## Abstract

Many state Medicaid officials are concerned about rising prescription drug spending, particularly drugs approved through the Food and Drug Administration's (FDA) accelerated approval pathway. The authors examined how much of Medicaid programs' accelerated approval spending is attributable to products that have demonstrated clinical benefits versus those that have not. Their findings provide support for states' concerns that pharmaceutical companies often fail to complete their required postapproval confirmatory studies within the FDA's requested timeline. But the findings also highlight one issue that policy stakeholders have not yet devoted substantial attention to: the use of surrogate endpoints involved in the postapproval confirmatory studies for most of the products in this study's sample. The granularity of the study's results enabled an analysis of the impact of different policy recommendations on both the accelerated approval pathway and Medicaid programs. These findings inform the current policy debate, suggesting that policy stakeholders might focus attention on products converting their approval on the basis of surrogate outcomes rather than on clinical outcomes.

The impact of drug coverage and pricing policies on state Medicaid programs and state budgets is an understudied area within prescription drug pricing debates. Unlike the federal government, nearly all states have a balanced budget requirement, though the strength of the requirements varies (Gamage 2010), meaning that unexpected increases in prescription drug expenditures may require balancing with cuts elsewhere, either within the Medicaid program or within other portions of state budgets (such as education or infrastructure). Yet state Medicaid programs are often constrained in their abilities to manage prescription drug coverage and spending, which could exacerbate budgetary challenges. Consequently, many states are concerned about rising prescription drug spending.

Under the “grand bargain” (Young and Norton 2018) of the 1990 Medicaid Drug Rebate Program, state Medicaid programs that choose to cover prescription drugs (as all have chosen to do) must cover essentially all Food and Drug Administration (FDA)-approved drugs, with limited statutory carve-outs for cosmetic products, drugs used to “promote fertility,” and erectile dysfunction drugs (42 U.S.C. § 1396r-8(d)). In exchange, pharmaceutical companies must provide Medicaid programs with two types of discounts for drugs sold under the program: a large mandatory minimum rebate (set in 1990 at 12.5% but currently set at 23.1%) from the drug's average manufacturer price as well as a rebate if the drug's price increases faster than inflation, to insulate Medicaid from those increases (42 U.S.C. § 1396r-8(c)). These rebates are shared between the federal government and the states. In most cases, these two rebates have served to protect the safety net program as it attempts to provide comprehensive care for its beneficiaries, and Medicaid is typically able to obtain much lower prices than Medicare can for the same drugs (CBO 2021).

Despite these savings, Medicaid lacks the ability to use key strategies often used by other insurers to manage spending on high-cost products. Medicaid programs cannot limit use of high-cost drugs among beneficiaries through closed formularies (by excluding some drugs from coverage entirely) or cost-sharing obligations (by requiring patients to pay significant out-of-pocket amounts for particular drugs). Patient cost-sharing is capped at nominal amounts where it is permitted at all; for many products and patients, cost-sharing is not charged. Although Medicaid programs do have the ability to use preferred drug lists that may mimic the function of other utilization management tools, such as prior authorizations or step therapy, these lists may be less impactful in driving utilization when drugs lack alternative treatments, as is the case with drugs approved under the accelerated approval pathway.

Products approved through the FDA's accelerated approval pathway have recently faced increased scrutiny by payers because of their high prices and variable or unproven benefits. First created in 1992 in response to the HIV/AIDS epidemic, this pathway aims to speed patient access to drugs for “serious or life-threatening” diseases by enabling the FDA to approve them on the basis of surrogate endpoints that are “reasonably likely” to predict clinical benefit rather than on the basis of true clinical outcomes (21 U.S.C. § 356(c)(1)(A)). For instance, many early HIV/AIDS drugs were approved on the basis of their ability to lower patients' viral loads or increase patients' immune function (FDA 2015). These surrogate endpoints were soon found to be highly predictive of the drugs' ability to achieve meaningful clinical benefits in slowing the progression of HIV and reducing mortality. For other products approved under the accelerated approval pathway, surrogate endpoints may not be predictive of clinical endpoints. Thus, manufacturers of accelerated approval drugs must conduct postapproval confirmatory studies to verify clinical benefits. However, state Medicaid programs are legally obligated to cover accelerated approval drugs (CMS 2018) while studies continue.

