Comparison of two self-rating instruments for medication adherence assessment in hypertension revealed insufficient psychometric properties.

OBJECTIVE In cases of insufficiently controlled blood pressure, it is important for practitioners to distinguish between "nonadherence" and "nonresponse" to antihypertensive drug treatment. A reliable and valid adherence measurement based on the patient's self-report may be helpful in daily practice. STUDY DESIGN AND SETTING In a primary care sample with 353 hypertensive patients, we applied two self-rating instruments to assess medication adherence (the "Hill-Bone Compliance to High Blood Pressure Therapy Scale" and Morisky's "Self-Reported Measure of Medication Adherence") and compared their psychometric properties. RESULTS Both scales showed low acceptability and insufficiency to moderate internal consistency (Cronbach's alpha=0.25 and 0.73, respectively). Their convergent validity as indexed by kappa=0.39 could be judged as "fair" at best. Testing the power to predict blood pressure >140/90mmHg, both scales showed an accuracy of 57% and 62%, respectively. The positive likelihood, that is, the increase in likelihood of high blood pressure in cases of nonadherence was 1.00 and 1.32, respectively. CONCLUSION The use of both scales cannot be recommended. They showed considerable floor effects, and their ability to identify medication adherence was inconsistent for nearly every third patient. The power of both scales to predict uncontrolled blood pressure was essentially a chance. The underlying conceptual framework of medication adherence therefore needs to be rethought.