RR 1.08; 95% CI 0.46-2.57), probably reflecting the small number of severe episodes in this directly discharged population (n=26).
Table 4.
Factors associated with the 30-day adverse event
| Variable | Crude RR (95% CI) | p | Adjusted RR (95% CI) | p |
|---|---|---|---|---|
| ≥2 exacerbations in 12 months | 2.47 (1.84–3.33) | <0.001 | 1.80 (1.15–2.83) | 0.010 |
| Moderate exacerbation (Rome) | 1.56 (1.14–2.14) | 0.005 | 1.52 (1.10–2.11) | 0.012 |
| Severe exacerbation (Rome) | 1.33 (0.60–2.98) | 0.482 | 1.08 (0.46–2.57) | 0.855 |
| ≥4 ED visits in 12 months | 2.18 (1.60–2.96) | <0.001 | 1.40 (0.94–2.07) | 0.097 |
| Hospital admission in previous 3 months | 2.21 (1.62–3.02) | <0.001 | 1.44 (0.98–2.13) | 0.063 |
| Charlson index ≥3 | 1.49 (1.10–2.02) | 0.010 | 1.29 (0.89–1.87) | 0.178 |
| Barthel index <90 | 1.35 (0.93–1.96) | 0.112 | 0.98 (0.63–1.52) | 0.921 |
| >3 antibiotic courses in 1 year | 2.17 (1.61–2.93) | <0.001 | 1.01 (0.66–1.54) | 0.972 |
| Antibiotic course in previous 3 months | 1.87 (1.39–2.53) | <0.001 | 1.00 (0.66–1.52) | 1.000 |
| Age (per year) | 1.01 (1.00–1.03) | 0.042 | 1.01 (0.99–1.03) | 0.219 |
| Male sex | 1.02 (0.74–1.42) | 0.895 | 1.00 (0.71–1.43) | 0.984 |
| Heart failure | 1.26 (0.87–1.82) | 0.214 | 0.89 (0.57–1.39) | 0.622 |
| Chronic kidney disease | 1.19 (0.78–1.82) | 0.418 | 0.94 (0.58–1.51) | 0.792 |
| Atrial fibrillation | 1.07 (0.72–1.59) | 0.749 | 0.79 (0.51–1.23) | 0.296 |
| Ischaemic heart disease | 1.07 (0.70–1.62) | 0.758 | 0.80 (0.50–1.29) | 0.362 |
| Diabetes mellitus | 1.00 (0.70–1.43) | 0.982 | 0.88 (0.60–1.30) | 0.530 |
| Antibiotic in the ED | 1.21 (0.89–1.65) | 0.228 | 1.11 (0.80–1.54) | 0.519 |
| Appropriate antibiotic at discharge | 0.64 (0.45–0.90) | 0.012 | 0.59 (0.41–0.85) | 0.004 |
RR: relative risk; 95% CI: 95% confidence interval.
Estimates obtained by Poisson regression, adjusted simultaneously for all variables in the table. The reference category for Rome severity is mild exacerbation. Model adjusted for heart failure, chronic kidney disease, atrial fibrillation, ischaemic heart disease, and diabetes mellitus. Significant adjusted estimates are shown in bold.
Figure 2.
Forest plot of factors associated with the 30-day adverse event in patients discharged directly from the ED on oral antibiotics
The breakdown by components of the composite (Table 5) showed no significant differences in any of them separately: exacerbation recurrence (13.9% vs 8.7%; p=0.056), ED revisit (10.1% vs 6.9%; p=0.206), hospitalisation (3.8% vs 4.6%; p=0.834), and death (2.5% vs 1.3%; p=0.436). The largest absolute differences, although not statistically significant, were observed in recurrence and revisit.
Table 5.
