This prospective cohort study covered 16 hospitals belonging to the Spanish National Health Service (Hospital Costa del Sol, Hospital Valme, Hospital de Motril, Corporació Sanitaria Parc Taulí, Hospital del Mar, Hospital Universitario de La Princesa, Hospital Universitario Gregorio Marañón, Hospital Universitario La Paz, Hospital de Móstoles, Hospital Marqués de Valdecilla, Hospital Santa Marina, Hospital San Eloy, Hospital Galdakao-Usansolo, Hospital Txagorritxu, Complejo Hospitalario Donostia, and Hospital Cruces). The Institutional Review Boards of the participating hospitals approved this project. Patients with eCOPD attending the EDs of any of these hospitals were informed of the goals of the study, and invited to voluntarily participate and sign an informed consent form. All information was kept confidential. Recruitment started in June 2008 and ended in September 2010. A description of the study protocol was published previously [15].
Patients were eligible for the study if they presented to the ED with symptoms consistent with eCOPD. COPD was confirmed if the patient had a forced expiratory volume in 1 second/forced vital capacity (FEV1/FVC) quotient of less than 70%. Exacerbation was defined as an event in the natural course of the disease characterized by a change in the patient’s baseline dyspnea, cough, and/or sputum that was beyond normal day to day variations and may have warranted a change in regular medication in a patient with underlying COPD [16]. For cases of COPD newly diagnosed in the ED to be included in the study, they had to be confirmed by spirometry within 60 days after the index episode at a time when the patient was stable [17]. Patients were excluded from the study if, at the time they were seeing at the ED, they had eCOPD complicated by a comorbidity such as pneumonia, pneumothorax, pulmonary embolism, lung cancer, or left cardiac insufficiency. Other exclusion criteria included a diagnosis of asthma, extensive bronchiectasis, sequelae of tuberculosis, pleural thickening, or restrictive disease. Patients who did not wish to participate were also excluded.
Data collected
Data collected upon arrival in the ED included socioeconomic data, information about the patient’s respiratory function (arterial blood gases, respiratory rate, dyspnea), consciousness level measured by the Glasgow Coma Scale (GCS; altered consciousness defined as a score of ≤15 points, unaltered consciousness as a score of >15) [18], and presence of other pathologies recorded in the Charlson Comorbidity Index [19]. Additional data collected in the ED at the time a decision was made to admit or discharge the patient included the patient’s symptoms, signs, and respiratory status at that time. All information regarding the ED evaluation was recorded as it was provided by the ED physician in charge of the patient.
For patients admitted to the hospital, we collected additional data from the patient’s medical record and from a direct interview with the patient on the first day after admission and on the day of discharge. For patients discharged from the ED to home, telephone interviews were conducted with the patient around 1 and 7 days after discharge to assess hospital readmission and vital status. We asked all patients to tell us about their physical activity, general health, and dyspnea level while in stable condition before the eCOPD index and at 24 hours after being admitted to the hospital or discharged from the ED to home. We used the Medical Research Council (MRC) breathlessness scale [20] to measure baseline dyspnea.
For all patients with known COPD, additional variables were collected from medical records, including baseline severity of COPD as measured by FEV1; hospital admissions for eCOPD during the previous 12 months; baseline therapy (inhaled short-acting or long-acting beta-agonist, short-acting or long-acting anticholinergics, oral or inhaled corticosteroid, theophyllines, and/or need for noninvasive mechanical ventilation (NIMV) or long-term home oxygen therapy (LTHOT)); and presence of diabetes, hypertension, ischemic heart disease and/or valve disease, cor pulmonale, hepatic disease, peptic ulcer disease, psychiatric disorders, rheumatic disease, and any history of stroke or deep-vein thrombosis, and of other conditions needed to determine the Charlson Comorbidity Index.
Reviewers were trained before data collection, and a precise manual was developed, which was closely followed for the collection of data.
