Cost-effectiveness of intensive atorvastatin treatment in high-risk patients compared with usual care in a postgeneric statin market: Economic analysis of the aggressive lipid-lowering initiation abates new cardiac events (ALLIANCE) study

Clinical Therapeutics/Volume 30, Theme Issue, 2008

Cost-Effectiveness of Intensive Atorvastatin Treatment in High-Risk Patients Compared with Usual Care in a Postgeneric Statin Market: Economic Analysis of the Aggressive Lipid-Lowering Initiation Abates New Cardiac Events (ALLIANCE) Study C. Daniel Mullins, PhD1; Gail B. Rattinger, PhD, PharmD1; Andreas Kuznik, PhD2; and MichaelJ. Koren, MD3

University ofMaryland) School ofPharmacy) Baltimore) Maryland; 2Pfizer Inc.) New York) New York; and Jacksonville Center for Clinical Research) Jacksonville) Florida 1

ABSTRACT Background: Intensive lipid-lowering therapy reduces cardiovascular (CV) event rates more than moderate therapy. Currently available generic statins cost less but provide less risk reduction. Objective: Our goal was ro calculate the incremental cost-effectiveness ratio (ICER) comparing an intensive atorvastatin-based regimen versus usual care, defined as prevalent statin therapy used at the time of the Aggressive Lipid-Lowering Initiation Abates New Cardiac Events (ALLIANCE) study-a trial of usual community practice that was conducted among patients enrolled in managed-care plans. Methods: A 7-year cost model incorporated patientlevel CV event rates and mortality data from the ALLIANCE study. To address the managed-care perspective, costs for clinical end points were estimated from administrative claims data for equivalent events. Direct medical costs included drugs, acute events, and a maximum of 36 months of follow-up costs. Average wholesale pricing was used for drug costs. Differences in patient copayments were incorporated and a 3% discount factor was applied. Scenario analyses explored 3- and 5-year time frames, varying discount rates, drug costs, and copayment assumptions. Results: The baseline ICER using all events examined in the ALLIANCE study and a 7-year time frame was US $10,344 (year-2007 value) per event avoided. Over 7 years, the number needed to treat to avoid a clinical end point was 11 patients. Using a realworld economic model that considered arorvastatin becoming generic in 3 years, starting patients on

2204

branded and converting them to generic atorvastatin would be a cost-saving strategy compared with current generic statin use. Conclusions: Intensive lipid-lowering using arorvastatin increased prescription drug costs of treating CV disease patients but reduced CV events. Medical cost offsets associated with reduced events, based on data from the ALLIANCE study, resulted in ICERs within accepted cost-effectiveness ranges comparing atorvastatin with a generic statin. Atorvastatin-based regimens produced cost savings from a managed-care perspective when the anticipated impact of the generic availability of arorvastatin was modeled. (Clin Ther. 2008;30[Theme Issue]:2204-2216) © 2008 Excerpta Medica Inc. Key words: cost-effectiveness, statins, cholesterollowering drugs, coronary heart disease.

INTRODUCTION Although the discovery of more effective pharmaceutical treatments is clinically welcome, these successes raise concerns as to how to implement and pay for these medical advances. The dilemma of funding evergreater pharmaceutical expenditures may become even more challenging in the future. Prescription drug expenditure increases will likely continue to accrue because of the following factors: (1) the aging of

- ------ - -

.........•..........•..•

Accepted for publication October 16, 2008. do;:l 0.1 016/j.c1inthera.2008.12.007 0149-2918/$32.00 © 2008 Excerpta Medica Inc. All rights reserved.

Volume 30 Theme Issue

CD. Mullins et al.

American and developed-world populations; (2) new users of medicines resulting from increased prevalence or diagnosis of disease; (3) expanded drug coverage under Medicare Part D and other plans; (4) pure price inflation because of price increases for existing products; and (5) product shifts to newer and often more expensive medicines in place of older therapies. 1- 8 Since at least the early 1990s, payers have sought various strategies to offset increases in drug expenditures. Pharmacy benefit managers (PBMs) and other insurers have developed pharmaceutical management programs aimed at curbing prescription drug growth. Some of their efforts, such as greater use of generic drugs, have reduced total prescription drug spending, while other pharmacy management strategies, such as tier placement and higher copayments, have merely shifted the costs from the PBM or insurer to the patientJ,9-14 Although clinicians and patients generally support greater generic drug use, one implementation of these policies-the use of therapeutic substitutionspresents vexing tradeoffs. 15 ,16 Under this circumstance, an older generic compound might be used in place of a newer proprietary compound that has proven advantages over its predecessor, but also costs more. Payers who wish to encourage therapeutic substitutions through incentives can reduce their acquisition costs for pharmaceuticals but may not know the downstream effects of these policies. 9,17-2o Cardiovascular (CV) drugs, notably including lipidlowering agents, have been an important part of the increase in drug expenditures over the past decade. The release of the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) guidelines in 2001, and the update in 2004, led to a marked expansion in the number of patients treated for hypercholesterolemia. In turn, greater use of lipidlowering therapies contributed to substantial increases in lipid-lowering therapy expenditures, particularly for statins, the most widely prescribed drug class. 21 - 2s The NCEP ATP III guidelines divide the general adult population into 3 different risk categories, each with different goals for low-density lipoprotein cholesterol (LDL-C) levels. Low-risk patients comprise those who do not have coronary heart disease (CHD), who have ::;1 CHD risk factor, and who have targets for LDL-C <160 mg/dL. Those with known CHD or CHD risk equivalents are considered high risk. Their LDL-C target is at least <100 mg/dL, with an optional goal of 70 mg/dL. This latter goal is based on studies that

