Utilization and spending trends for antiretroviral medications in the U.S. Medicaid program from 1991 to 2005

Background HIV/AIDS incidence and mortality rates have decreased in the U.S. since 1996. Accompanying the longer life spans of those diagnosed with the disease, however, is a tremendous rise in expenditures on medication. The objective of this study is to describe the trends in utilization of, spending on, and market shares of antiretroviral medications in the U.S. Medicaid Program. Antiretroviral drugs include nucleoside reverse transcriptase inhibitors (NRTIs), protease inhibitors (PIs), nonnucleoside reverse transcriptase inhibitors (NNRTIs), and fusion inhibitors (FIs). Methods Utilization and payment data from 1991 to 2005 are provided by the Centers for Medicare & Medicaid Services. Descriptive summary analyses were used to assess quarterly prescription numbers and amounts of payment. Results The total number of prescriptions for antiretrovirals increased from 168,914 in 1991 to 2.0 million in 1998, and 3.0 million in 2005, a 16.7-fold increase over 15 years. The number of prescriptions for NRTIs reached 1.6 million in 2005. Prescriptions for PIs increased from 114 in 1995 to 932,176 in 2005, while the number of prescriptions for NNRTIs increased from 1,339 in 1996 to 401,272 in 2005. The total payment for antiretroviral drugs in the U.S. Medicaid Program increased from US$ 30.6 million in 1991 to US$ 1.6 billion in 2005, a 49.8-fold increase. In 2005, NRTIs as a class had the highest payment market share. These drugs alone accounted for US$ 787.9 million in Medicaid spending (50.8 percent of spending on antiretrovirals). Payment per prescription for each drug, with the exception of Agenerase®, increased, at least somewhat, over time. The relatively expensive drugs in 2005 included Trizivir® ($1040) and Combivir® ($640), as well as Reyataz® ($750), Lexiva® ($700), Sustiva® ($420), Viramune® ($370), and Fuzeon® ($1914). Conclusion The tremendous growth in antiretroviral spending is due primarily to rising utilization, secondarily to the entry of newer, more expensive antiretrovirals, and, finally, in part to rising per-prescription cost of existing medications.


Background
Since the first reported case in June 1981, approximately 1.7 million people in the United States have been infected with HIV, including more than 550,000 who have already died and an estimated 1.2 million living with HIV/AIDS in 2005 [1][2][3][4]. AIDS cases increased rapidly in the 1980s and peaked in 1992 (an estimated 78,000 cases diagnosed) before stabilizing in 1998; since then, approximately 40,000 AIDS cases have been diagnosed annually, although, over the last several years, there is some indication that diagnoses are again on the rise [5]. The HIV/ AIDS-related mortality rate rose steadily through the 1980s, peaking in 1994-1995 [6] and declining since then.
There is no known cure for AIDS. Patients infected with HIV rely on antiretroviral treatment, a life-long disease management strategy, costing between US$10,000 and US$15,000 per year [7]. Once drugs from several antiretroviral drug classes were available beginning in 1996, the introduction of therapies incorporating combinations of drugs from two or three different drug classes has led to a wide range of possible antiretroviral therapy combinations. These new combinations, often referred to as highly active antiretroviral therapy (HAART), have been shown to have a significant impact both on markers of disease progression (viral load and CD4 T-cell counts) [8][9][10] and on HIV/AIDS-associated mortality and morbidity [6, [11][12][13][14]. National guidelines for the treatment of HIV infection recommend HAART as first-choice therapy [15].
Accompanying HAART and the longer survival rates offered by HAART has been a marked rise in expenditures on antiretroviral medications. The U.S. Medicaid Program bears a substantial burden in HIV/AIDS expense coverage, especially in the latter stages of the disease when individuals are too ill to work. The objective of this study is to describe the trends of utilization of and spending on antiretroviral drug classes as well as individual antiretroviral medications in the U.S. Medicaid Program. These results provide useful information to policy makers and health professionals interested in cost-effectiveness and cost-containment strategies along with their usual concerns of safety and efficacy.

