Study objectives: To examine potential threats to internal and external study validity caused by differential patient withdrawal from a randomized controlled trial evaluating pharmacist management of hypertension, to compare the characteristics of patients who withdrew with those of patients who completed the study, and to identify characteristics that predispose patients to withdraw from hypertension management.
Design: Prospective, randomized, comparative study.
Setting: Network of primary care clinics.
Patients: Four hundred sixty-three patients with a diagnosis of hypertension and a last documented systolic blood pressure of 160 mm Hg or greater and/or diastolic blood pressure of 100 mm Hg or greater.
Intervention: Patients were randomly allocated to the pharmacist intervention or usual-care (control) group. Those in the pharmacist intervention group were collaboratively managed by a primary care clinical pharmacy specialist and their primary care provider. Patients in the control group received usual care from only their primary care provider.
Measurements and main results: Of the 463 patients, 191 (41%) withdrew from the study after randomization and 272 (59%) completed the study. Patients who withdrew from the pharmacist intervention group were similar to patients who withdrew from the usual-care group with respect to age, sex, insurance status, and chronic conditions. Patients who smoked or had commercial insurance were more likely to withdraw from the study than the other participants. However, multivariate analysis of all variables, when adjusted for the effect of the intervention, revealed that insurance status was the only variable associated with a heightened probability of withdrawal (p=0.002).
Conclusion: Although this study had a high withdrawal rate, between-group patient characteristics remained balanced. Therefore, internal validity was preserved, and outcomes from the study groups could be reliably compared. A lack of significant differences between patients who withdrew versus those who completed, with the exception of insurance status, suggests that external validity was not jeopardized.