2008

show improved clinical outcomes attributable to intensive atorvastatin therapy in high-risk patients, even among those with modest baseline elevations of LDL-C. For example, in the Pravastatin or Atorvastatin Evaluation and Infection Therapy-Thrombolysis in Myocardial Infarction 22 trial,26 the strategy of treating all patients with acute coronary syndromes using highdose atorvastatin significantly outperformed using the now generically available drug pravastatin, which provides a moderate degree of LDL-C lowering. These study results and more aggressive guidelines now necessitate greater utilization and higher statin doses for many more patients. 27,28 Although one might conclude that these clinical recommendations would increase statin expenditures, the revised guidelines were released at the same time that generic statins became available. Generic lovastatin (available since 2004), pravastatin (available since 2006), and simvastatin (available since 2006) are indicated therapeutic options for many patients who need hyperlipidemia treatment. 22 ,29-31 The availability of these drugs has led to recent drug expenditure forecasts that predict a substantial decrease in spending on statins because of the lower acquisition cost of these generic drugs. 32-34 Unfortunately, these predictions concentrate almost exclusively on the impact of generic statins on the pharmacy budget, a phenomenon referred to as a silo budget mentality because they do not address questions regarding the impact of generic statins, compared with more potent statins, on total direct medical costs. Before the market entry of generic simvastatin in 2006, branded atorvastatin * generated $13.6 billion globally and was the world's top-selling pharmaceuticaP5 Research suggests that atorvastatin is more effective than simvastatin,36,37 which may explain why the branded product had a preferred status on many formularies before generic statins became available. 38 ,39 In 2008, many formulary decision-makers raised generic simvastatin to a first-line agent on their preferred drug lists and struggled to determine whether and when to continue to provide unlimited, rather than step-edit, access to atorvastatin (eg, second tier). To address this question, we developed a model and calculated the predicted differences in total direct medical costs to an insurer using branded atorvastatin versus a generic statin (ie, simvastatin) using the best available outcomes evidence from the Aggressive Lipid*Trademark: Lipitor® (Pfizer Inc., New York, New York).

2205

Clinical Therapeutics

Lowering Initiation Abates New Cardiac Events (ALLIANCE) study,40 a head-to-head randomized clinical trial and real-world study of managed-care patients. Furthermore, to probe the effect of market availability of generic atorvastatin, we performed scenario analyses assuming the availability of low-cost generic atorvastatin at the time of its patent expiration in 3 years.

METHODS Overview

The ALLIANCE study was a randomized, controlled, open-label, multicenter clinical trial that compared CV clinical outcomes in a total of 2442 patients with established CHD. Patients were treated to an LDL-C target of <80 mg/dL either with an intensive atorvastatin regimen (atorvastatin arm), up to the maximum 80-mg atorvastatin dose (based on physiciandirected titration), or with usual care with a lipidlowering regimen established and prescribed by the patient's physician (usual-care arm).40 In the trial, patients were followed for up to 7 years, with a mean follow-up time of 51.5 months. Although the trial's primary end point was time to first primary CV event, patients were followed after the primary event, and multiple end points were recorded. The ALLIANCE study end points are summarized in the figure. To model the pharmacoeconomic implications of the ALLIANCE study, direct medical costs for CV events from an administrative claims data set were utilized to determine the costs for the various CV end points that occurred in each arm of the ALLIANCE study.4o These trial-based CV end point expenses and the pharmaceutical costs for lipid-lowering therapy were used to develop a model of the cost per CV end point avoided, comparing the cost-effectiveness of intensive atorvastatin therapy with that of usual care. Scenario analyses were performed on our base model to explore how cost-effectiveness results changed over a range of practical variations in cost inputs. Most analyses used a payer perspective and estimated the costs excluding the patient's monthly prescription copayment. However, 1 scenario analysis specifically addressed cost-effectiveness from a patient perspective.

CV End Points The CV end points shown in Table I were used to build the model of cost per CV end point avoided. They corresponded to the CV end points observed 2206

and followed in the ALLIANCE study.40 With the exception of all-cause mortality, the end points required patient hospitalizations and are associated with the costs of acute and long-term care.

Base-Case Model Inputs and Assumptions The model perspective was a managed-care decisionmaker's viewpoint. Direct medical costs included both the cost of care for each CV end point occurrence, for an individual patient, and medication costs for lipidlowering therapy for all patients. The unit of analysis was an individual event, rather than an individual patient, because managed care organizations (MCOs) often examine the number needed to treat (NNT) to avoid an end point, and multiple end points can and often do occur in individual patients, each with corresponding discrete costs. Consequently, incremental cost-effectiveness ratios (ICERs) were calculated at the event level. The base-case model incorporated CV event-rate data over a maximum of 7 years of followup, with a mean of 51.5 months, as occurred in the trial. 4o The base-case model assumed a 3% discount rate across the years and a US $10 difference in patient monthly prescription copayments between the treatment arms. Total costs and CV end points were summed for each treatment arm and the discount rate was applied to all costs. The NNT and ICER calculations that used the base-case model inputs are summarized in Table II. Drug Costs For the atorvastatin arm, drug costs reflect utilization patterns from the trial, in which 17.7% of patients took 10 mg/d and 82.3% of patients took 20 mg/d or higher. The mean daily dose of atorvastatin in the trial was 40.5 mg. Atorvastatin costs were estimated using 2007 average wholesale price, minus 15% (AWP - 15%).41 Although some patients in the ALLIANCE study received drug coverage through the Veterans Administration (VA), the majority of patients were privately insured; furthermore, the perspective of the analysis presented in this article is managed care. Therefore, discounted AWPs, rather than VA drug prices, were used. Because generic statins are now available in the marketplace, and they represent the most likely initial lipid-lowering therapy a patient would receive in a managed care environment, the lipidlowering therapy drug costs for the comparator arm are based on the 2007 prices for generic simvastatin. Volume 30 Theme Issue

!'oJ

o o

CO

Total patients enrolled in ALLIANCE trial (N = 2442)

Patients enrolled in usual-care arm (n=1225)

Patients enrolled in intensive atorvastatin arm (n=1217)

-------~-

---~,

----

Patients experiencing end points during follow-up period (n = 408)

I

f

i

i !

["".

