. . . "Since these study designs using different types of controls are expected to yield quite different conclusions (effectiveness vs. comparative effectiveness), we separated them in our analyses." .