Source: US Congressional Budget Office
This week, five analysts from CBO’s Health Analysis Division are presenting their work at the 11th Annual Conference of the American Society of Health Economists (“ASHEcon”) in Austin, Texas. These presentations are part of our engagement with the broader research community, which improves the quality of our analysis and makes our methods and findings more transparent and available. They show the wide range and high quality of the Health Analysis Division’s work:
Outcomes Associated with Initiation of Psychosocial Therapy With and Without Medications for Opioid Use Disorder
Buprenorphine, Psychosocial Treatment and Recreational Marijuana Laws: New Evidence on State and Federal Policy Initiatives to Address the Opioid Crisis
Ryan Mutter (CBO), Donna Spencer (OptumLabs), Jeffrey McPheeters (OptumLabs)
Providing treatment to individuals with opioid use disorder (OUD) could reduce the demand for opioids and alter the course of the opioid crisis. Evidence-based treatment for OUD is available. Individuals with OUD can be treated with medications that have been approved by the Food and Drug Administration. Psychosocial treatment is also recommended for individuals with OUD.
Patients in treatment for OUD are at risk for numerous adverse outcomes, including opioid-involved overdose. Continued use of prescription opioids is also of potential concern because it can present a therapeutic challenge to the treatment of OUD. In addition, intentional self-harm is a potential adverse outcome because OUD is a risk factor for suicide. There is considerable variation in the treatment that patients with OUD receive. However, there has been little research on variations in outcomes by type of OUD treatment initiated. This study examines adverse outcomes among patients who initiate treatment for OUD with psychosocial therapy only compared to those who start treatment with both psychosocial therapy and pharmacotherapy.
Data and Methods
This study used a large, national claims database of individuals with commercial insurance or in Medicare Advantage plans with linked county-level Area Health Resource File (AHRF) data from the OptumLabs Data Warehouse (OLDW). Adults with OUD who started treatment with psychosocial therapy only or the combination of psychosocial therapy and pharmacotherapy were identified between July 2010 and March 2019 and were observed for 3 to 18 months following treatment start. Cox proportional hazards regression models were used to estimate the association of treatment type initiated with opioid overdose and self-harm (separately). Logistic regression was used to estimate the association of treatment type started with prescription opioids filled within 90 days of treatment initiation. All models controlled for patient demographic, clinical, and area characteristics.
Relative to patients who initiated treatment with psychosocial therapy only, patients who initiated treatment with both pharmacotherapy and psychosocial therapy had a lower hazard of having an overdose encounter (HR=0.86, p<0.03) or a self-harm encounter (HR=0.82, p<0.02). Relative to patients who initiated treatment with psychosocial therapy only, patients who initiated treatment with both pharmacotherapy and psychosocial therapy also had lower odds of having a prescription opioid filled in the first 90 days of treatment (OR=0.26, p<0.01).
Patients with certain behavioral and physical health comorbidities had higher hazards of having an overdose or self-harm encounter. Patients with moderate or high morphine milligram prescription opioid fills, potentially problematic opioid fills, or indicators of possible injection drug use in the baseline had higher odds of having a prescription opioid filled within the first 90 days of initiating treatment for OUD.
Lower risk of opioid overdose, intentional self-harm encounters, and prescription opioid fills are associated with initiation of treatment with evidence-based pharmacotherapy and psychosocial treatment compared to psychosocial treatment alone. Patients whose physical and behavioral health comorbidities and prescription opioid use put them at risk for adverse outcomes may benefit from additional support from providers.
Changes in Prescription Drug Sales and Net Prices in Part D Following First Generic Entry
Prescription Drug Prices and Manufacturer Behavior
Colin Baker, Rachel Fehr, Scott Laughery (Presenting Author)
To quantify changes in sales volumes, retail and net prices, and spending in Medicare Part D after brand-name prescription drugs are first subject to generic drug competition.
Consumers benefit from lower prices when a brand-name prescription drug faces competition from equivalent generic drugs, and hastening generic drug market entry is part of many proposals aimed at making drugs more affordable. Previous estimates of the price change when a brand-name drug faces generic competition have generally relied on retail data that omit rebates or discounts. This analysis uses yearly Part D data that include confidential drug-level manufacturer rebates and coverage gap discounts reported to CMS, allowing us to estimate average changes in net prices and net spending for our sample of drugs.