___. _.•_._._..•.J..

1"'-

Patients experiencing no end point during follow-up period

(n~809)

Patients experiencing end points during follow-up period (n = 443)

Patients experiencing 1 end point during follow-up period (n ~ 246)

Patients experiencing 1 end point during follow-up period

Patients experiencing 2 end points during follow-up period (n = 125)

Patients experiencing 2 end points during follow-up period (n=155)

Patients experiencing ~3 end points during follow-up period (n ~ 37)

Patients experiencing ~3 end points during follow-up period (n = 64)

.._•._ _......

Patients experiencing no end point during follow·up period (n=782)

(n~224)

('l

!J ~

!'oJ !'oJ

o .....

Figure. Patients reaching end points over 7-year follow-up in the Aggressive Lipid-Lowering Initiation Abates New Cardiac Events (ALLIANCE) study.4o

~

:i" III

~ ~

Clinical Therapeutics

Table I. Total clinical trial end points reached during the 7-year Aggressive Lipid-Lowering Initiation Abates New Cardiac Events study 40 follow-up period, used to build a model to compare cost per event avoided with branded atorvastatin or generic statin.

Clinical Trial End Point Cardiac death Nonfatal MI Resuscitated cardiac arrest Cardiac revascularization Unstable angina hospitalization Peripheral revascularization CHF hospitalization Stroke Death from all other causes Total end points Primary end points Secondary end points MI

=

myocardial infarction; CHF

~

Primary or Secondary

Atorvastatin (n = 1217)

Usual Care (n = 1225)

Primary Primary Primary Primary Primary Secondary Secondary Secondary Secondary

43

61 94

2

5

195 126 50 42 32 78 620 418 202

221 147 58

56 37 66 745 528

217

coronary heart failure.

Both arms reflect 51.5 months of therapy, which was the mean length of follow-up in the ALLIANCE study.40 Clinical End-Point Costs

Costs for CV end points were estimated from administrative claims data 42 for equivalent events and were adjusted to 2007 US dollars using the medical component of the consumer price index. 43 The data set included all patients irrespective of any comorbidities; therefore, the differences in costs between the patient group with the particular CV event and the patient group without the CV event represent the marginal costs of CV events. Acute end-point costs included the hospital costs and the first month's followup costs. In addition, follow-up hospitalization and other direct medical costs up to a maximum of 36 months of follow-up were included for each acute end point. In cases where an individual patient experienced ::0:2 end points in the same month, acute costs of both end points were included for the first month; thereafter, only the maximum monthly follow-up costs for the more expensive end point were included. In cases where individuals experienced additional events in future periods, the acute cost of the subsequent event was included in the month of the event and follow-up costs for subsequent months were based on hospitalization 2208

52

and other direct medical costs, up to a maximum of 36 months of follow-up, as with the initial event. To avoid double-counting, in cases where multiple end points occurred in an overlapping 36-month period, the maximum cost for the more expensive end point was applied for each month. In the base-case model, all patients' costs were estimated for a period of 7 years or until death. Scenario Analyses and Discounting

To determine the robustness of our base-case ICER, we performed a variety of scenario analyses of our base-case model. The time horizon was shortened from 7 years to 5 or 3 years to examine the effects of shorter time frames, in alignment with preferred time horizons of some MCOs. In the case of the 3-year models, drug-use costs were calculated to include 36 months of drug use, rather than the 51.5 months used in the 5- and 7-year models. The model discount rate was varied from 0% to 5% to determine variations in inflation rates. Another scenario analysis allowed the difference between monthly patient prescription copayment for branded atorvastatin versus generic statins to range from year-2007 US $5 to $20. This was done to estimate the variation in atorvastatin cost-effectiveness Volume 30 Theme Issue

CO. Mullins et al. r·············_··_····································

_ _

_.-._

_

.

Table II. Baseline input in a model to compare cost per event avoided with branded atorvastatin or generic statin, based on data from the Aggressive Lipid-Lowering Initiation Abates New Cardiac Events study.4o Cost data are given in year-2007 US dollars. Variable Drug cost per patient per day, $* Atorvastatin 10 mg, AWP - 15% Atorvastatin 20,40, or 80 mg, AWP -15% Generic statin Insurance copayment differentials for atorvastatin vs generic statin End-point costs, $t Cardiac death Nonfatal Mt Resuscitated cardiac arrest Cardiac revascularization Unstable angina hospitalization Peripheral revascularization CHF hospitalization Stroke Death from all other causes Other Follow-up time, mo Mean, Range Discount rate, %

Value 2.47 3.53 0.37 10 19,997 42,662 62,271 34,765 26,231 34,765 39,662 30,633

°

51.5 1-84 3

AWP - 15% ~ average wholesale price minus 15%; MI ~ myocardial infarction; CHF = congestive heart failure. *Drug costs are based on discounted AWP for branded atorvastatin and generic simvastatin reported in the 2007 Drug Topics Red Book. 41 tEvent costs were based on acute and a maximum of 36 months follow-up treatment and care costs (see Methods section).

likely to be observed with different MCO prescription drug plan benefit structures. One scenario, which incorporated the anticipated entry of generic atorvastatin in 2011 at price parity with generic simvastatin, was modeled to reflect atorvastatin's upcoming loss of exclusivity (LOE). A break-even calculation was per-

2008

formed to determine the differential monthly prescription copayment value between branded atorvastatin and a generic statin comparator that would lead to identical total medical and pharmacy costs across arms from an MCa perspective using our 7-year base-case model. Finally, we developed scenarios with ICERs calculated from the patient perspective based on varying patient monthly prescription copayment. For these scenarios, monthly copayments ranged from year2007 US $5 to $20.

