The American Journal of Managed Care November 2008 - Special Issue
How Similar Are States' Medicaid Preferred Drug Lists?
Objective: To compare the generosity and consistency of 10 states’ Medicaid preferred drug lists (PDLs) in high-volume therapeutic classes.
Study Design: Descriptive comparisons of 7 of the top 10 therapeutic classes by Medicaid sales and of the top 10 most populous states with Medicaid PDLs.
Methods: A PDL specifies which drugs are available to patients without receiving prior approval from the state. State PDLs were collected in January 2008 to determine the status (covered or not covered) of 110 different drugs in each state. The US Food and Drug Administration Orange Book provided patent status for each drug. States were compared for generosity and similarity of coverage overall, by patent status, and by therapeutic class.
Results: For 42 (38%) of the drugs, there was wide consistency in PDL design, with at least 9 states classifying the drug with the same PDL status. For the other 62% of drugs, there was greater variation, with 2 or more states classifying the drugs differently than the others. Generosity and consistency also varied by therapeutic class and patent status.
Conclusion: For most drugs, Medicaid PDLs are not implemented consistently across states, suggesting that states do not rely on common clinical evidence to make value-based coverage decisions. Greater involvement by the federal government in designing or regulating monopolistic Medicare Part D PDLs may result in similar inconsistencies.
(Am J Manag Care. 2008;14(11 Spec No.):SP46-SP52)
Medicaid programs use preferred drug lists (PDLs) to control prescription drug spending. Comparison of the generosity and consistency of PDL coverage in the 10 most populous states, with PDLs for 110 active ingredients in 7 of the top therapeutic classes, showed that:
- On average, states covered 67% of the active ingredients.
- For 62% of ingredients, at least 2 states differed from the others in their coverage decisions.
- States were most similar in their coverage of anticonvulsants, but disagreed most in their coverage of antidepressants and antipsychotics.
- States did not consistently apply clinical evidence to make value-based coverage decisions.
Preferred drug lists fall in the middle of the range of formulary design, with open formularies at one end and closed formularies at the other. Open formularies provide insurance coverage for any drug, whereas closed formularies provide coverage only for drugs on the formularies. The credible threat to exclude certain drugs from coverage is shared by both closed formularies and PDLs, and enhances payers’ ability to bargain for lower prices. In addition to lowering negotiated prices, closed formularies and PDLs can reduce spending by shifting prescribing decisions toward lower-cost alternatives.
Previous research indicates that PDLs have lowered states’ growth in prescription drug spending.1 Other work using microdata found that PDLs have altered prescribing decisions,2 and in some contexts this appears to have negative implications for quality.3 PDLs also appear to restrict patients’ access to medications, particularly for newer drugs that have been linked to higher quality,4,5 and to reduce patient adherence to continuing medication therapy.6 These effects of PDLs occur despite their coupling with PAs, likely because the unreimbursed administrative burden of requesting PA promotes greater prescribing of drugs covered by the PDL.7,8
Despite these concerns with Medicaid PDLs, state PDL design might reflect value-based purchasing principles if the cost savings exceed the costs associated with the negative effects of PDLs. If states’ pharmacy and therapeutics (P&T) committees rely on scientific evidence to weigh these tradeoffs, they can create PDLs that promote value-based prescribing, for example, by covering generic products but excluding much costlier on-patent therapeutic substitutes with potentially higher marginal effectiveness. If states’ P&T committees rely on a consistent base of national or international clinical evidence, then the resulting PDLs themselves should be largely consistent across states. Recent efforts have focused on promoting Medicaid programs’ use of clinical evidence; for example, the Foundation of Managed Care Pharmacy established a standardized format for drug manufacturers to report data regarding drugs’ value for use by P&T committees.9,10 Additionally, a number of state Medicaid agencies have joined the Drug Effectiveness Review Project, which combines existing research about the comparative value and outcomes of drugs in a given therapeutic class into a single source available to both member and nonmember states.11
We examined the level of generosity and extent of agreement among states’ prescription drug coverage decisions, as well as how coverage and consistency varied with patent status and therapeutic class. The purpose was to gain insight into whether states’ formulary designs rely on a consistent evidence base, or whether states differ in their criteria for determining coverage. Existing research indicates that states had diverse coverage prior to the recent growth in adoption of PDLs and the growth in evidence-based medicine and economic evaluations of drugs.12 Research on private insurers likewise indicates substantial variation in formulary design because of various organizational characteristics.13 Other researchers found that neither Florida’s Medicaid formulary nor a private formulary had evidence-based or value-based designs.14
Because state Medicaid PDLs are not available in an electronic, easily comparable format, we limited the scope of the analysis to the largest states and therapeutic classes. Specifically, we gathered state Medicaid PDLs in place as of January 2008 to examine 1100 formulary design decisions (110 drugs in 10 states). We determined which states had PDLs from a survey conducted by the Kaiser Family Foundation15 and, for nonrespondents, by examining each state’s Medicaid Web site. Of the 37 states that have implemented PDLs, we included in this study the 10 most populated: Illinois, Florida, Texas, New York, California, Ohio, Michigan, Georgia, Virginia, and Massachusetts.