Main Outcomes and Measures
We use Part D claims data paired with confidential drug-level data on manufacturer rebates to estimate changes in total quantities purchased, market shares for brand and generic products, retail and net prices, and spending for up to four years before and four years after brand-name drugs lose market exclusivity. We identify 326 drugs whose brand-name version experienced a loss of exclusivity (LOE) between 2011 and 2019. For that sample of compounds, we estimate brand-name and generic volumes, spending, and prices in each year.
We estimate that, for the average drug, prescription volume in Part D (including brand-name and generic versions) rose by roughly half during the first four years after LOE, driven by rising generic prescriptions in each of those years. Brand-name prescriptions (which include prescriptions for authorized generic versions under our definition), unsurprisingly, fell steadily. During the four years immediately preceding LOE, average estimated prices for brand-name drugs – both the retail prices paid at the point of sale and net prices accounting for manufacturer rebates and discounts – increased by about 10 percent per year. In the year that LOE occurred, prices for generic versions were similar to the net prices of their reference brand-name products, both averaging approximately 70 percent of the reference brand-name product’s retail price. During the first four years following LOE, average brand-name retail prices declined by about 3 percent annually, while brand-name net prices rose about 3 percent annually, and the average generic price fell by more than 25 percent annually.
Discussion and Conclusion
The estimates reveal different patterns for retail versus net drug prices around the time that brand-name drugs lose exclusivity. Our results offer a basis for estimating the effects of policies aimed at hastening generic drug market entry.
CBO’s Simulation Model of New Drug Development
Empirical Modeling of New Pharmaceutical Development
This paper presents the Congressional Budget Office’s simulation model for analyzing legislative proposals that may substantially affect new drug development. The model is used to analyze policies that may affect expected future profits, expected development costs, financing costs or the number of drug candidates available for human clinical trials. Given changes in decisions to enter at each stage, the model estimates when and by how much the number of new drugs entering the market will change.
The model is based on a stylized representation of the pharmaceutical decisionmaking process. A firm is projected to continue development of a drug if expected returns exceed expected costs. The model considers the decision problem of whether to enter one of four development stage for pharmaceuticals; pre-clinical, phase I, phase II and phase III. For simplicity the model assumes that the firm’s decision problem at each stage is independent of the previous stage. The model’s parameter values are derived from both estimation and calibration procedures. The model uses revenue estimates calculated using nonpublic Medicare Part D data. Those data include information on the rebates paid by the manufacturing firms, allowing those rebates to be netted out of revenue. Using data from 2010 to 2018, CBO estimates how drug revenue varies with time on market. CBO uses results from DiMasi, Grabowski, and Hansen (Journal of Health Economics 2016) to estimate development costs. The agency uses a Roy model to combine information on revenue and cost. That model accounts for selection and correlation in the observed revenue and cost data (Heckman and Honore, Econometrica, 1990). CBO calibrated entry probabilities and other parameters on the basis of results presented in Blume-Kohout and Sood (Journal of Public Economics, 2013), DiMasi (Impact Report, 2013), and Khmelnitskaya (Working Paper, 2020).
To illustrate the implications of the model, the paper considers a legislative change that lowers expected returns for the top-earning drugs and increases the financing costs of drugs. A 15 percent to 25 percent reduction in expected returns for drugs in the top quintile of expected returns is associated with a reduction of 4 drugs in the first decade following the introduction of the policy, 28 drugs in the second decade and 43 drugs in the third decade. CBO expects 440 new drugs to enter the market each decade. This corresponds to a long term elasticity of 0.55, that is for a 10% reduction in expected revenue there is 5.5% reduction in the number of drugs entering the market. Alternatively, CBO estimates that approximately 1 drug is lost for each $4.6b in lost expected revenue. In CBO’s assessment, those estimates are in the middle of a wide distribution of potential effects. The effects could be smaller if expenditures in late-phase human trials are larger, for example. Alternatively, the effects could be larger if the cost of capital is larger. CBO did not predict what kind of drugs would be affected or analyze the effects of forgone innovation on health outcomes.
