Conjoint Analysis in Pharmaceutical Research

Abstract

In this issue of JMCP, the topic of medication adherence is addressed in an article by White et al. that compares differences in prescription refill patterns of antihyperlipidemic agents between patient groups receiving medications via mail service or retail pharmacies. 1 While the objective of this naturalistic study was to compare refill patterns between two different types of prescription services utilized by plan members, the results and discussion from this article highlight an important measurement that may directly impact the clinical and economic outcomes related to the use of medications for treating chronic diseases. Measures of medication adherence, also termed " compliance, " add an important dimension to outcomes studies by providing decision-makers with a valuable metric for comparing the utiliza-tion patterns of selected therapeutic agents. When medications are prescribed rationally and patients follow their regimens appropri-ately, the likelihood of achieving positive patient outcomes increases dramatically. However, when patients don't adhere to their regimens, which may occur for numerous reasons, the opportunity for achieving positive health outcomes may be jeop-ardized. Measuring levels of adherence to medication therapies can help to explain or predict which plan design, educational intervention, or particular drug product has the potential to pro-vide optimal patient outcomes within an organized health care set-ting. In the article in this issue, the authors included patients in the study who had been newly started on an HMG-CoA reductase inhibitor based on a review of pharmacy claims data from a California HMO. Claims data were analyzed for each patient for a period of one year after initiating the therapy regimen. After con-trolling for some demographic differences between the study groups, the authors found that patients utilizing the mail service had significantly higher rates of adherence than patients using community pharmacies. The use of pharmacy claims data to measure patient adherence has gained widespread use within pharmacy benefit programs. Large prescription claims databases have provided researchers with convenient access to utilization patterns of large patient pop-ulations and the opportunity to investigate medication adherence. Perhaps the most comprehensive methodological work in the area of measuring medication compliance using administrative databases comes from Steiner et al. in two widely cited articles. 2,3 In an original research article and in a subsequent review piece, Steiner and his colleagues provided a rationale for using different methods to calculate the patient adherence rate. Different rates may be calculated depending upon the length of time that one chooses to observe adherence within a population. Utilizing a measurement based on Steiner' s earlier work, Sclar et al. described the now often-used " Medication Possession Ratio " (MPR) to calculate days of therapy between refills using prescrip-tion claims histories. 4 To calculate this ratio, the total days supply of the medication dispensed (not including the last prescription dispensed) is divided by the total number of days between the first prescription dispensed and the last prescription dispensed during an observation period. Typically, this calculated ratio is less than 1.0, owing to the fact that patients will exhibit gaps in medication days. However, if patients obtain refills before their supply has been exhausted, the calculation may exceed 1.0. As a benefit to readers, White et al. chose to calculate rates of adherence in more than one way. For example, the authors calcu-lated a mean MPR of 93% for patients using mail-service pharma-cy and an MPR of 82% for community pharmacy patients. However, over the one-year follow-up period, mail-service patients refilled prescriptions with an average total days supply that covered only 81% of the observed days, while community pharmacy patients' refills covered, on average, only 57% of the observed days. While mail-service patients had longer mean days of persistence compared to community pharmacy patients (280 days versus 214 days, respectively), the authors did not report rates of discontinuation within the study groups, which would have allowed for an insightful comparison. Previous research has shown discontinuation rates of 10% to 60% within this class of medications.