Each bubble represents 1 drug, with the area of the bubble proportional to the number of 30-day supplies of the drug dispensed in the US in 2015 and 2017. The x-axis represents the relative change in list price and the y-axis the change in net-price from 2015 to 2017. Outliers are labeled by generic name; corresponding brand names can be found in eTable 1 in the Supplement.
Each bubble represents 1 drug, with the area of the bubble proportional to the number of 30-day supplies of the drug dispensed in the US in 2015 and 2017. The x-axis represents the change in list (A) or net (B) price, and the y-axis represents the change in out-of-pocket spending from 2015 to 2017. Outliers are labeled by generic name; corresponding brand names can be found in eTable 1 in the Supplement. One drug, polyethylene glycol 3350/electrolytes (MoviPrep), had a median out-of-pocket spending of $0 in 2017, resulting in a 100% decrease.
Each bubble represents 1 drug, with the area of the bubble proportional to the number of 30-day supplies of the drug dispensed in the US in 2015 and 2017. The x-axis represents the change in list price (panels A and B) or net price (panels C and D), and the y-axis represents the median change in out-of-pocket spending from 2015 to 2017. Panels A and C include patients with any deductible or coinsurance, and panels B and D include patients without any deductible or coinsurance.
eFigure 1. Drugs in Cohort and Reasons for Exclusion
eTable 1. Drugs Included in Cohort
eTable 2. Changes in Out-of-Pocket Spending and Correlation With Changes in Drug Prices, Unweighted
eTable 3. Drug Prices and Out-of-Pocket Spending, Excluding Drugs With Higher Net Than List Price Estimates in Any Quarter
eTable 4. Changes in Out-of-Pocket Spending and Correlation With Changes in Drug Prices, Excluding Drugs With Higher Net Than List Price Estimates in Any Quarter
eFigure 2. Correlation Between Drug Prices and Out-of-Pocket Spending, by High Deductible Insurance Plan
eFigure 3. Correlation Between Drug Prices and Out-of-Pocket Spending, by Deductible
eFigure 4. Correlation Between Drug Prices and Out-of-Pocket Spending, by Coinsurance
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Rome BN, Feldman WB, Desai RJ, Kesselheim AS. Correlation Between Changes in Brand-Name Drug Prices and Patient Out-of-Pocket Costs. JAMA Netw Open. 2021;4(5):e218816. doi:10.1001/jamanetworkopen.2021.8816
Among commercially insured patients with varying pharmacy benefit designs, are changes in prices for brand-name prescription drugs correlated with changes in patient out-of-pocket costs?
In this cohort study of 79 brand-name drugs, there was no correlation between changes in drug prices and out-of-pocket spending. However, among the 57.3% of patients whose insurance required coinsurance or deductibles, changes in out-of-pocket spending were correlated with changes in list prices but not postrebate net prices.
Some commercially insured patients are insulated from paying more when drug list prices increase, but patients with deductibles and coinsurance pay more when manufacturers raise a drug’s list price and do not directly benefit from confidential rebates paid by manufacturers to insurers.
List prices set by manufacturers for brand-name prescription drugs in the US have been increasing faster than inflation, although confidential manufacturer rebates offset some of these increases. Most commercially insured patients pay at least some out-of-pocket costs for prescription drugs, and higher patient spending is associated with lower adherence and worse health outcomes.
To examine whether price changes for brand-name drugs are correlated with changes in patient out-of-pocket spending and whether this association varies by insurance benefit design.
Design, Setting, and Participants
A cohort study of 79 brand-name drugs with available pricing data from January 2015 to December 2017 was conducted, with data obtained from a national commercial insurance claims database.
Change in the list prices set by manufacturers and estimated net prices after rebates among non-Medicaid payers.
Main Outcomes and Measures
Change in median out-of-pocket spending among all patients and stratified by insurance pharmacy benefit design, including high-deductible insurance plans and plans with any amount of deductibles or coinsurance.
Among 79 drugs, median increases were 16.7% (interquartile range [IQR], 13.6%-21.1%) for list prices, 5.4% (IQR, −3.9% to 11.7%) for net prices, and 3.5% (IQR, 1.4%-9.1%) for out-of-pocket spending from 2015 to 2017. Changes in list prices were correlated with changes in net prices (r = 0.34; P = .002). Overall, changes in out-of-pocket spending were not correlated with changes in list prices (r = 0.14; P = .22) or net prices (r = 0.04; P = .71). Among 53.7% of patients who paid any drug deductible or coinsurance, median out-of-pocket spending increased by 15.0%, and changes were moderately correlated with changes in list prices (r = 0.38; P = .001) but not net prices (r = 0.06; P = .62).
Conclusions and Relevance
Some commercially insured patients who pay only prescription drug copayments appear to be insulated from increases in drug prices. However, more than half of patients pay deductibles or coinsurance and may experience substantial increases in out-of-pocket spending when drug prices increase. Among these patients, there was no evidence that manufacturer rebates to insurers are associated with patients’ out-of-pocket spending. Policies to rein in unregulated annual increases in list prices for brand-name drugs may have important consequences for patient out-of-pocket spending.