To choose which therapeutic classes to study, we ranked them using the Universal System of Classification (USC) based on their Medicaid sales in the 12 months before March 2008 as reported by the Wolters Kluwer Health’s Pharmaceutical Audit Suite tool.16 We included 7 of the top 10 3-digit USCs: analgesics-narcotics (USC 022), anticonvulsants (USC 202), antidepressants (USC 643), antihyperlipidemics (USC 321), antipsychotics (USC 641), antiulcerants (USC 234), and inhaled steroids (USC 284), and excluded antibiotics (USC 151), HIV antivirals (USC 821), and analeptics (USC 645). Because of legal requirements preventing states from limiting access to critical care, states have exempted certain therapeutic classes from their PDLs, essentially leaving the formularies open for those classes. Thus, we excluded HIV antivirals, which were exempted from all of the state PDLs in this study. We excluded antibiotics because states often covered some of a drug’s strengths and formulations but not others. We excluded the tenth highest-selling class, analeptics, because as central nervous system stimulants they were subject to additional restrictions beyond the PDL in Michigan and New York, and PDL coverage of a given drug varied by patient age in Georgia and California. We examined PDL status at the active-ingredient level, which we refer to as a “drug.” Because we considered only the drug level, we excluded differences across states by formulation, strength, brand name, or other specific aspects of a product. Together, the 7 therapeutic categories included 110 different drugs that accounted for 39% of Medicaid sales (by dollars) nationwide for the 12 months before March 2008.16 Each of these therapeutic classes had been reviewed previously by the Drug Effectiveness Review Project. Patent status for each drug was obtained from the US Food and Drug Administration (FDA) Orange Book. Drugs were characterized as off patent when the FDA Orange Book listed therapeutic-equivalent drugs that were sold by drug manufacturers other than the proprietary manufacturer, or when any form or strength of the drug became off patent. Table 1 lists the therapeutic classes chosen and the number of drugs, the proportion of drugs on patent, and which states exempted the category from their PDLs.
We examined each state’s PDL to determine whether each drug was covered. Drugs were deemed covered if they were listed with preferred status or did not require PA, or if the class was exempted from the PDL altogether. (The Massachusetts PDL occasionally designates for some specific drugs that PA is required if more than 30 units are required per month. In these cases, if PA was required only for supplies for more than 1 month, the drug was still counted as being on the PDL.)
Because states sometimes distinguish between a drug’s brand and generic versions or between different formulations (eg, extended release vs short acting) in their coverage decisions, we considered a drug to be covered if any version was available without PA. Combination products of 2 or more drugs were viewed as distinct drugs, and they were considered on patent if any of the drugs were on patent.
Using these data, we generated descriptive results and where appropriate performed Fisher’s exact tests or t tests to indicate the generosity of coverage and the degree of consistency in coverage across states. Specifically, generosity was measured by the proportion of drugs that were covered by each state. Consistency was measured by the number of states that had the same coverage for each drug. For example, states were considered unanimous on a given drug’s status either if none included it on the PDL or if all 10 did, but they had the greatest disagreement if 5 states included the drug and the other 5 did not.
The generosity of each state’s coverage overall and by patent status is shown in Table 2. States varied in their coverage of drugs in these high-volume classes, with a difference of 33 percentage points between the state covering the fewest (Georgia, 48.2%) and the most (Michigan, 80.9%). The states covered two thirds of these drugs on average; although with a standard deviation of 10.7 percentage points, there was a degree of variation across states. Overall, states were more generous in their coverage of off-patent drugs (70.1% vs 59.0%), and these differences were significant at P <.05 for 4 states individually. There was greater variation in coverage for off-patent drugs in terms of both standard deviation (14.5 percentage points) and the difference between the most and least generous states (42.7 percentage points).
Table 3 indicates the consistency in coverage across states at the drug level. States were unanimous in their coverage decisions for 20 (18.2% of 110) drugs, with only 2 not covered by any of the PDLs and 18 covered by all 10. States were in greatest disagreement for 31 (28.2%) drugs, with at least 4 states covering them and at least another 4 not covering them. A majority of drugs (54%) had a more moderate amount of disagreement, with 1 to 3 states choosing the PDL status opposite from the 7- to 9-state majority. No clear differences existed for consistency in coverage of on-patent versus off-patent drugs. States were less likely to be unanimous for on-patent drugs (14.3% vs 20.6%) but also were less likely to have the greatest degree of disagreement for them (26.2% vs 29.4%), and neither difference is statistically significant.