The ripple effects of a restrictive Medicaid formulary.

1 analyzes the effects of an interesting natural experiment— Maine's adoption of a restrictive Medicaid drug for-mulary in January 2001. The focus of the paper is on spillovers from this restrictive formulary to other sectors of the market. In particular, it investigates how the implementation of this formulary affected the prescribing behavior of physicians with respect to their non-Medicaid patients—those patients that pay out of pocket, have their prescriptions reimbursed by an insurance plan, and those that utilize a mail-order supplier. Medicaid-preferred drug programs differ from those employed by HMOs in that the prevalent 3-tier copayments are not a feasible approach for the Medicaid population. Accordingly, many states with severe pressures on their Medicaid budgets are now turning to essentially closed formularies where prior authorization is necessary to obtain nonpreferred drugs. The Wang paper focuses on the proton pump inhibitors (PPIs) class of drugs for gastroesophageal reflux disease. This is one of the therapeutic classes with the highest level of drug consumptive expenditures. In the case of the Maine program, the only preferred PPI drug on this formulary was pantoprazole. In the preformulary period, this drug had a market share of only a few percent. The authors structure their analysis as a before-and-after event study, with a 3-state comparison design (New Hampshire and Vermont serve as controls). The data set, while limited in scope in terms of insurance data, is an excellent one to consider some first-level hypotheses on spillover effects. Data on individual prescribing physicians are the basic unit of observation. Given the experimental design and data, the results are very persuasive with respect to the main hypotheses examined. In particular, the institution of the state's restrictive formulary has resulted in a dramatic change in the preferred drug's market share in Maine (79% increase in Maine compared to 1% to 2 % in New Hampshire and Vermont). Second, and more interesting , there are substantial spillovers to cash and third-party payer of the prescriptions. The market share of pantoprazole increased 10% in Maine among cash prescriptions (versus 3% in the control states) in the period after the Medicaid formulary was implemented. Similarly it increased 7% in Maine among other third-party prescriptions (versus 1% among the control states). The authors utilize a linear regression model to quantify the extent of spillovers. These spillovers are economically and statistically significant, and increase with the share of a prescriber's Medicaid practice. One concern …