University of Southern California Dissertations and Theses

The value of novel antihyperlipidemic treatments in the U.S. healthcare system: Reducing the burden of cardiovascular diseases and filling the gap of low adherence in statins

(USC Thesis Other)

The value of novel antihyperlipidemic treatments in the U.S. healthcare system: Reducing the burden of cardiovascular diseases and filling the gap of low adherence in statins

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1 The Value of Novel Antihyperlipidemic Treatments in the U.S. Healthcare System: Reducing the Burden of Cardiovascular Diseases and Filling the Gap of Low Adherence in Statins A dissertation by Wei-Han Cheng Faculty of the USC Graduate School in fulfillment of the requirements for the degree of Doctor of Philosophy in the subject of Health Economics University of Southern California Los Angeles, California August, 2018 2 1 TABLE OF CONTENTS INTRODUCTION ........................................................................................................................................ 4 Chapter 1: PCSK9 SHOW VALUE FOR PATIENTS AND THE U.S. HEALTHCARE SYSTEM .......... 7 BACKGROUND and INTRODUCTION ................................................................................................ 8 METHODS ............................................................................................................................................... 9 The Future Elderly Model ..................................................................................................................... 9 Simulations ......................................................................................................................................... 11 Uncertainty .......................................................................................................................................... 14 RESULTS ............................................................................................................................................... 14 Life expectancy, quality of life, and functional status ........................................................................ 14 Cost-effectiveness ............................................................................................................................... 15 Sensitivity analysis .............................................................................................................................. 16 DISCUSSION ......................................................................................................................................... 17 TABLES and FIGURES ......................................................................................................................... 21 REFERENCES ....................................................................................................................................... 29 SUPPLEMENTAL APPENDIX ............................................................................................................. 36 Scenarios ............................................................................................................................................. 36 The impact of PCSK9i on health and costs ......................................................................................... 38 Sensitivity analysis .............................................................................................................................. 40 Results of the sensitivity analysis ....................................................................................................... 42 Supplemental tables and figures ......................................................................................................... 45 References ........................................................................................................................................... 65 Chapter 2: Three-Part Pricing to Facilitate Early Access to Novel Drugs of Uncertain Efficacy .............. 67 ABSTRACT ............................................................................................................................................ 67 BACKGROUND and INTRODUCTION .............................................................................................. 67 METHODS ............................................................................................................................................. 68 RESULTS ............................................................................................................................................... 71 DISCUSSIONS ....................................................................................................................................... 71 TABLES and FIGURES ......................................................................................................................... 73 SUPPLEMENTAL APPENDIX ............................................................................................................. 79 The Future Elderly Model ................................................................................................................... 79 Simulations ......................................................................................................................................... 80 Scenarios ............................................................................................................................................. 81 3 The impact of PCSK9 inhibitors on health ......................................................................................... 82 The Price of PCSK9 inhibitors in the Three-part pricing model......................................................... 83 External Validation ............................................................................................................................. 84 References ........................................................................................................................................... 85 Chapter 3: Longitudinal trends and predictor of statin adherence in U.S Medicare Beneficiaries ............. 87 ABSTRACT ............................................................................................................................................ 87 INTRODUCTION .................................................................................................................................. 88 METHOD ............................................................................................................................................... 89 Study design and data sets .................................................................................................................. 89 Eligible Population .............................................................................................................................. 89 Outcome .............................................................................................................................................. 90 Explanatory Variables ......................................................................................................................... 90 Statistical Analysis .............................................................................................................................. 91 RESULTS ............................................................................................................................................... 91 Longitudinal trends in statin adherence and mean daily OOP costs ................................................... 92 Factors associated with statin adherence ............................................................................................ 92 DISCUSSIONS ....................................................................................................................................... 93 TABLES and FIGURES ......................................................................................................................... 96 REFERENCES ..................................................................................................................................... 103 4 INTRODUCTION Despite advances in medical technologies, cardiovascular disease (CVD) remains the leading cause of death—nearly 7.2million deaths annually—and a major cause of disability in the United States. A large body of evidence demonstrates that low-density lipoprotein cholesterol (LDL-C) is a principal driver of atherosclerotic cardiovascular disease(ASCVD)—the underlying cause of most clinical manifestations of CVD—and thus the primary target for CVD risk reduction interventions. Despite the use of statins, previous research has shown that a substantial proportion of treated high-risk patients fail to achieve target LDL-C levels, with estimated 12 out of 15 million of high-risk patients with LDL-C over 70 mg/dl. Stain intolerance, the poor adherence and high discontinuation rate are common concerns in clinical practice, which attenuate the efficacy of statins in individual lipid-lowering and risk reduction in CVD preventions. In August 2015, FDA approved two monoclonal antibodies targeting proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors, alirocumab and evolocumab, which are novel lipid-lowering approaches that inhibit the binding of PCSK9 to the LDL receptor, resulting in much more powerful LDL-C lowering potency compared with statins. The initial labels for both PCSK9 inhibitors as an adjunct therapy for patients with familial hypercholesterolemia or those who require additional lowering of LDL beyond maximum tolerated statins or other cholesterol treatments. In 2017, evolocumab’s label was expanded to adult patients with established cardiovascular disease who are at risk of myocardial infarction, stroke, and coronary revascularization, based on the results suggested by the FOURIER trial. Despite the substantial health benefits suggested by the clinical trials, payers and policymakers are concerned that this new class of expensive specialty medications poses a substantial economic burden given the drugs’ high price tags, ranging from$14,100 to $14,600 per patient per year in 2015. Given controversy surrounding long-term efficacy and uncertainty in real-world effectiveness, payers have shown limited willingness to cover this therapy, citing its potentially low cost-effectiveness, with only one third of patient initially prescribed PCSK9 inhibitors actually received it. The studies presented in this dissertation discuss the potential value of PCSK9 inhibitors from the U.S. healthcare system perspective and propose an innovative drug pricing strategy to address the current health outcome lost due to the access barriers of this new drug. We also 5 identified the trend of suboptimal use of statins in a national representative elderly population, which is helpful guide the policy makers for prioritizing the access to the PCSK9 inhibitors. To forecast the long-term benefit of PCSK9 inhibitors in a real-world U.S. population instead of a synthetic cohort, the first paper of my dissertation estimates the net monetary value of PCSK9 inhibitors in the older Americans (age 51 and older) under the FDA-approved eligibility when comparing with the status quo scenario —the scenario of the current treatment strategies to manage hypercholesterolemia without the introduction of PCSK9 inhibitors. We also consider an extended eligibility scenario which includes patients with no pre-existing cardiovascular disease (CVD) but at high-risk, to inform the potential value when the patients gain greater access to this new therapy. We conducted simulations using the Future Elderly Model (FEM), an established dynamic microsimulation model, to project the lifetime outcomes for the U.S. population aged 51 or older. Efficacy estimates of Drugs from published meta-analysis studies of the early Phase III clinical trials were used to project changes in life expectancy, quality-adjusted life-years, and lifetime medical spending resulting from use of PCSK9 inhibitors. While PCSK9 could add a substantial societal value in a long run, there is still a paramount concern that the high prices of innovative prescription drugs like PCSK9, which places economic burden under a short-term budget impact model. Utilization restriction by payers results in a slow adoption of PCSK9 inhibitors, which creates barriers for patients who could benefit from the new therapy and lead to worse clinical outcomes for patients. Therefore, the outcome-based pricing for prescription drugs has been recently taken hold in the U.S. to ensure the prices we pay for the new drugs reflect the benefits we receive. The payers must find new ways to achieve cost control without limiting access to beneficial services and at the same time ensure enough incentives to the innovation for the manufacturers. While substantial manufacturer discounts could address the dilemma, it is unclear whether those would be flexible enough to reflect real- world outcomes, especially if the therapy’s long-term effectiveness is better than initially expected or larger absolute risk reductions in major ASCVD event rates given the higher baseline event rate in the real world. Therefore, we propose a payment model - a three-part pricing model - incorporating the real-world evidence to encourage payers to pay for services that provide expected or better outcomes. The aim of our second paper is to compare the value of our three-part pricing model comparing with the current one-price model (status quo) by using 6 PCSK9 inhibitors as an example. The FEM is used to estimate the numbers of averted cardiovascular events in each year in those with existing CVD and/or FH and elevated LDL-C. We use the effectiveness estimates from a recently published clinical trial (FOURIER study) which focused on long-term cardiovascular outcomes in high risk patients. Among the eligible population, two PCSK9 uptake functions are used to estimate the numbers of patients actually be treated. To make the manufacturers be indifferent between the two pricing models, the drug price for our three-part pricing model will be calibrated to ensure the overall revenue during the patent protection period will be the same as the status quo. Given the access expansion of PCSK9 inhibitors is challenging and the hurdle of pricing strategy for this novel class of drugs hasn’t been solved, on the other hand it would be important to understand the societal value of improving the adherence to statins. The third paper of this dissertation aim to evaluate the longitudinal trends of statin adherence in the U.S elderly population and identify the factors which are associated with the better statin adherence. 7 CHAPTER 1: PCSK9 SHOW VALUE FOR PATIENTS AND THE U.S. HEALTHCARE SYSTEM Full Citation: Cheng WH, Gaudette É, Goldman DP. PCSK9 Inhibitors Show Value for Patients and the U.S. Healthcare System. Value Health. 2017 Dec;20(10):1270-1278. ABSTRACT OBJECTIVES: Proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors were approved by the U.S. Food and Drug Administration (FDA) as cholesterol-lowering therapies for patients with familial hypercholesterolemia or atherosclerotic cardiovascular disease. This study estimates the long-term health and economic value of PCSK9 inhibitors for older Americans (aged 51 and older). METHODS: We conducted simulations using the Future Elderly Model (FEM), an established dynamic microsimulation model, to project the lifetime outcomes for the U.S. population aged 51 or older. Health effects estimates and confidence intervals from published meta-analysis studies were used to project changes in life expectancy, quality-adjusted life-years, and lifetime medical spending resulting from use of PCSK9 inhibitors. We considered two treatment scenarios: 1) current FDA eligibility; and 2) an extended eligibility scenario which includes patients with no pre-existing cardiovascular disease (CVD) but at high-risk. We assumed the price of PCSK9 inhibitors was discounted by 35% in the first 12 years and by 57% thereafter, with gradual uptake of the drug in eligible populations. RESULTS: Utilization of PCSK9 inhibitors by individuals covered by current FDA approval would extend life-expectancy at age 51 by an estimated 1.1 years and would yield a lifetime net value of $5,800 per person. If utilization were extended to those at high-risk for CVD, PCSK9 inhibitors would generate a lifetime net benefit of $14,100 per person. CONCLUSION: Expanded access to PCSK9 inhibitors would offer positive long-term net value for patients and the U.S. healthcare system at the current discounted prices. 8 BACKGROUND AND INTRODUCTION Despite advances in medical technologies, cardiovascular disease (CVD) remains the leading cause of death—nearly 7.2 million deaths annually—and a major cause of disability in the United States.(1, 2) From 2011 to 2012, the estimated annual direct and indirect costs of CVD and stroke were more than $310 billion, (3) with annual costs projected to nearly triple from 2010 to 2030, from $273 billion to $818 billion. (4, 5) Additional interventions are needed for people at risk of CVD, focusing both on patient lifestyle changes and managing modifiable cardiovascular risk factors successfully. A large body of evidence demonstrates that low-density lipoprotein cholesterol (LDL-C) is a principal driver of atherosclerotic vascular disease (ASCVD)—the underlying cause of the majority of clinical manifestations of CVD—and thus the primary target for CVD risk reduction interventions.(6-9) For more than a decade, guidelines have indicated that patients with elevated LDL should use HMG-CoA reductase inhibitors, also known as statins, adjunct to diet to reduce cholesterol. Despite the use of statins, previous research has shown that a substantial proportion of treated high-risk patients fail to achieve target LDL-C levels.(10, 11) Moreover, statin intolerance is a common concern in clinical practice, with wide variation of individual lipid- lowering and risk reduction.(12-15) A substantial proportion of patients not meeting conventional LDL-C goals—more than 73 million U.S. adults (32%) experience elevated LDL- C(16)—suggests substantial benefits in reducing the burden of hypercholesterolemia. Recently, the FDA approved two monoclonal antibodies targeting proprotein convertase subtilisin/kexin type 9 (PCSK9), alirocumab and evolocumab, which are novel lipid-lowering approaches that inhibit the binding of PCSK9 to the LDL receptor, resulting in powerful LDL-C lowering potency.(17-19) Both agents are administered via subcutaneous injection and were approved as an adjunct to diet and maximally tolerated statin therapy for treatment of adults with familial hypercholesterolemia (FH) or clinical ASCVD who require additional lowering of LDL- C.(20) Preliminary Phase III clinical trials, though not powered to assess long-term cardiovascular outcomes because of the short study windows, showed approximately a 50% risk reduction in cardiovascular events and all-cause mortality while maintaining a favorable safety profile.(21- 23) If clinical benefits observed in trials are sustained long term, PCSK9 inhibitors could 9 become an important option for patients at high risk of ASCVD and potentially create substantial health benefit by preventing CVD events. Despite health benefits suggested by the literature, payers and policymakers are concerned this new class of expensive specialty medications poses a substantial economic burden given the drugs’ current prices, which range from $14,100-$14,600 per patient per year. A recent study suggested that even if the drug price of PCSK9 inhibitors could be covered by an annual $245 billion savings in prevented CVD events, the high price of PCSK9 inhibitors still poses a substantial economic burden to the U.S. healthcare system if only accounting for direct medical costs from avoided CVD. (24) Another study suggested these agents may not be cost-effective in patients with FH or ASCVD at current U.S. prices.(25) Answers are needed about whether PCSK9 inhibitors will significantly improve mortality and reduce CVD events in Americans beyond currently available diet-statin therapy and prove cost-effective over time.(26) Given uncertainty in long-term efficacy, the FDA approved use of PCSK9 inhibitors under strict criteria, and because of the drugs’ high price, payers have suggested that PCSK9 inhibitors should be targeted to a narrow population. (20, 27, 28) However, a larger population with elevated LDL-C could benefit from PCSK9 inhibitors because clinical trials have shown substantial LDL-C lowering effects in persons who failed to receive adequate health benefits from statins regardless of their history of ASCVD.(29-31) Therefore, it is essential to evaluate the value of PCSK9 inhibitors under broader indications. This article’s primary objective is to estimate the health benefits of PCSK9 inhibitors in the U.S. population with familial hypercholesterolemia or cardiovascular disease and to quantify the value of these gains, while taking into account the uncertainty surrounding the drugs’ clinical effectiveness. Secondarily, we estimate the long-term value of PCSK9 inhibitors if their utilization were extended to persons with no pre-existing cardiovascular disease but with high- CVD risks. METHODS The Future Elderly Model 10 We estimated potential health benefits and costs by using the Future Elderly Model (FEM), a dynamic microsimulation model that tracks cohorts older than age 50 to project their health and economic outcomes. Rather than aggregating health characteristics of a cohort, the FEM follows the evolution of individual-level health trajectories in a microsimulation framework. Initially, the FEM was developed to forecast long-term health and healthcare costs under different scenarios for medical technology and utilization.(32) In recent years, the FEM has been used to estimate the value of statin therapy in the obese population,(33) the value of aspirin,(34) and the value of delayed biological aging.(35, 36) The FEM also has been used to estimate the impacts of other health policy changes, such as the introduction of dietary sodium reduction policies,(37) tobacco control policies, (38) and U.S. pharmaceutical policy.(39) We describe the model and methods briefly here; complete technical information is available in the Appendix. The FEM simulates the lives of older Americans based on the Health and Retirement Survey (HRS), a nationally representative biennial survey of Americans aged 51 and older. The FEM has three core components. The first is the Health Transition module, which consists of a series of health and functional status transition equations and mortality equations to model the health of the 51+ population over their lifetimes. Health is described by the presence of certain chronic conditions, and functional status is measured by limitations in activities of daily living, instrumental activities of daily living, and nursing home residency, reported by the HRS data. All health conditions, functional states, and risk factors were modeled with first-order Markov processes that controlled for a set of baseline variables, including age, gender, education, race, body-mass index, smoking status, and health at the time of entry into the study. For the purpose of this study, we added biomarkers (specifically cholesterol and HbA1c levels), blood pressure measurements, and treatment status (including respondents’ current therapies for cholesterol, blood pressure control, and diabetes) in the health transition module. These variables were added to better identify the target population eligible for PCSK9 inhibitors and their CVD risk as the model moves forward. These variables were obtained from the HRS biomarker data available from 2006 to 2012. We added biomarker and blood pressure levels to the list of covariates predicting cardiovascular disease. In the simulations, biomarker and blood pressure transitions were modelled as a function of respondents’ social demographics, health status, and treatment status. 11 We computed quality-adjusted life-year (QALY) measures based on the EQ-5D, a standardized health-related quality-of-life instrument measuring a respondent’s general health status on five dimensions: mobility, daily activities, self-care, anxiety, depression, and pain.(40) EQ-5D scores are estimated with an ordinary least squares regression as a function of the chronic conditions and FEM-specified functional status, using Medical Expenditure Panel Survey (MEPS) data. The second FEM component is the Policy Outcomes module, which examines fiscal outcomes, including the costs of health entitlement programs—specifically federal and state spending for Medicare and Medicaid. The FEM predicts expected enrollment for Medicare and Medicaid, as well as expenditures for both entitlement programs and private medical expenditures, given a set of health, economic, and demographic states and characteristics. The predictions are based on MEPS data prior to age 65 and the Medicare Current Beneficiary Survey after age 65. The third component is a Replenishing Cohort module, which introduces new cohorts of 51-year- olds in each simulated year as the model progresses. The FEM predicts the demographic and health characteristics for these younger populations based on data from the National Health Interview Survey, the Current Population Survey, and the National Health Nutrition and Examination Survey. Simulations First, we conducted “cohort simulations,” tracking a 2016 cohort of Americans aged 51 to 52 until their death under alternative PCSK9 inhibitor scenarios. In addition, we conducted “population simulations” to investigate the population-wide trends implied by observing a representative cross-section of the older U.S. population in each period. We used the full FEM population to project outcomes (including the replenishing cohort) for the entire population of Americans aged 51 and older from year 2016 to year 2056. Population health outcomes were calculated for each time period by aggregating individual health measures. Scenarios We considered three scenarios—one representing the status quo and the other two representing scenarios introducing use of PCSK9 inhibitors. In PCSK9 inhibitor scenarios, we modified the status quo scenario by applying the health benefits and additional healthcare spending from PCSK9 inhibitor use to the current standard of care. First, a status quo scenario establishes a 12 baseline assuming treatment strategies for managing hypercholesterolemia in the U.S prior to the introduction of PCSK9 inhibitors. We next generated two scenarios to evaluate the benefits and costs of introducing PCSK9 inhibitors. Besides meeting the criteria described below, the patients are required to meet additional criteria to become eligible for PCSK9 inhibitors, including being aged 80 and younger, currently being on a cholesterol-lowering therapy, and having failed to reduce LDL-C to ≤ 70 mg/dl: a. Current Eligibility: In this scenario, the populations eligible for PCSK9 inhibitors were defined by current FDA-approved indications(20) and the first two groups of statin benefit groups as outlined by American College of Cardiology/American Heart Association (ACC/AHA) guidelines,(8) which included those with familial hypercholesterolemia (defined as LDL- C level higher than 190 mg/dL)(41, 42) and preexisting CVD. b. Extended Eligibility: This scenario extends access to patients without a history of CVD but with high-risk equivalents are also eligible for treatment with PCSK9 inhibitors. The CVD high-risk equivalents were defined as persons with diabetes aged 40 to 75 years, or with an estimated 10-year ASCVD risk >7.5%. This group corresponded with the statin benefit groups 3 and 4 in ACC/AHA guidelines.(8) Uncertainty surrounding long-term effectiveness and pricing concerns have served as barriers to widespread adoption of PCSK9 inhibitors, and their adoption has been gradual.(27, 43) Therefore, among the PCSK9 inhibitor-eligible population, we assumed a zero probability of actual PCSK9 inhibitor assignment in year 2014, and the probability linearly increases to one through year 2020. The process to identify PCSK9 inhibitor eligibility in FEM simulations is detailed in Appendix. We estimated that 13.8 million individuals were eligible for PCSK9 inhibitors under current FDA approval in 2016; eligible individuals increased to 28.5 million under extended eligibility. (Figure 1). The actual PCSK9 inhibitor assignment after phasing in adoption in a gradual linear manner is displayed in Figure A1. There were about 4.6 million and 9.5 million individuals assigned to use PCSK9 inhibitors under current and extended eligibilities, respectively, in 2016. 13 The impact of PCSK9 inhibitors on health and costs To reflect the health impacts of PCSK inhibitors reported in the literature, we modified health transitions and outcomes of eligible individuals in the PCSK9 inhibitor scenarios, specifically by reducing the risk of having the first cardiovascular disease and all-cause mortality, as well as applying additional drug costs and disutility weights for PCSK9 inhibitors. The key parameters and their ranges for sensitivity analysis are listed in Table 1. For PCSK9 inhibitor-eligible individuals and for those without prior heart disease, we decreased the probabilities of heart disease incidence by factors with a mean of 0.54, and for those receiving PCSK9 inhibitors, we decreased their probabilities of mortality by factors with a mean of 0.45, (21, 22), which correspond to the risk-ratios reported by published meta-analyses. The risk reduction in mortality was further adjusted to account for the interaction between heart disease prevention and mortality. (Appendix A). Since the effectiveness of PCSK9 inhibitors among subgroups with different characteristics are not yet well established, we assumed the same effect for populations taking PCSK9 inhibitors across their life-years covered by the drugs. Although research has found an elevated risk of a neurocognitive adverse event suspected to be associated with use of PCSK9 inhibitors, the FDA concluded that the association was insignificant. (44) Evidence has shown utility differences between different treatment modalities, with a subcutaneous injection leading to lower health utility compared with an oral therapy; therefore, we imposed a disutility of 0.004 per person receiving PCSK9 inhibitors.(45) Annual costs of PCSK9 inhibitors were assumed to be equal to their wholesale acquisition costs; assumed to be the mean of the 2015 annual costs of alirocumab ($14,600) and evolocumab ($14,100).(46) The price for a new branded drug usually falls by a discount rate ranging from 23% to 46%, after the entry of other branded competing drugs; and an additional discount of 33% will be applied due to the expiration of patent protection. (47, 48) Therefore, our analysis assumed that branded PCSK9 inhibitor drugs will be discounted by 35% after their first year on the market; after 2028, PCSK9 inhibitor drugs are discounted by an additional 33%, resulting in a 57% total discount.(47, 49) All costs outcomes were adjusted to 2015 US dollars. 