Health Spending By Age in the Nongroup and Small Group Markets
Enrollment and Spending Patterns in the Nongroup Market
Caroline Hanson and Alexandra Minicozzi
The Affordable Care Act (ACA) introduced regulations affecting the nongroup and small group markets, where individuals and small firms, respectively, purchase health insurance. Those regulations include modified community rating, guaranteed issue, mandated essential health benefits, and risk adjustment. Modified community rating meant that health insurance issuers were now allowed to vary rates only based on age (by no more than a 3:1 ratio), number of family members, geographic area, and tobacco use. Moreover, issuers that offer individual policies in a state now must offer all coverage products to all eligible individuals, regardless of health status. Policy makers were (and to some extent continue to be) concerned that those regulatory changes would lead to large premium increases, major shifts in who enrolled, and possible market instability. We provide a more complete understanding of the composition of these markets, in terms of who enrolls and the healthcare they consume, which is important for understanding how the market is functioning under current law and for accurately predicting how different policies – like changing the age rating curve or reducing the age of Medicare eligibility – might affect premiums and enrollment.
This research uses the EDGE Limited Data Set, which includes claims covering almost all enrollment in nongroup and small group plans participating in risk adjustment under the ACA. For our study period (2017-2019), it consists of almost 65 million member-years. While this data tells us nothing about markets or enrollees prior to the ACA, its extensive and accurate information on spending and enrollment makes it the ideal data source to estimate mean allowed spending, including both insurer-paid spending and patient cost-sharing, per member-year of enrollment by age, sex, and market segment. We present smoothed estimates using the Whittaker-Henderson graduation. We also explore spending on specific types of services by age, sex, and market to understand what drives the trends in average spending.
We find that a higher share of nongroup market enrollees are adults, aged 55 or older, and female than in the small group market. Younger male enrollees in the nongroup market spend relatively more than younger male enrollees in the small group market, both in terms of spending by age across markets and in terms of relative spending by age within each market. Younger male enrollees have higher average spending for a broad range of services. By contrast, female enrollees of all ages in the nongroup market spend less than in the small group market, a result that is partially driven by lower enrollment by females with obstetric claims.
Ultimately, we conclude that the sex composition of the nongroup market combined with the nongroup market having relatively high spending young and middle-aged male enrollees, results in a spending curve that is flatter than the curve for employment-based coverage, and for most ages, flatter than the age-rating curve used to set premiums under the ACA. As a result, a policy that reduced the age of Medicare eligibility would likely slightly increase premiums in the nongroup market but decrease premiums for employment-based coverage.
Changes in Utilization Associated with Local Increases in Insurance Coverage Post-ACA
Spillover Effects of Medicaid Expansions on Healthcare Access and Utilization
Rebecca Sachs, Noelia Duchovny, Josh Varcie, Chapin White
A substantial body of research has examined the impact of expanding health insurance coverage on the newly insured, including their utilization of health care services and health outcomes. However, to fully understand the economic impact and budgetary implications of future coverage reforms, it is critical to study the spillover impacts of expansions on the continuously insured.
In this research, we quantify whether and how the coverage expansions from the Affordable Care Act (ACA) affected the amount of care provided to people whose enrollment in health insurance was not directly affected by the ACA. We currently examine Medicare enrollees, using a 100% Medicare Claims File. Future work will expand our analysis to cover the subsets of Medicaid and commercial enrollees whose eligibility and enrollment did not change because of the ACA, using Medicaid claims and IBM Watson MarketScan data.
Specifically, we measure the quantity and intensity of inpatient, medical provider, and outpatient events for our sample of beneficiaries in a given public use microdata area (PUMA) each year. To measure the size of the ACA coverage expansions, we exploit variation both in eligibility for Medicaid and eligibility for nongroup premium tax credits.
Our results illustrate how every percentage point increase in enrollment in Medicaid and nongroup coverage affects utilization among Medicare enrollees. Additional analysis shows how these offsets vary among different types of services. Preliminary analysis explores how health care market characteristics affect the degree of utilization spillovers.
Phillip L. Swagel is CBO’s Director.