RESULTS NNT Results As reported in the ALLIANCE study,40 there were 620 events among the 1217 atorvastatin patients (0.509 event per patient) and 745 events among the 1225 usual-care patients (0.608 event per patient). The differences in CV event rates between atorvastatin and the comparator arms produced an NNT of 11 patients in the 7-year base-case model. An NNT of 11 means that, for every 11 patients treated with atorvastatin rather than usual care over the 7-year time horizon, there is 1 fewer CV event. When interpreting the NNT for the 7-year time horizon, it is important to recall that the majority of patients had <7 years of follow-up (mean follow-up time was 51.5 months) in the ALLIANCE study.4o Therefore, the actual 7-year NNT may be lower than predicted. Over shorter time horizons, the NNT value was 19 patients in the 3-year scenario and 9 patients in the 5-year scenario. Results are summarized in Table III.

Base-Case Model Cost and Cost- Effectiveness Results The base-case model predicted higher pharmacy costs and greater medical offsets associated with atorvastatin. The increase in pharmacy costs to an insurer over the 7-year time horizon was $3881 per patient, whereas the reduction in direct medical costs over that same 7-year time horizon was $2860 per patient. Thus, the net increase to an insurer was $1021 per patient over 7 years to achieve the net reduction in CV events. The ICER, defined as the cost per CV end point avoided, was calculated as the ratio of the difference in total costs and the difference in total clinical end points reached across arms. Results of these calculations are summarized in Table IV. An ICER of $10,344 was calculated for the base case, a 7-year cost-per-clinical-

2209

Clinical Therapeutics

Table III. Number needed to treat (NNT) to avoid clinical end point with branded atorvastatin or generic statin in a model analysis based on data from the Aggressive Lipid-Lowering Initiation Abates New Cardiac Events study.4o

Scenario Analyses and Break-Even Calculations

Total Events, No. Atorvastati n

Generic Statin

NNT

3 Years

394

463

19

5 Years

580

722

9

7 Years

620

745

11

Follow-Up Time

end-point-avoided model incorporating the ALLIANCE study's clinical events and drug costs. Drug costs in the atorvastatin arm reflected the atorvastatin AWP -15%, combined with a $10 patient copayment, whereas drug costs in the usual-care arm reflected generic statin acquisition cost without a patient copayment.

Pharmacy prescription copayment amounts, discount rates, and time horizons were varied in a number of scenarios. In 5- and 7-year scenarios, the anticipated impact of a generic atorvastatin entry in 2011 was incorporated. Results are summarized in Table IV. Varying the discount rate from 0% to 5% produced minimal changes across all time horizons, as shown in Table IV. With a $5 monthly prescription

Table IV. Sensitivity of cost-per-c1inical-end-point-avoided estimates to key model assumptions in an analysis based on data from the Aggressive Lipid-Lowering Initiation Abates New Cardiac Events stud y 40 and average wholesale price for branded atorvastatin or generic statin.

Model Scenario

Cost per Clinical End Point Avoided, Year-2007 US $

Atorvastatin vs generic statin for 7 years (base case) $5 Copayment differential $20 Copayment differential 3 Years of branded atorvastatin followed by generic atorvastatin 3 Years of branded atorvastatin followed by generic atorvastatin with $0 copayment differential for branded period 0% Discount 5% Discount

10,344 12,829 5375 -951 (dominant)

Atorvastatin vs generic statin for 5 years $5 Copayment differential $20 Copayment differential 3 Years of branded atorvastatin followed by generic atorvastatin 3 Years of branded atorvastatin pricing followed by generic atorvastatin with $0 copayment differential for branded period 0% Discount 5% Discount

8878 11,052 4530 -1006 (dominant)

Atorvastatin vs generic statin for 3 years $5 Copayment differential $20 Copayment differential 0% Discount 5% Discount

2210

2590 10,121 10,467

2093 8680 8989 24,200 27,424 17,751 24,450 24,038

Volume 30 Theme Issue

CO. Mullins et al.

copayment differential, lCERs ranged from $11,052 in the 5-year scenario to $27,424 in the 3-year scenario, respectively. The break-even calculation value was $31 per month, representing the difference in monthly prescription patient copayments between atorvastatin and a generic statin comparator that would lead to no net increase in total spending by an MCO. The anticipated impact of a generic atorvastatin entry in 2011 resulted in dominant scenarios when a monthly prescription copayment differential of $10 was utilized during the branded period for both 5- and 7-year time horizons. When the monthly prescription copayment differential was eliminated, scenarios incorporating the impact of a generic atorvastatin entry resulted in lCERs of $2093 and $2590 for the 5- and 7-year time horizons, respectively.

Finally, lCERs from the patient perspective were based on the CV end-point reduction benefits already described and cost differences across treatment arms that reflected only the patients' prescription copayment amounts. Results are summarized in Table V. Using a difference of $10 per month in copayments between branded atorvastatin and a generic statin comparator over 5- and 7-year time horizons, patient leERs were calculated at $4353 and $4969, respectively. Raising the copayment differential to $20 per month increased these values to $8696 and $9938, respectively. When considering the impact of generic atorvastatin after 3 years of branded atorvastatin pricing, values for the $20 copayment differential decreased to $6196 for the 5-year and $7083 for the 7-year model. For the $10 copayment scenario, these

Table V. Patient-perspective incremental cost-effectiveness ratio sensitivity of cost-per-c1inical-end-pointavoided estimates based on monthly patient prescription copayment in an analysis based on data from the Aggressive Lipid-Lowering Initiation Abates New Cardiac Events stud y 40 and average wholesale price for branded atorvastatin or generic statin. Cost per Clinical End Point Avoided, Year-2007 US $

Model Scenario Atorvastatin vs generic statin for 7 years (base case) $5 Copayment differential $20 Copayment differential 3 Years of branded atorvastatin followed by generic atorvastatin with $10 copayment differential for branded period 3 Years of branded atorvastatin followed by generic atorvastatin with $20 copayment differential for branded period 0% Discount 5% Discount Atorvastatin vs generic statin for 5 years $5 Copayment differential $20 Copayment differential 3 Years of branded atorvastatin followed $10 copayment differential for branded 3 Years of branded atorvastatin followed $20 copayment differential for branded 0% Discount 5% Discount Atorvastatin vs generic statin for 3 years $5 Co payment differential $20 Copayment differential 0% Discount 5% Discount

