This approach was taken to permit the assessment of possible drug-specific effects by minimizing potential confounding effects of differential glucose control. All the patients were followed until study closeout whenever possible, regardless of whether they were taking a study medication.