14 Uncertainty Three different approaches of sensitivity analysis were performed to evaluate the robustness of our results. Probability sensitivity analysis was used to adjust the parameter uncertainty surrounding clinical effectiveness, which was estimated with the wide confidence intervals of the meta-analyses results shown in Table 1. This uncertainty will likely diminish as more clinical trials are conducted and samples increase but needs to be accounted for in our analyses. To do so, we drew random values from the distributions of relative-risk estimates of effectiveness reported in the literature for each repetition of simulations. In detail, we first drew 200 sets of risk-ratio estimates from log-normal distributions and conducted separate simulations for each set of estimates; then we computed the results from the 200 simulations and sorted them for all outcomes of interest. As a result, the point estimates of our results correspond to the mean of each variable of interest across the 200 simulations. The bounds of the 95% confidence intervals correspond to the fifth lowest and highest results from the sorted estimates of the 200 simulations. These intervals can be interpreted as the 95% confidence intervals with regard to the clinical uncertainty of the effectiveness of PCSK9 inhibitors. Additional scenarios were generated to adjust the structural uncertainty of the model assumptions, in terms of eligibility criteria and effect size of PCSK9 inhibitors. One-way sensitivity analysis was performed to examine how the results would differ with changes in drug cost, disutility weights, and discount rate parameters. Details can be found in Appendix A. RESULTS Life expectancy, quality of life, and functional status Our cohort simulations revealed that PCSK9 inhibitor use would significantly increase life expectancy, disability-free life expectancy and quality of life, and reduce the incidence of heart disease. Table 1 summarizes how the scenarios would affect key health indicators over the life course of nationally representative 51-year-olds in comparison to the current standard of care (the status quo scenario).Compared to the status quo scenario, the cohort is expected to live on average 1.1 years longer under the FDA-approved PCSK9 inhibitor scenario and 1.9 years longer under the extended PCSK9 inhibitor eligibility criteria (Table 2). 15 The cumulative incident cases of cardiovascular disease by age 79 were predicted for 1,000 individuals without prior cardiovascular disease at age 51. Because of the efficacy of PCSK9 inhibitors on mortality reduction, individuals are estimated to live longer and develop chronic diseases. Given that the majority of people under FDA-approved eligibility have existing heart disease, the risk reduction of PCSK9 inhibitor on the first heart disease was diluted by the mortality effect, which resulted in only 7.7 less cases in every 1,000 people as compared to the status quo scenario, which was attributable to the population with FH or stroke but without existing heart disease. When the eligibility is extended to those without prior heart disease but are high-risk equivalent, PCSK9 inhibitors would lead to 40.3 fewer cases of heart disease. We would expect a greater incidence of other chronic conditions, including diabetes, hypertension, stroke, and cancer. Consequently, under the current eligibility scenario, among the 1.1 additional life years attributable to PCSK9 inhibitors, people live only 0.4-year in a healthy state. Figure 2 displays the results of population simulation analyses, where we predicted the effects of PCSK9 inhibitor use on the prevalence of heart disease (Figure 2A) and disability (Figure 2B) from year 2016 through the next four decades and compared the forecasted results in the status quo scenario with the PCSK9 inhibitor scenarios. Due to longer life expectancy of individuals with heart disease, in 2036, the prevalence of heart disease is projected to increase by 0.8 percentage points under current eligibility, compared to the status quo. In contrast, prevalence of heart disease is projected to fall by 1.5 percentage points relative to the status quo under extended eligibility, in which individuals use PCSK9 inhibitors to prevent the onset of heart disease. Due to the extension of life expectancy and the reduced risk of heart disease, we estimated 1.2 million to 2.5 million more members of the healthy population in each PCSK9 inhibitor scenario than those in the status quo scenario in 2036. Cost-effectiveness We projected the expected lifetime medical costs for our cohorts at age 51 and estimated the value of health benefits against medical costs. We used a 3% discount rate as base-case to compute present values from the age of 50 for both costs and benefits. We considered the value per QALY gained as $150,000, which is an acceptable threshold of three times the U.S. gross domestic product (GDP) per capita, as recommended by the World Health Organization.(50) Compared with the status quo, the values associated with QALYs gained are estimated to be 16 $45,900 higher in the FDA-approved eligibility scenario and $83,900 higher in the extended eligibility scenario. Under the FDA-approved indications, the additional healthcare costs, including the drug cost of PCSK9 inhibitors are estimated to be $40,100 compared to the status quo. In the extended eligibility scenario, we estimated a larger additional incremental healthcare cost ($69,800) due to the effect on longevity from more people receiving PCSK9 inhibitors. Compared to the status quo, PCSK9 inhibitors generated an average $132,000 and $125,900 per additional QALY gained in the FDA-approved eligibility and extended eligibility scenarios, respectively. The net value of PCSK9 inhibitors is estimated to be $5,800 in the FDA-approved eligibility scenario but is significant at a 0.10 level, while the value in extended eligibility scenario is $14,100 and significant at a 0.05 level. (Table 3). Incremental QALYs gained under the extended eligibility scenario relative to the FDA-approved eligibility scenario would cost on average $120,400 per QALY and provide $8,200 of net value per capita (Appendix Table A4). At the population level, the aggregate incremental medical costs in both PCSK9 inhibitor scenarios were estimated to outweigh the total value of QALYs gained in the first eight years (Figure A2 in the Appendix). However, the net values become positive since year 2026 in both scenarios and are projected to grow after the drugs go off patent. The FEM projected a cumulative net present value of $0.54 trillion in the PCSK9 inhibitor current eligibility scenario and $0.90 trillion in the extended eligibility scenario by 2036. Sensitivity analysis Five other scenarios were created to account for additional sources of uncertainty surrounding our simulations. The results of these analyses are presented in Appendix. Most scenarios resulted in similar or less favorable net values for PCSK9 inhibitors, as compared with our baseline results. The net values of PCSK9 inhibitors become negative if only 10% of the population eligible for PCSK9 inhibitors was statin-intolerant and used PCSK9 inhibitors. For the FDA- approved scenario, higher positive net values and favorable ICERs were observed when using the clinical effectiveness estimated from the relationship between the LDL-C reduction and the risks of event, instead of using estimates from the PCSK9 inhibitor meta-analysis. In general, despite that, the base-case estimate is sensitive to assumptions in eligibility criteria and the effect size of the drugs, with the average net value per capita ranging from -$3,600 to $11,400 under the FDA-approved eligibility and from $-900 to $12,900 under extended eligibility; the mean 17 ICER estimates in the scenarios were all below or equal to $150,000 per QALY gained (Table A9). In one-way sensitivity analyses varying drug cost, utility weight, and discount rate, estimates are most sensitive to the cost of PCSK9 inhibitors (Table A11). If patients have no access to the discount for the PCSK9 inhibitors during the patent protection period, the net value of PCSK9 inhibitors would be small under the FDA-approved eligibility. DISCUSSION Our study estimates the economic value of PCSK9 inhibitors for the United States over the next 40 years and demonstrates that use as indicated could substantially improve health outcomes among Americans over age 50, including reducing the incidence of cardiovascular disease and extending life expectancy. Assuming current discounted prices, treatment with PCSK9 inhibitors for older Americans under current indications yield substantial net benefits. When analyzing a cohort of 51-year-olds living in 2016, PCSK9 inhibitors used under FDA-approved indications are estimated to yield a lifetime net value of $5,800 per capita (which is significant at a 0.10 level). Interestingly, PCSK9 inhibitors are estimated to generate a higher lifetime net benefit, up to $14,100 per capita, when use is extended to people without a history of CVD but who are at risk of developing the disease. Given our comparator—the status quo scenario of treatment strategies to manage hypercholesterolemia, which include not only statins but also other interventions such as ezetimide, diet, or exercise under current clinical guidelines—our study informs potential values of PCSK9 inhibitors to the healthcare system in addition to the current standard of care. Adding PCSK9 inhibitors to the treatment arsenal is also cost-effective, even with broad eligibility criteria. We estimate an incremental cost-effectiveness ratio (ICER) of $132,200 under current FDA-approved indications and $125,900 under extended eligibility, which fall well within the accepted range for well-known recent innovative therapies. In the case of the new oral anticoagulant dibigatran for stroke prevention, the ICER was reported as $143,000 versus traditional warfarin therapy in elderly Americans. In another example, the novel hepatitis C treatment, sofosbuvir, showed ICERs ranging from $9,700 to $284,300 depending on the 18 patient's status with respect to treatment history, HCV genotype, and presence of cirrhosis. (51- 53) Although evidence of the effectiveness of PCSK9 inhibitors continues to accumulate, the drug poses a challenge for payers and policymakers who have raised concerns about price and short- term affordability.(26, 27, 54) Our results suggest that PCSK9 inhibitors would increase lifetime medical spending because of increased drug costs and the impact of PCSK9 inhibitors on extending life, ultimately increasing economic burdens from higher risk of comorbidities accompanying natural aging. We also find that the value of PCSK9 inhibitors is sensitive to the price of this new drug. Therefore, if consumers have access to a discounted price for the drug, the value of health gained could outweigh costs over the long term and more generous coverage criteria for PCSK9 inhibitors would potentially generate a higher net value, under the assumption that the efficacy of these drugs in CVD incidence and overall mortality are consistent with the current evidence among PCSK9 inhibitor users. In a recent study, Kazi et al. estimated an incremental cost of about $150,000 per QALY gained for PCSK9 inhibitors compared with adding ezetimibe to statins (25) when the price of PCSK9 inhibitor is $6,810 per year, which is close to the price level we assumed when the drug goes generic. The discrepancy in results also can be explained by the model parameters and assumptions. First, their model contains only CVD-associated health and health spending outcomes; they did not capture disability or costs attributable to hypertensive heart disease, peripheral arterial disease, or other non-cardiovascular outcomes associated with atherosclerotic cardiovascular disease. Therefore, they likely underestimated the cost savings and QALY gains associated with PCSK9 inhibitors, while the FEM medical spending module covers a wider range of spending from all associated comorbidities, mortality, and disabilities. Second, their study assumed perfect long-term compliance to therapy, potentially overestimating the outcomes of their control group and resulting in lower effectiveness in the PCSK9 inhibitor group. In contrast, our status quo scenario represents the current strategies and treatment for hypercholesterolemia in the United States, which is more representative of real-world practice. Our study also considers a gradual take-up of this new class of drug, which might cause lower incremental costs from PCSK9 inhibitors but is closer to reality when adopting a novel medical technology. 19 Our study has several limitations. First, as reflected by the wide confidence intervals around our results, there remains uncertainty regarding the clinical effectiveness of PCSK9 inhibitors. Our results reflect the uncertainty from existing meta-analyses that aggregate results across several trials, but the evidence is still deficient. For instance, a recently published trial, focusing on long- term cardiovascular outcomes (FOURIER study) raised questions about the validity of some of our parameters.(55) This new study found risk reductions on stroke, which we did not include, but did not find an all-cause mortality reduction among patients with clinically evident cardiovascular disease, even though significant reductions in aggregated major cardiovascular events were achieved. While the study adds important new information about the effectiveness of PCSK9 inhibitors, it was not sufficiently powered to estimate overall mortality within a short follow-up period (on average 2.5 years). In addition, the trial enrolled healthier and younger patients than the real world, who are at lower risk of CVD events and might benefit less from aggressive secondary prevention. Similar with other randomized controlled trials literature we used to estimate the health effects of PCSK9 inhibitors, the effectiveness might not be generalizable to the real world and might differ from clinical trial settings due to lower adherence with drug therapy. A recently published commentary also highlighted the importance of using real-world data to estimate the efficacy of the new technologies when performing cost- effectiveness analyses, to provide more valuable results for a comprehensive population of people who could benefit from the interventions.(56) Thus, more evidence is needed to clarify the long-term health benefits of PCSK9 inhibitors. In addition, since the FEM did not model second CVD events, we might underestimate health benefits of PCSK9 inhibitors for those with prior CVD. The potential cost reductions and quality of life improvements due to avoided secondary CVD events are ignored in a secondary prevention setting, while the additional costs we find are likely overestimated. In addition, in the FEM we didn’t estimate the medical spending due to a specific diagnosis or event; therefore, a limitation of our study is the inability to estimate costs of cardiovascular events avoided by the treatment. Lastly, we assumed that the relevant treatment population for PCSK9 inhibitors included persons on cholesterol-lowering therapies who were not at goal LDL-C. In reality, many patients are either not at maximally tolerated doses or are non-adherent with their current therapies.(57) Optimizing therapies may reduce the size of the population who could potentially benefit from PCSK9 inhibitors, and our estimates may be an upper bound of the value of PCSK9 inhibitors in these populations. Also, 20 other interventions besides PCSK9 inhibitors could be considered as comparators for cholesterol management, such as more aggressive lifestyle changes relevant to risk factors. In sum, CVD imposes an enormous health burden on the older population in the United States. Although PCSK9 inhibitors are expected to increase healthcare spending, our study estimates the value of health gains outweighs the costs, based on the best available evidence of the drugs’ health impact. In addition to current FDA-approved indications, our study also suggested potential greater value if more people gain access to this new class of drugs. 21 TABLES AND FIGURES Table 1. Key Input Parameters Used in the Model Parameters Base-case Reference Range for sensitivity Analysis Type of sensitivity analysis Effect size Relative risk of events for PCSK9 inhibitor versus standard of care based on the Phase III clinical trials 1 Probabilistic sensitivity analysis performed by drawing random values from the confidence intervals for relative-risk estimates of the base-case estimate for each repetition of simulations. Overall mortality 0.45 Navarese EP et al., 2015 [0.23 - 0.86] Major cardiovascular events 0.54 Lipinski MJ et al., 2016 [0.38 - 0.77] Costs (in 2015 dollars) Annual drug costs for PCSK9 inhibitors Original cost $14,350 Red Book Online [10,763, 17,938] One-way sensitivity analysis based on +/- 25% of the base- case estimates. Discounted cost due to rebate and the entry of competitors 2 $9,328 Tirrell M et al., 2015 [6,996 , 11,660] 22 Cost without patent protection (after year 2028) 3 $6,249 Conti RM et a., 2014 [4,686, 7,811] Utility weights Disutility due to injection, per patient 0.004 Matza LS et al., 2013 [0.003, 0.005] One-way sensitivity analysis Discount rate 3% [1%, 5%] One-way sensitivity analysis 1 The base-case assumed a constant relative reduction in the risk of overall mortality and major cardiovascular event, independent of how long the patients were treated by PCSK9 inhibitors. This was estimated from the most recent meta-analysis of PCSK9 trials that reported the end point. There were too few strokes in the short-term PCSK9 trials, so we didn’t consider the benefit in stroke with PCSK9 inhibitors. 2 The cost is calculated as $14,350 multiplied by 0.65 (a 35% discount). 3 The cost is calculated as $14,350 multiplied by 0.65, then multiplied by 0.67 (an additional 33% discount). 23 FIGURE 1. Projected Populations Eligible for PCSK9 Inhibitors by Statin-benefit Groups (SBGs), Year 2016 and 2036* *Individuals in statin-benefit groups (recommended by ACC/AHA guidelines) who failed to achieve a goal LDL-C level (≥ 70 mg/dL) and who regularly take lipid-lowering therapy are potentially eligible for PCSK9 inhibitor use until age 80. Current Elig refers to the current eligibility 0 5 10 15 20 25 30 35 40 45 2016 2036 Millions of PCSK9 Inhibitor Eligible Individuals CVD risk equivalents & LDL-C ≥70 mg/dL & Treated CVD & LDL-C ≥70 mg/dL & Treated FH 28.5 1.8 1.8 12.1 15.1 18.1 19.3 38.6 Current Elig Extended Elig Extended Elig Current Elig 24 criteria for PCSK9 inhibitors, corresponding to FDA approval. Extended Elig refers to the extended eligibility for PCSK9 inhibitors, using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. CVD risk equivalents refer to individuals with a clinical diagnosis of diabetes and estimated 10-year CVD risk higher than 7.5%. CVD: Cardiovascular disease, defined as any diagnosis of congestive heart failure, coronary heart disease, angina, heart attack, and any other heart diseases. FH: familial hyperlipidemia, defined as those with LDL-C levels ≥190 mg/dl. 25 TABLE 2. The Impact of PCSK9 Inhibitor Use on Health Outcome at Age 51 Current Eligibility refers to the current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. *The result differences compared with the status quo scenario. **The results were calculated from age 51 until death. †Total incidence before age 79 for a 1,000 Status quo PCSK9 inhibitor Scenario I - Current Eligibility PCSK9 inhibitor Scenario II - Extended Eligibility Mean Mean Difference * 95% CI Mean Difference * 95% CI Health outcomes, at age 51 ** Life expectancy, years 30.9 32.0 1.1 [0.9, 1.4] 32.7 1.9 [1.5, 2.3] Disability-free life expectancy, years 22.9 23.2 0.4 [0.3, 0.5] 23.6 0.8 [0.6, 1.0] Quality-adjusted life years (QALYs) 25.5 26.1 0.6 [0.4, 0.8] 26.7 1.2 [0.9, 1.5] Cumulative disease incidence at age 79 (per thousand)† Heart Disease 419.3 411.6 -7.7 [-0.9, -14.7] 379.0 -40.3 [-35.5, -69.8] Stroke 207.2 219.0 11.8 [5.0, 18.9] 222.3 15.0 [6.3, 23.4] Diabetes 473.8 479.5 5.7 [1.7, 9.9] 483.0 9.2 [2.4, 14.5] Hypertension 732.4 737.6 5.2 [0.3, 11.8] 742.2 9.8 [1.7, 14.9] Cancer 285.5 294.8 9.4 [4.0, 16.3] 299.9 14.4 [7.0, 21.3] 26 individuals without the disease at age 51. Disability-free life expectancy refers to reporting no instrumental activity of daily living or activity of daily living limitations and not living in a nursing home. Quality-adjusted life-years adjust length of life for quality based on a person’s chronic conditions and functional status. 95% confidence intervals of the difference estimates are presented in brackets, indicating the uncertainty of the effectiveness of PCSK9 inhibitors. The 95% confidence intervals for the mean values in each scenario can be found in the Supplemental Appendix. 27 FIGURE 2. The Long-term Health Impacts of PCSK9 Inhibitor Use on U.S. Population Aged over 50* Figure 3A compared the projected prevalence of heart disease, and Figure 3B compared the projected millions of healthy individuals with the status quo and two PCSK9 inhibitor scenarios. * Individuals in statin-benefit groups (recommended by ACC/AHA guidelines) who failed to achieve a goal LDL-C level (≥ 70 mg/dL) and who regularly take lipid-lowering therapy are potentially eligible for PCSK9 inhibitor use until age 80. Among the PCSK9 inhibitor-eligible population, the actual assignment of PCSK9 inhibitor use was phased-in with a probability of zero starting from year 2014 and linearly increased to 1 through year 2020. PCSK9 inhibitor Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. Healthy individuals refers to those reporting no instrumental activity of daily living or activity of daily living limitations and not living in a nursing home. Heart disease includes any diagnosis of congestive heart failure, coronary heart disease, angina, heart attack, and any other heart diseases. A. Prevalence of Heart Disease (%) B. Millions of Healthy Individuals 86.2 96.3 101.5 97.5 102.6 98.8 103.8 85 90 95 100 105 110 Millions of Healthy Individuals 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 2060 Year Status quo PCSK9i Current Eligibility PCSK9i Extended Eligibility 24.5 31.3 31.5 32.1 32.7 29.8 30.3 24 26 28 30 32 Prevalence of Heart Disease (%) 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 2060 Year Status quo PCSK9i Current Eligibility PCSK9i Extended Eligibility 28 TABLE 3. The Net Benefits of PCSK9 Inhibitor Use at Age 51 ($2015 thousands) Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. *The result differences compared with the status quo scenario. **Calculated by multiplying QALY-gains with a willingness-to-pay of $150,000. †PCSK9 inhibitor medication costs were calculated based on a discount rate of 35% from 2016 to 2027, with an additional 33% discount since year 2028. All amounts are in present value at age 51, computed with a 3% discount rate. 95% confidence intervals of the mean differences are presented in brackets, indicating the uncertainty of the effectiveness of PCSK9 inhibitors. The 95% confidence intervals for the mean values in each scenario can be found in the Supplemental Appendix. All the costs were adjusted to year 2015 US thousand dollars. Status quo PCSK9 Inhibitor Scenario I – Current Eligibility PCSK9 Inhibitor Scenario II – Extended Eligibility Mean Mean Difference * 95% CI Mean Difference * 95% CI Value of expected quality-adjusted life-years gained (Discounted) 2465.6 2511.5 45.9 [33.5, 58.9] 2549.5 83.9 [66.5, 102.5] Per Capita Life-time healthcare and medication costs (Discounted) Healthcare costs 538.7 567.9 29.3 [21.4, 39.6] 587.4 48.6 [36.7, 59.4] PCSK9i medication costs† 0.0 10.8 10.8 [10.1, 11.4] 21.3 21.3 [20.4, 22.3] Total 538.7 578.7 40.1 [32.3, 50.7] 608.5 69.8 [59.3, 82.3] Incremental Cost-Effectiveness Ratio (ICER) 132.2 [110.8, 153.9] 125.9 [111.4, 140.1] Net value per capita (in thousands) ** 5.8 [-1.1, 13.7] 14.1 [4.2, 25.3] 29 REFERENCES 1. The global burden of disease: 2004 update. Available from: http://www.who.int/healthinfo/global_burden_disease/GBD_report_2004update_full.pdf?ua=1, 2006. [Accessed December 15, 2016] 2. WHO. Prevention of cardiovascular disease: guidelines for assessment and management of cardivascular risk. Geneva: World Health Organization 2007. Available from: http://www.who.int/cardiovascular_diseases/guidelines/PocketGL.ENGLISH.AFR-D- E.rev1.pdf. [Accessed December 01, 2016] 3. Mozaffarian D, Benjamin EJ, Go AS, et al. Heart disease and stroke statistics--2015 update: a report from the American Heart Association. Circulation. 2015; 131: e29-322. 4. Weintraub WS, Daniels SR, Burke LE, et al. Value of primordial and primary prevention for cardiovascular disease: a policy statement from the American Heart Association. Circulation. 2011; 124: 967-90. 5. Heidenreich PA, Trogdon JG, Khavjou OA, et al. Forecasting the future of cardiovascular disease in the United States: a policy statement from the American Heart Association. Circulation. 2011; 123: 933-44. 6. Boekholdt SM, Arsenault BJ, Mora S, et al. Association of LDL cholesterol, non-HDL cholesterol, and apolipoprotein B levels with risk of cardiovascular events among patients treated with statins: a meta-analysis. Jama. 2012; 307: 1302-9. 7. Sniderman AD, Williams K, Contois JH, et al. A meta-analysis of low-density lipoprotein cholesterol, non-high-density lipoprotein cholesterol, and apolipoprotein B as markers of cardiovascular risk. Circulation Cardiovascular quality and outcomes. 2011; 4: 337-45. 8. Stone NJ, Robinson JG, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Journal of the American College of Cardiology. 2014; 63: 2889-934. 30 9. Guideline N. Cardiovascular disease: risk assessment and reduction, including lipid modification. Available from: https://www.nice.org.uk/guidance/CG181. [Accessed December 15, 2016] 10. Rodriguez F, Olufade T, Heithoff K, et al. Frequency of high-risk patients not receiving high-potency statin (from a large managed care database). The American journal of cardiology. 2015; 115: 190-5. 11. Mann D, Reynolds K, Smith D, et al. Trends in statin use and low-density lipoprotein cholesterol levels among US adults: impact of the 2001 National Cholesterol Education Program guidelines. The Annals of pharmacotherapy. 2008; 42: 1208-15. 12. Zhang H, Plutzky J, Skentzos S, et al. Discontinuation of statins in routine care settings: a cohort study. Annals of internal medicine. 2013; 158: 526-34. 13. Glueck CJ, Budhani SB, Masineni SS, et al. Vitamin D deficiency, myositis-myalgia, and reversible statin intolerance. Current medical research and opinion. 2011; 27: 1683-90. 14. Hsia SH, Desnoyers ML, Lee ML. Differences in cholesterol management among states in relation to health insurance and race/ethnicity across the United States. Journal of clinical lipidology. 2013; 7: 675-82. 15. Waters DD, Brotons C, Chiang CW, et al. Lipid treatment assessment project 2: a multinational survey to evaluate the proportion of patients achieving low-density lipoprotein cholesterol goals. Circulation. 2009; 120: 28-34. 16. Mozaffarian D, Benjamin EJ, Go AS, et al. Executive Summary: Heart Disease and Stroke Statistics--2016 Update: A Report From the American Heart Association. Circulation. 2016; 133: 447-54. 17. Agabiti Rosei E, Salvetti M. Management of Hypercholesterolemia, Appropriateness of Therapeutic Approaches and New Drugs in Patients with High Cardiovascular Risk. High blood pressure & cardiovascular prevention : the official journal of the Italian Society of Hypertension. 2016; 23: 217-30. 18. Gouni-Berthold I. PCSK9 antibodies: A new class of lipid-lowering drugs. Atherosclerosis Supplements. 2015; 18: 21-7. 31 19. Li C, Lin L, Zhang W, et al. Efficiency and safety of proprotein convertase subtilisin/kexin 9 monoclonal antibody on hypercholesterolemia: a meta-analysis of 20 randomized controlled trials. Journal of the American Heart Association. 2015; 4: e001937. 20. Amgen. Repatha (evolocumab) Package Insert. Available from: http://pi.amgen.com/united_states/repatha/repatha_pi_hcp_english.pdf. [Accessed September 15, 2016] 21. Navarese EP, Kolodziejczak M, Schulze V, et al. Effects of Proprotein Convertase Subtilisin/Kexin Type 9 Antibodies in Adults With Hypercholesterolemia: A Systematic Review and Meta-analysis. Annals of internal medicine. 2015; 163: 40-51. 22. Lipinski MJ, Benedetto U, Escarcega RO, et al. The impact of proprotein convertase subtilisin-kexin type 9 serine protease inhibitors on lipid levels and outcomes in patients with primary hypercholesterolaemia: a network meta-analysis. European heart journal. 2016; 37: 536- 45. 23. Robinson JG, Farnier M, Krempf M, et al. Efficacy and safety of alirocumab in reducing lipids and cardiovascular events. The New England journal of medicine. 2015; 372: 1489-99. 24. Glueck CJ, Shah P, Goldenberg N, et al. Eligibility for PCSK9 treatment in 734 Hypercholesterolemic patients referred to a regional cholesterol treatment center with LDL cholesterol >/= 70 mg/dl despite maximal tolerated cholesterol lowering therapy. Lipids in health and disease. 2016; 15: 55. 25. Kazi DS, Moran AE, Coxson PG, et al. Cost-effectiveness of PCSK9 Inhibitor Therapy in Patients With Heterozygous Familial Hypercholesterolemia or Atherosclerotic Cardiovascular Disease. Jama. 2016; 316: 743-53. 26. Weintraub WS, Gidding SS. PCSK9 Inhibitors: A Technology Worth Paying For? PharmacoEconomics. 2016; 34: 217-20. 27. Mehr SR. Will the PCSK9 Inhibitors Be Employers' "Line in the Sand"? American health & drug benefits. 2016; 9: 171-4. 28. Lyons K. High Cholesterol PCSK9 Drugs Could be Next To Bust Budgets, Stretch Pocketbooks. 2015. Available from: http://www.prnewswire.com/news-releases/high- 32 cholesterol-pcsk9-drugs-could-be-next-to-bust-budgets-stretch-pocketbooks-300096091.html. [Accessed October 30, 2016] 29. Colhoun HM, Robinson JG, Farnier M, et al. Efficacy and safety of alirocumab, a fully human PCSK9 monoclonal antibody, in high cardiovascular risk patients with poorly controlled hypercholesterolemia on maximally tolerated doses of statins: rationale and design of the ODYSSEY COMBO I and II trials. BMC cardiovascular disorders. 2014; 14: 121. 30. Kereiakes DJ, Robinson JG, Cannon CP, et al. Efficacy and safety of the proprotein convertase subtilisin/kexin type 9 inhibitor alirocumab among high cardiovascular risk patients on maximally tolerated statin therapy: The ODYSSEY COMBO I study. American heart journal. 2015; 169: 906-15 e13. 31. Roth EM, Taskinen MR, Ginsberg HN, et al. Monotherapy with the PCSK9 inhibitor alirocumab versus ezetimibe in patients with hypercholesterolemia: results of a 24 week, double- blind, randomized Phase 3 trial. International journal of cardiology. 2014; 176: 55-61. 32. Goldman DP, Shang B, Bhattacharya J, et al. Consequences of health trends and medical innovation for the future elderly. Health affairs (Project Hope). 2005; 24 Suppl 2: W5R5-17. 33. Gaudette E, Goldman DP, Messali A, et al. Do Statins Reduce the Health and Health Care Costs of Obesity? PharmacoEconomics. 2015; 33: 723-34. 34. Agus D, Gaudette É, Messali A, et al. The long-term benefits of increased aspirin use by at-risk Americans aged 50 and older. PLOS ONE, 2016. 