Observational, medical record review study of patients admitted for AECOPD. Patients admitted to hospital or in hospital-at-home programs were included. Participating hospitals were assigned a quota of patients to include in the study, based on their annual discharge rates. Patients were consecutively evaluated from 15 January 2011 until the quota had been reached.

All public hospitals in Catalonia that treated more than 50 cases admitted for COPD exacerbation in 2010 were invited to participate in the study by the Catalan Department of Health (CatSalut).

Hospitals that agreed to participate were classified according to the number of discharges for COPD exacerbation recorded in the CatSalut minimum basic data set16 for the year 2010. Hospitals were classified into 3 groups:

• •

  Level I hospital: fewer than 100 discharges/year for AECOPD.

• •

  Level II hospital: between 100 and 400 discharges/year for AECOPD.

• •

  Level III hospital: more than 400 discharges/year for AECOPD.

Two hundred variables were studied, grouped into 7 categories:

• 1.

  Demographic and descriptive data on the patient's condition in the 12 months prior to admission using a specifically designed questionnaire, which included questions on previous diagnosis, clinical follow-up, admissions for COPD, and number of chronic diseases described in the primary and secondary diagnosis in the discharge reports.

• 2.

  Pharmacological treatment prior to hospital admission, including home oxygen therapy and other home respiratory therapies.

• 3.

  Data related with care in the emergency department as the source of the referral, clinical status on admission and care process in the emergency department.

• 4.

  Data related with the hospital stay, admission from the emergency department to the ward, use of non-invasive mechanical ventilation (NIV), department that treated the patient or early mobilization, among others.

• 5.

  Information on the discharge process, such as clinical assessment, information received by the patient in the discharge report and destination on discharge.

• 6.

  Outcome variables related with admission and follow-up at 30 and 90 days, including mean stay, mortality and readmissions.

• 7.

  Comparison between hospitals by level.

COPD severity was assessed on the basis of FEV1, expressed as a percentage of predicted, when spirometric data were available. BODE cut-off points were used to define the categories, combining the severe and very severe categories; mild COPD was considered as FEV1 between 65% and 80%, moderate between 50% and 64%, and severe ≤49%.

Each hospital appointed a member of staff to review the hospital medical records. All professionals who carried out the fieldwork had received appropriate training and followed a procedure manual. During the data collection period, they were also provided with a contact telephone number for the research team in order to resolve any problems. One of the investigators (ET) was responsible for monitoring and supporting each site throughout the study.

A specific database was created for the study. Quality control was carried out to detect possible inconsistent values, in which case the person appointed to review the data was contacted to correct errors.

This study was part of the healthcare quality control strategies put in place by CatSalut. The study design complied with international ethical guidelines, and patient and data confidentiality was guaranteed in accordance with Spanish organic law on Personal Data Protection (15/1999, 13 December). Those responsible for analyzing the information had no access to data that might allow them to identify the patients.

A descriptive study of the data was performed using SPSS v18.

Of the 48 hospitals contacted, 30 (62.5%) agreed to take part in the study. A total of 910 patients were included, which accounts for more than 70% of the target population. The distribution of responses represents all healthcare regions, in accordance with the population. Mean patient age was 74.3 (+10.1) years; 83% were men. Current smokers made up 45% of cases, and of these, 9.8% continued to smoke. A total of 88% had been previously diagnosed with COPD, although previous spirometric data were recorded in only 65% of medical records. Table 1 describes the general characteristics of the patients evaluated.

The patient's first assessment in the emergency department was performed by a nurse in 52% of cases and by a doctor in 28.6%; this information was not recorded in 19.4%.

Information on dyspnea was not collected in 32% of cases at the time of admission. The respiratory rate was recorded in 52% of patients and the temperature in 85%. Arterial blood gas (ABG) measurement was performed in 80% of cases. The median time from arrival at the emergency department until the ABG was performed was 1h (IQR: 0.3–2.1). FiO2 was recorded at the time the sample was taken in 52% of the ABGs performed; this was 21% in 70% of the ABGs. The mean values obtained were pH 7.41±0.07, PO2 60±10mmHg and PCO2 48±15mmHg. The pH was less than 7.35 in 14.5% of cases. Pulse oximetry was recorded in 74% of patients. Only 25 patients (2.5%) did not have ABG or pulse oximetry on admission.