The FDA may be beginning to increase enforcement under the pathway: between December 2020 and June 2022, 14 cancer indications have been withdrawn under the program for failure to confirm a clinical benefit. Before that time, only 7 cancer indications had been withdrawn throughout the program's life span (FDA 2022).

In response to these shortcomings of the accelerated approval pathway, a range of policy actors have begun to investigate potential reforms. In 2017, the Massachusetts Medicaid program applied for a waiver permitting it to exclude accelerated approval drugs from coverage, though that waiver was denied (Bagley and Sachs 2018). In early 2021, the Medicaid and CHIP Payment and Access Commission (MACPAC) recommended altering the terms on which Medicaid provides reimbursement for accelerated approval drugs, by increasing the mandatory minimum Medicaid rebates for accelerated approval drugs either while confirmatory studies are in progress or after a particular period of time has elapsed without the confirmatory studies' completion, to create an incentive for manufacturers to complete confirmatory studies (MACPAC 2021).

In earlier work, some of the authors of this article demonstrated that although accelerated approval drugs represent a small proportion—less than 1%—of Medicaid drug utilization, they represented between 6.37% and 9.07% of annual Medicaid drug spending net of rebates between 2015 and 2019 (Sachs et al. 2021). This article further extends the body of research on the accelerated approval pathway in the context of Medicaid by classifying spending by product category, determining how much of Medicaid programs' accelerated approval spending is attributable to products that have demonstrated clinical benefits versus those that have not. The granularity of these results enables us to analyze the impact of different policy recommendations on both the accelerated approval pathway and Medicaid programs, and to inform the current policy debate.

## Methods

### Sample

We reviewed the FDA's Center for Drug Evaluation and Research (CDER) Drug and Biologic Accelerated Approvals report (as of December 31, 2021) (FDA 2021a) to identify products that were granted accelerated approval between January 2009 and December 2016. The sample included accelerated approvals of both new products and new indications for previously approved products (Naci, Smalley, and Kesselheim 2017). The sample was restricted to product-indication pairs approved prior to and including December 2016, to allow at least five years for the completion of postapproval confirmatory studies.

### Identification of Postapproval Confirmatory Studies

We reviewed approval letters in the Drugs@FDA database (FDA 2021a) to identify postapproval confirmatory studies requested for each product-indication pair at the time of accelerated approval. Consistent with earlier research (Naci, Smalley, and Kesselheim 2017), we excluded postapproval requirements that were solely focused on safety (and not efficacy) evaluations. For each postapproval confirmatory study, we extracted the primary and secondary endpoints specified by the FDA (when available) and the “Final Report Submission” date (when the day was not reported in the approval letter, we assigned the last day of the month). Next, we used the trial name and number (when reported in the approval letter), product name, and medical condition to screen ClinicalTrials.gov for each applicable postapproval confirmatory study. We documented the primary and secondary endpoints for each registered postapproval confirmatory study.

### Identification of Published Postapproval Confirmatory Studies

To identify publications of each completed postapproval confirmatory study, we reviewed the linked publications indexed on ClinicalTrials.gov (Naci, Smalley, and Kesselheim 2017; NIH 2021). To supplement publications indexed on ClinicalTrials.gov, we searched PubMed and Google with the ClinicalTrials.gov identification number and/or a combination of the product name and medical condition (Gyawali, Rome, and Kesselheim 2021; Naci, Smalley, and Kesselheim 2017). We reviewed each publication to confirm the primary and secondary endpoints that were studied and to determine which endpoints (clinical, surrogate, or both) were met.

### Identification of Changes in Approval Status

We reviewed the CDER Drug and Biologic Accelerated Approvals report (as of December 31, 2021) (FDA 2021b) to identify the current approval status—converted, withdrawn, or not yet converted—of each product-indication pair. For each product-indication pair that had a change in status, we extracted the “Full Approval Conversion-Withdrawal Date” from the CDER report and the date when the manufacturer submitted their application to the FDA from the corresponding approval letter in the Drugs@FDA database. Submissions were considered “on time” if the application was submitted on or before the “Final Report Submission” date specified in the original accelerated approval letter.

