Table 7.

Multivariate analysis for the risk factors of COPD exacerbation

Hosmer-Lemeshow goodness-of-fit showed chi-square = 3.48 and P = 0.90, showing good fit of the models; C-statistics value was 0.70. Clinico-physiologic indices with a univariate P value < 0.2 were considered candidates. Initial variables for multiple logistic regression were age, current smoking, hospitalization due to exacerbation during past one year, Charlson index score, body mass index, modified MRC dyspnea scale, 6 minute walk distance, St. George Rspiratory Questionnaires score, FEV₁, % predicted. All radiologic indices irrespective of the result of univariate analysis were included in the multiple logistic regression.