The effect of a three-tier formulary on antidepressant utilization and expenditures

J Ment Health Policy Econ. 2008 Jun;11(2):67-77.


Background: Health plans in the United States are struggling to contain rapid growth in their spending on medications. They have responded by implementing multi-tiered formularies, which label certain brand medications 'non-preferred' and require higher patient copayments for those medications. This multi-tier policy relies on patients' willingness to switch medications in response to copayment differentials. The antidepressant class has certain characteristics that may pose problems for implementation of three-tier formularies, such as differences in which medication works for which patient, and high rates of medication discontinuation.

Aims of the study: To measure the effect of a three-tier formulary on antidepressant utilization and spending, including decomposing spending allocations between patient and plan.

Methods: We use claims and eligibility files for a large, mature nonprofit managed care organization that started introducing its three-tier formulary on January 1, 2000, with a staggered implementation across employer groups. The sample includes 109,686 individuals who were continuously enrolled members during the study period. We use a pretest-posttest quasi-experimental design that includes a comparison group, comprising members whose employer had not adopted three-tier as of March 1, 2000. This permits some control for potentially confounding changes that could have coincided with three-tier implementation.

Results: For the antidepressants that became nonpreferred, prescriptions per enrollee decreased 11% in the three-tier group and increased 5% in the comparison group. The own-copay elasticity of demand for nonpreferred drugs can be approximated as -0.11. Difference-in-differences regression finds that the three-tier formulary slowed the growth in the probability of using antidepressants in the post-period, which was 0.3 percentage points lower than it would have been without three-tier. The three-tier formulary also increased out-of-pocket payments while reducing plan payments and total spending.

Discussion: The results indicate that the plan enrollees were somewhat responsive to the changed incentives, shifting away from the drugs that became nonpreferred. However, the intervention also resulted in cost-shifting from plan to enrollees, indicating some price-inelasticity. The reduction in the proportion of enrollees filling any prescriptions contrasts with results of prior studies for non-psychotropic drug classes. Limitations include the possibility of confounding changes coinciding with three-tier implementation (if they affected the two groups differentially); restriction to continuous enrollees; and lack of data on rebates the plan paid to drug manufacturers.

Implications for health care provision and use: The results of this study suggest that the impact of the three-tier formulary approach may be somewhat different for antidepressants than for some other classes.

Implications for health policy: Policymakers should monitor the effects of three-tier programs on utilization in psychotropic medication classes.

Implications for further research: Future studies should seek to understand the reasons for patients' limited response to the change in incentives, perhaps using physician and/or patient surveys. Studies should also examine the effects of three-tier programs on patient adherence, quality of care, and clinical and economic outcomes.

Publication types

  • Comparative Study
  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Antidepressive Agents / economics*
  • Antidepressive Agents / therapeutic use
  • Cost Sharing / economics*
  • Depression / psychology
  • Drug Costs / statistics & numerical data
  • Drug Prescriptions / economics*
  • Drug Prescriptions / statistics & numerical data
  • Drug Utilization / economics
  • Female
  • Formularies as Topic*
  • Humans
  • Insurance Claim Review / economics
  • Insurance Claim Review / statistics & numerical data
  • Logistic Models
  • Male
  • Managed Care Programs / economics
  • Managed Care Programs / statistics & numerical data
  • Patient Compliance / statistics & numerical data
  • Reimbursement, Incentive / economics
  • Research Design
  • United States


  • Antidepressive Agents