Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Adherence to a prescription drug treatment regimen for a chronic condition is generally considered to be an important factor in maintaining and improving a patient's health status. Higher levels of maintenance prescription drug compliance have been associated with better clinical outcomes for individuals with a chronic condition (Steiner and Prochazka 1997). Numerous studies addressing the consumption of maintenance drugs have found that compliance rates are generally suboptimal and vary by medication class. In most studies researchers have attempted to explain differences in refill behavior using personal or clinical characteristics like age or gender, which have produced inconsistent results. This study examines refill adherence patterns for three classes of maintenance medications commonly used to treat chronic illness: oral antidiabetic medications, cholesterol-lowering statins and thyroid medications. We examine the healthcare experience of employees in one large firm from the MarketScan Database who were continuously enrolled for 20 calendar quarters, the third quarter of 1997 through second quarter 2002. Employees receiving any of these medication classes in the pre-period were identified using a combination of therapeutic classification and National Drug Codes: oral antidiabetic medications (2,127 employees), statins (6,389 employees) and thyroid hormones (2,611 employees)

This study explores the effects of prior consumption and habit persistence on current consumption using a linear feedback model, a dynamic count panel data model. This allows us to examine two aspects of persistence: an accumulated stock from a 7-quarter presample period (3Q97 through 1Q99) and a short-term effect from utilization in the prior quarter during a thirteen quarter study period (2Q99 through 2Q02). This model adopts a presample mean to measure a patient's unobserved heterogeneity directly. Explanatory variables included sociodemographic characteristics, health plan type, copayments, medication variables and time.

Persistence played a large role in explaining refill adherence. The marginal effect of one prescription filled in the previous quarter resulted in a 0.59 increase in the number of oral antidiabetic prescription fills, 0.41 thyroid drug prescription fills and .37 statin prescription fills in the current quarter (all p < 0.01). A one unit increase in consumption in the presample period (from an average of 1 to 2 prescription fills per quarter in the seven presample quarters, for example) had about half or slightly larger than half the effect as a one unit increase in the number of prescriptions filled in the previous quarter (p < 0.01). A rank-ordering of persistence effects by medication class matched an a priori ranking (diabetes medications, thyroid medication and statins) based on the time frame (immediate to long term) in which medication benefits were likely to be discerned by the patient.

We found that previous refill behavior strongly associated with current filling patterns and previous levels of prescription drug utilization can be used to identify patients who will be compliant or non-compliant.