2008

4969 2485 9938 3542 7083 5217 4818 4353 2172 8696

by generic atorvastatin with period by generic atorvastatin with period

3103 6196 4565 4216 6456 3228 12,894 6641 6329

2211

Clinical Therapeutics

values decreased to $3103 and $3542, respectively. Because these calculations were based solely on prescription copayment amounts, direct and indirect costs that patients incur for hospitalization and lost work time would be expected to further favorably offset these expenses. DISCUSSION Many reports have focused on the clinical efficacy and cost-effectiveness of statins. 26 ,39,44-56 Atorvastatin, in particular, has been studied extensively, with outcomes benefits reported in more than a dozen trials. 57- 6o The ALLIANCE trial, which was designed to reflect usual community practice with generic statins, explored secondary prevention advantages attributable to using aggressive lipid-lowering therapy with atorvastatin. 4o

Patients enrolled in the trial had established CHD and had access to managed care or US VA formularies for prescription drugs. 4o The insured group in the ALLIANCE trial should better predict health care utilization rates in an MCO setting compared with groups in other US-based clinical trials, which often enroll substantial numbers of uninsured or indigent patients. Furthermore, the addition of costs of CV events from routine clinical practice makes the results more applicable to MCOs. Many studies have examined the cost-effectiveness of statins based on clinical trial outcomes. 38 This literature is divided into studies focused on primary51-53,61 or second ary 39,48,53-55 prevention of CV events. Moreover, many previous economic analyses for secondary prevention 39 ,48,53-55 were performed utilizing a societal perspective,39,54 quality-adjusted life years (QALYs), or life-years gained (or some combination thereof) as their units of analysis,39,48,53-55 and extrapolated benefits beyond the trial time horizon. 39 ,53-55 Although Russell et aJ53 performed analyses from a health care-payer perspective, their study was based on low statin doses and clinical data from the Comparative Dose Efficacy Study of Atorvastatin Versus Simvastatin, Pravastatin, Lovastatin, and Fluvastatin in Patients with Hypercholesterolemia (CURVES) srudy,62 which preceded the newer NCEP ATP III treatment guidelines. 22 At the time of the ALLIANCE study, there were no generic statins in the marketplace, so an economic analysis of the trial outcomes without detailed individual patient information regarding therapeutic regimens and insurance coverage in the usual-care arm would have required a num2212

ber of drug cost assumptions to perform an analysis from a managed care perspective. In the current environment, MCOs are trying to assess the costs and risk-benefit ratios of using generic rather than branded statins to maximize both clinical and financial outcomes. Using the ALLIANCE data,40 we found an acceptable cost-effectiveness profile for atorvastatin compared with generic statins. Calculated ICERs to avoid CV end points in high-risk patients were highly cost-effective and robust using a variety of models and time frames. ICERs were particularly favorable over the 5- and 7-year horizons (less than US $11,000 per end point avoided, in year-2007 values) and somewhat more modestly appealing over 3 years (less than US $25,000 per end point avoided, in year2007 values). However, when the model was revised to acknowledge that atorvastatin will undergo LOE in 2011 and will then be available at generic prices, the cost-effectiveness profile of atorvastatin in high-risk patients improved dramatically in our calculations and became a dominant strategy in 5- and 7-year models. Consequently, MCOs may wish to consider a policy that maintains a favorable formulary status for atorvastatin in patients with established CHD, based on both positive clinical outcomes and cost-effectiveness. Statin-switching is a commonplace practice by MCOs to reduce costs. 63 - 65 Unfortunately, this practice can have some unintended effects, such as negatively influencing medication adherence. 66 Switching strategies can also inadvertently lead to inaccurate doses in patients who are already controlled or introduce drugdrug interactions. 67 ,68 Maintaining more effective pharmacotherapy might enable higher risk CHD patients to attain more stringent NCEP ATP III treatment guideline goals. Beyond MCOs, the formulary placement of various statins has relevance to both employers and public payers, such as Medicare and Medicaid. Arguably, the low NNTs to avoid additional CV end points (NNTs <20 in all analyses) should be compelling to private employers interested in reducing the risk of absenteeism or employee turnover. We believe that our study has 3 key advantages: the model uses an end-point analysis based on actual data from a head-to-head clinical trial rather than predicted CV outcomes; the analysis incorporates generic statin costs to represent the current real-world, usualcare lipid-lowering therapy as a comparator; and the various scenario analyses provide different costeffectiveness results using different assumptions that Volume 30 Theme Issue

CD. Mullins et al.

MCOs may wish to consider. Using the MCO perspective, we identified a target population that derives significant benefits from an atorvastatin option. The CV end points that occurred in the ALLIANCE trial are clinically relevant to managed care populations and have important economic impact. Finally, using generic statins as a comparator represented a stiff test by evaluating the starkest drug cost differential, yet atorvasta tin therapy still represented an attractive choice. Our study also has some limitations. The primary analysis in this study was cost per event avoided, which is commonly requested by some payers but may provide limited guidance to decision-makers who prefer cost! QALY calculations. 69 The model assumes that usualcare patients in the ALLIANCE study were taking generic statins. In fact, the specific lipid-lowering medications utilized by the individual patients in the usual-care arm of the ALLIANCE study varied considerably at the discretion of treating physicians,40 which makes the comparator arm in the ALLIANCE trial less than ideal. The details of these regimens, and related patient adherence, counseling, and individual lab data, are not known. In addition, the patient follow-up was fragmented and patients frequently changed plans and medications, according to their physicians. Nonetheless, roughly two thirds of usual-care patients were receiving lipid-lowering therapy at baseline and the additional mean 23.3% LDL-C reduction beyond the on-treatment baseline of 146 to 111 mg/dL during the course of the trial suggests that physicians used reasonably aggressive means in this comparator group. The usual-care LDL-C reduction in the ALLIANCE usual-care arm is comparable with that reported for simvastatin 20 mg or pravastatin 40 mg, based on other trials, such as the CURVES study.62 Finally, the analysis was based on the outcomes achieved by established CHD patients, defined in the ALLIANCE study as including those patients with the following: (l) a history of acute myocardial infarction, coronary artery bypass graft surgery, unstable angina >3 months previously, or percutaneous transluminal angioplasty >6 months previously; and (2) LDL-C levels between 110 and 200 mg/dL on initial lipid-lowering therapy or between 130 and 250 mg/dL if not initially on lipidlowering therapy.40Whether these results can be generalized to a broader population of individuals at risk for CV events is uncertain. Also uncertain is how patient copayment differences between branded and generic statins would alter the corresponding outcomes observed in the ALLIANCE study. 2008