Increasing prescription drug prices are a source of concern in the US. In 2018, the US spent $476 billion on prescription drugs, with 80% of spending on brand-name products.1,2 Brand-name manufacturers freely set list prices (akin to the sticker price for a car) for their drugs at whatever level they choose, and these prices increased by 9.1% per year on average over the past decade.3 For many drugs, manufacturers have offset increasing list prices by providing confidential rebates negotiated with pharmacy benefit managers and insurers.4,5 Although rebate levels vary greatly by drug class, some have estimated that despite continued year-over-year increases in list prices, growth in net prices after accounting for rebates has slowed or halted since 2015, particularly for certain drug classes.1-3
However, rebates negotiated by insurers are not necessarily reflected in the out-of-pocket prescription drug costs borne by patients. Patient spending at the pharmacy counter—including copayments, coinsurance, and deductibles—accounted for 14% of all US prescription drug spending in 2018.2 For patients in Medicare Part D plans, cost-sharing is typically higher and is often determined as a fixed percentage of a drug’s list price (25% of drug costs until patients reach the catastrophic coverage phase).6 As a result, Part D beneficiaries may be adversely affected by increasing list prices and may not benefit directly from rebates.7
Half of individuals in the US are insured by employer-based commercial insurance plans,8 which traditionally charge patients flat copayments for prescription drugs but have increasingly shifted costs to patients by adding deductibles and/or coinsurance.9 Yet few studies have examined the association between increasing list prices and patient out-of-pocket costs for commercially insured patients. One recent study noted that list prices for 14 top-selling drugs doubled from 2010 to 2016, while median out-of-pocket costs increased by 53%.10 Another study reported that, although insulin list prices more than doubled from 2006 to 2017, patient out-of-pocket spending was relatively stable over the same period.11 The latter study stratified patients based on whether they were in a high-deductible insurance plan, but to our knowledge, no study has systematically examined the relationship between drug prices and patient out-of-pocket costs stratified by commercial pharmacy benefit design.
Given the limited understanding of how prices affect patient out-of-pocket costs for commercially insured individuals, we examined the correlation between changes in list prices, net prices, and out-of-pocket spending among a cohort of brand-name drugs used by commercially insured patients from January 2015 to December 2017, overall and among patients with different insurance benefit designs.
In this cohort study, we identified the drug cohort and obtained list and net price estimates from SSR Health, which estimates prices and rebates for more than 1000 brand-name drugs manufactured by publicly traded companies.12 We measured median patient out-of-pocket spending using claims data from a national sample of approximately 30 million patients with employer-sponsored commercial insurance at any given time (MarketScan; IBM Truven Health Analytics).
We obtained approval from the Mass General Brigham Institutional Review Board to use deidentified claims data for this study. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.
SSR Health provides data for drugs with at least 1 net price estimate from 2007 to 2019, but most drugs do not have estimates for the entire period. We included only drugs with continuous net price estimates in each quarter from 2015 to 2017 and drugs with at least 1000 units reported per quarter to minimize the instability of net price estimates. We excluded drugs administered by clinicians (eg, vaccines, intravenous preparations) because pharmacy claims may not accurately capture out-of-pocket spending for these products. We also excluded brand-name drugs with generic competition before January 2018 because insurers may increase patient cost-sharing for brand-name products after generic versions are released. Generic entry date was obtained from SSR Health, supplemented by a review of when the first generic National Drug Codes were listed in a compendium of all drug products published by First Databank.13 We also excluded contraceptives, which all commercial insurers have generally been required to cover without cost-sharing since passage of the Affordable Care Act.14
Pricing estimates in SSR Health are given for an entire drug product, representing a weighted average of unit prices across all formulations and dosage forms. A given product may have variable definitions of units (eg, tablets vs milliliters of solution) and different prices per unit. Consequently, variation in the relative use of different dosage forms may result in changes in the product’s unit price over time that do not reflect variations in the price of the drug. To narrowly focus on the association between a drug’s price and out-of-pocket costs, we included only drug products that had the same unit price across all dosage forms. In addition, we included only drugs for which we could identify a standard number of units per 30-day supply based on the dosing guidelines on the drug’s US Food and Drug Administration–approved labeling. For drugs with prices that vary based on the dose prescribed, out-of-pocket spending may be influenced both by price and dose, and our aim was to focus solely on the association between price and out-of-pocket spending.
For each drug included in the cohort, we used the US Food and Drug Administration–approved labeling to determine the drug’s route of administration and therapeutic category. We also estimated annual US revenue in 2015 and 2017 as a product of net price multiplied by estimated use among all patients in the US from SSR Health.
