Validity of the adherence estimator in the prediction of 9-month persistence with medications prescribed for chronic diseases: a prospective analysis of data from pharmacy claims

Clin Ther. 2009 Nov;31(11):2584-607. doi: 10.1016/j.clinthera.2009.11.030.


Objective: The aim of this article was to assess the predictive validity of the Adherence Estimator--a 3-item instrument designed to estimate a patient's propensity to adhere to medications prescribed for chronic disease.

Methods: The Adherence Estimator was a 3-item part of a larger survey mailed to adults aged >or=40 years who had a qualifying index prescription filled in June 2008. A qualifying prescription was defined as one for a medication indicated for the treatment of 1 of 5 chronic diseases (cardiovascular disease, dyslipidemia [lipid-lowering drugs], diabetes [oral antihyperglycemics], osteoporosis [oral bisphosphonates], or asthma). Outcomes were compared between the adherence risk groups derived from the Adherence Estimator (low risk = score of 0, medium risk = score of 2-7, and high risk = score of 8-36). Treatment persistence over a period of 9 months was measured using pharmacy claims data. The primary outcome was the median proportion of days covered (PDC) by >or=1 medication during the first 9 months after the index fill. Secondary outcomes included adherence to the index medication, defined as PDC dichotomized to >or=0.80 or <0.80; rate of obtaining a second fill within 30 days of the index fill; and medication possession ratio (MPR) for refill adherence.

Results: There were 1676 usable responses. Ages ranged from 40 to 88 years, with a mean of 64.6 years. Almost two thirds (1076/1676 [64.2%]) of the sample were female, and 1483/1676 (88.5%) were white. Statistically significant associations for all 3 pairwise comparisons (low vs medium risk, low vs high risk, and medium vs high risk) were observed between the Adherence Estimator risk groups for: (1) median PDC (0.655, 0.598, and 0.484 in the low-, medium-, and high-risk groups, respectively [all, false discovery rate [FDR] <0.05]); (2) PDC categorized (293/711 [41.21%], 200/588 [34.01%], and 105/377 [27.85%] [all, FDR <0.05]); and (3) rate of obtaining a second fill for the index medication within 30 days (489/711 [68.78%], 374/588 [63.61%], and 207/377 [54.91%] [all, FDR <0.05]). The low- and high-risk groups differed from one another on: (1) persistence with the index medication at 9 months (265/711 [37.27%] and 95/377 [25.20%]); (2) persistence with >1 medication at 9 months (291/711 [40.93%] and 108/377 [28.65%]); and (3) obtaining a second fill for any medication within 30 days (501/711 [70.46%] and 219/377 [59.09%]) (all, P < 0.05). The low- and high-risk groups differed significantly from one another in MPR for refill adherence (0.912 vs 0.866). Results observed within diseases mirrored those for the total sample, but with less precision.

Conclusion: In the present analysis of the validity of the Adherence Estimator in predicting adherence, baseline propensity to adhere to medications prescribed for chronic diseases was statistically associated with several measures of adherence and persistence, as derived from pharmacy claims data, over a 9-month period.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Age Factors
  • Aged
  • Algorithms*
  • Chronic Disease / drug therapy*
  • Databases, Factual
  • Drug Prescriptions / statistics & numerical data
  • Female
  • Forecasting
  • Health Surveys
  • Humans
  • Insurance Claim Review
  • Male
  • Massachusetts / epidemiology
  • Middle Aged
  • Patient Compliance / statistics & numerical data*
  • Pharmacies / statistics & numerical data
  • Prospective Studies
  • Reproducibility of Results
  • Sample Size
  • Selection Bias
  • Sex Factors
  • Socioeconomic Factors