35. Goldman DP, Gaudette E, Cheng WH. Competing Risks: Investing in Sickness Rather Than Health. American journal of preventive medicine. 2016; 50: S45-50. 36. Goldman DP, Cutler D, Rowe JW, et al. Substantial health and economic returns from delayed aging may warrant a new focus for medical research. Health affairs (Project Hope). 2013; 32: 1698-705. 37. Vidyanti I B-DR. Projecting long-term impact of modest sodium reduction in Los Angeles County. Proceedings of the 2015 Winter Simulation Conference. Huntington Beach, CA, 2015. 33 38. Hurd M ZY, Girosi F, Goldman DP. The Effects of Tobacco Control Policy on the Social Security Trust Fund, in After tobacco : what would happen if Americans stopped smoking? In: P.S. Bearman KMN, and L. Wright, Editors., ed.: Columbia University Press: New York, 2011. 39. Lakdawalla DN, Goldman DP, Michaud PC, et al. U.S. pharmaceutical policy in a global marketplace. Health affairs (Project Hope). 2009; 28: w138-50. 40. Dolan P, Roberts J. Modelling valuations for Eq-5d health states: an alternative model using differences in valuations. Medical care. 2002; 40: 442-6. 41. Gidding SS, Champagne MA, de Ferranti SD, et al. The Agenda for Familial Hypercholesterolemia: A Scientific Statement From the American Heart Association. Circulation. 2015; 132: 2167-92. 42. Singh S, Bittner V. Familial hypercholesterolemia--epidemiology, diagnosis, and screening. Current atherosclerosis reports. 2015; 17: 482. 43. Lauer MS. PCSK9 Inhibitors: Lots of Work Done, Lots More to Do. Annals of internal medicine. 2016; 164: 624-5. 44. FDA. FDA Briefing Document: Endocrinologic and Metabolic Drugs Advisory Committee (EMDAC). 2015. Available from: http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/Endo crinologicandMetabolicDrugsAdvisoryCommittee/UCM450072.pdf, [Accessed October 10 2016] 45. Matza LS, Cong Z, Chung K, et al. Utilities associated with subcutaneous injections and intravenous infusions for treatment of patients with bone metastases. Patient preference and adherence. 2013; 7: 855-65. 46. Truven Health Analytics. Red Book Online. Available from: http://www.redbook.com/redbook/about. [Accessed September 15, 2016] 47. Grabowski HG VJ. Brand loyalty, entry, and price competition in pharmaceuticals after the 1984 Drug Act. J Law Econ. 1992; 35. 34 48. Tirrell M. Pricing wars heat up over hepatitis C drugs., 2015. Available from: http://www.cnbc.com/id/102396903. [Accessed November 30, 2016]. 49. Conti RM BE. Specialty drugs prices and utilization after loss of U.S. patent exclusivity, 2001 - 2007. Available from: http://www.nber.org/papers/w20016, 2014. [Accessed September 15, 2016] 50. Edejer T T BR, Adam T. Making Choices in Health: WHO Guide to Cost-Effectiveness Analysis. Geneva, Switzerland: World Health Organization; 2003. Available from: http://www.who.int/choice/publications/p_2003_generalised_cea.pdf. [Accessed August 10, 2016]. 51. Kasmeridis C, Apostolakis S, Ehlers L, et al. Cost effectiveness of treatments for stroke prevention in atrial fibrillation: focus on the novel oral anticoagulants. PharmacoEconomics. 2013; 31: 971-80. 52. Najafzadeh M, Andersson K, Shrank WH, et al. Cost-effectiveness of novel regimens for the treatment of hepatitis C virus. Annals of internal medicine. 2015; 162: 407-19. 53. Chhatwal J, Kanwal F, Roberts MS, et al. Cost-effectiveness and budget impact of hepatitis C virus treatment with sofosbuvir and ledipasvir in the United States. Ann Intern Med. 2015; 162: 397-406. 54. Shrank W. LA, Singh S., Brennan T. In the debate about cost and efficacy, PCSK9 inhibitors may be the biggest challenge yet. . Health Affairs Blog, 2015. Available from: http://healthaffairs.org/blog/2015/02/17/in-the-debate-about-cost-and-efficacy-pcsk9-inhibitors- may-be-the-biggest-challenge-yet/. [Accessed November 01, 2016] 55. Sabatine MS, Giugliano RP, Keech AC, et al. Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. The New England journal of medicine. 2017; 376: 1713- 22. 56. Toth PP, Stevens W, Chou JW. Why have published studies of the cost effectiveness of PCSK-9 inhibitors yielded such markedly different results. Journal of medical economics. 2017: 1-6. 35 57. Hirsh BJ, Smilowitz NR, Rosenson RS, et al. Utilization of and Adherence to Guideline- Recommended Lipid-Lowering Therapy After Acute Coronary Syndrome: Opportunities for Improvement. Journal of the American College of Cardiology. 2015; 66: 184-92. 36 SUPPLEMENTAL APPENDIX Scenarios The purpose of this appendix is to provide a more detailed overview of the scenarios presented in the Methods section, with regard to treatment eligibility criteria and the impact of PCSK9 inhibitors on health transitions and costs. 1. Status quo scenario: In the status quo, or baseline, scenario, we did not change the incidence of disease. This scenario can be seen as current treatment strategies managing hypercholesterolemia and as the standard of care in controlling cholesterol—primary and secondary prevention protocols for cardiovascular diseases. Therefore, this scenario includes not only statin treatments but also other interventions, including ezetimide, diet, or exercise as suggested by current clinical guidelines. In the baseline scenario, spending estimates can be interpreted as resources consumed by an individual, given how medicine currently is practiced in the United States. 2. PCSK9 inhibitor scenarios: To evaluate the effects under different eligibility criteria of using PCSK9 inhibitor, we generated one scenario to assign PCSK9 inhibitor use to persons eligible as defined by current FDA-approved indications and the other scenario with wider indication for PCSK9 inhibitor use. In these two scenarios, we modified the status quo scenario by applying the health benefits and additional healthcare spending from PCSK9 inhibitor to the current standard of care. The PCSK9 inhibitor eligibility was assigned based on the FEM population using the HRS data. a) Current Eligibility: In this scenario the populations eligible for PCSK9 inhibitor were defined to approximate the current FDA-approved indications and the first two groups of statin benefit groups (SBGs ), as outlined by ACC/AHA guidelines:(1) Heterozygous FH: a LDL-C level higher than 190 mg/dL (4.91 mmol/L), based on definitions published by the current guidelines.(2, 3) Preexisting CVD: persons diagnosed at any point in their lives with any cardiovascular disease, including congestive heart failure, coronary heart disease, angina, heart attack, stroke, and other heart diseases. 37 b) Extended Eligibility: Because the long-term effectiveness is yet to be confirmed, PCSK9 inhibitor use is currently approved under a more restrictive criteria. Still, results from Phase III clinical trials reported tremendous effectiveness of PCSK9 inhibitor among persons who were not diagnosed with CVD but who were at high risk of developing CVD.(4, 5) In addition to individuals diagnosed with FH and pre-existing CVD, we assumed that people without a history of CVD who currently possess high-risk equivalents are eligible for PCSK9 inhibitors. The CVD high-risk equivalents were defined as persons with diabetes aged 40 to 75 years, or with an estimated 10-year ASCVD risk >7.5%. The 10-year CVD risk scores were calculated for each patient aged 40 to 75 years without a CVD diagnosis. The 10-year ASCVD risk was calculated based on the gender and race-specific equations provided by the Report on the Assessment of Cardiovascular Risk created by the National Heart, Lung, and Blood Institute Full Work Group. These equations have been used to support the cholesterol therapy recommendations in ACC/AHA guidelines. Details can be found in Table A1. This group corresponded with SBG 3 and 4 stated in ACC/AHA guidelines.(1) Given that PCSK9 inhibitors have been proven as a second-line adjunctive therapy for high cholesterol, to qualify for PCSK9i use, populations with pre-existing CVD and CVD high-risk equivalents have to be on current cholesterol-lowering therapy and have failed to reach the LDL- C goal of <70 mg/dL. Based on the current evidence from phase III clinical trials, the efficacy and safety of PCSK9 inhibitors has not been established in populations older than 80 years old; therefore, we only considered persons eligible to receive a PCSK9i under the age of 80. Realistically, not all the PCSK9 inhibitor-eligible individuals would receive the drug as soon as it is commercially launched. In fact, uncertainty surrounding its long-term effectiveness, as well as pricing concerns, have served as barriers to widespread adoption.(6, 7) Therefore, we assumed that PCSK9 inhibitors are initially taken by a small proportion of the eligible population and use will gradually increase through the years. Thus among the PCSK9i-eligible population, the actual assignment of PCSK9 inhibitor use was phased in, with a zero probability starting from year 2014 (which means none of the PCSK9i-eligible individuals uses these new drugs), and linearly increasing to 1 (which means 100 percent of eligible individuals are taking the drugs) through year 2020 in both PCSK9i scenarios. 38 The impact of PCSK9i on health and costs To reflect the health impact of PCSK9i use reported in the literature, we modified health transitions and outcomes of eligible individuals in the PCSK9i scenarios, specifically by reducing the risk of having the first cardiovascular disease and all-cause mortality, as well as applying the PCSK9 inhibitor drug cost and the disutility weight due to the adverse effects of PCSK9 inhibitors. Both health benefits and side effects of PCSK9 inhibitors reported by clinical trials were considered. Table A2 summarizes the most important findings in clinical effectiveness on major CVD events and overall mortality from recent meta-analyses. While relationships between LDL-C reductions and CVD outcomes have been established in patients on statins, they may differ in patients receiving PCSK9 inhibitors, and definitive clinical evidence is not yet available. Therefore, instead of using reductions in LDL-C for PCSK9 inhibitor-eligible individuals, we directly decreased the probabilities of contracting heart disease based on the published risk-ratios from the meta-analyses. However, we generated a scenario in sensitivity analysis to use the effectiveness estimated from the relationships between LDL-C reductions and health outcomes established by the statin studies, instead of using the estimates from PCSK9 inhibitor clinical trials. Also, given that the current meta-analyses of PCSK9 inhibitors didn’t have enough power to calculate the effectiveness of the drugs on risk of CVD and mortality among subpopulations with different characteristics, we assumed that the populations eligible for PCSK9 inhibitors received the same effect size and the effect size does not change across their lifetime covered by the drugs. a. Heart disease: We decrease the incidence of having heart disease by factors with a mean of 0.54, for those eligible for PCSK9 inhibitors and without prior heart disease. The heart disease was defined as any diagnosis with myocardial infarction, angina, chronic heart failure, and any other coronary heart disease. These risk reductions were based on the risk ratios for total CVD events reported by a meta-analysis study.(8) Given that there is no separate evidence showing different risk ratios between the primary prevention and secondary prevention settings, we assumed that risk ratios are the same among all PCSK9i users. Clinical trial and meta-analysis studies report estimates ranging from 0.49 to 0.54 (Table A5). We used the most conservative 39 estimate from meta-analysis studies and long-term clinical trials to estimate the benefits of PCSK9 inhibitors. b. Mortality: Similar to the risk reduction in CVD, we assumed the risk ratios of mortality were the same among all PCSK9 inhibitor users. We decreased the all-cause mortality rate by factors with a mean of 0.45, based on the estimates reported by Navarese EP et al., 2015.(9) However, given PCSK9 inhibitors would impact mortality indirectly from a reduction in heart disease incidence, we adjusted the mortality risk-reduction rate by accounting for the interaction between heart disease and mortality. Therefore, the mortality adjustments are modulated to account for the indirect impact of PCSK9i on mortality through its prevention of non-fatal heart disease incidence, which would impact the probability of mortality in subsequent periods. We first conducted a cohort simulation that enabled PCSK9 inhibitors to impact heart disease, and we compared the ratio of mortality rates in this scenario to the status quo to estimate a ratio indicating how the mortality rate changed compared with the status quo scenario; then, we adjusted the mortality risk-reduction rate by the estimated ratio to eliminate the indirect effect from the incidence reduction in heart disease. Specifically, the Navarese EP et al (2015) meta-analysis study included clinical trials with populations with a mean age of 58.6 at baseline and who were followed for a maximum of 26 months, showing PCSK9 inhibitor use was associated with an all-cause mortality risk ratio point estimate of 0.45. With FEM, we allowed PCSK9 inhibitors to impact incidence of heart disease (by using the risk ratio of 0.54 from Lipinski MJ et al., 2016) but not mortality at ages 59 to 61 in a cohort simulation.(8) This scenario resulted in an all-cause mortality ratio with the status quo of 0.998 over these ages in the current eligibility population and 0.983 in the extended eligibility population. We thus impose a reduction of mortality using factors with a mean of 0.45/0.998 = 0.451 in the PCSK9 inhibitor current eligibility scenario and 0.45/0.983 = 0.46 in the extended eligibility scenario. c. The drug cost of PCSK9 inhibitors: Since medical costs in the MEPS and MCBS include prescription drugs, the direct costs of current lipid-lowering therapy are included in the status quo and PCSK9 inhibitor scenarios. Because PCSK9 inhibitors are currently approved as a second-line adjunct therapy, we added the PCSK9 40 inhibitor drug costs for those assigned to the PCSK9 inhibitor scenarios. Annual drug costs were assumed to be equal to their wholesale acquisition costs; in the case of PCSK9 inhibitors, this was assumed to be the mean of the 2015 annual costs of alirocumab ($14,600) and evolocumab ($14,100).(10) The price for a new branded drug usually drops after the entry of other branded competing drugs. For example, in the case of a new hepatitis C branded drug, Medicaid receives a discount in the range of 23%, and the prices fall by as much as 46%, as documented in the press.(11, 12) Therefore, our analysis took the average of 23% and 46% and assumed that branded PCSK9 inhibitors are under negotiation and discounted by 35% after their first year on the market, due to the entry of other branded competitors; PCSK9 inhibitors are further discounted by an additional 33% due to a loss in patent protection, which results in a 57% total discount in the years after 2028.(11, 13) d. Impacts of adverse events: In the literature, the elevated risk of a neurocognitive adverse event was suspected to be associated the PCSK9 inhibitors. However, as reported in an FDA briefing document, the study authors concluded that “it is challenging to definitely attribute the neurocognitive adverse event to evolocumab/alirocumab versus another drug or pre-existing condition.(14)” Evidence has shown type of administration impacts patient utility, with an injection leading to a lower QALY compared to an oral therapy; therefore, we imposed a disutility of roughly 0.004 per person receiving PCSK9 inhibitors.(15) Sensitivity analysis This article aims to develop and study realistic scenarios in which PCSK9 inhibitors were introduced to the market in 2015 and have been used under appropriate indications since then. However, different criteria might be applied to identify the population eligible for PCSK9 inhibitors because it is clinically challenging to identify the population on maximum tolerated dose of their current cholesterol therapies, and decisions are usually made based on subjective judgment from physicians (i.e. a higher goal LDL-C level might be set before assigning PCSK9 inhibitors). In addition, the long-term effectiveness of PCSK9 inhibitors is unclear, and the available clinical trials were not powered to evaluate the effects on mortality. Therefore, two different types of sensitivity analyses were performed to evaluate the robustness of the health 41 impacts and net values of PCSK9 inhibitors: the structural sensitivity analysis and one-way sensitivity analysis. We conducted five scenario analyses to adjust the structural uncertainty surrounding the assumptions we made, specifically on eligibility criteria for PCSK9 inhibitor users and the clinical effect size for the drugs: a. Increase the LDL-goal to 100 mg/dL: The conventional LDL-C goal is 70 mg/dl. However, in clinical practice, physicians might have more generous LDL-C goals and may recommend for patients to remain on their current lipid-lowering therapy until their LDL-C measurements reach 100 mg/dL. Therefore, in both our PCSK9i current eligibility and extended eligibility scenarios, we designated the LDL-C goal as 100 mg/dL. b. Given that most of the clinical trials of PCSK9 inhibitors didn’t allow a long-term follow-up to provide definitive evidence of the effectiveness on CVD incidence and mortality, we estimated effectiveness based on the relationship between the reduction in LDL-C level and the risk ratios of CVD and mortality, which were documented by the most recent The Cholesterol Treatment Trialists' Collaboration (CTTC) meta-analyses of statin trials. The parameters we used from these studies are listed in Table A7. These studies included trials with a longer follow-up, a larger population, and designed to measure cardiovascular outcomes. The CTTC study in 2010 reported the relative risk of event per 1-mmol/L reduction in LDL- C among those with prior cardiovascular disease was 0.90 [95% CI 0.87 – 0.93] in any death.(16) Based on the meta-analysis of PCSK9 inhibitor trials, Navarese EP (2015), PCSK9 inhibitors showed an average 57% lower reduction in LDL-C compared with the control group, which corresponded to about 1.79 mmol/L absolute difference in LDL-C.(9) Therefore, for those who are eligible for PCSK9 inhibitors, we reduced mortality by a risk-ratio of 0.821 [95% CI 0.767 – 0.875] in the simulations for those with prior CVD. The other CTTC study published in 2012 comprising the lower-risk population showed a relative risk of 0.91 [95% CI 0.88 – 0.93] in overall mortality and 0.72 [95% CI 0.66 – 0.78] in any major vascular event per 1-mmol/L LDL-C reduction.(17) Thus we reduced the overall mortality by 0.839 [95% CI 0.785 – 0.875] and reduced the probability of having heart disease by 0.499 [95% CI 0.391 – 0.606] for those without prior CVD. 42 c. In the base-case we assumed the effect of PCSK9 inhibitors remain unchanged over the lifetime of the patients. However, there has been no clear clinical evidence supporting this assumption, and how effectiveness changes over a lifetime also is unknown. As the maximum follow-up period in the PCSK9 inhibitor trials was about 2 years, in this scenario we assumed PCSK9 inhibitors have the full effect size on mortality and incidence of heart disease for the first 2 years they receive the drugs, and afterward, the effect size decreases to half of the base-case estimate. d. Pessimistic scenario: According to FDA labeling, PCSK9 inhibitors are indicated for adults with maximally tolerated statin therapy for the treatment of adults with heterozygous FH or clinical atherosclerotic CVD who require additional lowering of LDL-C. However, the survey data pose challenges to defining the population of adults with maximally tolerated statin therapy. We assumed that only 10% of patients were statin-intolerant and, therefore, eligible for PCSK9 inhibitors; that is, these patients had either FH or a history of CVD, were taking lipid-lowering therapy, and failed to reach the LDL-C goal of 70 mg/dL. This assumption corresponded to the criteria for PCSK9i target populations proposed by Kazi et al.(18) e. Optimistic scenario: In this scenario we assumed that all potentially eligible individuals commenced PCSK9i inhibitor treatment as soon as the drug hit the market in 2015, without phasing in a linear probability for PCSK9 inhibitor adoption. Results of the sensitivity analysis The results for the structural sensitivity analysis are presented in Tables A8 to A9. Compared to the base-case analysis results shown in Table 2, the PCSK9 scenarios with restrained eligibility to those with LDL > 100 mg/dL (instead of LDL > 70 mg/dl in the base-case analyses) results in lower life expectancy, QALY gains, and lower medical spending, leading to a similar net value per capita in both the FDA-approved eligibility ($4,400) and extended eligibility scenarios ($12,900), with a significance at a 5% level under the extended eligibility. The scenario using the effectiveness estimated by the relationships between the LDL-C level and the risks of CVD event 43 from the CTTC meta-analyses provided much smaller effect sizes on mortality, and therefore resulted in much lower life expectancy and QALY gains. However, compared with the base- case, this scenario reduced incremental medical spending by more than half given shorter life expectancy but similar risk reduction in heart disease due to PCSK9 inhibitors, which resulted in a higher positive net value, with $11,400 under FDA-approved eligibility and more favorable ICERs in both scenarios. When assuming the effectiveness of PCSK9 inhibitors are reduced by half of the base-case estimates after the first 2 years receiving the drugs, the net value decreases in both scenarios. When assuming only 10% of those eligible for PCSK9 inhibitors are using the maximally tolerated dose of statins and use PCSK9 inhibitors, there is only a 0.1 - 0.2-year increase in life expectancy, and a mean of negative QALY gain under the FDA-approved eligibility, which means that the disutility due to the adverse effects of PCSK9 inhibitors slightly outweighs the utility gained by the benefits of the drug; thus, as indicated in Table A9, the estimates of net value become negative, although non-significant, in both PCSK9 scenarios. When assuming all eligible patients start taking PCSK9 inhibitors once they are available on the market— without gradual take-up—the most optimistic scenario would lead to only 0.1-year gain in life expectancy and lower net values per capita at age 51 due to higher spending from PCSK9 inhibitors. The results of mean ICER estimates in our scenario analyses range from $85,300 to $140,100 per discounted QALY gained under FDA-approved eligibility and from $111,700 to $150,600 per discounted QALY gained under extended eligibility. At the population level, the cumulative net values of PCSK9 inhibitors by year 2036 are projected to be positive in most of the sensitivity scenarios (Table A10), except for the scenario limiting the use of PCSK9 inhibitors to 10% of the eligible population, which corresponds to the results from the cohort simulation. One-way sensitivity analysis was performed to determine the robustness of the model to changes in key parameters (Table A11). Given that the uncertainty of clinical effectiveness was already adjusted by the probabilistic and structural sensitivity analysis, the key parameters we tested here were specifically drug costs, the disutility weight of PCSK9 inhibitors, and the discount rate. We varied one input parameter at a time (plus or minus 25% of the base-case for costs and disutility weights; 1% or 5% for the discount rate) while holding others constant at their base-case estimates. Based on the results from both cohort and population simulations, the results of one- 44 way sensitivity analysis showed that the base-case results are most sensitive to the drug costs of PCSK9 inhibitors, compared with the discount rate and the disutility weight. The mean net values would change from $3,100 to $8,500 in the FDA-approved scenario and from $8,700 to $19,300 in the extended eligibility scenario, when the drug costs change from -25% to + 25%. The magnitudes of change in ICERs are relatively small, over which ranges from $123,000 to $141,300 per QALY gains under FDA-approved eligibility and from $116,100 to $135,500 under extended eligibility when compared with the standard of care (status quo) scenario. Given that the price of PCSK9 inhibitors has been cited as an important factor in the cost-effectiveness of the drug and the final price is still under negotiation, we added a worst-case scenario that assumed PCSK9 inhibitors remain at their list price under the patent protection period and then receive a 33% discount after the drug goes to generic. The result showed that if patients don’t have access to the drug discount, the net value would be $300 and $2,600 under the FDA- approved eligibility and the extended eligibility, respectively, with an ICER of $154,700 and $146,700 per QALY gain. 45 Supplemental tables and figures Table A1. Equation Parameters of the Pooled Cohort Equations for Estimation of 10-Year Risk for Hard ASCVD* and Specific Examples for Each Race and Sex Group (Goff DC et al. 2013 Report on the Assessment of Cardiovascular Risk) Women White African-American Coefficient † Coefficient † Ln Age (y) –29.799 17.114 Ln Age, Squared 4.884 N/A Ln Total Cholesterol (mg/dL) 13.540 0.940 Ln Age×Ln Total Cholesterol –3.114 N/A Ln HDL–C (mg/dL) –13.578 –18.920 Ln Age×Ln HDL–C 3.149 4.475 Log Treated Systolic BP (mm Hg) 2.019 29.291 Log Age×Log Treated Systolic BP N/A –6.432 Log Untreated Systolic BP (mm Hg) 1.957 27.820 Log Age×Log Untreated Systolic BP N/A –6.087 Current Smoker (1=Yes, 0=No) 7.574 0.691 Log Age×Current Smoker –1.665 N/A Diabetes (1=Yes, 0=No) 0.661 0.874 Individual Sum of Coefficient× Value Calculated Calculated Mean (Coefficient× Value) –29.18 86.61 Baseline Survival 0.9665 0.9533 46 *Defined as first occurrence of nonfatal MI or CHD death, or fatal or nonfatal stroke. †Coefficient: For age, lipids, and BP, defined as the natural log of the value, have to be multiplied by the parameter coefficient estimate. When an age interaction is present with lipids or BP, the natural log of age is multiplied by the natural log of the lipid or BP, and the result is multiplied by the coefficient estimate. “N/A” indicates that that specific covariate was not included in the model for that sex-race group; “–” indicates that this value was not included in the example (e.g., this example used untreated systolic BP, not treated systolic BP). Men White African-American Coefficient † Coefficient † Log Age (y) 12.344 2.469 Log Total Cholesterol (mg/dL) 11.853 0.302 Log Age×Log Total Cholesterol –2.664 N/A Log HDL–C (mg/dL) –7.990 –0.307 Log Age×Log HDL–C 1.769 N/A Log Treated Systolic BP (mm Hg) 1.797 1.916 Log Untreated Systolic BP (mm Hg) 1.764 1.809 Current Smoker (1=Yes, 0=No) 7.837 0.549 Log Age×Current Smoker –1.795 N/A Diabetes (1=Yes, 0=No) 0.658 0.645 Individual Sum of Coefficient× Value Calculated Calculated Mean (Coefficient× Value) 61.18 19.54 Baseline Survival 0.9144 0.8954 47 Calculation of the 10-year risk estimate for hard ASCVD can best be described as a series of steps. The natural log of age, total cholesterol, HDL– C, and systolic BP are first calculated with systolic BP being either a treated or untreated value. Any appropriate interaction terms are then calculated. These values are then multiplied by the coefficients from the equation for the specific race-sex group of the individual. The sum of the “Coefficient×Value” is then calculated for the individual and used to calculate the 10-year risk by using the equation below. The estimated 10-year risk of a first hard ASCVD event is formally calculated as 1 minus the survival rate at 10 years (“Baseline Survival” in Table A1), raised to the power of the exponent of the “Coefficient×Value” sum (IndX’β in the below equation) minus the race and sex specific overall mean “Coefficient×Value” sum (Mean X’β in the below equation); or, in equation form: 𝑃𝑃 𝑃𝑃 𝑃𝑃 𝑃𝑃 𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃 𝑃𝑃 𝑜𝑜 𝑃𝑃 𝑜𝑜𝑃𝑃𝑃𝑃 𝑓𝑓 𝑃𝑃 ℎ𝑃𝑃𝑃𝑃 𝑎𝑎 𝐴𝐴 𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆 𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒 𝑃𝑃 𝑤𝑤 𝑃𝑃 𝑃𝑃 ℎ 𝑃𝑃𝑒𝑒 10 𝑃𝑃 𝑒𝑒𝑃𝑃𝑃𝑃 𝑓𝑓 = 1 − 𝑆𝑆 1 0 𝑒𝑒 � 𝐼𝐼 𝐼𝐼 𝐼𝐼 𝑋𝑋 ′ 𝛽𝛽−𝑀𝑀 𝑀𝑀𝑀𝑀 𝐼𝐼 𝑋𝑋 ′ 𝛽𝛽� 48 4.1 8.5 27.5 13.4 28.5 14.6 27.8 13.9 0 5 10 15 20 25 30 35 40 Prevalence of Populations Assigned to PCSK9 inhibots (%) 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 Year PCSK9 Inhibitor Current Eligibility PCSK9 Inhibitor Extended Eligibility 9.5 4.6 32.5 15.9 38.6 19.8 41.6 20.8 0 10 20 30 40 50 60 Millions of Individuals Assigned to PCSK9 Inhibitors 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 Year PCSK9 Inhibitor Current Eligibility PCSK9 Inhibitor Extended Eligibility FIGURE A1. The Projected Proportions and Populations in Millions Assigned to Use PCSK9 Inhibitors in U.S. Population over Age 50, from Year 2016 to 2056* Figure A1 shows the projected prevalence of the population actually assigned to use PCSK9 inhibitors, and Figure 2B shows the projected total of millions of individuals who actually use PCSK9 inhibitors. *Individuals in statin-benefit groups (recommended by ACC/AHA guidelines) who failed to achieve a goal LDL-C level (≥ 70 mg/dL) and who regularly take lipid-lowering therapy are potentially eligible for PCSK9 inhibitor use until age 80. Among the PCSK9 inhibitor-eligible population, the actual assignment of PCSK9 inhibitor use was phased in with a probability of zero starting from year 2014 and linearly increased to 1 through year 2020. PCSK9 Inhibitor Current Eligibility refers to the current FDA- approved eligibility criteria for PCSK9 inhibitors, while Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. A. Prevalence (%) B. Millions 49 FIGURE A2. The Net Benefits of PCSK9 Inhibitor Use in the Overall U.S. Population over Age 50, 2016 to 2056 * The cumulative net benefit by 2036 in the U.S. population aged 51 and older is $0.54 trillion in the PCSK9 inhibitor scenario under current eligibility, and $0.90 trillion in PCSK9 Inhibitor Scenario II (in 2015 dollars). *All individuals over age 50 in each year. Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. The value of QALY gained was calculated by multiplying QALY-gains with a willingness-to-pay of $150,000. All amounts are computed with a 3% discount rate. PCSK9 inhibitor Current Eligibility Scenario -100 0 100 200 300 400 500 600 2015 billion 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 Year Value of QALY Gained Total Incremental Medical Costs Net Present Value PCSK9 inhibitor Extended Eligibility Scenario -100 0 100 200 300 400 500 600 2015 billion 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 Year Value of QALY Gained Total Incremental Medical Costs Net Present Value 50 TABLE A2. The Impact of PCSK9 Inhibitor Use on Health Outcome at Age 51 (With 95% Confidence Intervals of the Estimates) 1 The 95% confidence intervals of the mean for the status quo were calculated based on the fifth lowest and highest results from the sorted estimates of the 200 simulations, which reflects the uncertainty of the Monte-Carlo repetitions. 