Components of the 30-day composite adverse event according to antibiotic appropriateness
| Component | Inappropriate regimen (n=158) | Appropriate regimen (n=829) | p |
|---|---|---|---|
| Composite adverse event | 33 (20.9) | 110 (13.3) | 0.019 |
| Exacerbation recurrence | 22 (13.9) | 72 (8.7) | 0.056 |
| ED revisit | 16 (10.1) | 57 (6.9) | 0.206 |
| Hospitalisation | 6 (3.8) | 38 (4.6) | 0.834 |
| 30-day death | 4 (2.5) | 11 (1.3) | 0.436 |
Data expressed as n (%).
Discussion
This study yields three main findings. The first is that the decision to prescribe antibiotics does align with the patient’s clinical profile, although a quarter of those treated lacked markers of infection. The second is that, by contrast, the selection of the specific agent does not appear to be individualised according to the patient’s characteristics. The third is that appropriateness of antibiotic treatment to national therapeutic guideline recommendations is associated with fewer adverse events among patients discharged from the ED on antibiotic therapy.
The decision to start antibiotic treatment
The decision whether to prescribe antibiotics was associated with a specific clinical profile. Those treated were older, more comorbid, with more severe episodes and higher infection markers, with a median CRP five times higher (30.5 vs 6.0 mg/L). In this regard, an interesting finding was that not receiving antibiotics was not associated with prognosis after adjustment (adjusted RR 1.09; 95% CI 0.86-1.37), reflecting an indication reasonably matched to risk.
Previous studies have already shown that the benefit is concentrated in episodes with purulence or severity [9,16,17], and biomarker-guided strategies point in the same direction [7]. The 63.5% prescribing rate in the ED is an appropriate figure when contrasted with the classic estimate that between 50% and 70% of exacerbations are due to bacterial infection [2]. The fact that 1 in 4 treated patients had neither fever nor elevated CRP might suggest a margin of overprescribing, but should be interpreted with caution, as fever and elevated CRP are known to be absent in elderly or frail patients owing to immunosenescence [18].
Choice of antibiotic treatment
A striking finding of the study is that the selection of the agent does not appear to be individualised according to the patient’s clinical profile. When comparing the five therapeutic groups defined, none of the fifteen variables analysed showed significant differences (age, sex, comorbidity, frailty, CRP, exacerbation severity, previous exacerbations or visits, recent antibiotic exposure). This finding is consistent with European audits showing guideline adherence of around 60% [7]. The high comorbidity burden of this population, well characterised in Spanish EDs [19], likewise does not appear to guide the choice.
Levofloxacin was the most frequently prescribed antibiotic, both in the ED (32.6%) and at discharge (39.3%). All current Spanish documents and the Spanish and European regulatory agencies [13,14] relegate fluoroquinolones to an alternative rather than a first-line role, reserving their indication for cases with a risk of P. aeruginosa infection or when a beta-lactam cannot be used. The extent of this use is difficult to attribute in full to the reserved indications (the risk of P. aeruginosa or the inability to use a beta-lactam), whose expected prevalence in patients with mild-to-moderate exacerbations discharged from the ED is low. In this respect, the choice of agent was not associated with the patient’s clinical profile (Table 3), which does not support fluoroquinolones having been reserved in a targeted manner for the highest-risk patients. Nonetheless, the registry did not record Pseudomonas risk factors or beta-lactam allergy, so this interpretation cannot be demonstrated directly.
Fluoroquinolones have been relegated for three main reasons: (1) safety, owing to the serious adverse effects warned of by the FDA and the EMA [13,14], particularly relevant in elderly and highly comorbid patients such as those in our cohort [11,13]; (2) ecological impact, as they alter the microbiota more intensely and increase the risk of Clostridioides difficile infection [11]; and (3) preservation of the therapeutic armamentarium, as they are the only oral option, together with ciprofloxacin, against P. aeruginosa, and their indiscriminate use may compromise their future efficacy. It should be recalled that Spain is the European country with the highest fluoroquinolone consumption and that almost all levofloxacin is used for respiratory indications off-label [11].