CONCLUSIONS Cost-effectiveness results based on the ALLIANCE trial's outcomes provide insight into the economic implications of a head-to-head clinical trial comparing branded atorvastatin with a generic statin comparator. Over the 7-year maximum follow-up time of the trial, the ICER of branded atorvastatin to a generic comparator was year-2007 US $10,344 per CV end point avoided. Scenario analysis exploring 3- and 5-year time horizons, monthly prescription copayment differentials, and the impact of a generic atorvastatin suggest a range of ICERs from year-2007 US $27,424 to a dominant strategy, whereby atorvastatin would be more effective and less costly, from a managed-care perspective. These ICERs suggest that patients with established CHD constitute a population in which branded atorvastatin provides good economic value, even when compared with generic statins. ACKNOWLEDGMENTS This research and its publication were funded by Pfizer Inc. Drs. Mullins, Rattinger, and Koren received consulting fees from Pfizer for research design, data analysis and/or interpretation, and manuscript preparation. In addition, Dr. Mullins has previously received grants from Pfizer; he and Dr. Koren have both previously consulted for Pfizer. REFERENCES 1. Mullins CD, Wang j, Palumbo FB, Stuart B. The impact of pipeline drugs on drug spending growth. Health Aff (Millwood). 2001 ;20:21 0-215. 2. Morgan SG. QuantifYing components of drug expenditure inflation: The British Columbia seniors' drug benefit plan. Health 5erv Res. 2002;37: 1243-1266. 3. Banthin jS, Miller GE. Trends in prescription drug expenditures by Medicaid enrollees. Med Care. 2006;44 (Suppl 5):127-135. 4. Heffler 5, Smith 5, Keehan 5, et al. US health spending projections for 2004-2014. Health Aff (Millwood). 2005; Suppl(Web Exciusives):W5-74-W5-85. 5. Hoffman jM, Shah ND, Vermeulen LC, et al. Projecting future drug expenditures-2008. Am J Health 5yst Pharm. 2008;65:234-253. 6. Huh 5, Rice T, Ettner SL. Prescription drug coverage and effects on drug expenditures among elderly Medicare beneficiaries. Health 5erv Res. 2008;43:810-832. 7. Lichtenberg FR, Sun sx. The impact of Medicare Part D on prescription drug use by the elderly. Health Aff (Millwood). 2007;26:1735-1744.

2213

Clinical Therapeutics 8. Zuvekas SH, Cohen JW. Prescription drugs and the changing concentration of health care expenditures. Health Aff(Millwood). 2007;26: 249-257.

Intern Med.

2005;143:844-845].

Ann Intern Med. 2005;142:891-897.

18. Schulman KA, Rubenstein LE, Abernethy DR, et al. The effect of phar-

26. Cannon CP, Braunwald E, McCabe CH, et ai, for the Pravastatin or Atorvastati n Eval uation and Infection Therapy-Thrombolysis in Myo-

maceutical benefits managers: Is it

cardial Infarction 22 Investigators.

9. Joyce GF, Escarce JJ, Solomon MD, Goldman DP. Employer drug bene-

being evaluated? Ann Intern Med. 1996;124:906-913.

fit plans and spending on prescrip-

19. Murphy JE. Generic substitution and optimal patient care. Arch Intern Med. 1999; 159:429-433.

Intensive versus moderate lipid lowering with statins after acute coronary syndromes [published correc-

tion drugs [published correction appears in )AMA. 2002;288:2409]. )AMA. 2002;288: 1733-1739. 10. Motheral B, Sheth J. Financial consequences of drug benefit plans.)AMA. 2003;289:423-424; author reply 424. 11. Motheral BR, Henderson R, Cox ER.

tion appears in N Engl) Med. 2006; 354:778]. N Engl) Med. 2004;350: 1495-1504.

20. Mager DE, Cox ER. Relationship between generic and preferredbrand prescription copayment dif-

27. Grundy SM, C1eeman JI, Merz CN, et ai, for the Coordinating Commit-

ferentials and generic fill rate. Am)

tee of the National Cholesterol

Manag Care. 2007;13:347-352.

Education Program. A summary of implications of recent clinical trials for the National Cholesterol Educa-

Plan-sponsor savings and member experience with point-of-service

21. The Expert Panel on Detection, Evaluation, and Treatment of High

prescription step therapy. Am) Manag Care. 2004; 10:457-464. 12. Fairman KA, Motheral BR, Hender-

Blood Cholesterol in Adults. Execu-

tion Program Adult Treatment Pan-

tive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in

el III guidelines. Arterioscler Thromb Vasc Bioi. 2004;24: 1329-1330. 28. Grundy SM, Cleeman JI, Merz CN, et ai, for the National Heart, Lung, and Blood Institute; the American

Adults (Adult Treatment Panel III).

College of Cardiology Foundation; and the American Heart Association. Implications of recent clinical trials for the National Cholesterol

son RR. Retrospective, long-term follow-up study of the effect of a three-tier prescription drug copayment system on pharmaceutical and other medical utilization and costs. Gin Ther. 2003;25:3147-3161; discussion 3144-3146. 13. Klepser DG, Huether JR, Handke LJ, Williams CEo Effect on drug utiliza-

)AMA. 2001 ;285:2486-2497.

