In order to determine the lung function of a patient at the start of the model, some of the baseline patient characteristics are assumed to be correlated. In order to understand and quantify the correlation between these baseline characteristics, patient-level data from three multinational, multi- center, randomized, double-blind, placebo-controlled studies evaluating the efficacy of indacaterol in over 3,800 patients with moderate to severe COPD were made available . A correlation matrix was constructed employing the Cholesky decomposition method on this pooled dataset . The distribution of each individual variable was tested and all continuous variables were found to be Normally distributed. The following variables were regressed against each other to test for correlations: forced expiratory volume in 1 second (FEV1)% predicted, age, gender, height and smoking status. The co-variance matrix linked gender and age to height, smoking status and lung function in order to generate the FEV1 value at the start of each simulation for each patient.