Methods
We study utilization, spending, and market share for each of the antiretrovirals listed in Table 1 [16]. Each drug is classified as a nucleoside reverse transcriptase inhibitor (NRTI), a protease inhibitors (PI), a nonnucleoside reverse transcriptase inhibitor (NNRTI), or a fusion inhibitor (FI). Its manufacturer, approval date, and time to approval (after new drug application submission) are all identified. Note the relatively short pre-approval periods for all of the drugs due to the AA priority status for AIDS therapy applications [17]. (As of 2003, the average time to approval for all drugs was around one and a half years [18].) Retrospective descriptive summary analyses were conducted with the purpose of describing the trends in utilization of and spending on antiretroviral medications over the last 15 years. Pharmacy utilization and expenditure data, from 1991 quarter 1 through 2005 quarter 4, were taken from the national Medicaid pharmacy files provided by the Centers for Medicare & Medicaid Services (CMS). These files contain number of outpatient prescriptions and payment amounts for all National Drug Code (NDC) drug forms paid for by any state (except Arizona, but including the District of Columbia) Medicaid program [19]. The national files are huge databases representing aggregation across all the states and are subject to occasional coding error, which we corrected to the best of our ability. Prescription and expenditure data were then aggregated across all NDCs for each of the drugs listed in Table 1. Per-prescription spending (which is referred to loosely as "price" throughout this article) was calculated as total expenditure for the drug divided by total number of prescriptions. Note that per-prescription spending exceeds actual acquisition cost to Medicaid due to federal and state rebates received by Medicaid from the drug manufacturers.
Quarterly market shares for four classes of antiretroviral medications are calculated as both the percentage of total antiretroviral prescription numbers and the percentage of total antiretroviral expenditures in the U.S. Medicaid market. We refer to these market shares as the prescription market share and the payment market share, respectively. All expenditure values are expressed in current U.S. dollars.

Results
The total number of antiretroviral prescriptions paid for by Medicaid increased from 168,914 in 1991 to 3.0 million in 2005, showing a 16.7-fold increase over 15 years (see Table 2). In 1996, and again in 1997, utilization of antiretrovirals increased by more than 100 percent in a single year. The number of prescriptions for NRTIs increased from 168,914 in 1991 to 1.      (Figure 4, 5, or 6) by number of prescriptions (Figure 1, 2  Other NRTIs a gradual increase over time), and Norvir® (gradual decrease in price, followed by an abrupt price increase in 2004 quarter 1 and then followed by again a gradual decline). In 2005 quarter 4, prices for a drug in the PI class ranged from a little less than US$ 300 (Agenerase®) to US$ 926 (Aptivus®, the latest entry to the PI class in 2005 quarter 3). (Neither of these drugs is featured individually but contributes to the category "Other PIs.") Similarly to the NRTIS, the later PI entrants (such as Lexiva®, Kaletra®, and Reyataz®, and certainly Aptivus®) have entered at higher price points than their earlier counterparts. Finally, the data tell a similar story for the NNRTIs. Sustiva®, market leader and latest entrant to the market, entered at a higher price than either Rescriptor® (ignoring its price in the first quarter for Medicaid) or Viramune®. Each of the three drugs has shown a price increase over time. Since 1998 quarter 3, Rescriptor® has had a price increase of 37.3 percent, Sustiva® of 24.2 percent, and Viramune® of 62.0 percent, as of 2005 quarter 4.

Discussion
In the preceding section, we showed a substantial rise in Medicaid expenditures on antiretroviral medications. While most of that rise can be explained by the rise in utilization of antiretroviral medicines (this utilization increase is observed in the face of decreasing or stable HIV/AIDS incidence rates in the U.S.), some is certainly attributable to both the entry of newer, more expensive drugs in all of the antiretroviral drug classes and price increases of drugs once they are being marketed.
The first factor driving the increase in Medicaid prescriptions over the past decade and a half is the move from monotherapy to combination therapy. In a short period of time after the approval of the PIs, HAART became the standard treatment for those infected with HIV, implying an increase in the use of double-and, more recently, triple-combination antiretroviral therapy regimens, with drugs across antiretroviral drug classes, among HIVinfected persons. Medicaid patients consistently, however, have lower use rates for the newer antiretroviral drugs than the general population [20,21]; estimates for various states indicate that HAART use among HIV-positive patients on Medicaid in 1998 ranged from 37 percent in Texas to almost 70 percent in New Jersey [22,23].
Secondly, rising utilization can be explained by declining mortality rates, leading to individuals' requiring more prescriptions over life's course. Using data from a randomassignment clinical trial, it was found that those assigned HAART therapy had a 58 percent lower mortality rate than those in the control group [24]. Survival data for patients in non-experimental settings have demonstrated that patients using HAART therapy have substantially lower mortality rates than those not using it.
During the study period, more than twenty marketed antiretrovirals were supplied by nine manufacturers. The data indicate that new entry occurs over time at higher and higher price points. Price differences among the antiretrovirals are explained to some degree by differences in effectiveness and safety profiles. Newer drugs that offer added dosing convenience and improved safety profiles are priced higher than previously popular drugs that are being used less often as their drawbacks become better defined and drug resistance is developed [15]. The data indicate as

