Variation in and Savings from Therapeutic Substitution Across State Medicaid Programs

With an 18-20% prevalence rate compared to 8-10% for the general population, depression is particularly prevalent among Medicaid recipients and represents a significant cost burden.  Prozac®, the first of a number of successful selective serotonin reuptake inhibitors (SSRIs), was considered a major advance in the treatment of depression.  In August 2001, the patent for Prozac® expired. Although all state Medicaid programs eventually benefited from the lower-priced generic substitutes that entered the market, they varied considerably in speed of generic uptake and consistent updating of their reimbursement limits as generic prices dropped over time.  As a result, there were significant uncaptured savings during the transition period from a single- to multi-firm market (Kelton, Chang, Kreling, 2013).

An additional source of savings was Medicaid’s ability to shift market share from other branded SSRIs to generic fluoxetine, a practice referred to as therapeutic substitution.  In 2001, there was limited intra-SSRI-class substitutability partly because of differences in their pharmacokinetic properties but primarily because of the effectiveness of manufacturers’ direct-to-consumer advertising and sales force detailing to physicians (Elzinga, Mills, 1997). Although payers, including Medicaid programs, could have insisted on therapeutic substitution of generic fluoxetine once the patent for branded fluoxetine expired, they varied in their efforts and abilities to do so.          

We first document the variation across states in the loss of fluoxetine’s market share following patent expiration for Prozac®.  Preliminary results reveal substantial variation:  between the calendar quarter immediately prior to generic entry and the end of 2005, the share of fluoxetine dropped from 23.38% to 14.31% for New York Medicaid beneficiaries but only from 22.94% to 21.55% in Michigan.  In a few states, the share actually rose, for example, in Wisconsin, from 20.18% to 25.97%.

Secondly, we compute the uncaptured savings that each state could have recovered from 2001 quarter 3 through 2005 quarter 4 had it adopted a moderately aggressive policy toward SSRI cost savings, involving the encouragement of fluoxetine prescribing to all newly diagnosed patients with depression but not requiring patients currently on other branded SSRIs to switch to fluoxetine.  The savings calculations assume three different scenarios regarding generic uptake and reimbursement rates for generic fluoxetine. Again, preliminary results showed substantial variation in total savings per SSRI prescription: $3.47, $6.64, and $9.23 (in 2005 quarter 4 dollars) for Michigan, Wisconsin, and New York, respectively.     

Finally, to explain the variation across states in both market-share change and per-prescription uncaptured savings, we explore whether state-specific policies aimed at lowering costs have an effect. For instance, a program’s tiered copayments for generic and brand-name drugs might lead to greater therapeutic substitution, with less costly generic fluoxetine’s being preferred to branded alternatives. Physician education on the availability and use of generic drugs could also lead to enhanced substitutability.  Furthermore, the availability of a prior authorization program or preferred drug list might enhance a state’s ability to take advantage of therapeutic substitution provided they were updated regularly after generic entry  and monitored to ensure access to other SSRIs should fluoxetine prove ineffective for a particular patient.