For product-indication pairs that were not yet converted, we used the “Final Report Submission” date listed in the accelerated approval letter to determine whether a postapproval confirmatory study was within or past FDA timelines. Next, we searched the FDA Postmarket Requirements and Commitments database (Gyawali, Hey, and Kesselheim 2019; Gyawali, Rome, and Kesselheim 2021; Naci, Smalley, and Kesselheim 2017) to identify and document the current status (ongoing, submitted, delayed, or terminated) of postapproval confirmatory studies. Similar to previous studies (Naci, Smalley, and Kesselheim 2017), we identified and recorded the status (recruiting, withdrawn, suspended, completed, terminated) of confirmatory studies registered in ClinicalTrials.gov for those requirements that were not reported in the FDA database (three studies for two product-indication pairs).

### Medicaid Spending

We used state drug utilization data (CMS 2021) to estimate gross and net (postrebate) Medicaid spending from 2015 through 2019. We focused our analysis on 2015 through 2019 to account for changes to the size and composition of Medicaid enrollment following the implementation of the Affordable Care Act (ACA), which expanded Medicaid eligibility to populations previously ineligible for coverage. States had the option of adopting the ACA's Medicaid expansion starting in 2014. Thus far, 38 states and the District of Columbia have expanded their Medicaid programs (KFF 2022), significantly increasing the number of adult beneficiaries in those states. The ACA also made several changes to the Medicaid Drug Rebate Program, including increasing the mandatory minimum branded drug rebate from 15.1% to 23.1% and extending rebates to products covered by managed care organizations (Sachs et al. 2021).

We defined gross spending as the total amount reimbursed for included products as reported in the national state drug utilization data. We used a three-step approach to estimate net spending. First, we applied the mandatory minimum rebate of 23.1% to the products included in our sample or 17.1% to the products indicated for blood clotting or exclusively for pediatric populations (42 U.S.C. § 1396r-8(c)). Second, we identified the first year in which each product was observed (from 1992, the inception of the accelerated approval pathway, through 2019) and estimated the median unit price. The median unit price was inflation-adjusted by 2% in each subsequent year. Third, we accounted for the Medicaid inflation penalty by deducting any amount reimbursed above the inflation-adjusted unit price, weighted by total unit fills (Sachs et al. 2021). Gross and net spending amounts were adjusted to 2019 dollars using the Consumer Price Index for All Urban Consumers (Dunn, Grosse, and Zuvekas 2018; Sachs et al. 2021).

### Statistical Analysis

We used descriptive statistics to assess characteristics of postapproval confirmatory studies and gross and net Medicaid spending. We estimated quarterly spending across three product categories—not yet converted, converted based on surrogate endpoints alone, and converted based on clinical endpoints (including products that met clinical endpoints alone and both clinical and surrogate endpoints)—which were based on a product's approval status in each given quarter and year (quarter 1 2015 through quarter 4 2019). We then summarized spending for each year and for each product category. Since indication-level spending is not available in the state drug utilization data, we estimated spending based on the earliest accelerated approval and conversion dates for products with multiple indications. All analyses were conducted in Stata version 14 and Microsoft Excel.

### Sensitivity Analysis

Because drug prices do not vary by indication, in sensitivity analyses we accounted for products that had both accelerated approval and nonaccelerated approval indications (e.g., accelerated approval indication was converted and/or an indication was approved via the traditional pathway) by stratifying gross and net Medicaid spending by quarters in which a product was exclusively marketed for an accelerated approval indication versus quarters in which a product was marketed for at least one nonaccelerated approval indication. Similar to our primary analysis, we then summarized spending for each year.

## Results

### Characteristics of Accelerated Approval Products

Between January 2009 and December 2016, 47 products with 55 indications were granted accelerated approval (table 1). Thirty-six products were indicated for the treatment of cancer, of which 14% (n = 5) had multiple cancer indications (appendix tables 1–3). Eleven products were approved for conditions including chronic iron overload, multidrug-resistant tuberculosis, primary biliary cholangitis, and Duchenne muscular dystrophy (appendix tables 1 and 3).

### Characteristics of Products with a Change in Approval Status

Thirty-eight product-indication pairs (34 cancer indications and four noncancer indications) had completed their confirmatory studies and were converted as of December 31, 2021 (table 1). Time from accelerated approval to conversion ranged from 0.48 years to 9.55 years, with a median of 3.08 years (interquartile range [IQR] 1.89, 4.35) across included products. Cancer product-indication pairs had faster median time to conversion compared to noncancer product-indication pairs (2.85 years [IQR 1.86, 4.10] versus 6.41 years [IQR 4.61, 8.01]). Forty-eight postapproval confirmatory studies were completed for all 38 product-indication pairs. Of these, 73% (n = 35) were submitted to the FDA within the agency's requested time frame. Time from accelerated approval to confirmatory study submission ranged from 0.18 years to 8.72 years, with a median of 2.41 years (IQR 1.38, 4.02). Compared to noncancer product-indication pairs, cancer product-indication pairs had a faster median time to submission (2.27 years [IQR 1.33, 3.57] versus 5.67 years [IQR 4.49, 6.67]) but a smaller proportion of on-time submissions (71% versus 86%).

