We observed a significant delay in the filling of clopidogrel prescriptions under the PA system compared with the new AP process, and this delay placed patients at risk. Indeed, we observed two repeat revascularizations that were likely due to these delays. A more global problem was that fewer than three-quarters of patients in both groups filled their clopidogrel prescriptions within 60 days of discharge.
Given the rising costs of pharmaceuticals and the significant proportion of the population covered by some form of drug benefit insurance, there is a great deal of interest in the impact of drug coverage on drug use, health care costs and health outcomes. Unfortunately, the majority of studies exploring these questions suffer from methodological flaws, including poor data quality and availability, selection bias, lack of applicability and failure to consider clinical outcomes.
The PA process restricts the use of prescription drugs by requiring approval from the insurer before dispensing, and it is widely used in Canada and the United States. In general, the PA process targets high-cost drugs restricted for use in specific clinical situations or patient criteria (such as clopidogrel post-stent insertion); specific, effective drugs for which there are less costly therapeutic alternatives; and step therapy programs that require a previous trial of another agent (3
). It is frequently stressed that the PA process must be timely and that it may not be appropriate for all classes of medications or medical conditions. The traditional program, which was used by ABC in the present study, requires staff adjudication of individual cases according to established criteria. Online adjudication with computer decision trees has been explored as a means of improving efficiency (3
There have been several previous studies assessing PA processes. However, they have tended to focus on medications for chronic conditions, such as nonsteroidal anti-inflammatory drugs (6
) and selective serotonin reuptake inhibitors (9
). Most of these did not assess clinical outcomes, and those that did were hampered by poor survey response rates. In our study, the PA process was used for acute indication, and we demonstrated significant potential for unintended negative clinical outcomes.
The change to an AP process improved patient access to therapy. Nevertheless, only three-quarters of patients filled their prescription for clopidogrel, leaving at least one-quarter at unacceptably high risk for subacute stent thrombosis. This finding was unexpected and requires further investigation.
There are a number of limitations associated with the present study. The partially controlled design does not provide the same level of causal inference as a randomized controlled trial. A comparison group that did not have this change in coverage was not available, so exogenous political or social changes affecting the outcomes of interest would not have been detected. The present retrospective study did not examine either patient education or discharge instructions, nor was the socioeconomic status of each patient ascertained. However, there were no other known initiatives concerning these clinical areas identified during the study period. As in all nonrandomized studies, there was potential for a selection bias. However, the ABC policy change provided the unique opportunity to study the impact of PA procedures in a real-life setting, in which a randomized trial is unlikely to be conducted. Although the inclusion of patients older than 65 years of age may have limited the applicability of our results, this population is a major user of health care resources, is restricted by a fixed income and is commonly covered by government-based insurance. It was not possible to ascertain whether any patients who delayed their prescription fill were provided samples or used a spouse’s prescription, or whether patients were provided a portion of their prescription at no charge by the community pharmacy until authorization was received. Chart review determined that no patient had been supplied with pass or discharge clopidogrel. The availability of private supplemental health insurance coverage was not determined. However, in the majority of cases, this coverage is not applied until ABC rejects a claim or a portion thereof. Finally, the ABC database only supplied information on the patient collecting the prescription, with no measure of adherence thereafter.