2 The 95% confidence intervals of the mean for the PCSK9 inhibitor scenarios account for the clinical uncertainty of the effectiveness of PCSK9 inhibitors by drawing random values from the confidence intervals for relative-risk estimates of the effectiveness reported in the literature for each repetition of simulations. Scenario Status quo PCSK9i Scenario I - Current Eligibility PCSK9i Scenario II - Extended Eligibility Mean 95% CI 1 Mean 95% CI 2 Mean 95% CI 2 Health outcomes, at age 51 Life expectancy, years 30.9 [30.3, 31.6] 32.0 [31.4, 32.7] 32.8 [32.2, 33.4] Disability-free life expectancy, years 22.9 [22.3, 23.4] 23.2 [22.7, 23.8] 23.6 [23.2, 24.2] Quality-adjusted life years (QALYs) 25.5 [25.9, 26.0] 26.1 [25.6, 26.6] 26.7 [26.2, 27.1] Cumulative disease incidence at age 79 (per thousand) Heart Disease 419.3 [386.6, 446.4] 411.6 [380.9, 438.8] 379.0 [334.7, 393.7] Stroke 207.2 [183.4, 230.4] 219.0 [192.5, 241.8] 222.3 [196.9, 245.1] Diabetes 473.8 [440.5, 501.7] 479.5 [445.2, 507.8] 483.0 [450.3, 511.1] Hypertension 732.4 [696.0, 767.2] 737.6 [703.1, 772.0] 742.2 [706.8, 778.7] Cancer 285.5 [258.3, 310.1] 294.8 [269.1, 319.0] 299.9 [277.2, 323.7] 51 TABLE A3. The Net Benefits of PCSK9 Inhibitor Use at Age 51 (Mean Estimates with 95% Confidence Intervals) 1 The 95% confidence intervals of the mean for the status quo were calculated based on the fifth lowest and highest results from the sorted estimates of the 200 simulations, which reflects the uncertainty of the Monte-Carlo repetitions. 2 The 95% confidence intervals of the mean for the PCSK9 inhibitor scenarios account for the clinical uncertainty of the effectiveness of PCSK9 inhibitors by drawing random values from the confidence intervals for relative-risk estimates of the effectiveness reported in the literature for each repetition of simulations. Status quo PCSK9 Inhibitor Scenario I - Current Eligibility PCSK9 Inhibitor Scenario II- Extended Eligibility Mean 95% CI 1 Mean 95% CI 2 Mean 95% CI 2 Value of expected quality-adjusted life- years gained (Discounted) 2465.6 [2431.8, 2504.0] 2511.5 [2478.4, 2547.1] 2549.5 [2520.0, 2583.8] Per Capita Life-time healthcare and medication costs (Discounted) Healthcare costs 538.7 [520.8, 559.0] 567.9 [549.1, 585.6] 587.4 [569.7, 606.3] PCSK9i medication costs 0.0 10.8 [10.1, 11.4] 21.3 [20.4, 22.3] Total 538.7 [520.8, 559.0] 578.7 [558.2, 595.9] 608.5 [587.5, 627.6] 52 TABLE A4. The Pairwise Comparisons of the Economic Outcomes between the Three Scenarios PCSK9i Current Eligibility vs. Status quo PCSK9 Extended Eligibility vs. Current Eligibility Mean Difference 1 95% CI Mean Difference 2 95% CI Value of expected quality-adjusted life-years gained (Discounted) 2511.5 45.9 [33.5, 58.9] 2549.5 38.0 [25.6, 50.8] Per Capita Life-time healthcare and medication costs (Discounted) Healthcare costs 567.9 29.3 [21.4, 39.6] 587.4 19.3 [13.4, 26.1] PCSK9i medication costs† 10.8 10.8 [10.1, 11.4] 21.3 10.5 [9.6, 11.2] Total 578.7 40.1 [32.3, 50.7] 608.5 29.8 [23.8, 36.8] Incremental Cost-Effectiveness Ratio (ICER) 132.2 [110.8, 153.9] 120.4 [96.8, 148.0] Net value per capita (in thousands) ** 5.8 [-1.1, 13.7] 8.2 [0.2, 16.6] Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. 1 The result differences compared the estimates of PCSK9 inhibitor current eligibility scenario with the status quo scenario. 2 The result differences compared the estimates of PCSK9 inhibitor current eligibility scenario with the extended eligibility scenario. **Calculated by multiplying QALY-gains with a willingness-to-pay of $150,000. †PCSK9 inhibitor medication costs were calculated based on a discount rate of 35% from 2016 to 2027, with an additional 33% discount since year 2028. All amounts are in present value at age 51, computed with a 3% discount rate. 95% confidence interval of the mean differences are presented in brackets, indicating the uncertainty of the effectiveness of PCSK9 inhibitors. The 95% confidence intervals for the mean values in each scenario can be found in the Supplemental Appendix. All the costs were adjusted to year 2015 US thousand dollars. 53 TABLE A5. Health Impact of PCSK9 Inhibitors Reported by Large Meta-Analyses and Long-term Clinical Trials Study Study Design Effect of PCSK9 Inhibitor Therapy vs. Standard of Care (Placebo, Statins, or Ezetimibe) Navarese EP et al., 2015 Meta-analysis All-cause mortality: OR (95% CI) = 0.45 ( 0.23 - 0.86) Myocardial infarction events: OR (95% CI) = 0.49 (0.26 - 0.93) Lipinski MJ et al., 2016 Meta-analysis All-cause mortality: OR (95% CI) = 0.43 ( 0.22 - 0.82) Major cardiovascular events: OR (95% CI) = 0.54 (0.38 - 0.77) Koren MJ et al., 2014 Long-term RCT All-cause mortality: OR (95% CI) = 0.33 (0.09 - 1.18) Major cardiovascular events: OR (95% CI) = 0.47 (0.28 - 0.78) Robinson JG et al., 2015 Long-term RCT All-cause mortality: OR (95% CI) = 0.40 (0.16 - 1.02) Major cardiovascular events: OR (95% CI) = 0.52 (0.31 - 0.90) Only findings relevant to the outcomes of interest are shown. Both point estimates and 95% confidence interval (CI) were included in the simulations. Findings used in the simulations are shown in bold font. OR: odds ratio. RCT: randomized controlled trial. 54 TABLE A6. Prevalence of Potential PCSK9 Inhibitor Eligible Populations, Observed Results in National Health and Nutrition Examination Survey (NHANES) 2011 - 2012 and Projected Results in the Future Elderly Model (FEM) for Population over Age 50 *Source: Observed results were the percentage points of the population aged 51+ in NHANES 2011-2012 respondents (authors’ calculations), and the projected results were the projected prevalence estimates from FEM in year 2012 and 2016. *Individuals in statin-benefit groups (recommended by ACC/AHA guidelines) who failed to achieve a goal LDL-C level (≥ 70 mg/dL) and who regularly take lipid-lowering therapy are potentially eligible for PCSK9 inhibitor use until age 80. Scenario I refers to the scenario with a goal LDL-C level of <100 mg/dl, and Scenario II refers to the scenario with a goal LDL-C level of <100 mg/dl. CVD: Cardiovascular disease, defined as any diagnosis of congestive heart failure, coronary heart disease, angina, heart attack, and any other heart diseases. Population over age 50 NHANES FEM 2011-2012 2016 Scenarios Goal LDL-C level Prevalence (%) Prevalence (%) Current Eligibility 70 mg/dL 15.4 12.5 100 mg/dL 9.3 8.8 Extended Eligibility 70 mg/dL 32.6 25.4 100 mg/dL 20.7 17.9 55 TABLE A7. Input Parameters Used in the Structural Sensitivity Analysis: The Scenario (2) Using the Effectiveness Estimated from the Statin Meta-analyses Parameters Base-case Range for sensitivity Analysis Reference Type of sensitivity analysis Effect size Relative risk of events for PCSK9 inhibitors versus standard of care based on the reduction of LDL-C levels Probabilistic sensitivity analysis performed by drawing random values from the confidence intervals for relative- risk estimates of the base-case estimate for each repetition of simulations. Reduction in LDL-C, for PCSK9 inhibitors relative to placebo 1 1.79 mmol/L. Lipinski MJ et al., 2016 Relative risk of event per 1-mmol/L reduction in LDL-C For those with prior cardiovascular events 2 Overall mortality 0.9 [0.87−0.93] Cholesterol Treatment Trialists’ Collaborators (CTTC), 2010 Any vascular events 0.79 [0.76–0.82] For those without prior cardiovascular events but in higher cardiovascular risk group 3 Overall mortality 0.91 [0.88 - 0.93] CTTC, 2012 Any vascular events 0.72 [0.66 - 0.78] 56 1 Consistent with the LDL hypothesis, the base case assumed a constant relative reduction in the risk of overall mortality and major vascular event for each millimole per liter (mmol/L) of reduction in LDL-C, independent of how long the patients were treated by the agent. This was estimated from the most recent meta-analysis from the Cholesterol Treatment Trialists in statins. There were too few strokes in the short-term PCSK9 trials, so we didn’t consider the benefit in stroke with PCSK9 inhibitors. 2 The estimates were obtained from the relative risks in the sub-group with prior coronary heart or vascular disease in the most recent meta- analysis from the Cholesterol Treatment Trialists in statins. 3 The high cardiovascular risk population was defined as those who without prior cardiovascular disease and with a five-year risk of major vascular event ≥10%, estimated from the 174,149 participants in 27 trials of statin therapy. This group corresponds to individuals without history of vascular disease but with multiple risk factors that confer a 10-year risk of major coronary events > 10%, which is considered moderately high or high risk and were recommended to consider drug therapy for cholesterol modification by The Adult Treatment Panel III (ATP III) of the National Cholesterol Education program in the US. 57 TABLE A8. Structural Sensitivity Analysis: The Impact of PCSK Inhibitor Use on Health Outcome at Age 51 in Various Scenarios under PCSK9 Inhibitor Current Eligibility and Extended Eligibility The first scenario includes patients who failed to achieve a higher LDL-C goal, which is above 100 mg/dL. The second scenario used the effect size estimated from the relative risk of event per 1-mmol/L reduction in LDL-C based on the CTTC studies in statin users. The third scenario assumes the effects of PCSK9 inhibitors on the mortality and incidence of cardiovascular disease maintains for the first 2 years since they receive the drug and the effect sizes decrease to half afterwards. The fourth scenario refers to the most pessimistic scenario where only 10% of the population eligible to PCSK9 inhibitors was statin-intolerant, and used PCSK9 inhibitors. The last scenario refers to the most optimistic scenario where the eligible population starts using PCSK9 inhibitors upon commercial availability. Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. *The result differences compared with the status quo scenario. **The results were calculated from age 51 Structural Sensitivity Analysis PCSK9 Inhibitor Current Eligibility PCSK9 Inhibitor Extended Eligibility Life Expectancy ** , years Quality-adjusted life years ** Life Expectancy ** , years Quality-adjusted life years ** (QALYs) (QALYs) Difference * 95% CI Difference * 95% CI Difference * 95% CI Difference * 95% CI Higher LDL-C goal - 100 mg/dL 0.7 [0.5, 0.9] 0.3 [0.1, 0.5] 1.2 [0.9, 1.4] 0.7 [0.5, 0.9] Scenario with CTTC estimates 0.5 [0.4, 0.7] 0.3 [0.1, 0.5] 0.8 [0.6, 1.0] 0.6 [0.4, 0.7] Scenario assuming the effect size half-down after 2-years 0.8 [0.6, 1.1] 0.4 [0.2, 0.6] 1.4 [1.1, 1.8] 0.8 [0.5, 1.1] Pessimistic scenario 0.1 [0.0, 0.2] -0.05 [-0.2, 0.1] 0.2 [0.1, 0.3] 0.05 [-0.1, 0.1] Optimistic scenario 1.2 [0.9, 1.4] 0.6 [0.4, 0.8] 1.9 [1.5, 2.3] 1.2 [0.9, 1.5] 58 until death. Quality-adjusted life-years adjust length of life for quality based on a person’s chronic conditions and functional status. 95% confidence intervals are presented in brackets, indicating the uncertainty of the effectiveness of PCSK9 inhibitors. 59 TABLE A9. Structural Sensitivity Analysis: The Net Benefits of PCSK9 Inhibitor Use at Age 51 ($2015 thousands) under PCSK9 Inhibitor Current Eligibility and Extended Eligibility The first scenario includes patients who failed to achieve a higher LDL-C goal, which is above 100 mg/dL. The second scenario used the effect size estimated from the relative risk of event per 1-mmol/L reduction in LDL-C based on the CTTC studies in statin users. The third scenario Structural Sensitivity Analysis PCSK9 Inhibitor Current Eligibility PCSK9 Inhibitor Extended Eligibility Total Healthcare and Medication Costs Net value per capita ICER Total Healthcare and Medication Costs Net value per capita ICER (in thousands) (in thousands) (in thousands) (in thousands) (in thousands) (in thousands) Diff * 95% CI Diff * 95% CI Diff * 95% CI Diff * 95% CI Diff * 95% CI Diff * 95% CI Higher LDL- C goal - 100 mg/dL 24.1 [17.6, 32.2] 4.4 [5.1, 12.5] 138.9 [107.3, 196.5] 45.5 [36.6, 53.6] 12.9 [1.8, 25.5] 131.1 [107.7, 158.3] Scenario with CTTC estimates 13.9 [7.3, 21.0] 11.4 [1.5, 20.6] 85.3 [42.8, 132.9] 30.2 [22.6, 37.1] 10.9 [1.6, 20.7] 111.7 [83.0, 140.0] Scenario assuming the effect size half-down after 2-years 33.2 [25.6, 43.1] 3.8 [-2.6, 10.9] 138.6 [113.2, 164.1] 60.3 [49.5, 74.4] 6.7 [-2.4, 16.5] 136.4 [118.7, 156.9] Pessimistic scenarios 4.3 [2.2, 6.9] -3.6 [-10.9, 0.2] 140.1 [-1403.6, 1104.5] a 6.6 [3.8, 9.5] -0.9 [-6.2, 3.2] 150.6 [-291.1, 418.8] b Optimistic scenarios 42.1 [35.3, 54.2] 4.9 [-1.9, 12.4] 135.8 [114.9, 157.7] 73.6 [62.6, 86.5] 12.4 [2.9, 24.5] 129.3 [115.3, 143.4] 60 assumes the effects of PCSK9 inhibitors on the mortality and incidence of cardiovascular disease maintains for the first 2 years since the drug they receive the drug and the effect sizes decrease to half afterwards. The fourth scenario refers to the most pessimistic scenario where only 10% of the population eligible to PCSK9 inhibitors was statin-intolerant, and used PCSK9 inhibitors. The last scenario refers to the most optimistic scenario where the eligible population starts using PCSK9 inhibitors upon commercial availability. Current Eligibility refers to the current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. a The mean estimate of the negative ICER indicating the status quo is the dominant option versus the PCSK9 inhibitor current eligibility scenario. b The lower bound of the ICER indicating the status quo is the dominant option versus the PCSK9 inhibitor extended eligibility scenario. All amounts are in present value at age 51, computed with a 3% discount rate. 95% confidence intervals of the mean differences are presented in brackets, indicating the uncertainty of the effectiveness of PCSK9 inhibitors. All the costs were adjusted to year 2015 US thousand dollars. 61 TABLE A10. Structural Sensitivity Analysis: The Cumulative Net Benefits of PCSK9 Inhibitor Use in the Overall U.S. Population over Age 50, by 2036* Current Eligibility Extended Eligibility Structural sensitivity analysis Cumulative Net present value (in 2015 U.S. trillion) Higher LDL-C goal - 100 mg/dL 0.31 0.52 Scenario with CTTC estimates 0.85 0.59 Scenario assuming the effect size half-down after 2 years 0.18 0.38 Pessimistic scenarios -0.042 -0.04 Optimistic scenarios 0.32 0.71 The first scenario includes patients who failed to achieve a higher LDL-C goal, which is above 100 mg/dL. The second scenario used the effect size estimated from the relative risk of event per 1-mmol/L reduction in LDL-C based on the CTTC studies in statin users. The third scenario assumes the effects of PCSK9 inhibitors on the mortality and incidence of cardiovascular disease maintains for the first 2 years since the drug they receive the drug and the effect sizes decrease to half afterwards. The fourth scenario refers to the most pessimistic scenario where only 10% of the population eligible to PCSK9 inhibitors was statin-intolerant, and used PCSK9 inhibitors. The last scenario refers to the most optimistic scenario where the eligible population starts using PCSK9 inhibitors upon commercial availability. Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. *All individuals over age 50 in each year. Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. The net value was calculated as the total value of QALY gained minus the total incremental cost, which the value of value of QALY gained was calculated by multiplying QALY-gains with a willingness-to-pay of $150,000. All amounts are computed with a 3% discount rate. All the costs were adjusted to year 2015 US thousand dollars. 62 TABLE A11. One-way Sensitivity Analysis: The Net Benefits of PCSK9 Inhibitor Use at Age 51 ($2015 thousands) under PCSK9 Inhibitor Current Eligibility and Extended Eligibility The one-way sensitivity analyses for drug costs and utility parameters were performed by varying one input at a time (plus or minus 25% for cost and disutility weight, and 1 – 5% for discount rate) while holding others constant at their base-case estimates. We added a worst-case scenario that assumes the patients don’t have access to the discounted drug price under the patent protection period and the drug price drop by 33% after going to generic. The Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. *The result differences compared with the status quo scenario. ** Calculated by multiplying QALY-gains with a willingness-to-pay of $150,000. All amounts One-way Sensitivity Analysis PCSK9 Inhibitor Current Eligibility PCSK9 Inhibitor Extended Eligibility Net value per capita ** ICER Net value per capita ** ICER (in thousands) (in thousands) (in thousands) (in thousands) Difference * 95% CI Difference * 95% CI Difference * 95% CI Difference * 95% CI PCSK9 inhibitor drug costs - 25% 8.5 [1.7, 16.4] 123 [102.3, 143.6] 19.3 [9.6, 30.6] 116.1 [103.1, 129.7] PCSK9 inhibitor drug costs +25% 3.1 [-3.8, 10.9] 141.3 [119.1, 166.7] 8.7 [0.1, 19.8] 135.5 [119.3, 152.3] No discount for drug costs under patent protection period 0.3 [-6.9, 7.8] 154.8 [124.6, 182.1] 2.6 [-7.1, 13.6] 146.7 [128.6, 165.4] Discount rate 1% 6.8 [-8.9, 20.8] 138.4 [112.6, 168.1] 22.2 [5.6, 41.0] 126.6 [111.7, 142.9] Discount rate 5% 4.1 [0.1, 9.0] 130.1 [110.7, 149.1] 8.0 [1.6, 15.3] 126.2 [113.6, 144.1] PCSK9 inhibitor disutility - 25% 6.2 [-0.8, 14.0] 131 [110.1, 152.3] 14.7 [4.9, 26.0] 128.2 [110.7, 138.6] PCSK9 inhibitor disutility +25% 5.6 [-1.4, 13.0] 133.0 [111.5, 155.8] 13.4 [3.5, 24.7] 124.9 [112.5, 141.1] 63 are in present value at age 51, computed with a 3% discount rate. 95% confidence intervals of the mean differences are presented in brackets, indicating the uncertainty of the effectiveness of PCSK9 inhibitors. All the costs were adjusted to year 2015 US thousand dollars. 64 TABLE A12. One-way Sensitivity Analysis: The Cumulative Net Benefits of PCSK9 Inhibitor Use in the Overall U.S. Population over Age 50, by 2036* The one-way sensitivity analyses for drug costs and utility parameters were performed by varying one input at a time (plus or minus 25% for cost and disutility weight, and 1 – 5% for discount rate) while holding others constant at their base-case estimates. Current Eligibility refers to current FDA-approved eligibility criteria for PCSK9 inhibitors. *All individuals over age 50 in each year. Current Eligibility refers to current FDA- approved eligibility criteria for PCSK9 inhibitors. Extended Eligibility refers to using PCSK9 inhibitors as primary prevention therapy for those without clinical CVD but who possess CVD high-risk equivalents. The net value was calculated as the total value of QALY gained minus the total incremental cost, which the value of value of QALY gained was calculated by multiplying QALY-gains with a willingness-to-pay of $150,000. All amounts are computed with a 3% discount rate. All the costs were adjusted to year 2015 US thousand dollars Current Eligibility Extended Eligibility One-way sensitivity analysis Net present value (in $2015 trillion) PCSK9 inhibitor drug costs -25% 0.91 1.69 PCSK9 inhibitor drug costs +25% 0.15 0.20 Discount rate 1% 0.85 1.49 Discount rate 5% 0.33 0.50 PCSK9 inhibitor disutility -25% 0.68 1.11 PCSK9 inhibitor disutility +25% 0.41 0.70 65 References 1. Stone NJ, Robinson JG, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Journal of the American College of Cardiology. 2014; 63: 2889-934. 2. Gidding SS, Champagne MA, de Ferranti SD, et al. The Agenda for Familial Hypercholesterolemia: A Scientific Statement From the American Heart Association. Circulation. 2015; 132: 2167-92. 3. Singh S, Bittner V. Familial hypercholesterolemia--epidemiology, diagnosis, and screening. Current atherosclerosis reports. 2015; 17: 482. 4. Colhoun HM, Robinson JG, Farnier M, et al. Efficacy and safety of alirocumab, a fully human PCSK9 monoclonal antibody, in high cardiovascular risk patients with poorly controlled hypercholesterolemia on maximally tolerated doses of statins: rationale and design of the ODYSSEY COMBO I and II trials. BMC cardiovascular disorders. 2014; 14: 121. 5. Roth EM, Taskinen MR, Ginsberg HN, et al. Monotherapy with the PCSK9 inhibitor alirocumab versus ezetimibe in patients with hypercholesterolemia: results of a 24 week, double- blind, randomized Phase 3 trial. International journal of cardiology. 2014; 176: 55-61. 6. Mehr SR. Will the PCSK9 Inhibitors Be Employers' "Line in the Sand"? American health & drug benefits. 2016; 9: 171-4. 7. Lauer MS. PCSK9 Inhibitors: Lots of Work Done, Lots More to Do. Annals of internal medicine. 2016; 164: 624-5. 8. Lipinski MJ, Benedetto U, Escarcega RO, et al. The impact of proprotein convertase subtilisin-kexin type 9 serine protease inhibitors on lipid levels and outcomes in patients with primary hypercholesterolaemia: a network meta-analysis. European heart journal. 2016; 37: 536- 45. 9. Navarese EP, Kolodziejczak M, Schulze V, et al. Effects of Proprotein Convertase Subtilisin/Kexin Type 9 Antibodies in Adults With Hypercholesterolemia: A Systematic Review and Meta-analysis. Annals of internal medicine. 2015; 163: 40-51. 66 10. Truven Health Analytics. Red Book Online. Available from: http://www.redbook.com/redbook/about. [Accessed September 15, 2016]. 11. Grabowski DC, Lakdawalla DN, Goldman DP, et al. The large social value resulting from use of statins warrants steps to improve adherence and broaden treatment. Health affairs (Project Hope). 2012; 31: 2276-85. 12. Tirrell M. Pricing wars heat up over hepatitis C drugs., 2015. Available from: http://www.cnbc.com/id/102396903. [Accessed November 30, 2016]. 13. Conti RM BE. Specialty drugs prices and utilization after loss of U.S. patent exclusivity, 2001 - 2007 2014. Available from: http://www.nber.org/papers/w20016, 2014. [Accessed September 15, 2016] 14. FDA. FDA Briefing Document: Endocrinologic and Metabolic Drugs Advisory Committee (EMDAC). 2015. Available from: http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/Endo crinologicandMetabolicDrugsAdvisoryCommittee/UCM450072.pdf, [Accessed October 10 2016] 15. Matza LS, Cong Z, Chung K, et al. Utilities associated with subcutaneous injections and intravenous infusions for treatment of patients with bone metastases. Patient preference and adherence. 2013; 7: 855-65. 16. Cholesterol Treatment Trialists C, Baigent C, Blackwell L, et al. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet (London, England). 2010; 376: 1670-81. 17. Cholesterol Treatment Trialists C, Mihaylova B, Emberson J, et al. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta- analysis of individual data from 27 randomised trials. Lancet (London, England). 2012; 380: 581-90. 18. Kazi DS, Moran AE, Coxson PG, et al. Cost-effectiveness of PCSK9 Inhibitor Therapy in Patients With Heterozygous Familial Hypercholesterolemia or Atherosclerotic Cardiovascular Disease. Jama. 2016; 316: 743-53. 67 CHAPTER 2: THREE-PART PRICING TO FACILITATE EARLY ACCESS TO NOVEL DRUGS OF UNCERTAIN EFFICACY Full Citation: Chassang S. Cheng W, Hlavka J, Pani L, Snowberg E, Van Nuys K, Goldman DP. Three-Part Pricing to Facilitate Early Access to Novel Drugs of Uncertain Efficacy. ABSTRACT Paying for value has risen to the top of the health policy agenda as society seeks ways to improve quality without substantial additional spending. Medicare in particular has made considerable progress in tying reimbursement to outcomes to pay providers. 1 However, there has not been similar progress in pharmaceuticals. For decades, the United States has relied on a price- per-dose model, limiting access to innovative but expensive therapies. 2 Tying reimbursement to outcomes offers one intriguing solution. 3 However, there are some fundamental challenges, especially for newly launched drugs where long-term efficacy is unknown, and where the benefits might accrue long after a patient has changed health plans. In this paper, we propose a novel pricing strategy – three-part pricing, or TPP – to overcome these hurdles. We demonstrate how the pricing might work for an important class of newer cardiovascular agents, PCSK9 inhibitors, where uncertainty about long-term efficacy and concern about prices has hindered access. BACKGROUND AND INTRODUCTION Proprotein convertase subtilisin/kexin type 9 inhibitors (PCSK9 inhibitors) for high cholesterol were introduced in 2015 with some controversy. The drugs could potentially benefit millions of patients who cannot manage their low-density lipoprotein cholesterol (LDL-C) using statins and other medications alone. The initial labels of evolocumab (Repatha®; Amgen) and alicrocumab (Praluent®; sanofi/Regeneron) were somewhat narrow; covering patients with familial hypercholesterolemia or those who required additional lowering of LDL-C beyond standard of care. 2,4 The use of evolocumab was expanded in 2017 to include adult patients with established cardiovascular disease who were at risk of myocardial infarction, stroke, and coronary revascularization. 3,5 An expansion of alirocumab’s label is expected this year. 6,7 68 Immediately upon launch, payers raised concerns about the overall costs of these drugs – with some projecting annual spending in the range of $50 to $100 billion. 8 The Institute for Clinical and Economic Review (ICER) in 2015 predicted the uptake for PCSK9 inhibitors could reach over 2.5 million patients within 5 years, 9 while several million patients could benefit from the new drugs under current labels. Uptake has been slow, despite positive trial results. This has been directly linked to coverage restrictions by payers, who cite their potentially low cost-effectiveness. 10 As a result, only half of patients initially prescribed PCSK9 inhibitors in their first year of availability received coverage approval from payers, and one-third of approved prescriptions were not filled due to high patient copays. 11 Amgen reported $60 million in U.S. Repatha® revenues in the second quarter of 2017. Assuming a 34% discount to the list price, this amounts to only 25,000 patients treated in that quarter. Much of the controversy surrounds long-term efficacy. 12 The FOURIER clinical trial, which resulted in expansion of the evolocumab label, did not fully resolve uncertainty about its long-term survival benefits, given the short (26-month) median follow-up. Efforts to address slow uptake have been piloted by payers such as Cigna 11 and Harvard Pilgrim 12 who have entered into outcome- based refund agreements with manufacturers; however, there is no evidence that these have led to higher uptake or made the drugs more cost-effective 13 The current state of affairs leaves everyone worse off. Slower adoption of PCSK9 inhibitors results in worse clinical outcomes for patients and limits the rate at which evidence about real-world outcomes can be collected. 11 This, in turn, hinders future negotiations to improve access to this therapy. Even substantial manufacturer discounts may not address the dilemma, especially if the drugs’ long-term effectiveness is better than observed in clinical trials. Cardiovascular risk increases with the length of exposure to high LDL-C 14,15 and cost-effectiveness improves in real- world populations with higher baseline risk. METHODS Three-part pricing model (TPP). We propose a three-part pricing (TPP) schedule to address the significant unmet need, high budget impact, and uncertainty about long-term effectiveness of 69 PCSK9 inhibitors. 16 Compared to the status quo of price-per-dose, where prices are set high at launch until patent expiration, the pricing schedule under TPP has three phases during the drug’s exclusivity period: Phase 1. Evaluation. During an evaluation phase, the price is set low to encourage adoption and develop real-world evidence rapidly; Phase 2. Reward. In the reward phase, prices are set based on the effectiveness established in the evaluation phase, which rewards manufacturers for their innovation; Phase 3. Access. In the access phase, prices are lowered to facilitate widespread adoption. Price schedule under the status quo and TPP. Figure 1 illustrates the status quo for PCSK9 inhibitors, starting from 2016 until expected exclusivity expiration in 2030. It shows a constant annual price (for a monthly or bi-monthly dosing regimen) of $9,598 from 2016 until generic competition reduces the price to $2,181 in 2030. The former is based on an assumed 34% discount to the list price, equal to the industry-wide discount average estimate, and the latter on an expected 85% reduction in price when exclusivity is lost. 17 In our modeling of TPP, we assume a three-year Evaluation Phase, 1 the drug is priced at 50% off list price to encourage rapid adoption. Pricing in the seven-year Reward Phase depends on the drug’s performance in terms of reduced myocardial infarction and stroke risk during evaluation, and is calibrated to result in the same cost per event avoided (a composite of both events) under three different efficacy scenarios. The price in the Access Phase is set at $3,635 (75% off the list price), to ensure the total cash flow to the manufacturer over the exclusivity period (2016 - 2030) is equal for expected efficacy in both TPP and status quo at a 3% discount rate. Based on our calibration, Reward Phase prices are $5,281 for low efficacy, $11,761 for expected efficacy, and $18,982 for high efficacy. Efficacy here is defined based on the reduction in risk of myocardial infarction (MI) and stroke observed in the pivotal FOURIER trial. 5 We assume a drug meets expected efficacy criteria when the risk reduction of MI ranges from 0.660 to 0.795 (median 1 Although we assume three years in this example, phase lengths may vary; for medicines with less uncertainty, shorter Evaluation Phases may be sufficient. 70 0.730) or the risk reduction of stroke ranges from 0.735 to 0.845 (median 0.790). If the drug demonstrates real-world risk reduction that is better than this range, we classify it as high efficacy. We classify a drug as low efficacy when the real-world risk reduction for MI ranges from 0.795 to 1.000 (median 0.86) or for stroke from 0.845 to 1.00 (median 0.90). Modeling outcomes. We compare the results of TPP with the status quo pricing using the Future Elderly Model (FEM), an economic demographic microsimulation model, to estimate the health benefits of PCSK9 inhibitors among Americans aged 51 and older for each efficacy scenario. The FEM uses initial demographic characteristics and health conditions for each individual to project their medical spending, health conditions and behaviors, disability status, and quality of life. The model has been developed over time with support from the National Institute on Aging, the Department of Labor, the MacArthur Foundation, and the Centers for Medicare & Medicaid Services to study health care innovation in a wide variety of contexts. 18-20 Of particular note, the model has been used to study the benefits of innovation in heart failure treatment, 21 statin use, 22 and reduction in cardiovascular risk factors. 