Manufacturers studied a variety of endpoints to confirm benefit (table 1 and appendix table 1). For example, most of the cancer product-indication pairs (59%; n = 20) assessed multiple endpoints: disease response, overall survival and progression-free survival. Among noncancer product-indication pairs, half examined either disease response endpoints alone (50%, n = 2) or a combination of disease response and other surrogate endpoints (50%, n = 2). Of the 38 converted product-indication pairs, 42% (n = 16) confirmed clinical endpoints (three confirmed clinical endpoints alone, and 13 confirmed both clinical and surrogate endpoints), and 58% (n = 22) confirmed surrogate endpoints alone. Findings were similar for cancer product-indication pairs; however, all noncancer product-indication pairs confirmed clinical endpoints (two confirmed clinical endpoints alone, and two confirmed both clinical and surrogate endpoints).

In February 2020, April 2021, and July 2021, three cancer product-indication pairs included in our sample (olaratumab, atezolizumab, and romidepsin, respectively) were voluntarily withdrawn from the market because confirmatory studies failed to meet primary endpoints (appendix table 2).

### Characteristics of Products Not Yet Converted

As of December 31, 2021, 16% (n = 7) and 64% (n = 7) of cancer and noncancer product-indication pairs, respectively, had not yet been converted. Overall, 64% (n = 9) of these product-indication pairs were past FDA timelines, with most postapproval confirmatory studies delayed (failed to meet submission or completion milestones) (appendix table 3). Of the five product-indication pairs that were within FDA timelines, three had confirmatory studies that were ongoing (one for bedaquiline, one for idursulfase, and two for obeticholic acid), and four had postapproval requirements that were delayed (one each for idelalisib, idursulfase, and obeticholic acid because of enrollment challenges and delayed report submissions) or terminated (one for nivolumab as a result of the changing treatment landscape for classical Hodgkin lymphoma).

### Medicaid Spending

Gross Medicaid spending on products receiving accelerated approval between 2009 and 2016 increased from $1.2 billion in 2015 to$2.9 billion in 2019. As expected, gross spending on products not yet converted declined from $1.0 billion to$707.1 million as products completed required follow-up studies. In 2019, spending on products converted based on surrogate endpoints alone reached $1.6 billion, while spending on products converted based on clinical endpoints reached$567.8 million (figure 1, panel A). Patterns were similar when considering net spending, with net Medicaid spending on products receiving accelerated approval increasing from $923.8 million in 2015 to$2.2 billion in 2019 (figure 1, panel B). When assessing gross and net spending on converted products only, those that received approval based on surrogate endpoints alone accounted for more than 70% of all spending on converted products in each year (figure 2).

### Sensitivity Analysis

When restricting our spending analysis to quarters in which the drug was only available via accelerated approval indications, we find that gross spending was $440.9 million in 2015 and$377.7 million in 2019 (appendix figure 1, panels A and B).

## Discussion

These findings provide important context for ongoing analyses of the accelerated approval program and in particular its relationship to state Medicaid programs. To date, key policy stakeholders have primarily expressed concern publicly about companies' incentives to complete confirmatory studies and the limited evidence of efficacy available in the meantime, when coupled with the high prices of the products in question. If companies are able to set prices with only limited constraints before they have demonstrated clinical benefit, their incentives to complete confirmatory studies may be muted, particularly as those studies may fail to confirm a clinical benefit and, in at least some cases, lead to the product's or indication's withdrawal from the market (Mezher 2021). Our findings provide support for these concerns, given the magnitude of state spending on products that have been on the market for several years without completing postapproval confirmatory studies. We observed that one in four product-indication pairs in our sample of converted products was past the FDA's specified submission timeline, suggesting that concerns about timely completion of postapproval confirmatory studies are warranted.

