Variation in medication adherence across patient behavioral segments: a multi-country study in hypertension.

Abstract

ObjectivesThis study determines the following for a hypertensive patient population: 1) the prevalence of patient worldview clusters; 2) differences in medication adherence across these clusters; and 3) the adherence predictive power of the clusters relative to measures of patients’ concerns over their medication’s cost, side effects, and efficacy.MethodsMembers from patient panels in the UK, Germany, Italy, and Spain were invited to participate in an online survey that included the Medication Adherence Report Scale-5 (MARS-5) adherence instrument and a patient segmentation instrument developed by CoMac Analytics, Inc, based on a linguistic analysis of patient talk. Subjects were screened to have a diagnosis of hypertension and treatment with at least one antihypertensive agent.ResultsA total of 353 patients completed the online survey in August/September 2011 and were categorized against three different behavioral domains: 1) control orientation (n=176 respondents [50%] for I, internal; n=177 respondents [50%] for E, external); 2) emotion (n=100 respondents [28%] for P, positive; n=253 respondents [72%] for N, negative); and 3) agency or ability to act on choices (n=227 respondents [64%] for H, high agency; n=126 [36%] for L, low agency). Domains were grouped into eight different clusters with EPH and IPH being the most prevalent (88 respondents [25%] in each cluster). The prevalence of other behavior clusters ranged from 6% (22 respondents, INH) to 12% (41 respondents, IPL). The proportion of patients defined as perfectly adherent (scored 25 on MARS-5) varied sharply across the segments: 51% adherent (45 of 88 respondents) for the IPH vs 8% adherent (2 of 25 respondents) classified as INL. Side effects, being employed, and stopping medicine because the patient got better were all significant determinants of adherence in a probit regression model.ConclusionBy categorizing patients into worldview clusters, we identified wide differences in adherence that can be used to prioritize interventions and to customize adherence messages. Also, the predictive power of segments was greater than that for variables measuring concerns over cost, side effects, and efficacy.