23 A detailed discussion of our methods can be found in the supplemental appendix; here we provide an overview. We first identify the population eligible for PCSK9 inhibitors based on the FDA label and inclusion criteria for the pivotal FOURIER trial: those with familial hypercholesterolemia (defined as LDL cholesterol level higher than 190 mg/dl) and those with an existing cardiovascular condition and LDL cholesterol level of at least 70mg/dL while receiving cholesterol-lowering therapy. We do not assume that everyone is treated initially, consistent with the data observed to date, and we allow uptake to change once the real-world experience becomes available. Specifically, we estimate that uptake will rise gradually during the first two phases of TPP, reaching about 5% of eligible patients treated annually within 6 years of launch. During the final phase, we assume that uptake adjusts based on efficacy data: under low efficacy, uptake gradually decreases to 2.5% of the eligible patient population, and under high efficacy, uptake increases to 10% of the eligible population. The key metric for evaluation of PCSK9 inhibitors is “cost per event avoided.” The FOURIER trial demonstrated reduced incidence of myocardial infarction and stroke, both of which are simulated in the FEM. We simulate the entire population aged 51 years and older from 2016 71 onwards, accumulating information about individuals’ health conditions (including incident disease cases) and total drug spending in each year—taking into account the disease risk reduction and the price of treatment, which reflects the annual per-patient cost of PCSK9 inhibitors. RESULTS Spending on PCSK9 inhibitors. Figure 2 shows the present discounted value of spending on PCSK9 inhibitors under different scenarios. As we show, TPP produces benefits for manufacturers, patients, and payers compared to the status quo. Under the status quo, spending varies modestly with efficacy, ranging from a low of $64.3 billion to a high of $84.4 billion. Spending with TPP varies more widely, from $34.7 billion to $114.7 billion. Importantly, if PCSK9 inhibitors exhibit low real-world efficacy, TPP avoids $30 billion in spending relative to the status quo. If the drug performs similarly as in the clinical trials, TPP generates comparable cash flow, but delivers revenues faster, enabling more immediate investment in further R&D. If real-world effectiveness is better than the clinical trials predict, spending is higher under TPP, reflecting the higher uptake in the high efficacy scenario with TPP relative to the status quo. Cost per event avoided. The ultimate question, however, is how much health benefit we get for the PCSK9 inhibitor spending shown in Figure 2. In Figure 3, we examine the cost per event avoided. TPP lowers the cost per event avoided under both low and expected efficacy scenarios, and results in comparable cost per event avoided under high efficacy. In this way, TPP leaves society better off in terms of value per dollar spent, relative to status quo pricing. 2 DISCUSSIONS A three-part pricing approach offers several advantages over the status quo. If the drug’s real- world efficacy is better than in trials, more patients are treated, with the same cost per event avoided as the status quo. However, if real-world efficacy is worse than expected, payers spend less on the drug (by about $30 billion), with a 50% lower cost per event avoided than under the 2 We do not model cost-effectiveness in our model, given ongoing discussion about appropriate thresholds for different patient populations, but rather assume a generic discount to the list price to frame our price setting under Status quo pricing and associated spending on PCSK9 inhibitors. 72 status quo. Therefore, a pre-negotiated TPP price schedule would accelerate access and real-world evidence development at lower cost to society. Some issues would need to be resolved in practice. First, TPP shifts some efficacy risk to manufacturers, and it exposes payers to higher spending if the drug’s effectiveness exceeds expectations. Hence, manufacturers may not accept the deal unless they are confident about their drug’s performance, and payers may not want to risk greater spending even if the drug performs as expected. While both uncertainties exist in the status quo, TPP’s design is conditional on the parties accepting higher risk exposure. Second, collection of clinical outcomes during the Evaluation Phase could be hindered by several challenges, including reporting issues, adherence to therapy and data-sharing challenges among providers, payers and manufacturers. Provisions related to extra-low efficacy, under which coverage would be stopped, should be in place. Such details have already been worked out in private agreements with plans like Harvard Pilgrim 24 and Cigna, 25 so this concern is likely not prohibitive. Finally—and of most concern—a longer-term agreement such as TPP may weaken product competition and reduce benefits if another innovative therapy enters the market and provides greater value for money. Specific contractual provisions would be required to ensure the theoretical benefits of TPP are realized by all parties. In sum, pay-for-performance has been difficult to achieve in the pharmaceutical sector. Most importantly, it has failed because manufacturers have little incentive to launch at a low price to encourage use as more evidence is collected about real-world, long-term effectiveness. We propose a three-part price schedule which could provide access at a lower price during an evaluation phase, and limit the period during which the innovator is rewarded in proportion to the clinical benefits delivered by a new therapy. The result is better access while still rewarding innovators. While TPP is not a panacea, in the case of PCSK9 inhibitors it offers lower cost per event avoided, and distributes risk between manufactures and payers based on the drug’s performance. As such, it provides a promising alternative to existing payment models for high- cost therapies. 73 TABLES AND FIGURES Figure 1: Three-Part Pricing of PCSK9 inhibitors Figure 1 displayed the framework of the Status Quo and the Three-Part Pricing model. The Status Quo Pricing refers to the fee-per-dose reimbursement program. The initial price in the Status Quo Pricing scenario is determined based on a 34% discount of the list price ($14,542 in 2017). The price for the Evaluation Phase in the Three-Part Pricing model is set at 50% discount of the list price, followed by a negotiated price varied by performance of the drug and a price lower than the prices in the prior two phases. 74 Figure 2: Discounted revenues from PCSK9 inhibitors under different efficacy scenarios (using 3% p.a. discounting under both payment approaches) Figure 3: Cost per event avoided under different efficacy scenarios for PCSK9 inhibitors $604,358 $336,851 $214,266 $308,153 $286,612 $215,768 - 100,000 200,000 300,000 400,000 500,000 600,000 700,000 Low Efficacy Expected Efficacy High Efficacy Status Quo Three-Part Pricing $ spent per event avoided 75 REFERENCES 1 Muhlestein, D., Burton, N. & Winfield, L. The Changing Payment Landscape Of Current CMS Payment Models Foreshadows Future Plans. Health Affairs (2017). <https://www.healthaffairs.org/do/10.1377/hblog20170203.058589/full/>. 2 Goldman Dana, P. & Lakdawalla, D. Moving Beyond Price-Per-Dose In The Pharmaceutical Industry. Health Affairs (2015). <https://www.healthaffairs.org/do/10.1377/hblog20150930.050865/full/>. 3 Blumenthal, D. M., Jena, A. B. & Goldman Dana, P. Outcomes-Based Pricing as a Tool to Ensure Access to Novel but Expensive Biopharmaceuticals. Annals of Internal Medicine 166, 219-220, doi:10.7326/M16-1847 (2017). 4 The FDA Label for REPATHA. (2015). <https://www.accessdata.fda.gov/drugsatfda_docs/label/2015/125522s000lbl.pdf>. 5 Sabatine, M. S. et al. Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. New England Journal of Medicine 376, 1713-1722, doi:10.1056/NEJMoa1615664 (2017). 6 Farnier, M. Alirocumab for the treatment of hyperlipidemia in high-risk patients: an updated review. Expert Review of Cardiovascular Therapy 15, 923-932, doi:10.1080/14779072.2017.1409115 (2017). 7 Schwartz, G. G. et al. Effect of alirocumab, a monoclonal antibody to PCSK9, on long- term cardiovascular outcomes following acute coronary syndromes: rationale and design of the ODYSSEY outcomes trial. American Heart Journal 168, 682-689, doi:10.1016/j.ahj.2014.07.028 (2014). 8 Shrank, W., Lotvin, A. M., Singh, S. & Brennan, T. In The Debate About Cost And Efficacy, PCSK9 Inhibitors May Be The Biggest Challenge Yet. Health Affairs 2018 (2015). <https://www.healthaffairs.org/do/10.1377/hblog20150217.044597/full/>. 9 Tice, J. A., Ollendorf, D. A., Cunningham, C., Pearson, S. D. & Kazi, D. S. PCSK9 Inhibitors for Treatment of High Cholesterol: Effectiveness, Value, and Value Based Price 76 Benchmarks. (2015). <https://icer-review.org/wp-content/uploads/2016/01/Final-Report- for-Posting-11-24-15-1.pdf>. 10 Arrieta, A. et al. Updated Cost-effectiveness Assessments of PCSK9 Inhibitors From the Perspectives of the Health System and Private Payers: Insights Derived From the FOURIER Trial. JAMA Cardiology 2, 1369-1374, doi:10.1001/jamacardio.2017.3655 (2017). 11 Navar, A. M. et al. Association of Prior Authorization and Out-of-pocket Costs With Patient Access to PCSK9 Inhibitor Therapy. JAMA Cardiology 2, 1217-1225, doi:10.1001/jamacardio.2017.3451 (2017). 12 Blumenthal, D. M., Goldman, D. & Jena, A. B. Tying Reimbursement to Outcomes Is an Ideal Strategy for PCSK 9 Inhibitors. JAMA Cardiology, doi:10.1001/jamacardio.2017.2959 (2017). 13 Hernandez, I. Revisiting Outcomes-Based Pricing Propositions for the PCSK9 Inhibitor Evolocumab. JAMA Internal Medicine 177, 1388-1390, doi:10.1001/jamainternmed.2017.3143 (2017). 14 Ference, B. A. et al. Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies. A consensus statement from the European Atherosclerosis Society Consensus Panel. European Heart Journal 38, 2459- 2472, doi:10.1093/eurheartj/ehx144 (2017). 15 Braunwald, E. Reduction of LDL-cholesterol: important at all ages. European Heart Journal 37, 1982-1984, doi:10.1093/eurheartj/ehw100 (2016). 16 Chassang, S., Mantua, V., Snowberg, E., Xoxi, E. & Pani, L. Sustainable Reimbursements: Towards a Unified Framework for Pricing Drugs with Significant Uncertainties. CESifo Working Paper No. 6846 (2018). 17 Divan, V., Young, A., Walton, J. & Weston, M. Global Pharmaceuticals: Industry Overview/Analysis. Credit Suisse (2016). <https://research-doc.credit- suisse.com/docView?language=ENG&format=PDF&sourceid=csplusresearchcp&docum 77 ent_id=1061226501&serialid=WA0fY0Lhy75fzAbFCzsOZ9eQdrH78BmqbR32eAb6m1 s%3D>. 18 Goldman, D. P. et al. Consequences of health trends and medical innovation for the future elderly. Health Affairs 24, W5R5-W5R17 (2005). 19 Goldman, D. P. et al. Substantial health and economic returns from delayed aging may warrant a new focus for medical research. Health Affairs 32, 1698-1705, doi:10.1377/hlthaff.2013.0052 (2013). 20 Goldman, D. P. & Orszag, P. R. The Growing Gap in Life Expectancy: Using the Future Elderly Model to Estimate Implications for Social Security and Medicare. American Economic Review: AEA Papers and Proceedings 104, 230-233, doi:http://dx.doi.org/10.1257/aer.104.5.230 (2014). 21 Van Nuys, K., Xie, Z., Tysinger, B., Hlatky, M. A. & Goldman Dana, P. Innovation in Heart Failure Treatment. JACC: Heart Failure Forthcoming (2018). 22 Gaudette, E., Goldman, D. P., Messali, A. & Sood, N. Do Statins Reduce the Health and Health Care Costs of Obesity? Pharmacoeconomics, doi:10.1007/s40273-014-0234-y (2015). 23 Goldman, D. P. et al. The benefits of risk factor prevention in Americans aged 51 years and older. American Journal of Public Health 99, 2096-2101, doi:10.2105/AJPH.2009.172627 (2009). 24 Harvard Pilgrim Health Care. Harvard Pilgrim Signs Outcomes-Based Contracts with AstraZeneca for Brilinta and Bydureon, <https://www.businesswire.com/news/home/20170530005665/en/Harvard-Pilgrim- Signs-Outcomes-Based-Contracts-AstraZeneca-Brilinta> (2017). 25 Cigna Corporation. Cigna's Two New Value-Based Contracts With Pharma For PCSK9 Inhibitor Cholesterol Drugs Tie Financial Terms To Improved Customer Health, <https://www.cigna.com/newsroom/news-releases/2016/cignas-two-new-value-based- 78 contracts-with-pharma-for-pcsk9-inhibitor-cholesterol-drugs-tie-financial-terms-to- improved-customer-health> (2016). 79 SUPPLEMENTAL APPENDIX The Future Elderly Model To estimate the potential health benefits of PCSK9 inhibitors, we use the Future Elderly Model (FEM), a dynamic microsimulation model developed by Goldman et al. (2004) that has been used to study a wide variety of health policy questions with support from National Institutes of Aging, the MacArthur Foundation, and the Department of Labor. FEM tracks patients aged 50 years and older and projects their health and economic outcomes. Unlike many other social benefit simulation models, FEM estimates individual health trajectories rather than aggregating health characteristics of a cohort. FEM was developed to forecast the implications of different medical technology interventions on long-term health and estimate the impacts of health policy changes, including estimating the value of PCSK9 inhibitors in hypercholesterolemia patients, 1 the impacts of statins in an obese population, 2 delayed biological aging, 3,4 and delayed Alzheimer’s Disease. 5 The model and methods are briefly described here; complete technical information is available in the FEM technical Appendix. The FEM simulates the outcomes of older Americans based on the Health and Retirement Survey (HRS), a nationally representative biennial survey of Americans aged 51 and older which are used to compute the health transition models at the core of FEM. And to characterize the population that goes into the simulations. The FEM has three core modules. The Health Transition module calculates transition probabilities across various health states to model the health of the 51+ population over its lifetime. Health transitions include the presence of certain chronic conditions, functional status, mortality, biomarkers, and respondents’ current therapies for cholesterol, diabetes, and blood pressure control. All transition probabilities are modeled with first-order Markov processes controlling for age, gender, education, race, body-mass index, smoking behavior, and health at the time of entry into the simulation. Most chronic conditions are treated as absorbing, meaning that once an individual receives an initial diagnosis, they have the condition until death. Two non-absorbing conditions – myocardial infarction (heart attack) and re-occurring stroke – were modelled to estimate the probabilities of having these conditions in every simulated year. Biomarker values (specifically cholesterol, HbA1c, and blood pressure) 80 were added to the list of covariates predicting cardiovascular diseases, diabetes, and hypertension. We predicted quality-adjusted life-year (QALY) measures using EQ-5D, a standardized and well-validated health-related quality-of-life instrument measuring a respondent’s general health status based on five dimensions: mobility, daily activities, self-care, anxiety, depression, and pain. 6 The Medical Expenditure Panel Survey (MEPS) data were used to estimate EQ-5D scores with an ordinary least squares regression as a function of a patient’s chronic conditions and their FEM-specified functional status. The second FEM component is the Policy Outcomes module, which predicts an individual’s health care spending and other economic outcomes (i.e. social security and disability benefits). The FEM first predicts expenditures for beneficiaries enrolled in Medicare and Medicaid, as well as expenditures by private insurers. Medical spending includes the costs of medical provider visits, hospital events, inpatient stays, outpatient visits, emergency department visits, dental care, home health care, optometry, other medical equipment and services, prescribed medicines, and nursing home stays. The Medical Expenditure Panel Survey was used to estimate the spending for individuals prior to age 65 and the Medicare Current Beneficiary Survey was used for the spending on individuals aged 65 and older. The estimates are based on pooled least squares regressions of each type of spending on risk factors, self-reported conditions, and functional status, with spending inflated to current dollars using the medical component of the consumer price index. The third model component is a Replenishing Cohort module, which uses data from the Health and Retirement Study to introduce new cohorts of 51-year-olds in each simulated year. The National Health Interview Survey, the Current Population Survey, and the National Health Nutrition and Examination Survey data were used to predict the demographic, social economic status, and the health status for these younger populations. Simulations We conducted a population-wide simulation to investigate population-wide trends observed in a representative cross-section of the specific U.S. population in each period from 2016 onwards. A 81 full FEM population (including the replenishing cohort) was used to project the outcomes for the entire population of Americans aged 51 years and older, with population health outcomes calculated by aggregating individual health measures in each simulated year. Scenarios We first created a baseline scenario assuming treatment strategies for managing hypercholesterolemia in the U.S prior to the introduction of PCSK9 inhibitors. Next, we generated three scenarios representing the introduction of PCSK9 inhibitors with differing assumptions about their uptake and efficacy. We reflect the health impact of PCSK9 inhibitors reported in clinical trials by adjusting health transitions and outcomes of eligible individuals accordingly, together with estimates about expected healthcare spending due to PCSK9 inhibitors, and contrast these to the standard of care. We identified the population eligible for PCSK9 inhibitors based on the original FDA-approved indications, as well as the inclusion criteria in the FOURIER trial, which was the first published randomized clinical trial to evaluate the efficacy of PCSK9 inhibitors on hard clinical endpoints pertinent to cardiovascular risk reduction. 7,8 Individuals aged 51 and older with an existing diagnosis of a cardiovascular condition and LDL cholesterol level of at least 70mg/dL while receiving cholesterol-lowering therapy were deemed eligible for PCSK9 inhibitors. Those with familial hypercholesterolemia (defined as LDL cholesterol level higher than 190 mg/dl) with or without a cardiovascular condition were also eligible in our model. We assumed that the eligibility for PCSK9 inhibitors is an absorbing state given that hypercholesterolemia is a chronic condition that requires lifetime treatment to maintain LDL-C level within an acceptable range. We estimated that in 2016, 14 million individuals would be eligible for PCSK9 inhibitors in the U.S. population aged 51 and older. Uncertainty surrounding long-term effectiveness and high retail cost concerns has served as barriers to widespread adoption of PCSK9 inhibitors. 9,10 Health care payers commonly respond to such uncertainty by limiting access to the drug using prior authorization, high patient copays and other utilization management tools. As a result, the adoption of PCSK9 inhibitors in the real- world setting has been very slow. 9,11 Therefore, to reflect the uncertainty about the drugs’ real- 82 world effectiveness and their slow uptake, three scenarios were designed with varying uptake functions and assumptions about the drugs’ effectiveness: 1) An Expected Efficacy scenario assumes that the real-world effectiveness of PCSK9 inhibitors in cardiovascular outcomes is consistent with the efficacy reported FOURIER. In this scenario, we assumed the number of PCSK9-treated patients begins with 0% of the eligible population in 2014 and gradually rises to a maximum of 5% of the eligible population in 2022. This pattern is consistent with the evidence presented in Cannon et al. 2017, who estimate that 14% of the high-risk population would be eligible for PCSK9 inhibitors, with between 31% - 37% of these eventually receiving the therapy. 9,11 The shape of our uptake function is based on Robey S et al. 2017, who find that, on average, biologics take approximately 6 years to reach peak sales, following an S-shape trajectory. 12 Under these assumptions, we estimated the number of treated patients reaches approximately 0.5 million in 2019 and the cumulative number of patients by 2021 is 2.54 million, which is consistent with the estimates reported by the Institute of Clinical and Economic Review. 13 2) A Low Efficacy scenario assumes that PCSK9 inhibitors have real-world effectiveness below that reported in the FOURIER. We assumed the same uptake function between 2016-2025 as under Expected Efficacy but decrease the treated population to 2.5% of the eligible patients by the end of the exclusivity period. 3) A High Efficacy scenario assumes that PCSK9 inhibitors have real-world efficacy above that reported in FOURIER, with the proportion of treated patients increasing from 5% in 2025 to 10% by the time the drug loses exclusivity. By the end of the exclusivity period, we expect a cumulative 11.9 million patients would receive PCSK9 inhibitors under the Expected Efficacy scenario, and 9.6 million and 16.3 million patients treated under the Low and High Efficacy scenarios, respectively. The impact of PCSK9 inhibitors on health To reflect the health impacts of PCSK9 inhibitors, we modified the transition probabilities of treated individuals in each PCSK9 scenario by reducing the probabilities of myocardial infarction or stroke based on FOURIER results. In the Expected Efficacy scenario, we associate 83 PCSK9 inhibitors with a risk reduction of 0.73 for myocardial infarction (MI) and of 0.79 for stroke. These estimates correspond to the mean estimates reported in FOURIER, with an assumption of a constant hazard ratio over time. 7 In the Low Efficacy scenario, we assume a risk reduction of 0.86 in MI and of 0.90 in stroke. In the High Efficacy scenario, we assume a risk reduction of 0.59 for MI of 0.68 for Stroke. For each simulated year, we estimated the total numbers of incident MI and stroke cases for each uptake and efficacy scenario as well as for the baseline scenario (without the use of PCSK9 inhibitors). Next, we calculated the numbers of events avoided under each scenario compared to the baseline scenario. We estimated a cumulative 340 thousand events would be avoided in the Expected Efficacy scenario by the end of exclusivity period; and 160 thousand and 740 thousand events would be avoided in Low and High Efficacy scenarios, respectively. The Price of PCSK9 inhibitors in the Three-part pricing model As illustrated in Figure 1, we established a novel Three-Part Pricing (TPP) Model which includes a three-year Evaluation Phase, followed by a seven-year Reward Phase, during which the price will depend on the real-world effectiveness demonstrated in the Evaluation Phase. The price will decrease substantially in the Access Phase to enable broader access to the therapy. In our model, we compared the results of TPP model to a standard fixed-fee-per-dose payment (or FFPD), which assumes a constant rebate rate for PCSK9 inhibitors over the entire exclusivity period. Based on its list price in 2017, the annual per-patient cost of PCSK9 inhibitors would be $14,542, 14 assuming a flat 34% discount off the list price under FFPD. 15 Under TPP, several decisions were made to calibrate the price in each phase. First, we impose a 50% discount off the list price in the Evaluation Phase. In the Reward Phase, to ensure that the price is proportional to the drug’s efficacy, the price is calibrated such that the average cost per event avoided is equal across all three efficacy scenarios. In addition, we categorize a drug as having Expected efficacy when the risk reduction ranges from 0.66 to 0.795 (with a median 0.73) for MI or 0.735 to 0.845 (with a median 0.79) for stroke in the treated population. We categorize a drug as having High efficacy when the risk reduction falls below those ranges, and as having Low efficacy when the risk reduction ranges above those ranges. Finally, we varied the price in the Access Phase in the Expected Efficacy scenario to ensure the cumulative cash flow of the drug to the manufacturer is the same under both TPP and FFPD over the exclusivity period if the drug’s effectiveness meets 84 expectations. This ensures that the manufacturer will be indifferent between the two pricing models ex ante. We impose a uniform 75% discount to the list price in the Access Phase under all scenarios, which results in the same total cashflow in expected efficacy under both payment models. External Validation We conducted a “cohort simulation” by disabling the Replenishing Cohorts Module, and thereby followed a representative cohort of Americans aged 51-80 in 2016 until their deaths. We tracked the effect of the treatment intervention on their lifetime outcomes with and without PCSK9 inhibitors. In the Expected Efficacy scenario, we use the parameters reported in Fonarow et al. 16 2017 (including the direct medical costs associated with MI and stroke incidence, and the disutility of injection site reaction due to PCSK9 treatment) to model the average incremental costs per quality-adjusted life years gained per person among those eligible for the treatment. Our cost per event avoided and cost per QALY gained estimates are comparable with Fonarow et al.’s results (which reported an incremental $268,637 per QALY gained at the list price of PCSK9 inhibitors). 16 85 References 1 Cheng, W. H., Gaudette, E. & Goldman, D. P. PCSK9 Inhibitors Show Value for Patients and the US Health Care System. Value in health : the journal of the International Society for Pharmacoeconomics and Outcomes Research 20, 1270-1278, doi:10.1016/j.jval.2017.05.014 (2017). 2 Gaudette, E., Goldman, D. P., Messali, A. & Sood, N. Do Statins Reduce the Health and Health Care Costs of Obesity? PharmacoEconomics 33, 723-734, doi:10.1007/s40273-014- 0234-y (2015). 3 Goldman, D. P., Gaudette, E. & Cheng, W. H. Competing Risks: Investing in Sickness Rather Than Health. American journal of preventive medicine 50, S45-50, doi:10.1016/j.amepre.2015.12.005 (2016). 4 Goldman, D. P. et al. Substantial health and economic returns from delayed aging may warrant a new focus for medical research. Health affairs (Project Hope) 32, 1698-1705, doi:10.1377/hlthaff.2013.0052 (2013). 5 Zissimopoulos, J., Crimmins, E. & St Clair, P. The Value of Delaying Alzheimer's Disease Onset. Forum for health economics & policy 18, 25-39, doi:10.1515/fhep-2014-0013 (2014). 6 Dolan, P. & Roberts, J. Modelling valuations for Eq-5d health states: an alternative model using differences in valuations. Medical care 40, 442-446 (2002). 7 Sabatine, M. S. et al. Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. The New England journal of medicine 376, 1713-1722, doi:10.1056/NEJMoa1615664 (2017). 8 The FDA Label for REPATHA. (2015). <https://www.accessdata.fda.gov/drugsatfda_docs/label/2015/125522s000lbl.pdf>. 9 Navar, A. M. et al. Association of Prior Authorization and Out-of-pocket Costs With Patient Access to PCSK9 Inhibitor Therapy. JAMA cardiology 2, 1217-1225, doi:10.1001/jamacardio.2017.3451 (2017). 86 10 Hlatky, M. A. & Kazi, D. S. PCSK9 Inhibitors: Economics and Policy. Journal of the American College of Cardiology 70, 2677-2687, doi:10.1016/j.jacc.2017.10.001 (2017). 11 M, C. Data Show it’s Hard to Fill PCSK9 Prescriptions, Confirming Cardiologists’ Complaints., <http://www.ajmc.com/conferences/acc-2017/Data-Show-Its-Hard-to-Fill-PCSK9- Prescriptions-Confirming-Cardiologists-Complaints.> (2017). 12 Robey, S. & David, F. S. Drug launch curves in the modern era. Nature reviews. Drug discovery 16, 13-14, doi:10.1038/nrd.2016.236 (2017). 13 Tice, J. A., Ollendorf, D. A., Cunningham, C., Pearson, S. D. & Kazi, D. S. PCSK9 Inhibitors for Treatment of High Cholesterol: Effectiveness, Value, and Value Based Price Benchmarks. (2015). <https://icer-review.org/wp-content/uploads/2016/01/Final-Report-for- Posting-11-24-15-1.pdf>. 14 Kazi, D. S. et al. Updated Cost-effectiveness Analysis of PCSK9 Inhibitors Based on the Results of the FOURIER Trial. Jama 318, 748-750, doi:10.1001/jama.2017.9924 (2017). 15 Global pharmaceuticals industry overview/analysis. (2016). <https://research-doc.credit- suisse.com/docView?language=ENG&format=PDF&sourceid=csplusresearchcp&document_id= 1061226501&serialid=WA0fY0Lhy75fzAbFCzsOZ9eQdrH78BmqbR32eAb6m1s%3D>. 16 Fonarow, G. C. et al. Cost-effectiveness of Evolocumab Therapy for Reducing Cardiovascular Events in Patients With Atherosclerotic Cardiovascular Disease. JAMA cardiology 2, 1069-1078, doi:10.1001/jamacardio.2017.2762 (2017). 87 CHAPTER 3: LONGITUDINAL TRENDS AND PREDICTOR OF STATIN ADHERENCE IN U.S MEDICARE BENEFICIARIES ABSTRACT Aim: Suboptimal use of statins has been identified as a serious problem associated with the worse health outcome and increased morbidity and mortality among elderly patients with cardiovascular diseases. Our study aims to understanding the longitudinal trends in statin adherence which would be helpful for informing strategies of adherence improvement. Methods: We identified the elderly statin users from a National Representative Medicare Beneficiaries by using the sample in Medicare Current Beneficiary Survey (MCBS) data. We calculated their 180-day statin adherence since their first fill date in each year. Proportion of days covered (PDC) was calculated to measure statin adherence and those who considered as having “Good adherence” was defined by PDC > 80%. We also examined the trend of patient average out-of-pocket costs per day patient and examined the factors associated with statin adherence with a multivariate logistic model. Results: From 2006 to 2012, the prevalence of good statin adherent increase from 66% to 77% with an increment of 8% per year (p for trend <0.001). On the other hand, the mean OOP costs for the statins decreased from $0.54 per day in 2006 to $0.26 in 2012. Statin OOP costs decreased disproportionately for White patients and those who with higher annual income, while the increment of statin adherence in those groups were not significantly different when compared with the minorities and low-income group. Factors associated with good statin adherence included those with older age, White, higher income, obesity, low income subsidy. Conclusion: As more statins go off patent over time, the disproportionate drop in copayment for relatively advantaged groups didn’t reflect a significantly disproportionate increment in statin adherence. Policies other than decreasing cost-sharing might be needed to help guide future intervention to increase adherence. 88 INTRODUCTION Statin therapy is known to be effective in preventing major adverse cardiovascular events in patients with established atherosclerosis cardiovascular disease (CVD) and in subjects with a high cardiovascular risk and elevated cholesterol level. In the past two decades, a majority of patients with previous ASCVD are benefit significantly from guideline recommended statin use, which was estimated to reduce the mortality by 50%.(1-3) However, substantial treated high-risk patients fail to achieve target LDL-C levels in the U.S, with estimated 12 out of 15 million with LDL > 70 mg/dl and 6 million with LDL > 100 mg/dl in 2015.