No changes were recorded in the quantity or color of sputum in 28% and 27% of cases, respectively. Sputum gram staining was carried out in 32%, and cultured in 94% of these cases. Chest X-ray was performed in 99.5% of cases, ECG in 69% and routine laboratory tests in 93%.

Patients were admitted from the emergency department to internal medicine (47%), respiratory medicine (37%) and other departments (12%), and 4% were included in a hospital-at-home program. The median time from arrival at the emergency department to admission was 10h (IQR: 5–23); 22% of patients remained there for more than 24h. A total of 9.6% of patients were treated with NIV, which was started in the emergency department in 57% of cases.

Mean hospital stay was 8.3 (7.7) days, with a median of 7 days (IQR: 4–10). The mean stay was longer than 8 days in 33.4% of patients, between 5–8 days in 37.4% and less than 5 days in 29.2%.

Administration of a sedative or hypnotic between arrival at the emergency department and discharge was recorded in 19% of cases. According to the information available, the patient was confined to bed for a median of 2 days (IQR: 1–3).

On discharge, the mean number of respiratory drugs prescribed was 3.8 (+1.8), and 1.4 (+1.4) in the case of non-respiratory drugs. At discharge, 62% of patients were receiving 4 or more drugs related with their respiratory disease (at the time of admission, 21.7% of patients were receiving 4 or more respiratory drugs). There were no significant differences between admission and discharge as regards the number of non-respiratory pharmacological treatments.

Of the patients admitted to hospital for AECOPD, 80.2% were discharged home, 7% continued their hospital-at-home program, 5.6% were transferred to a nursing home, 2.8% to other care systems and 4.4% died during admission.

Ninety-five percent of patients received a discharge report. An appointment for a follow-up visit within 2 months was recorded in 35% of patients, 40% of these with the hospital pulmonologist, 34% with their general practitioner, and 26% to others. The discharge report recommended smoking cessation in all of the active smokers (100%), specifying recommendations on diet in 21%, physical activity in 13%, alcohol consumption in 3%, and on dealing with exacerbations in 11%. The inhalation technique was noted as reviewed in 24% of discharges, detection of social problems was noted in 20%, and there was a formal mechanism to ensure continuity of care in 24% of cases.

ABG was performed at discharge in 21% of patients; test conditions were indicated in 83% of cases, and in 77% of these it was performed in baseline conditions. The mean ABG values available at discharge were: pH 7.42±0.04, PO2 64.5±13mmHg and PCO2 49±9.7mmHg. Pulse oximetry at discharge was recorded in 61%. Home NIV was maintained in 3% of patients after discharge, and 46 patients were prescribed home oxygen therapy for the first time.

The overall mortality, from admission in the emergency department to 90 days after discharge, was 12.4%, while hospital readmissions were 25% at 1 month and 49% at 3 months, 74% due to respiratory causes.

Of the patients studied, 190 (21%) were admitted to level I hospitals, 238 (26%) to level II hospitals and 482 (53%) to level III centers. The mean lung function did not differ significantly between different hospital levels, although patients admitted to level III hospitals were more frequently on home oxygen therapy (P=.01). Objective assessment at discharge (ABG or forced spirometry) was significantly lower in type III hospitals. Level III hospitals attended patients with a higher number of comorbidities (P<.001), and significantly longer stays (linear P=.020). This difference remained after adjusting for hospital level, age, admissions in the previous year and number of comorbidities in a multivariate linear regression model, with the stay increasing by 0.8 days as the hospital level increased (P=.049). The other variables analyzed are detailed in Table 2.

Analysis of the 30- and 90-day readmission variables for respiratory problems showed an inverse linear relationship between the number of readmissions and length of stay. As the length of the hospital stay increased, the number of admissions at 90 days for respiratory causes decreased (r=−0.152; P=.039). The logistic regression model also showed that patients with a previous diagnosis of COPD were less likely to be readmitted (OR=0.406; P=.015); this was also true of those on long-term home oxygen therapy (OR=0.406; P=.015) and with better FEV1 values (OR=0.977; P=.010), while age increased the likelihood of readmission (OR=1.028; P=.051). Regular physical activity was more frequently found in patients who were not readmitted than in the rest (71% vs 58%; P=.028). We did not observe any relationship between the risk of readmission and living alone. Inclusion in a hospital-at-home program reduced the probability of readmission at 30 days (5.7% vs 13%; P=.002), but not at 90 days.

Factors that increased the probability of dying at 30 and 90 days were age (OR=1.110; P=.004) and length of hospital stay (OR=1.214; P≤.001).