At a time that Medicaid programs nationally are severely pressed to afford their current drug benefits, these findings demonstrate a source of potential savings from increased use of generic medications. Although the potential savings from the two scenarios presented range between a quarter billion dollars and almost a half billion dollars, formidable amounts for programs that are seeking to contain costs in any way possible, they represent a modest proportion of total Medicaid spending on prescription drugs. The data provide some important insights for the design of a prescription drug benefit for the elderly, either under Medicare or in a private insurance context.
Much of the excess costs found result from prescriptions for which the physician specifically requested a brand name drug as “medically necessary.” The pharmacologic rationale for such a decision in most cases is dubious. One method of reducing excess spending on brand name drugs is to target physician behavior in this regard.
Prior research has shown considerable variation in physician beliefs and practices regarding generic substitution, as well as poor understanding among physicians of the FDA regulations for generic products (Banahan and Kolassa 1997
; Murphy 1999
). Physicians may be influenced by marketing information to believe that brand name drugs are somehow more effective or are held to a higher manufacturing standard, even in the absence of data supporting this belief. The important role of the physician in the decision to use a brand name or a generic drug has been described previously (Hellerstein 1998
), and some interventions have attempted to target physician behavior directly (Ahluwalia et al.
). Nevertheless, given that Medicaid already has financial regulations favoring the use of generic drugs (by limiting payment for the brand name version), this portion of potential savings may be difficult to achieve through reimbursement policy and may require additional changes in physician prescribing practices. One important limitation of our analysis was our inability to incorporate manufacturer rebates into the calculations. The rebates are calculated based on the average manufacturer's price for a drug (National Pharmaceutical Council 1998
); that price is not easily available and was not practical to use as part of this analysis.
Variations in MAC price levels are also an important source of unrealized savings. The drug-by-drug analysis () demonstrates that most potential savings come from a small number of medications. Indeed, in the main analysis, clozapine alone accounts for over 10 percent of the potential savings. The best MAC price available analysis provides evidence that some states are more successful at controlling costs for specific medications. The large increase in potential savings for some of the medications implies that much lower prices for these medications are available in some states than others. There is considerable heterogeneity among states in the proportion of drug spending which could have been saved by greater use of generic medications (). Variation in how states implement their MAC programs for prescription drug price limits may account for such differences. Future research comparing the details and operationalization of MAC programs across states may provide important information for the design of a Medicare prescription drug benefit and could help realize some of the potential savings. Policy efforts that target specific medications may be able to address areas of excess spending in an efficient manner.
In important ways, the data presented above represent a conservative estimate of the potential savings from more widespread use of generic drugs. By limiting the savings calculation to tablets and capsules, we assumed that there were no potential savings for inhaled medications, topical medications, transdermal patches, and several other classes of widely used drugs.
There has been much interest in more aggressive substitution of different drugs within a class (McAlister et al. 1999
), (therapeutic substitution, e.g., switching one ACE inhibitor for another), and this practice is currently mandated in many private insurance plans. A recent study of a reference-pricing system under which a health plan would only pay for lower priced ACE inhibitors demonstrated cost reductions without adverse effects on patients or other health care expenditures (Schneeweiss et al. 2002
). We did not attempt to incorporate this source of savings into our analysis; all potential savings described here result from the replacement of a brand name product with a generic drug of identical chemical composition and duration of action.
One important question that has been addressed in prior research (Bae 1997
; Stolberg and Gerth 2000
) is the pattern over time of market entry and prescribing of generic drugs after a medication goes off patent. Our research examined unrealized savings only for drugs for which generics were already in the marketplace. More information on the patterns of adoption of generic alternatives as they become available would help clarify one cause of the excess costs from use of brand name drugs described above. The rate at which physicians begin writing for generic names of drugs or stop invoking medical necessity for brand name drugs will likely correlate with the extent and effectiveness of marketing of the original brand name drugs. Further research is also needed on the causes of delays in commercial availability of generic alternatives after patents have expired on the original brand name products (Stolberg and Gerth 2000
These findings provide evidence that important savings could be realized in the Medicaid drug program through more widespread use of generic medications. The estimates presented are conservative, and the true potential savings is likely larger. As both public and private insurers struggle to accommodate increases in the cost of drug coverage, multiple strategies will be needed to control expenditures. This analysis points to one area in which excess expense can be avoided with no compromise in clinical outcomes.