Also, noteworthy is the use of cefixime, cefuroxime, and azithromycin monotherapy, regimens not recommended by the national guidelines. Cefixime and cefuroxime have relevant limitations against S. pneumoniae, especially in strains with reduced penicillin susceptibility, which represent around 20–25% of isolates [7,11], whereas cefditoren maintains high microbiological activity [7,10,11,20,21]. In addition, cefuroxime achieves low pulmonary concentrations and its absorption may be reduced by proton-pump inhibitors [10,11], commonly used in this population. These findings highlight that not all oral cephalosporins are equivalent and that regarding them as interchangeable is an error with potential clinical repercussions. Macrolide monotherapy was the leading cause of inappropriate prescribing (82/158 regimens). Given the resistance of S. pneumoniae to macrolides in Spain (20-25%), the use of azithromycin monotherapy implies a high probability of insufficient coverage against one of the main pathogens of COPD exacerbation [2,11].
The selection of the oral antibiotic should pursue not only clinical cure but also bacterial eradication, as greater eradication prolongs the exacerbation-free interval [7,22]. This requires antibiotics with high microbiological activity and the least possible impact on the microbiota [11,23]. In addition, the median duration of seven days observed in our cohort exceeds the five days recommended for mild and moderate exacerbations, without evidence of additional clinical benefit [11,12].
The importance of antibiotic appropriateness
In patients discharged directly from the ED, inappropriate prescribing was associated with an absolute difference of 7.6 percentage points in 30-day adverse events (20.9% versus 13.3%), equivalent to one additional adverse event for every 13 patients receiving an inappropriate regimen. This association persisted after adjustment for age, sex, comorbidity, frailty, episode severity, previous history, and antibiotic administered in the ED, and appropriateness was the only factor independently associated with a lower risk. In our study, appropriateness of antibiotic treatment at ED discharge in patients with an exacerbation of COPD, following the therapeutic guidelines of the scientific societies, was associated with a 41% lower risk of 30-day adverse events.
The excess of events was concentrated in exacerbation recurrence and early ED revisit, the components that reflect treatment failure, whereas the safety components (hospitalisation and death) did not differ relevantly. This pattern is consistent with the proposed mechanism: in patients considered fit for discharge, insufficient empirical coverage should not translate into early death but into treatment failure and early relapse, in line with the described relationship between bacterial eradication and the exacerbation-free interval [7,22]. It should be emphasised that the composite outcome was prespecified in the registry protocol, so its composition was not selected post hoc. Nonetheless, no component reached significance in isolation, which is to be expected given the small number of events per component, and recurrence and revisit are, moreover, the outcomes most sensitive to ascertainment bias.
Inappropriate antibiotic prescribing in the ED is frequent according to various studies, although almost all are descriptive and do not evaluate its clinical consequences. In a study conducted in the United States, 27.6% of prescriptions were inappropriate [24]; in an Australian study overall appropriateness was 70.4%, and in COPD exacerbation it fell to 54.3% (the lowest of all indications) [25]; and in another Australian series 34.2% of prescriptions were inappropriate (32.5% in respiratory infection) [26]. However, these studies quantify the magnitude of the problem, not its clinical repercussion. Our study addresses this gap by being able to estimate the risk of poor clinical outcomes according to inappropriate prescribing.
Limitations
This study has several limitations. First, its retrospective observational design precludes establishing causal relationships and does not exclude the influence of unrecorded factors, although the analysis was adjusted for the main prognostic factors. Nonetheless, it cannot be ruled out that unrecorded factors (suspected aetiology, allergies, previous colonisation, interactions, presence of bronchiectasis) influenced both the choice and the prognosis simultaneously. Second, the absence of information on sputum purulence and the Anthonisen criteria precluded assessment of the appropriateness of the antibiotic indication. Lastly, the size of the analysed cohort limited subgroup analyses, and the under-representation of severe exacerbations restricts extrapolation of the results to these patients. Finally, the study was conducted in a single autonomous community, so prescribing patterns might differ in other settings.