Our analysis also illuminates the ways in which different groups of accelerated approval drugs may contribute to Medicaid spending. Our findings cast light on one issue that policy stakeholders have not yet devoted substantial attention to: the use of surrogate endpoints involved in the postapproval confirmatory studies for a significant number of these products. We found that most indications (22 out of 38) completed their confirmatory studies on the basis of evaluating surrogate endpoints, typically progression-free survival for oncology indications, rather than clinical endpoints (e.g., overall survival). Importantly, we also found that a significant share of all state spending on accelerated approval drugs is attributable to this set of products. Whether progression-free survival is a marker of overall survival among individuals with cancer is controversial and likely varies by product and indication (Prasad et al. 2015). However, it is concerning that overall survival is not an endpoint in many of these studies, given the severity of illness and the clear importance of overall survival for patients. Yet policy stakeholders have identified few, if any, strategies that would address Medicaid spending in this category.

This study is not without some limitations. First, five years of follow-up for the most recently approved drugs may not have been sufficient for manufacturers to complete and submit postapproval confirmatory studies, especially since the FDA's own timelines often exceeded five years for the products studied. Second, our analysis relied on publicly available documents, which do not capture discussions between the FDA and manufacturers regarding the planning of or changes to confirmatory studies (e.g., study design, endpoints, or timing) (Gyawali, Hey, and Kesselheim 2019; Naci, Smalley, and Kesselheim 2017). Third, we did not assess postapproval confirmatory study methodology or potential biases; rather, we reviewed study findings to determine the endpoints and benefits that were met (Naci, Smalley, and Kesselheim 2017). Fourth, state drug utilization data files are restricted to medications dispensed in outpatient settings, which may result in the underestimation of Medicaid spending on certain accelerated approval products included in our sample. Last, since the average manufacturer prices and product-specific rebates are not publicly available, we estimated these values using the median unit price and applied the mandatory minimum and inflation-based rebates to determine trends in spending. Unit prices may be less reliable for infused or injected drugs as compared with orally administered drugs.

Nonetheless, our findings have nuanced implications for policy reforms to the FDA accelerated approval pathway. First, many of the products in our sample had multiple indications, some having received accelerated approval and others receiving approval through the traditional pathway. Because state drug utilization data files do not report spending at the product-indication level, our primary analysis estimated spending based on the earliest accelerated approval and converted indication for products with multiple indications. However, we also completed a sensitivity analysis focusing on products that exclusively had accelerated approval indications. The need for this sensitivity analysis should signal to policy makers that it may be difficult to reform the payment system for products with multiple (including traditional) indications, unless those policy makers are able to obtain the data required to separate spending by indication, limiting the potential savings from these reform proposals.

Second, policy options that have been explored by MACPAC and other experts have included increasing rebates on accelerated-approval products during the period when confirmatory studies are being conducted to incentivize more timely completion of studies. Indeed, we find some evidence for the potential of such policies to reduce Medicaid spending on accelerated approval drugs, given the significant share of product-indication pairs exceeding the FDA's specified submission timeline. However, we also find that proposals that link changes in reimbursement to the completion of FDA-required confirmatory studies would only address a subset of Medicaid spending on accelerated-approval products.

Finally, in contrast to proposals that target products that have not yet completed their confirmatory studies, relatively few, if any, proposals would impact the class of products converting their approval on the basis of surrogate outcomes rather than clinical outcomes. Existing literature has questioned this approach in the oncology context in regard to progression-free survival, noting that the agency's allowance of this practice means that “patients and physicians continue to lack information about whether the cancer drug improves survival or quality of life, which is essential in the benefit-risk evaluation for clinical decisionmaking” (Gyawali, Hey, and Kesselheim 2019: 910). More policy attention might be paid to products with accelerated approval confirmed by a surrogate outcome, with consideration given to whether potential reforms to the FDA's operation of the accelerated approval process or to insurance reimbursement for the resulting products would be better fits for this category.