22. National

Cholesterol

Education

Program (NCEP) Expert Panel on Detection, Evaluation, and Treat-

Education Program Adult Treat-

ment of High Blood Cholesterol in

ment Panel III guidelines [published

tion and expenditures of a costshare change from copayment to

Adults (Adult Treatment Panel III). Third Report of the National Cho-

coinsurance. ) Manag Care Pharm. 2007;13:765-777. 14. Miller DP, Furberg CD, Small RH, et al. Controlling prescription drug expenditures: A report of success. Am) ManagCare. 2007;13:473-480. 15. Reiffel JA. Formulation substitution: A frequently overlooked variable in cardiovascular drug management. Prog Cardiovasc Dis. 2004;47:3-10.

lesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. CifC/.i1ation.2002;106:3143-3421. 23. Morrison A, Glassberg H. Determinants of the cost-effectiveness of statins.) Manag Care Pharm. 2003;9: 544-551.

correction appears In Circulation. 2004;110:763]. Circulation. 2004; 110:227-239. 29. Pravachol [package insert]. Princeton, NJ: Bristol-Myers Squibb Company; 2007. 30. Mevacor [package insert]. White-

16. Andermann F, Duh MS, Gosselin A,

24. Thorpe KE. Factors accounting for

achieves economic benefits from ge-

Paradis PE. Compulsory generic

the rise in health-care spending in the United States: The role of rising dis-

neric substitution but fails to do the same fortherapeutic substitution. Br

ease prevalence and treatment in-

) Gin Pharmacal. 2007;64:680-685.

switching of antiepileptic drugs: High switchback rates to branded compounds compared with other drug classes. Epilepsia. 2007;48:464-469 17. Haas JS, Phillips KA, Gerstenberger EP, Seger AC. Potential savings from substituting generic drugs for brandname drugs: Medical expenditure panel survey, 1997-2000 [published comment and reply appear in Ann

2214

tensity. PublicHealth. 2006; 120: 10021007. 25. Thorpe KE, Howard DH. The rise in

house Station, NJ: Merck; 2008. 31. Zocor[packageinsert]. Whitehouse Station, NJ: Merck; 2008. 32. Gumbs PD, Verschuren WM, Souvereln PC, et al. Society already

33. Straka RJ, Mamdani M, DamenJ, et al. Economic impacts attributable to the early clinical benefit of ator-

spending among Medicare beneficiaries: The role of chronic disease prevalence and changes In treat-

vastatin therapy-a US managed care perspective. CUff Med Res Opin.

ment intensity. Health Aff(Millwood). 2006;25:w378-w388.

34. Meissner B, Dickson M, Shinogle J, et al. Drug and medical cost effects

2007;23:1517-1529.

Volume 30 Theme Issue

CD. Mullins et aI. of a drug formulary change with therapeutic interchange for statin drugs in a multistate managed Medicaid organization [published cor-

35.

36.

37.

38.

39.

rection appears in J Manag Care Pharm. 2006;12:490]. J Manag Care Pharrn.2006;12:331-340. Reuters. Examine the Lipitor market analysis and forecast report, 20072012,2017 and 2023. http://www. reuters.com/ article/ press Release/ idUS244565+25-jun-2008+ MW20080625. Accessed November 11,2008. Schaefer Ej, McNamara jR, Tayler T, et al. Comparisons of effects of statins (atorvastatin, fluvastatin, lovastatin, pravastatin, and simvastatin) on fasting and postprandial lipoproteins in patients with coronary heart disease versus control subjects. Am J Cardiol. 2004;93:3139. Olsson AG, Eriksson M,johnson 0, et ai, for the 3T Study Investigators. A 52-week, multicenter, randomized, parallel-group, double-blind, double-dummy study to assess the efficacy of atorvastatin and simvastatin in reaching low-density lipoprotein cholesterol and triglyceride targets: The Treat-to-Target (3T) Study. Clin Ther. 2003;25: 119-138. Plosker GL, Lyseng-Williamson KA. Atorvastatin: A pharmacoeconomic review of its use in the primary and secondary prevention of cardiovascular events. PharmacoEconomics. 2007;25:1031-1053. Lindgren P, Graff j, Olsson AG, et ai, for the IDEAL Trial Investigators. Cost-effectiveness of high-dose

atorvastatin compared with regular dose simvastatin. Eur Heart). 2007; 28:1448-1453. 40. Koren Mj, Hunninghake DB, for the ALLIANCE Investigators. Clinical outcomes in managed-care patients with coronary heart disease treated aggressively in lipid-lowering disease management clinics: The ALLIANCE study. J Am Call Cardiol. 2004;44: 1772-1779.

2008

41. 2007 Drug Topics Red Book. Montvale, Nj: Thomson Medical Economics; 2007. 42. Thompson D, RubinjL, Nyambosej, et al. A flexible tool to estimate medical-care costs for study events In cardiovascular endpoint trials. Presented at: 13th Annual Meeting of the International Society for Pharmacoeconomics and Outcomes Research; May 2008; Toronto, Canada. 43. Bureau of Labor Statistics. Consumer price index. http://www.bls.gov/cpi. Accessed june 11, 2008. 44. Sever PS, Dahlbf B, Poulter NR, et ai, for the ASCOT Investigators. Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, In the AngloScandinavian Cardiac Outcomes Trial-Lipid Lowering Arm (ASCOTLLA): A multicentre randomised controlled trial. Lancet. 2003;361: 1149-1158. 45. Colhoun HM, Betteridge Dj, Durrington PN, et ai, for the CARDS Investigators. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): Multicentre randomised placebo-controlled trial. Lancet. 2004;364:685-696. 46. Pedersen TR, Faergeman 0, Kastelein jj, et ai, for the Incremental Decrease in End Points Through Aggressive Lipid Lowering (IDEAL) Study Group. High-dose atorvastatin vs usual-dose simvastatin for secondary prevention after myocardial infarction: The IDEAL study: A randomized controlled trial [published correction appears in JAMA. 2005; 294:3092].JAMA. 2005;294:24372445. 47. laRosa jC, Grundy SM, Waters DD, et ai, for the Treating to New Targets (TNT) Investigators. Intensive lipid lowering with atorvastatin in patients with stable coronary disease. N Engl J Med. 2005;352:1425-1435.