Conclusions
In COPD exacerbations managed in hospital emergency departments, the decision to indicate antibiotic therapy appears reasonably matched to the patient’s clinical profile, although a quarter of treated patients lack markers of infection. However, the choice of antibiotic is independent of the patient’s clinical characteristics and shows poor adherence to national guideline recommendations. In patients discharged from the ED, inappropriate prescribing was independently associated with a higher risk of 30-day adverse events. These results identify optimisation of antibiotic selection at discharge as a potentially relevant opportunity to improve the prognosis of these patients, being one of the few modifiable variables.
Annex 1
Respiratory Group of the Spanish Society of Emergency Medicine (SEMES): Gabriel José González Salazar (Hospital Central de la Defensa Gómez Ulla); Carmen Tejero Pallarés (Hospital Comarcal El Escorial); Isabel Salinas Bellón (Fundación Jiménez Díaz); Luis Ernesto Calderón Jave (Hospital General del Tajo); Ana Isabel Nogales García (Hospital General Infanta Sofía); Izaskun Verdejo Ledesma (Hospital Universitario del Sureste); Ana Belén Carlavilla Martínez (Hospital Universitario 12 de Octubre); Paloma Mora Raimundo (Hospital Universitario Clínico San Carlos); Isabel Gutiérrez Martín (Hospital Universitario de Fuenlabrada); Ana Herrera Rodríguez (Hospital Universitario de Getafe); Gloria Figueroa Carrasco (Hospital Universitario de Guadalajara); Licia de la Calle de la Rosa (Hospital Universitario del Henares); María Ulloa Argiz (Hospital Universitario Fundación Alcorcón); Elisa Casado Suela (Hospital General Universitario Gregorio Marañón); Virginia Lozano Araujo (Hospital Universitario de Móstoles); Álvaro Pineda Torcuato (Hospital Universitario Puerta de Hierro); Leonard Mihaita Roman (Hospital Universitario Ramón y Cajal); Stefanía Salazar Martínez (Hospital Universitario Rey Juan Carlos); Sergio de Santos (Hospital Universitario Infanta Leonor); Beatriz García Fernández-Bravo (Hospital Universitario Clínico San Carlos).
Footnotes
AI statement
I declare that no artificial intelligence tools (such as ChatGPT, Copilot, Gemini, or others) have been used in the writing, analysis, or review of this article.
Funding
No funding was received for the research, authorship and/or publication of this article.
Conflicts of interest
The authors declare that there is no commercial or financial conflict of interest for this research.
References
- 1.Global Initiative for Chronic Obstructive Lung Disease . Global strategy for prevention, diagnosis and management of COPD: 2024 report. 2024. Available from: https://goldcopd.org/2024-gold-report/ [cited 8 Aug 2026].
- 2.Piñera Salmerón P, Pulido Herrero E, Perales Muñoz R, Huerta García A, Alonso Avilés R, Cinesi Gómez C, et al.; SEMES Working Groups on Respiratory Pathology, Noninvasive Ventilatory Therapies and Infections in the Emergency Department. Management of COPD exacerbations in hospital emergency departments. Expert Rev Respir Med. 2025;19(12):1181–91. Available from: 10.1080/17476348.2025.2530207 [DOI] [PubMed] [Google Scholar]
- 3.Alonso Avilés R, del Pozo Vegas C, de Santos Castro PA, Guerrero Tejada R, Torres Gutiérrez RP, López Izquierdo R; Proyecto EPOC URG CyL . Diferencias entre pacientes con diagnóstico confirmado o con sospecha de enfermedad pulmonar obstructiva crónica que consultan en urgencias por síntomas respiratorios. Emergencias. 2025;37(1):31–8. Available from: 10.55633/s3me/110.2024 [DOI] [PubMed] [Google Scholar]
