Discrete variables are expressed as counts (percentages) and continuous variables as mean ± SD or median (interquartile range [IQR]). Comparability of groups was analyzed by χ2 test, two-sampled t test, Mann-Whitney U test, Kruskal-Wallis analysis of variance, or Wilcoxon matched-pair test, as appropriate. To analyze the relationship among different variables and copeptin levels on hospital admission, a multiple linear regression model including age, cardiopathy, Pao2, leukocyte counts, CRP, procalcitonin, and FEV1 percentage of predicted was used. To analyze the relationship among different clinical parameters and duration of hospital stay, a multiple linear regression model including age, cardiopathy, arterial hypertension, renal failure, diabetes mellitus, Pao2, Paco2, and FEV1 percentage of predicted was used. To assess the influence of age, hospitalization due to AECOPD in the previous year, cardiopathy, Pao2, leukocyte counts, CRP, procalcitonin, and FEV1 percentage of predicted and antimicrobial therapy during the index exacerbation on clinical failure, a Cox regression univariate and multivariate analysis were performed. Correlation analyses were performed using Spearman rank. The time to clinical failure was analyzed by Kaplan-Meier survival curves and compared by the log-rank test. All test were two tailed; p < 0.05 was defined as significant. Data were analyzed using statistical software (Statistical Package for Social Sciences, version 14 for Windows; SPSS; Chicago IL).