Definitions of outcome measure
The main outcomes measured were death occurring during the hospital admission or within 1 week of discharge to home from the ED, and death within 1 month of the index ED visit. Additional outcomes reported in this study were admission to the hospital and, if admitted, length of hospital stay; admission to the intensive care unit (ICU); need for invasive mechanical ventilation (IMV); need for NIMV for 2 or more days when mechanical ventilation was not used at home before admission; and admission to an intermediate respiratory care unit (IRCU) for 2 or more days (a minimum of 2 days was chosen to include only those patients needing more intensive and prolonged therapeutic interventions).
Patients were followed by phone or direct interview to reduce losses to a minimum.
Statistical analysis
The unit of the analysis was the patient. For patients who had more than one eCOPD requiring an ED visit during the recruitment period, only the first visit was considered for the analysis. Assumptions about how missing data were handled in this study have been described elsewhere. In general, missing values were imputed. In the case of missing data on basal level of dyspnea (MRC classification) it was imputed because class 5 comparison of the mortality rate of this group of patients with the other MRC categories (1 to 4) gave P values of less than 0.0001, whereas the P value within the MRC group 5 was 0.63.
The total sample was randomly divided in two to give a derivation sample and a validation sample. Descriptive analyses for both samples included frequency and percentages for categorical variables, and mean and standard deviations for continuous variables. The χ2and Fisher’s exact tests were used to test for statistical significance between proportions. For continuous variables, the Wilcoxon U-test was used.
In order to identify risk factors associated with short-term mortality in COPD, we performed univariate analyses in the derivation sample using logistic regression. Variables that were significant at P = 0.20 were entered into a multivariate logistic regression model. We performed logistic regression models in the derivation sample to select separately the variables for prediction of death. Final predictive factors in the multivariate analysis were those with a significance level of 0.05. Beta estimates, odds ratios (ORs) and 95% confidence intervals (95% CIs) were provided for the multivariate analysis. We developed a score by assigning a weight to each risk factor category based on the β parameter from the multivariate logistic regression. From the continuous score, four risk categories were created (mild, moderate, severe, and very severe). We considered the optimal classifier point as the point that maximized the sum of sensitivity and specificity. Final models were also adjusted by the treating hospital to see if that affected the results.
The predictive accuracy of each model was determined by calculating the area under the receiver operating characteristic (ROC) curve (discrimination) (AUC), and the models were calibrated by means of the Hosmer and Lemeshow test. We validated all AUCs—that is, those from the model and those from the continuous and categorical scores—in the validation sample by deriving the AUC in this sample [21].
Additional multivariate logistic regressions models were performed to evaluate the impact on short-term mortality of arterial blood gas values (pH, PCO2 and PO2) measured both at the time the patient arrived in the ED and at the time a decision was made to hospitalize the patient or discharge them to home from the ED. These were adjusted by our categorical score.
We compared various outcomes between the four risk classes of our categorical score. These included ICU admission; need for IMV; admission to an IRCU; admission to the hospital and, if admitted, length of hospital stay; readmissions within 10 days, 1 month, and 2 months after the index ED visit; and subsequent ED visits in the 2 months following the index ED visit.
To compare the predictive ability of our score, we applied information from all patients to previously created predictor scores for mortality in patients with stable COPD. These included the ADO (age, baseline dyspnea, and airflow obstruction measured by FEV1%) index [22], the HADO (health, activity, dyspnea, and airflow obstruction) score [23], baseline FEV1% classified according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) standards (FEV1% alone) [16], and the GOLD COPD combined assessment (baseline dyspnea plus previous exacerbations over the previous year) [24]. For the latter, we used hospitalizations during the previous year, as we did not collect the variable 'previous exacerbations.' Descriptive statistics for the previous mortality predictor scales (FEV1% classifications) were computed, and each AUC and its CI were determined and compared with our score.
All effects were considered significant at P < 0.05, unless otherwise stated. All statistical analyses were performed using SAS for Windows statistical software, v9.2 (SAS Institute, Inc., Carey, NC) and R© software v2.13.0.