(4) Failure of LDL achievement has been widely documented to be associated with a higher risk of major adverse cardiovascular events. Evidence suggested that suboptimal use of statin therapy may contribute to poor LDL management and consequently contribute to worsening of diseases and deaths, as several studies reported suboptimal long or short-term adherence to statins in a variety of patient populations (with a 1-year adherence rates of 45% to 85% for elderly patients initiated on lipid-lowering therapy).(5-10) Drug costs have been discussed as the main factors that contribute to medication underuse in treating chronic diseases.(11) Increased cost sharing is associated with lower rates of drug treatment, worse adherence among existing users, and more frequent discontinuation of therapy. (12) The other study done by Choudhry in 2011 showed that eliminating out-of-pocket costs for patient prescription improved medication adherence and rates of first major vascular events.(13) Chen SY et al (14) also showed that moving branded statins to the lowest copay tier, therefore lowering copayments, on patient adherence in non-LIS Medicare Part D beneficiaries. As the statins has been used over the decades, the statin adherence has likely been dynamic, changing over time and the trend could be diverse in different subgroups. The racial and social- economic disparities in statin use has been reported in earlier literature.(15,16) Karaca-Mandic P et al. demonstrated the greater OOP costs for statins are associated with reductions in statin adherence among Medicare Part D beneficiaries. (12) They reported that an increase in annual statin OOP from $200 to $240 was associated with a reduction in the rate of adherent beneficiaries from 67percent to 56 percent (p < .001). 89 In this paper, we describe longitudinal trends in statin use for national representative Medicare beneficiaries using statins while Medicare Part D were in effect (2006 – 2012). We also investigate the longitudinal trends in statin cost-sharing to evaluate how it correlated with the trends in statins adherences. In addition, we also examine patient demographics and health characteristics associated with statin adherence across all the users. METHOD Study design and data sets We used Medicare Current Beneficiary (MCBS) data Cost and Use data and linked claims data from 2006 to 2012 for this study. MCBS is a longitudinal rotating panel data surveying a representative sample of the Medicare beneficiaries enrolled in Part A or Part B. MCBS surveyed approximately 4,000 to 5,000 new enrollees each year and follow the participants up to four consecutive years or until death or loss of follow-up. During that period, they are interviewed in the fall of their induction year and then three times annually during the second through fourth survey years. Information captured includes use and expenditures for health services, insurance coverage, access to care, health and functional status, socioeconomic status, and demographic characteristics. The Cost and Use file estimates total annual healthcare utilization and expenditures for services covered by or not, including prescription drugs. In addition, the MCBS provides all Part A and B claims for each beneficiary including hospital inpatient, outpatient, physician, skilled nursing facility, home health agency, durable medical equipment and hospice claims. Since 2006, the MCBS added Medicare Part D utilization from Part D claims to the Prescribed Medication Events (PME) questionnaire. Medicare Part D claims also provided total payment, copayment, fill date, National Drug Code (NDC) code, and days of supplied for each prescription. Eligible Population We identified statin users in each year by using the Medicare Part D claims as part of the PME data. Eligible participants were aged 65 years or older with at least two Part D claims for a statin on different fill dates. For example, if a patient had two separate statin prescription fills in 2012, this patient was considered in the outcome measurements in 2012. The participants were required 90 to have continuous enrollment for 6 months after the index date. The index date for each year was defined as the first date of the statin fill in that year. Outcome Our primary outcome was 180-day statin adherence among the statin users measured in each year. The measure of adherence was calculated by the proportion of days covered (PDC), which the PDC numerator is number of days covered by all statin prescription fills during the first six months since the index date and the denominator is 180 days.(17,18) An advantageous property of PDC is that it avoids the double counting overlapped days between two prescription fills. A patient was defined as achieving a “Good Adherence” if the PDC was 80% or higher.(18) Clinical benefits are most likely to occur when this threshold is exceeded. The PDC was calculated from the utilization of statin medications and did not include switches to other cholesterol-lowering therapies, which may have underestimated the level of adherence to all cholesterol lowering medications. Use of other lipid-lowering agents occurred infrequently, in around 5% of out sample. The outcome variable is the average daily out-of-pocket (OOP) cost for patient’s statin prescriptions. The MCBS data reported the dollar amount that the beneficiary paid for each prescription drug event without being reimbursed by a third party. The amount includes copayments, coinsurance, deductible or other patient payment amounts, which contributes to a beneficiary’s true OOP costs for a Part D covered drug. The average daily OOP cost per patient was calculated by summing of the OOP cost for all statin prescription fills within 180 days and divided by number of days covered by statin prescription fills. Explanatory Variables We assessed the following variables for each year for each statin user: age, gender, race/ethnicity, education level, household income, marital status, obesity (identified by their BMI) and smoking status, self-reported diagnosis of health conditions and comorbidities, functional status (whether a person have any limitation in activities of daily livings or instrumental activities of daily livings), whether a person was enrolled in a Medicare-Advantage Prescription Drug Plan (MA-PD), whether a person received a low income subsidy from Medicare. All the patient characteristics were all obtained by using the MCBS Cost and Use data. 91 Statistical Analysis We compared the baseline characteristics in the fiscal year 2006 and 2012 among the users with poor adherence and the users with good adherence. Descriptive statistics were obtained for the baseline characteristics, and all outcomes, including frequency for categorical variables and means and SDs for continuous variables. Analysis of variance and Chi-square tests were used to compare good adherent and poor adherent patients. To example the factors and the yearly trends associated with a good statin adherence among the users, we modeled statin good adherence as a function of the explanatory variables using a time- variant pooled logistic regression. We also modeled the patient average daily OOP costs by using a time-variant pooled Ordinary least squares (OLD) linear regression to examine the factors and yearly trend associated with the OOP costs. All models were performed with standard errors clustered at a patient level. To examine the year trends in adherence, we created a yearly-trend variable as the number of years since 2006. We allowed the yearly-trend variable interacts with race and household income, to examine the whether the trend disproportionate between the minority or low-income group. All analyses were conducted by using Stata, v.14 (Stata Corp., College Station, TX) statistical software. The significance level threshold was set at p<0.05. All analyses were performed by adjusting the sample weights to present the national representative Medicare beneficiaries. RESULTS The number of beneficiaries using statins ranges from 1,866 in year 2006 to 2,895 in year 2012. The prevalence of good adherent statin users was 66% in 2006 and 77% in 2012, with an average age of 73 year-olds. In both 2006 and 2012, the good adherence group has a significantly lower rate of African Americans and Hispanic, and has a higher rate of users with an annual income higher than $40,000. In 2006, good adherent patients reported a higher prevalence of comorbidities, including heart disease, stroke, myocardial infarction, cancer, Alzheimer’s disease and lung disease; however, in 2012, there is no significant difference in the rates of comorbidities between poor and good adherent group, while those with good adherence had a better function status (with lower rate of any limitation in ADL). Additionally, good adherent 92 statin users were less likely to receive a lower-income subsidy compared with poor adherent users. The average daily OOP cost for the statin prescriptions are significantly lower among the good adherent patients versus those with poor adherent, while the cost become similar between the groups in 2012. Longitudinal trends in statin adherence and mean daily OOP costs Statin adherence increased from 66.2% in 2006 to 76.9% in 2012 (Figure 1). Patients were 8% more likely to become good adherent in each succeeding year (adjusted p-value for trend <0.05) (Table 3). Statin use increased for patients of all races and income levels, although the amounts of the improvement were slightly higher in minorities versus White, as well as in the higher income groups versus the lowest income group, however, the yearly trend was not significantly different (Table 3). On the other hand, the mean OOP costs for the statins decreased from $0.54 per day in 2006 to $0.26 in 2012. There is a substantial drop from 2011 to 2012 (about 23% decrease from the cost in 2011) which could be explained by the patent expiration in November, 2011 for Lipitor --- the statin with the largest market share. Although the OOP costs decreased for patients of all races and income levels, the amount of decrement differed. Both Black and Hispanic had a slower decreasing trend in their OOP costs as compared with White (Table 3), while those who in a higher income group had accelerated decreasing trend in their OOP costs as compared with the lowest income group. In addition, it appears that as statins go off patent in 2011, this is associated with a drop in copayment disproportionately for relatively advantaged groups, including high-income and/or non-minority race group (Figure 2). As a sensitivity analysis, we examine the 180-day adherence for every six months and the results shows similar trends in statin adherence and OOP statin costs (Supplemental Table 1). Factors associated with statin adherence The beneficiaries with older age are associated with higher likelihood of having good adherence to their statin prescriptions (Table 2). Compared to White, Black and Hispanic are almost 50% less likely to adhere to statins (adjusted OR = 0.51; 95% CI, 0.44– 0.59 for black and OR = 0.53; 95% CI, 0.46– 0.61 for Hispanic). The users with a higher annual income are associated with a 93 higher likelihood of being good adherent (adjusted OR = 1.15; 95% CI, 1.03 – 1.29 for those with annual income 20-40K and OR = 1.36; 95% CI, 1.18 – 1.57 for those with annual income > 40K). There is also a geographical difference where those who live in North (adjusted OR = 0.85; 95% CI, 0.74 – 0.98) and South region (adjusted OR = 0.72; 95% CI, 0.64 – 0.81) are less likely to adherence to their statin medications when comparing with those who living in East. Obesity is associated with increased odds ratio of good statin adherence (adjusted OR = 1.22; 95% CI, 1.09 – 1.37), while the current smokers are 13% less likely to adhere to statins (adjusted OR = 0.87; 95% CI, 0.76 – 1.00). Statin uses who had any ADL and/or any IADL were also significantly with 10% less likely to have good adherence (adjusted OR = 0.90; 95% CI 0.82 – 0.99). The use receiving a low income subsidy is associated with 21% more likely to be adherent (adjusted OR = 1.21; 95% CI, 1.08 – 1.36). Consistent with the results in yearly trend, there is a gradually increasing likelihood for each fiscal year since 2006. DISCUSSIONS Our study is the first describing the longitudinal trend in statin adherence among Medicare Part D beneficiaries. Statin adherence in Medicare beneficiaries increased gradually with a yearly trend of 8% increment from 2006 to 2012, while the average daily OOP statin costs reduced by $0.05 per patient per year. Although patients in minority group are associated with a lower likelihood of having good statin adherence (including Black, Hispanic, or those with a lower income), there is no significant difference in the yearly decreasing trend when comparing with White and/or high income group. However, the magnitude of annual reduction in patient OOP statin costs was significantly lower in White and high income group. In addition to race/ethnicity and income, we also identified the factors associated with good statin adherence, including obesity, any limitation in ADL and/or IADL, and receiving low income subsidy. From 2011 to 2012, we observed a larger magnitude of drop in patient OOP statin costs. This could be explained by the availability of generic Atorvastatin (Lipitor®) in the November 2011.(19) Lipitor® was the top selling prescription medication in the United States in 2010 with over $7 billion in total revenue.(19) Our result shows that as a commonly-prescribed statin going off patent, this is associated with a drop in statin OOP costs; however, the drop is disproportionate for relatively advantaged groups, which the users with high-income and/or non- 94 minorities experienced a more substantial reduction in their copayment. This result might be explained by the fact that the relatively advantaged groups were more likely to be using on- patent drugs in the first place. However, it appears no reduction in disparities in good adherence by income or by race/ethnicity. This result shows that in a very short-run, statins going off patent may not help to resolve health disparities in medication adherence. This result is also consistent to the theory that the price-sensitivity to drug use is expected to be lower in high-income groups. The main strength of our study is we used a national representative sample for the Medicare Part D beneficiaries, which provides detailed patient social-economic status and self-reported health and functional characteristics. In literature, statin adherence was mostly examined by using the administrative claims data with limited patient characteristics and limiting to the beneficiaries covered by a single payer.(6,10,20,21) We used the MCBS data, a unique claim-linked survey database, which allows us to accurately examine the medication adherence by identifying whether each day was filled by their statin prescriptions during the follow-up period and to calculate the average OOP cost per day filled; and at the same time, we were able to observe the detailed patient characteristics for each respondent every year. However, there are limitations in our study. First, the sample size ranges from 1,866 to 2,895 for each year, which is relatively small comparing with the previous studies and might explain the wide confidence intervals for the yearly trend estimates in minority groups when modeling the likelihood of good adherence. However, all our results were adjusted by the person-level sample weight and represents the beneficiaries of the entire U.S elderly population. Second, it is difficult to identify the new statin users in the data. Some elderly beneficiaries might have taken statins for a long time before they enrolled in Medicare. How long the patients were on the drug before they enrolled may impact their adherence. Nevertheless, those who were on statins for a longer time are assumed to be with higher CVD risk or worse health conditions, which those characteristics were captured in the data and taken into account in our models. Our study raises two policy issues: First, although the overall statin adherence has increased year by year in the Medicare Part D beneficiaries, more efforts will be needed to improve the disparities in statin adherence, particularly in minorities and low income sub-groups. Our results are consistent with previous literature that nonwhite patients are at increased risk of statin non- adherence and discontinuations.(15,16,18,22,23) To achieve the ultimate goal of reduce the 95 racial and social disparities in CVD management in the U.S elderly, more health service interventions might be required to improve patient statin adherence for the minority patients. Second, it was documented that statin refill adherence was higher among patients who exposed to generic substitution compared to those who were not.(12,14) Even though we observed a gradual decreasing trend in copayment for statins in overall as more generic statins are available since 2006, our results show that as drugs going off patent may not help resolving the disparities in medication adherence for minority groups in the very short-run. However, we assumed that in the long run, having quality drugs available at cheap prices should eventually help to address the disparities. Future studies is needed to examine when does the long-run health disparity effect start to kick in (where the “flow” of new users becomes larger than the "stock" of old users with most of whom were relatively advantaged). This may be useful for health plan administrators and policy makers to inform the design of strategies to improve adherence and health management in the long run for patients. 96 TABLES AND FIGURES Tables 1. Patient Characteristics for the statins users in 2006 and 2012, stratified by their adherence status 2006 2012 All (N=1,86 6; weighted populatio n size = 6,702,84 9) Poor Adheren ce (N=626, 33.8%) Good Adheren ce (N=1,24 1, 66.2%) All (N=2,895; weighted population size =12,604,18 5) Poor Adheren ce (N=670, 23.1%) Good Adheren ce (N=2,22 5, 76.9%) Age (Mean) 73.09 72.12 73.58 * 73.42 73.2 73.49 Male (%) 38.24 39.35 37.67 43.82 40.79 44.74 Race (%) Black 10.41 16.32 7.39 *** 8.91 12.88 7.72 *** Hispanic 9.44 12.81 7.71 *** 11.74 15.49 10.62 *** Other 4.28 3.98 4.44 3.82 3.92 3.79 Education level (%) HS or less 71.77 74.37 70.43 62.73 65.48 61.9 College or some college 22.32 20.55 23.22 29.59 26.12 30.64 Graduate School 5.92 5.09 6.34 7.67 8.4 7.46 Income (Mean) (%) <$20K 56.68 61.79 54.07 42.27 48.38 40.43 $20-40K 27.14 25.21 28.13 29.92 29.04 30.18 >$40K 16.18 13 17.8 ** 27.81 22.58 29.39 ** Marital Status (%) Married 47.53 46.61 47.99 48.27 43.95 49.56 ** Widowed 32.11 33.13 31.59 27.17 26.29 27.43 Health status (%) Overweight 37.27 36.85 37.48 37.77 37.21 37.94 Obese 31.46 31.76 31.3 34.18 32.08 34.82 Current smoking 12.46 12.11 12.64 12.47 11.74 14.93 * Ever had smoked 56.23 55.99 56.36 60.4 58.03 61.1 Health Conditions (%) Heart disease 52.39 47.29 54.99 * 52.83 51.19 53.32 Diabetes 37.76 39.91 36.66 41.81 41.86 41.62 Hypertension 77.61 78.22 77.3 80.83 79.69 81.18 Stroke 17.33 21.19 15.37 * 14.79 16.64 14.23 Myocardial infarction 21.18 18.59 22.49 * 19.29 19.5 19.23 Cancer 16.59 13.93 17.95 * 19.2 18.1 19.53 Psychological disease 22 23.13 21.42 7.62 6.99 7.14 Alzheimer's disease 5.4 3.74 6.25 * 5.26 5.15 5.6 Lung disease 19.78 23.05 18.11 * 23.11 24.22 22.78 Parkinson's disease 1.55 0.98 1.84 1.48 1.62 1.44 Functional Status (%) Any ADL 32.11 34.68 30.8 38.21 41.98 37.08 * Any IADL 41.53 42.66 40.96 43.74 46.37 42.95 97 Insurance status (%) MA-PD 28.97 27.87 29.53 37.11 38.54 36.68 Have Low Income Subsidy 37.95 39.78 37.01 31.19 36.44 29.61 *** Average daily out-of- pocket cost for statins, $ Mean 0.54 0.62 0.5 *** 0.264 0.292 0.257 *P<0.05, **p<0.01, *** P<0.005, based on the comparisons between poor adherence versus good adherence group. 98 Figure 1. Longitudinal trends in statin adherence and average daily out-of-pocket costs for statin prescriptions among the statin users from 2016 to 2012, all users, and stratified by race/ethnicity and annual income group. Table 2. .5 .55 .6 .65 .7 .75 .8 .85 .9 .95 1 Prevalent Statin Users with Good Adherence 2006 2007 2008 2009 2010 2011 2012 Year 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 Average Daily Out-of-Pocket Cost for Statins, $ 2006 2007 2008 2009 2010 2011 2012 Year .5 .55 .6 .65 .7 .75 .8 .85 .9 .95 1 Prevalent Statin Users with Good Adherence 2006 2007 2008 2009 2010 2011 2012 Year White Black Hispanic 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 Average Daily Out-of-Pocket Cost for Statins, $ 2006 2007 2008 2009 2010 2011 2012 Year White Black Hispanic .5 .55 .6 .65 .7 .75 .8 .85 .9 .95 1 Prevalent Statin Users with Good Adherence 2006 2007 2008 2009 2010 2011 2012 Year <20K 20-40K >40K 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 Average Daily Out-of-Pocket Cost for Statins, $ 2006 2007 2008 2009 2010 2011 2012 Year <20K 20-40K >40K (A) Prevalence of Good Adherence (B) Mean Daily Out-of-Pocket Statin Costs 99 Table 2. The odds ratio for having good statin adherence and the coefficient for the average daily out-of-pocket costs the patients spent their statin prescriptions Logit Model for Good Adherence Linear Regression Model for Average Daily OOP Costs Odds Ratio [95% Conf. Interval] Coefficient [95% Conf. Interval] age 1.01 *** [ 1.01 1.02 ] 0.000 [ -0.001 0.001 ] Male 1.00 [ 0.91 1.11 ] -0.009 [ -0.036 0.018 ] Black 0.51 *** [ 0.44 0.59 ] -0.018 [ -0.048 0.012 ] Hispanic 0.53 *** [ 0.46 0.61 ] -0.034 [ -0.069 0.000 ] Other non-white 0.84 [ 0.67 1.06 ] -0.035 [ -0.088 0.018 ] Education level High School ref ref College or some college 1.01 [ 0.91 1.12 ] 0.009 [ -0.021 0.039 ] Graduate School 1.01 [ 0.82 1.23 ] 0.046 [ -0.021 0.114 ] Income <$20K ref ref $20-40K 1.15 ** [ 1.03 1.29 ] 0.039 ** [ 0.010 0.069 ] >$40K 1.36 *** [ 1.18 1.57 ] 0.130 *** [ 0.087 0.173 ] Geographical Region East ref ref North 0.85 * [ 0.74 0.98 ] 0.051 ** [ 0.011 0.090 ] Midwest 0.93 [ 0.81 1.07 ] 0.023 [ -0.016 0.062 ] South 0.72 * [ 0.64 0.81 ] 0.049 ** [ 0.015 0.083 ] Marital Status Married 1.00 [ 0.88 1.13 ] 0.011 [ -0.019 0.040 ] Widowed 1.00 [ 0.87 1.14 ] 0.019 [ -0.012 0.049 ] Health status at the baseline Overweight 1.09 [ 0.99 1.21 ] -0.018 [ -0.046 0.010 ] Obese 1.22 *** [ 1.09 1.37 ] -0.024 [ -0.055 0.007 ] Current smoking 0.87 * [ 0.76 1.00 ] -0.008 [ -0.042 0.026 ] Ever had smoked 1.06 [ 0.97 1.17 ] -0.031 [ -0.059 -0.004 ] Heart disease 1.05 [ 0.95 1.15 ] 0.042 ** [ 0.015 0.069 ] Diabetes 1.07 [ 0.98 1.17 ] 0.013 [ -0.012 0.038 ] Hypertension 1.09 [ 0.98 1.21 ] 0.019 [ -0.012 0.050 ] Stroke 0.97 [ 0.86 1.08 ] 0.047 *** [ 0.012 0.083 ] Myocardial infarction 0.97 [ 0.87 1.09 ] 0.048 *** [ 0.012 0.083 ] Cancer 1.03 [ 0.93 1.15 ] 0.026 [ -0.008 0.059 ] Psychological disease 0.96 [ 0.84 1.09 ] 0.016 [ -0.018 0.050 ] Alzheimer's disease 1.08 [ 0.91 1.27 ] -0.009 [ -0.049 0.031 ] Lung disease 0.97 [ 0.87 1.08 ] -0.022 [ -0.050 0.006 ] Parkinson's disease 1.22 [ 0.88 1.67 ] 0.022 [ -0.093 0.137 ] 100 Functional Status Any ADL 0.90 * [ 0.82 0.99 ] 0.007 [ -0.018 0.032 ] Any IADL 0.91 * [ 0.83 1.00 ] -0.017 [ -0.041 0.007 ] Medicare Advantage Prescription Drug Plan (MA-PD) 1.05 [ 0.96 1.16 ] -0.131 *** [ -0.156 -0.107 ] Low Income Subsidy 1.21 *** [ 1.08 1.36 ] -0.404 *** [ -0.430 -0.377 ] Statin fill year 2006 ref ref 2007 1.11 [ 0.99 1.25 ] -0.092 *** [ -0.119 -0.065 ] 2008 1.16 ** [ 1.02 1.31 ] -0.150 *** [ -0.182 -0.119 ] 2009 1.34 *** [ 1.17 1.54 ] -0.191 *** [ -0.224 -0.158 ] 2010 1.39 *** [ 1.21 1.58 ] -0.207 *** [ -0.240 -0.172 ] 2011 1.52 *** [ 1.33 1.74 ] -0.240 *** [ -0.277 -0.204 ] 2012 1.63 *** [ 1.42 1.86 ] -0.311 *** [ -0.343 -0.281 ] * p<0.05, ** p<0.01, *** p<0.005 101 Table 3. The Yearly Trend for Good Adherence and the Average Daily Out-of-Pocket Costs, with the Interactions with Race and Income Level All the models are adjusted for age, gender, race, social economic status, health status, functional status, Medicare-Advantage Prescription Drug Plan, and low income subsidy. * p<0.05, ** p<0.01, *** p<0.005 Logit Model for Good Adherence (PDC ≥ 80) (1) Yearly trend since 2006 (2) Yearly trend interacts with race (3) Yearly trend interacts with income Odds Ratio [95% Conf. Interval] Odds Ratio [95% Conf. Interval] Odds Ratio [95% Conf. Interval] Yearly trend (from 2006) 1.08 *** [ 1.06 1.11 ] 1.08 *** [ 1.06 1.11 ] 1.08 *** 1.05 1.11 Yearly trend*Black 1.02 [ 0.96 1.09 ] Yearly trend*Hispanic 0.99 [ 0.93 1.05 ] Yearly trend*Income 20-40K 1.01 0.97 1.06 Yearly trend*Income >40K 1.01 0.96 1.06 Linear Regression Model for the Average Daily Out-of-Pocket Costs, $ (1) Yearly trend since 2006 (2) Yearly trend interacts with race (3) Yearly trend interacts with income Coefficient [95% Conf. Interval] Coefficient [95% Conf. Interval] Coefficient [95% Conf. Interval] Yearly trend (from 2006) -0.045 *** [ -0.050 -0.040 ] -0.0490 *** [ -0.055 -0.043 ] -0.028 *** -0.034 -0.023 Yearly trend*Black 0.023 *** [ 0.009 0.037 ] Yearly trend*Hispanic 0.018 ** [ 0.005 0.031 ] Yearly trend*Income 20-40K -0.031 *** -0.042 -0.019 Yearly trend*Income >40K -0.034 *** -0.048 -0.020 102 Supplemental Table. The Every-six-monthly Trend for Good Adherence and the Average Daily Out-of-Pocket Costs, with the Interactions with Race and Income Level All the models are adjusted for age, gender, race, social economic status, health status, functional status, Medicare-Advantage Prescription Drug Plan, and low income subsidy. * p<0.05, ** p<0.01, *** p<0.005 Logit Model for Good Adherence (PDC ≥ 80) (1) Six-month trend since 2006 (2) Six-month trend interacts with race (3) Six-month trend interacts with income Odds Ratio [95% Conf. Interval ] Odds Ratio [95% Conf. Interval ] Odds Ratio [95% Conf. Interval ] Every six-month trend (from 2006) 1.05 *** 1.03 1.05 1.04 *** 1.03 1.06 1.04 *** 1.03 1.06 Six-month trend*Black 1.01 0.98 1.04 Six-month trend*Hispanic 0.99 0.97 1.02 Six-month trend*Income 20- 40K 1.00 0.98 1.03 Six-month trend*Income >40K 1.00 0.98 1.03 Linear Regression Model for the Average Daily Out-of-Pocket Costs, $ (1) Six-month trend since 2006 (2) Six-month trend interacts with race (3) Six-month trend interacts with income Coefficie nt [95% Conf. Interval ] Coefficie nt [95% Conf. Interval ] Coefficie nt [95% Conf. Interval ] Every six-month trend (from 2006) -0.022 *** -0.025 -0.020 -0.024 *** -0.027 -0.021 -0.014 *** -0.017 -0.012 Six-month trend*Black 0.012 *** 0.005 0.018 Six-month trend*Hispanic 0.009 *** 0.003 0.015 Six-month trend*Income 20- 40K -0.015 *** -0.020 -0.009 Six-month trend*Income >40K -0.016 *** -0.024 -0.009 103 REFERENCES 1. Bansilal S, Castellano JM, Garrido E et al. Assessing the Impact of Medication Adherence on Long-Term Cardiovascular Outcomes. Journal of the American College of Cardiology 2016;68:789-801. 2. Silverman MG, Ference BA, Im K et al. Association Between Lowering LDL-C and Cardiovascular Risk Reduction Among Different Therapeutic Interventions: A Systematic Review and Meta-analysis. Jama 2016;316:1289-97. 3. Salami JA, Warraich H, Valero-Elizondo J et al. National Trends in Statin Use and Expenditures in the US Adult Population From 2002 to 2013: Insights From the Medical Expenditure Panel Survey. JAMA cardiology 2017;2:56-65. 4. Jena AB, Blumenthal DM, Stevens W, Chou JW, Ton TG, Goldman DP. Value of improved lipid control in patients at high risk for adverse cardiac events. The American journal of managed care 2016;22:e199-207. 5. Cherry SB, Benner JS, Hussein MA, Tang SS, Nichol MB. The clinical and economic burden of nonadherence with antihypertensive and lipid-lowering therapy in hypertensive patients. Value in health : the journal of the International Society for Pharmacoeconomics and Outcomes Research 2009;12:489-97. 6. Chowdhury R, Khan H, Heydon E et al. Adherence to cardiovascular therapy: a meta- analysis of prevalence and clinical consequences. European heart journal 2013;34:2940- 8. 7. De Vera MA, Bhole V, Burns LC, Lacaille D. Impact of statin adherence on cardiovascular disease and mortality outcomes: a systematic review. British journal of clinical pharmacology 2014;78:684-98. 8. Ho PM, Magid DJ, Shetterly SM et al. Medication nonadherence is associated with a broad range of adverse outcomes in patients with coronary artery disease. American heart journal 2008;155:772-9. 9. Pittman DG, Chen W, Bowlin SJ, Foody JM. Adherence to statins, subsequent healthcare costs, and cardiovascular hospitalizations. The American journal of cardiology 2011;107:1662-6. 104 10. Rodriguez F, Maron DJ, Knowles JW, Virani SS, Lin S, Heidenreich PA. Association Between Intensity of Statin Therapy and Mortality in Patients With Atherosclerotic Cardiovascular Disease. JAMA cardiology 2017;2:47-54. 11. Goldman DP, Joyce GF, Escarce JJ et al. Pharmacy benefits and the use of drugs by the chronically ill. Jama 2004;291:2344-50. 12. Karaca-Mandic P, Swenson T, Abraham JM, Kane RL. Association of Medicare Part D medication out-of-pocket costs with utilization of statin medications. Health services research 2013;48:1311-33. 13. Choudhry NK, Avorn J, Glynn RJ et al. Full coverage for preventive medications after myocardial infarction. The New England journal of medicine 2011;365:2088-97. 14. Chen SY, Shah SN, Lee YC, Boulanger L, Mardekian J, Kuznik A. Moving branded statins to lowest copay tier improves patient adherence. Journal of managed care pharmacy : JMCP 2014;20:34-42. 15. Lewey J, Shrank WH, Bowry AD, Kilabuk E, Brennan TA, Choudhry NK. Gender and racial disparities in adherence to statin therapy: a meta-analysis. American heart journal 2013;165:665-78, 678 e1. 16. Mann DM, Woodward M, Muntner P, Falzon L, Kronish I. Predictors of nonadherence to statins: a systematic review and meta-analysis. The Annals of pharmacotherapy 2010;44:1410-21. 17. Hess LM, Raebel MA, Conner DA, Malone DC. Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures. The Annals of pharmacotherapy 2006;40:1280-88. 18. Schultz JS, O'Donnell JC, McDonough KL, Sasane R, Meyer J. Determinants of compliance with statin therapy and low-density lipoprotein cholesterol goal attainment in a managed care population. The American journal of managed care 2005;11:306-12. 19. Jackevicius CA, Chou MM, Ross JS, Shah ND, Krumholz HM. Generic atorvastatin and health care costs. The New England journal of medicine 2012;366:201-4. 20. Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term persistence in use of statin therapy in elderly patients. Jama 2002;288:455-61. 105 21. Vinogradova Y, Coupland C, Brindle P, Hippisley-Cox J. Discontinuation and restarting in patients on statin treatment: prospective open cohort study using a primary care database. BMJ (Clinical research ed) 2016;353:i3305. 22. Lauffenburger JC, Robinson JG, Oramasionwu C, Fang G. Racial/Ethnic and gender gaps in the use of and adherence to evidence-based preventive therapies among elderly Medicare Part D beneficiaries after acute myocardial infarction. Circulation 2014;129:754-63. 23. Young JC, Sturmer T, Lund JL, Funk MJ. Predictors of prevalent statin use among older adults identified as statin initiators based on Medicare claims data. Pharmacoepidemiology and drug safety 2016;25:836-43.