Year and Quarter Medicaid Payment in Millions of U.S. Dollars
Rescriptor® Sustiva® Viramune® well that payment per prescription has risen over time for existing drugs in the market. The rising demand generated by both combination therapy and higher survival rates has overwhelmed any downward pressures on price.
Once a drug goes off patent, generic versions of the drug quickly enter the market. The number of entrants depends on the size of the market. A large number of generic producers will ideally drive price down to a level that is close to marginal cost. However, while didanosine (since 2004 quarter 4) and zidovudine (since 2005 quarter 4) are now available as generic drugs, low-cost versions of these two NRTIs have not yet made a big impact on the expenditure on antiretroviral medication for Medicaid. At this point, there are very few generic manufacturers. As of 2005 quarter 4, only one generic producer supplied didanosine to Medicaid. In 2005 quarter 4, the number of prescriptions for Videx ® (including Videx EC ® ) was still 12,128 (payment per prescription was $218). Comparatively, the number of prescriptions for the generic form of Videx ® , didanosine, was 16,085; spending per prescription was $201, not much lower than the brand-name price. Four generic producers sold zidovudine in 2005 quarter 4. That quarter, the average price of zidovudine, across the four manufacturers and all their NDCs, was $211. Compared to the price of Retrovir ® , $248, the generic producers were selling their drug for 85 percent of the brand-name price.

Study limitations
Drug use was inferred from aggregate prescription data, themselves based on adding up prescriptions and payments from Medicaid claims data; it was not possible to review patient-specific clinical information. We were unable to analyze the use of antiretrovirals as a function of disease treated, and unable to link the utilization with patient demographics or history of clinical conditions and comorbidities. We did not link the utilization of antiretrovirals to the prevalence rate of HIV or AIDS.

Policy implications
Medicaid plays a critical role in financing AIDS care by providing a comprehensive benefit package that includes prescription drugs [25]. The rising cost of antiretroviral medications has been a significant challenge for all state Medicaid programs in the United States. Because the antiretrovirals are protected by patents, with the recent exceptions of didanosine and zidovudine, aggressive generic substitution policies cannot be implemented to reduce Medicaid spending for this class of drugs. Other policies, however, include preferred drug lists, prior authorization, copays, and generally tighter controls on high-cost drugs [26]. Some states, including California and Florida, both of which are among the top three states in terms of numbers of people with HIV/AIDS, are looking to increase substantially copays for certain groups of individuals on Medicaid. With the relatively high number of prescriptions people with HIV/AIDS need to fill, however, such a copay increase could constitute a significant barrier to obtaining all the drugs they need [26].
From 2001 through 2004, there were 5,660 new HIV/ AIDS diagnoses reported to the Centers for Disease Control for persons 60 years old or older [3]. Moreover, one consequence of the expanded life span offered by antiretroviral treatment is that more elderly individuals will be coping with HIV/AIDS in the future. With the January 2006 implementation of a Medicare Part D prescription drug benefit, older people living with HIV/AIDS will be able to receive drug coverage from Medicare. Though drugs will be paid for differently under Medicare, Medicaid's experience with antiretrovirals during the last 15 years may shed some light on what will happen to Medicare spending over the next decade and a half, at least for this relatively small component of the overall Medicare budget.

Conclusion
In this paper, we have documented the rise in spending on antiretrovirals by the U.S. Medicaid Program since 1991. We have shown that although most of the rise can be explained by rising utilization, some is due to rising prices of both newer medications and those already established in the market. Medicaid will struggle for many years to come to keep costs contained to the greatest extent possible, making sure that access to drugs is not compromised. While there may be some relief from generic medications, the rate of innovation remains strong in the antiretroviral drug class, and we predict that, at least for the near future, costs will keep rising.