- 4.Piñera Salmerón P. Paciente con síndrome de agudización de la enfermedad pulmonar obstructiva crónica (EPOC) atendido en urgencias y su riesgo de reconsulta: hay que disminuir la incertidumbre. Emergencias. 2025;37(4):241–3. Available from: 10.55633/s3me/068.2025 [DOI] [PubMed] [Google Scholar]
- 5.Alonso Avilés R, Del Pozo Vegas C, López Izquierdo R, Lozano Alonso JE, Jimeno Asensio M, Rodríguez Borrego R, et al.; Proyecto EPOC URG CyL. REPOCUR: modelo de riesgo a 30 días en los pacientes adultos con exacerbación aguda de enfermedad pulmonar obstructiva crónica (EPOC) dados de alta de urgencias. Emergencias. 2025;37(4):251–8. Available from: 10.55633/s3me/051.2025 [DOI] [PubMed] [Google Scholar]
- 6.Alonso Avilés R, Del Pozo Vegas C, Lozano Alonso JE, Jimeno Asensio M, Gómez de Diego A, Caballero García A, et al. Complejidad del tratamiento de mantenimiento como predictor de deterioro a 30 días en exacerbaciones de la enfermedad pulmonar obstructiva crónica (EPOC) en urgencias. Emergencias. 2026;38(3):179–85. Available from: 10.55633/s3me/037.2026 [DOI] [PubMed] [Google Scholar]
- 7.González del Castillo J, Candel FJ, de la Fuente J, Gordo F, Martín-Sánchez FJ, Menéndez R, et al. Manejo integral del paciente con exacerbación aguda de la enfermedad pulmonar obstructiva crónica. Rev Esp Quimioter. 2018;31(5):461–84. Available from: https://seq.es/wp-content/uploads/2018/10/gonzalez09oct2018.pdf [PMC free article] [PubMed] [Google Scholar]
- 8.del Campo F, Álvaro de Castro TM, López-Izquierdo R, Moreno Torrero F, Álvarez D, Gutiérrez-Tobal GC, et al. Material particulado y exacerbaciones de la enfermedad pulmonar obstructiva crónica: análisis de las visitas a urgencias en un área de baja contaminación. Emergencias. 2025;37(4):267–73. Available from: 10.55633/s3me/052.2025 [DOI] [PubMed] [Google Scholar]
- 9.Anthonisen NR, Manfreda J, Warren CP, Hershfield ES, Harding GK, Nelson NA. Antibiotic therapy in exacerbations of chronic obstructive pulmonary disease. Ann Intern Med. 1987;106(2):196–204. Available from: 10.7326/0003-4819-106-2-196 [DOI] [PubMed] [Google Scholar]
- 10.Menéndez R, Cantón R, García-Caballero A, Barberán J. Tres claves para seleccionar el antibiótico oral adecuado en las infecciones respiratorias. Rev Esp Quimioter. 2019;32(6):497–515. Available from: https://seq.es/abstract/rev-esp-quimioter-2019-december-497-515/ [PMC free article] [PubMed] [Google Scholar]
- 11.Cantón R, Barberán J, Linares M, Molero JM, Rodríguez-González-Moro JM, Salavert M, et al. Decálogo para la selección del antibiótico oral en las infecciones respiratorias de vías bajas. Rev Esp Quimioter. 2022;35(1):16–29. Available from: 10.37201/req/172.2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Soler-Cataluña JJ, Piñera P, Trigueros JA, Calle M, Casanova C, Cosío BG, et al. Spanish COPD Guidelines (GesEPOC) 2021 update. Diagnosis and treatment of COPD exacerbation syndrome. Arch Bronconeumol. 2022;58(2):159–70. Available from: 10.1016/j.arbres.2021.05.011 [DOI] [PubMed] [Google Scholar]
- 13.European Medicines Agency . Disabling and potentially permanent side effects lead to suspension or restrictions of quinolone and fluoroquinolone antibiotics. EMA/175398/2019. Amsterdam: European Medicines Agency; 2019. Available from: https://www.ema.europa.eu/en/medicines/human/referrals/quinolone-fluoroquinolone-containing-medicinal-products [Google Scholar]
- 14.U.S. Food and Drug Administration . Fluoroquinolone antimicrobial drugs information. Silver Spring (MD): U.S. Food and Drug Administration. Available from: https://www.fda.gov/drugs/information-drug-class/fluoroquinolone-antimicrobial-drugs-information [Google Scholar]