48. Athyros VG, Papageorgiou AA, Mercouris BR, et al. Treatment with atorvastatin to the National Cholesterol Educational Program goal versus 'usual' care in secondary coronary heart disease prevention. The GREek Atorvastatin and Coronaryheart-disease Evaluation (GREACE) study. Curr Med Res Opin. 2002; 18: 220-228. 49. Schwartz GG, Olsson AG, Ezekowitz MD, et ai, for the Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering (MIRACL) Study Investigators. Effects of atorvastatin on early recurrent ischemic events in acute coronary syndromes: The MIRACL study: A randomized controlled trial. JAMA. 2001 ;285: 1711-1718. 50. Cannon CP, Steinberg BA, Murphy SA, et al. Meta-analysis of cardiovascular outcomes trials comparing intensive versus moderate statin therapy. J Am Coli Cardiol. 2006;48: 438-445. 51. Davies A, Hutton j, O'Donnell j, Kingslake S. Cost-effectiveness of rosuvastatin, atorvastatin, simvastatin, pravastatin and fluvastatin for the primary prevention of CHD in the UK. BrJCardiol. 2006;13:196-202. 52. Lindgren P, Buxton M, Kahan T, et ai, for the ASCOT Investigators. Cost-effectiveness of atorvastatin for the prevention of coronary and stroke events: An economic analysis of the Anglo-Scandinavian Cardiac Outcomes Trial-Lipid-Lowering Arm (ASCOT-LLA). Eur J Cardiovasc Prev Rehabil. 2005; 12:29-36. 53. Russell MW, Huse DM, Miller jD, et al. Cost effectiveness of HMG-CoA reductase inhibition in Canada. Can J Clin Pharmacol. 2001 ;8:9-16. 54. Chan PS, Nallamothu BK, Gurm HS, et al. Incremental benefit and cost-effectiveness of high-dose statin therapy in high-risk patients with coronary artery disease. Circulation. 2007; 115:2398-2409. 55. Elliott Wj, Weir DR. Comparative cost-effectiveness of HMG-CoA re-

2215

Clinical Therapeutics ductase inhibitors in secondary prevention of acute myocardial infarction. Am) Health 5yst Pharm. 1999; 56:1726-1732. 56. Robinson jG. Models for describing relations among the various statin drugs, low-density lipoprotein cholesterol lowering, pleiotropic ef-

57.

58.

59.

60.

61.

62.

63.

fects, and cardiovascular risk. Am ) Cordial. 2008;101:1009-1015. Arca M. Atorvastatin efficacy in the prevention of cardiovascular events in patients with diabetes mellitus and/or metabolic syndrome. Drugs. 2007;67(suppl 1):43-54. Arca M, Gaspardone A. Atorvastatin efficacy in the primary and secondary prevention of cardiovascular events. Drugs. 2007;67(suppll ):29-42. Bybee KA, Lee jH, O'Keefe jH. Cumulative clinical trial data on atorvastatin for reducing cardiovascular events: The clinical impact of atorvastatin. Curr Med Res Opin. 2008;24: 1217-1229. Gresser U, Gathof Bs. Atorvastatin: Gold standard for prophylaxis of myocardial ischemia and strokecomparison of the clinical benefit of statin on the basis of randomized controlled endpoint studies Eur) Med Res. 2004;9: 1-17. Ramsey SO, Clarke LO, Roberts CS, et al. An economic evaluation of atorvastatin for primary prevention of cardiovascular events in type 2 diabetes. PharmacoEcanamics. 2008; 26:329-339. jones P, Kafonek 5, Laurora I, Hunninghake O. Comparative dose efficacy study of atorvastatin versus simvastatin, pravastatin, lovastatin, and fluvastatin in patients with hypercholesterolemia (the CURVES study) [published correction appears in Am) Cordial. 1998;82: 128]. Am) Cardial. 1998;81 :582-587. Cox ER, Kulkarni A, Henderson R. Impact of patient and plan design factors on switching to preferred statin therapy. Ann Pharmacather. 2007;41: 1946-1953.

2216

64. Usher-Smith j, Ramsbottom T, Pearmain H, Kirby M. Evaluation of the clinical outcomes of switching patients from atorvastatin to simvastatin and losartan to candesartan in a primary care setting: 2 Years on. Int) Clin Pract. 2008;62:480-484. 65. Usher-Smith jA, Ramsbottom T, Pearmain H, Kirby M. Evaluation of the cost savings and clinical outcomes of switching patients from atorvastatin to simvastatin and losartan to candesartan in a primary care setting. Int) Clin Pract. 2007;61: 15-23. 66. Thiebaud P, Patel BV, Nichol MB, Berenbeim OM. The effect of switching on compliance and persistence: The case of statin treatment. Am) ManagCare.2005;11:670-674.

67. Hess G, Sanders KN, Hill j, Liu LZ. Therapeutic dose assessment of patient switching ITom atorvastatin to simvastatin [published correction appears in Am) Manag Care. 2007;13: 432]. Am) Manag Care. 2007;13 (suppI3):580-s85. 68. Molden E, Andersson KS, jacobsen O. Interactions between statins and macrolide antibiotics [in Norwegian]. Tidsskr Nor Laegefaren. 2007; 127: 1660-1661. 69. Mandelblattjs, Fryback OG, Weinstein MC, et al. Assessing the costeffectiveness of health interventions. In: Gold MR, Siegel jE, Russell LB, Weinstein MC, eds. Cost-Effectiveness in Health and Medicine. New York, NY: Oxford University Press; 1996: 135175.

Address correspondence to: C. Daniel Mullins, PhD, University of Maryland, School of Pharmacy, 220 Arch Street, 12th Floor, Baltimore, MD 21201. E-mail: [email protected] Volume 30 Theme Issue