Abstract (if available)

Abstract The studies presented in this dissertation discuss the potential value of PCSK9 inhibitors from the U.S. healthcare system perspective and propose an innovative drug pricing strategy to address the current health outcome lost due to the access barriers of this new drug. We also identified the trend of suboptimal use of statins in a national representative elderly population, which is helpful guide the policy makers for prioritizing the access to the PCSK9 inhibitors. ❧ To forecast the long-term benefit of PCSK9 inhibitors in a real-world U.S. population instead of a synthetic cohort, the first paper of my dissertation estimates the net monetary value of PCSK9 inhibitors in the older Americans (age 51 and older) under the FDA-approved eligibility when comparing with the status quo scenario -- the scenario of the current treatment strategies to manage hypercholesterolemia without the introduction of PCSK9 inhibitors. We also consider an extended eligibility scenario which includes patients with no pre-existing cardiovascular disease (CVD) but at high-risk, to inform the potential value when the patients gain greater access to this new therapy. We conducted simulations using the Future Elderly Model (FEM), an established dynamic microsimulation model, to project the lifetime outcomes for the U.S. population aged 51 or older. Efficacy estimates of Drugs from published meta-analysis studies of the early Phase III clinical trials were used to project changes in life expectancy, quality-adjusted life-years, and lifetime medical spending resulting from use of PCSK9 inhibitors. ❧ While PCSK9 could add a substantial societal value in a long run, there is still a paramount concern that the high prices of innovative prescription drugs like PCSK9, which places economic burden under a short-term budget impact model. Utilization restriction by payers results in a slow adoption of PCSK9 inhibitors, which creates barriers for patients who could benefit from the new therapy and lead to worse clinical outcomes for patients. Therefore, the outcome-based pricing for prescription drugs has been recently taken hold in the U.S. to ensure the prices we pay for the new drugs reflect the benefits we receive. The payers must find new ways to achieve cost control without limiting access to beneficial services and at the same time ensure enough incentives to the innovation for the manufacturers. While substantial manufacturer discounts could address the dilemma, it is unclear whether those would be flexible enough to reflect real-world outcomes, especially if the therapy’s long-term effectiveness is better than initially expected or larger absolute risk reductions in major ASCVD event rates given the higher baseline event rate in the real world. Therefore, we propose a payment model -- a three-part pricing model -- incorporating the real-world evidence to encourage payers to pay for services that provide expected or better outcomes. The aim of our second paper is to compare the value of our three-part pricing model comparing with the current one-price model (status quo) by using PCSK9 inhibitors as an example. The FEM is used to estimate the numbers of averted cardiovascular events in each year in those with existing CVD and/or FH and elevated LDL-C. We use the effectiveness estimates from a recently published clinical trial (FOURIER study) which focused on long-term cardiovascular outcomes in high risk patients. Among the eligible population, two PCSK9 uptake functions are used to estimate the numbers of patients actually be treated. To make the manufacturers be indifferent between the two pricing models, the drug price for our three-part pricing model will be calibrated to ensure the overall revenue during the patent protection period will be the same as the status quo. ❧ Given the access expansion of PCSK9 inhibitors is challenging and the hurdle of pricing strategy for this novel class of drugs hasn’t been solved, on the other hand it would be important to understand the societal value of improving the adherence to statins. The third paper of this dissertation aim to evaluate the longitudinal trends of statin adherence in the U.S elderly population and identify the factors which are associated with the better statin adherence.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Disease modifying treatments for Alzeimer's disease: modeling, value assessment and eligible patients