## References

Bagley, Nicholas, and Sachs, Rachel E..
2018
. “
Limiting State Flexibility in Drug Pricing
.”
New England Journal of Medicine
379
, no.
11
:
1002
4
.
CBO (Congressional Budget Office)
.
2021
. “
A Comparison of Brand-Name Drug Prices Among Selected Federal Programs
.”
February
18
. https://www.cbo.gov/publication/56978.
CMS (Centers for Medicare and Medicaid Services)
.
2018
. “
State Medicaid Coverage of Drugs Approved by the FDA Under Accelerated Approval Pathway
.”
June
27
CMS (Centers for Medicare and Medicaid Services)
.
2021
. “
State Drug Utilization Data
.” https://www.medicaid.gov/medicaid/prescription-drugs/state-drug-utilization-data/index.html (accessed
June
14
,
2022
).
Dunn, Abe, Grosse, Scott D., and Zuvekas, Samuel H.
2018
. “
Adjusting Health Expenditures for Inflation: A Review of Health Services Research in the United States
.”
Health Services Research
53
, no.
1
:
175
96
.
.
2015
. “
Human Immunodeficiency Virus-1 Infection: Developing Antiretroviral Drugs for Treatment—Guidance for Industry
.”
November
.
2021a
. “
CDER Drug and Biologic Accelerated Approvals Based on a Surrogate Endpoint
.”
December
31
.
2021b
. “
Drugs@FDA: FDA-Approved Drugs
.” https://www.accessdata.fda.gov/scripts/cder/daf/index.cfm (accessed
June
14
,
2022
).
.
2022
. “
Withdrawn: Cancer Accelerated Approvals
.” https://www.fda.gov/drugs/resources-information-approved-drugs/withdrawn-cancer-accelerated-approvals (accessed
June
17
,
2022
).
Gamage, David.
2010
. “
Preventing State Budget Crises: Managing the Fiscal Volatility Problem
.”
California Law Review
98
, no.
3
:
749
812
.
Gyawali, Bishal, Hey, Spencer Phillips, and Kesselheim, Aaron S..
2019
. “
Assessment of the Clinical Benefit of Cancer Drugs Receiving Accelerated Approval
.”
JAMA Internal Medicine
179
, no.
7
:
906
13
.
Gyawali, Bishal, Rome, Benjamin N., and Kesselheim, Aaron S.
2021
. “
Regulatory and Clinical Consequences of Negative Confirmatory Trials of Accelerated Approval Cancer Drugs: Retrospective Observational Study
.”
British Medical Journal
,
September
8
. https://doi.org/10.1136/bmj.n1959.
KFF (Kaiser Family Foundation)
.
2022
. “
Status of State Medicaid Expansion Decisions: Interactive Map
.” https://www.kff.org/medicaid/issue-brief/status-of-state-medicaid-expansion-decisions-interactive-map/ (accessed
June
14
,
2022
).
MACPAC (Medicaid and CHIP Payment and Access Commission)
.
2021
. “
High-Cost Specialty Drugs: Moving Towards Recommendations
.”
March
. https://www.macpac.gov/publication/high-cost-specialty-drugs-moving-towards-recommendations/.
Mezher, Michael.
2021
. “
Industry-Wide Accelerated Approval Review Yields Four Withdrawals
.” Regulatory Affairs Professionals Society,
March
10
. https://www.raps.org/news-and-articles/news-articles/2021/3/industry-wide-accelerated-approval-review-yields-f.
Naci, Huseyin, Smalley, Katelyn R., and Kesselheim, Aaron S.
2017
. “
Characteristics of Preapproval and Postapproval Studies for Drugs Granted Accelerated Approval by the US Food and Drug Administration
.”
JAMA
318
, no.
7
:
626
36
.
NIH (National Institutes of Health)
.
2021
. “
ClinicalTrials.gov Background
June
14
,
2022
).
Prasad, Vinay, Kim, Chul, Burotto, Mauricio, and Vandross, Andrae.
2015
. “
The Strength of Association between Surrogate End Points and Survival in Oncology: A Systematic Review of Trial-Level Meta-Analyses
.”
JAMA Internal Medicine
175
, no.
8
:
1389
98
.
Sachs, Rachel E., Gavulic, Kyle A., Donohue, Julie M., and Dusetzina, Stacie B.
2021
. “
Recent Trends in Medicaid Spending and Use of Drugs with US Food and Drug Administration Accelerated Approval
.”
JAMA Health Forum
2
, no.
10
: e213177. https://doi.org/10.1001/jamahealthforum.2021.3177.
Young, Colin A., and Norton, Michael P.
2018
. “
Baker, States Active in Quest to Slow Rising Drug Prices
.”
Associated Press
,
October
18
. https://apnews.com/article/d3109ded91a34b07a5f1ef2e9ac508c6.