- 15.Celli BR, Fabbri LM, Aaron SD, Agusti A, Brook R, Criner GJ, et al. An updated definition and severity classification of chronic obstructive pulmonary disease exacerbations: the Rome proposal. Am J Respir Crit Care Med. 2021;204(11):1251–8. Available from: 10.1164/rccm.202108-1819PP [DOI] [PubMed] [Google Scholar]
- 16.Miravitlles M, Moragas A, Hernández S, Bayona C, Llor C. Is it possible to identify exacerbations of mild to moderate COPD that do not require antibiotic treatment? Chest. 2013;144(5):1571–7. Available from: 10.1378/chest.13-0518 [DOI] [PubMed] [Google Scholar]
- 17.Stockley RA, O’Brien C, Pye A, Hill SL. Relationship of sputum color to nature and outpatient management of acute exacerbations of COPD. Chest. 2000;117(6):1638–45. Available from: 10.1378/chest.117.6.1638 [DOI] [PubMed] [Google Scholar]
- 18.Norman DC. Fever in the elderly. Clin Infect Dis. 2000;31(1):148–51. Available from: 10.1086/313896 [DOI] [PubMed] [Google Scholar]
- 19.Calvet-Pujol A, Ris-Romeu J, Albacar-Ingla M, Vicente-Díez F, Navarro-Arranz P, Puig-Campmany M. Análisis de la comorbilidad atendida en un servicio de urgencias a través de la codificación diagnóstica: el índice de Elixhauser. Emergencias. 2026;38(3):173–8. Available from: 10.55633/s3me/018.2026 [DOI] [PubMed] [Google Scholar]
- 20.Méndez-Lage S, Losada-Castillo I, Agulla-Budiño A; Grupo de trabajo del neumococo de los hospitales de Galicia . Streptococcus pneumoniae: distribución de serotipos, sensibilidad antibiótica, factores de riesgo y mortalidad en Galicia en un periodo de 2 años. Enferm Infecc Microbiol Clin. 2015;33(9):579–84. Available from: 10.1016/j.eimc.2015.01.010 [DOI] [PubMed] [Google Scholar]
- 21.Blasi F, Tarsia P, Mantero M, Morlacchi LC, Piffer F. Cefditoren versus levofloxacin in patients with exacerbations of chronic bronchitis: serum inflammatory biomarkers, clinical efficacy, and microbiological eradication. Ther Clin Risk Manag. 2013;9:55–64. Available from: 10.2147/TCRM.S41131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Miravitlles M, Espinosa C, Fernández-Laso E, Martos JA, Maldonado JA, Gallego M. Relationship between bacterial flora in sputum and functional impairment in patients with acute exacerbations of COPD. Chest. 1999;116(1):40–6. Available from: 10.1378/chest.116.1.40 [DOI] [PubMed] [Google Scholar]
- 23.Mezzatesta ML, Gona F, Marchese G, Nicolosi D, Toscano MA, Stefani S, et al. Cefditoren versus community-acquired respiratory pathogens: time-kill studies. J Chemother. 2009;21(4):378–82. Available from: 10.1179/joc.2009.21.4.378 [DOI] [PubMed] [Google Scholar]
- 24.Ladines-Lim JB, Fischer MA, Linder JA, Chua KP. Appropriateness of antibiotic prescribing in US emergency department visits, 2016-2021. Antimicrob Steward Healthc Epidemiol. 2024;4(1). Available from: 10.1017/ash.2024.79 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zosky-Shiller L, Thursky K, Park S, James R, Hall L, Chen C, et al. Antimicrobial prescribing quality in Australian emergency departments: an analysis of the Hospital NAPS data set. Antimicrob Steward Healthc Epidemiol. 2025;5(1). Available from: 10.1017/ash.2024.483 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Denny KJ, Gartside JG, Alcorn K, Cross JW, Maloney S, Keijzers G. Appropriateness of antibiotic prescribing in the emergency department. J Antimicrob Chemother. 2019;74(2):515–20. Available from: 10.1093/jac/dky447 [DOI] [PMC free article] [PubMed] [Google Scholar]