Adjusting for confounded variables: Pulmonary function and smoking in a special population