Dispense as written requests may have important implications for patient adherence. Patients are more adherent when they are initiated on generic or lower-cost medications.9 Although dispense as written requests would seem to reflect a conscious decision by patients or their physicians to use a specific agent, the increased cost-sharing that results from a dispense as written designation may decrease the likelihood that patients actually fill their prescriptions.

Little is known about how frequently physicians or patients request dispense as written, the medications for which dispense as written is most commonly used, and the physician and patient characteristics associated with dispense as written requests. Previous descriptive analyses of this practice were conducted when generic use was far less common and have limited applicability to current practice.10, 11, 12 No prior studies have investigated the effect of dispense as written requests on patients’ likelihood to fill their prescriptions or on overall health system costs. By using transactional data from a large national pharmacy benefits manager, we assessed rates and correlates of dispense as written requests and the relationship between these requests and rates of filled prescriptions.

Materials and Methods

Sample
We identified all patients enrolled in employer-sponsored health plans who received pharmacy benefits from CVS Caremark in calendar years 2008 and 2009. From this sampling frame, we selected all members with a) continuous eligibility for pharmacy benefits between July 1, 2008, and January 31, 2009, and b) a valid entry for gender and date of birth in the administrative record. We limited our sample to patients who submitted a prescription claim from any retail or mail-order pharmacy for adjudication between January 1 and 31, 2009, the identification period. We excluded all clients who were enrolled in a plan that imposed penalties for dispense as written requests, because dispense as written requests in these settings may not reflect those in the general population.

For patients who submitted multiple eligible prescriptions within a class, we selected the prescription with the latest date during the identification period, which likely eliminates accidental or erroneous prescriptions that were delivered. If a member submitted prescriptions in multiple therapeutic classes, each was deemed eligible and included in the analysis. We included all therapeutic categories defined by MediSpan’s Generic Product Identifier (GPI-2).

We identified the dispense as written assignment and the brand status from the submitted claim record. We categorized claim records into 1 of 3 mutually exclusive categories: a physician-assigned dispense as written (Physician Dispense as Written); a member-assigned dispense as written (Patient Dispense as Written); or no Dispense as Written. Brands were listed as either single-source brand, indicating no generic equivalent was available at the time of the study, or multi-source brands, which are branded medications with a generic equivalent. For every eligible prescription, we also determined whether the prescription was purchased (“filled”) or reversed. Reversed prescriptions are those that the patient chose not to purchase, and therefore were not filled.