PDF

Documenting the impact of under-treatment for chronic disorders

PDF

The impact of treatment decisions and adherence on outcomes in small hereditary disease populations

PDF

Three essays on estimating the effects of government programs and policies on health care among disadvantaged population

PDF

Three essays in health economics

PDF

Characterization of health outcomes in patients with hemophilia A and B: Findings from psychometric and health economic analyses

PDF

Understanding primary nonadherence to medications and its associated healthcare outcomes: a retrospective analysis of electronic medical records in an integrated healthcare setting

PDF

The price and clinical impact of specialty drug management in the US healthcare system

Pdf

Alzheimer's disease: risk factors, value, and alternative payment models

PDF

Health economics and outcomes research for informed decision making in rapidly evolving therapeutic areas

PDF

Evaluating treatment options for metastatic, castration-resistant prostate cancer: a comprehensive value assessment

PDF

Crohn’s disease: health outcomes and resource utilization in the biologic era

PDF

Chronic eye disease epidemiology in the multiethnic ophthalmology cohorts of California study

PDF

Molecular signature of aggressive disease and clonal diversity revealed by single-cell copy number analysis of prostate cancer cells across multiple disease states

PDF

Investigating the modulatory mechanisms of ILC2s as a therapeutic strategy in the context of health and disease

Asset Metadata

Creator Cheng, Wei-Han (author)

Core Title The value of novel antihyperlipidemic treatments in the U.S. healthcare system: Reducing the burden of cardiovascular diseases and filling the gap of low adherence in statins

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Health Economics

Publication Date 07/26/2018

Defense Date 05/15/2018

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag adherence,cardiovascular disease,OAI-PMH Harvest,PCSK9 inhibitors,statins

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Goldman, Dana (committee chair), Joyce, Geoffrey (committee member), Myerson, Rebecca (committee member)

Creator Email loriwendyld@gmail.com,weihanch@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-33824

Unique identifier UC11671508

Identifier etd-ChengWeiHa-6519.pdf (filename),usctheses-c89-33824 (legacy record id)

Legacy Identifier etd-ChengWeiHa-6519.pdf

Dmrecord 33824

Document Type Dissertation

Format application/pdf (imt)

Rights Cheng, Wei-Han

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA