Objective: To validate six indirect methods of identifying patients who do not comply with their treatment with hypolipaemiant drugs.
Design: A prospective study.
Setting: Primary care centres in the province of Alicante.
Patients: 107 lipaemic patients, on the lists of 5 General Medical practices and on drugs treatment.
Measurements and main results: The most accurate way to assess compliance was the surprise counting of pills in patients' homes. Patients who had between 80 and 110% compliance were defined as compliant. The six indirect methods validated were: Communication of self-compliance (CS), Attendance at appointments (AA), Doctor's judgment (DJ), Information about the illness (II), the Morisky-Green test (MG) and the grade of control (GC). AA, DJ and CS were the methods with highest specificity (91.2%-89.5%). GC and II were the most sensitive (88%-82%). GC obtained the greatest negative predictive value (77.7%), and DJ the greatest positive predictive value (73.6%). The concordance index (kappa) ranged from 0.23 for GC and -0.002 for II.
Conclusions: GC, DJ and CS are the methods with the best validity indicators and concordance. They could, therefore, be used together in clinical practice to identify patients not complying with their hypolipemiant treatment.