We obtained quarterly wholesale acquisition cost (ie, the list price set by manufacturers) and net prices from SSR Health, which estimates net prices by comparing publicly reported manufacturer revenue with estimates of total number of units sold in the US.12 These estimates approximate the average US price per unit after manufacturer rebates and other price concessions. We relied on SSR Health’s non-Medicaid rebate estimates because Medicaid rebates are statutory and usually higher than those for Medicare or commercial plans.15,16
Because drug sales volume to wholesalers can vary over a calendar year (eg, the drug can be stocked in January to last several months), we calculated annual averages for list and non-Medicaid net prices in 2015 and 2017, weighting each quarter equally. All annual average prices were converted to 2017 US dollars using the Consumer Price Index for all urban consumers.17 We also converted unit prices to the price for a typical 30-day supply of the drug based on the dosing instructions in the US Food and Drug Administration–approved labeling information. For drugs typically dispensed for 1-time use (eg, bowel preparation regimens), we converted to the price for a single use.
In some cases, SSR Health’s quarterly net price estimates exceeded list prices, usually because the number of units sold may be underreported.3 When net price exceeded list price in a given quarter, we presumed rebates of 0, so that the net price was equal to the list price for that quarter. Owing to unequal distribution of drugs throughout the year, a negative rebate in one quarter might offset a high rebate in a surrounding quarter. Therefore, assuming 0 rebates in these cases may lead to underestimates of annual average net prices. We also performed a sensitivity analysis in which we excluded drugs with a negative rebate estimate in any quarter.
For each brand-name drug, we identified separate cohorts of patients in 2015 and 2017 with at least 1 claim for the drug. We excluded patients with claims paid under capitated payment models because cost data for these claims are not reliably recorded.
We calculated the annual sum of out-of-pocket costs (copayments, coinsurance, and deductibles) paid across all filled prescriptions for the drug of interest in each calendar year. We also summed the annual quantity of drug prescribed (the number of units), which we converted to 30-day supplies to calculate each patient’s average annual out-of-pocket spending per 30-day supply (or 1-time use). We then calculated the median out-of-pocket spending among all users of each drug in each year, adjusted to 2017 US dollars. To ensure that unit prices were measured similarly in SSR Health and claims data, we calculated median total drug cost per unit in the claims data (including patient and payer components) and excluded drugs if this cost differed from the drug’s list price by more than 10%.
For each drug, we calculated the relative change in inflation-adjusted list price, non-Medicaid net price, and median out-of-pocket spending from 2015 to 2017 (eg, 2017 price – 2015 price / 2015 price). We calculated Pearson correlation coefficients between relative changes in drug prices and out-of-pocket spending to identify significant associations (r ≠ 0) at a 2-sided significance level of P < .05.
To account for varying associations between changes in drug prices and out-of-pocket spending among patients with different insurance benefit designs, we performed stratified analyses in which we correlated drug prices with median out-of-pocket spending among patients in high-deductible insurance plans and those who paid a deductible and/or coinsurance toward any prescription drug claim in the given year.
In the primary analysis, we weighted each drug based on its use, defined as the sum of 30-day supplies dispensed in the US in 2015 and 2017, obtained from SSR Health. The weighting was performed to ensure that our results were reflective of the highest-used drugs with the greatest public policy importance. In a secondary analysis, we calculated unweighted estimates. To test for bias resulting from our assumption of 0 rebates in any quarter when net price estimates exceeded list prices, we performed a sensitivity analysis that excluded the 21 drugs for which estimated net prices exceeded list price in any quarter. Analyses were performed using SAS Software, version 9.4 (SAS Institute Inc).
Of 1117 brand-name products with any SSR Health net price estimates from 2007 to 2019, 541 drugs had continuous net price estimates in each quarter from 2015 to 2017. Ultimately, 79 self-administered brand-name drugs without generic competition met the entry criteria for the cohort (eFigure 1 in the Supplement). A list of included drugs is provided in eTable 1 in the Supplement.
Of the 79 drugs in our cohort, 64 (81%) were administered orally, 8 (10%) by injection, 3 (4%) by inhalation, and the remaining 4 (5%) via topical, rectal, or ophthalmic routes. The most common therapeutic categories were oncologic (11 [14%]), antimicrobial (9 [11%]), endocrine/diabetes (9 [11%]), neurologic (9 [11%]), cardiovascular (8 [10%]), and gastrointestinal (8 [10%]).
Median US annual revenue per drug was $330 million (interquartile range [IQR], $120 million-$706 million) in 2015 and $401 million (IQR, $120 million-$888 million) in 2017. Total revenue for all 79 drugs was $63 billion in 2015 and $67 billion in 2017. In 2017, 18 drugs (23%) had annual revenue exceeding $1 billion.
Median list prices per 30-day supply increased from $333 (IQR, $229-$346) in 2015 to $386 (IQR, $271-$390) in 2017, with a median change in list price of 16.7% (IQR, 13.6%-21.1%) (Table 1). The unweighted median list prices were higher, but the percent change was similar at 17.2% (IQR, 12.3%-23.3%). List prices increased for all but 4 drugs; these drugs each had a 3.3% decrease in price from 2015 to 2017 related to inflation (ie, no change in nominal list price).
The median net price per 30-day drug supply was $173 (IQR, $139-$223) in 2015 and $166 (IQR, $144-$234) in 2017. The median change in net price was 5.4% (IQR, −3.9% to 11.7%). Unweighted results revealed a similar median increase in net price of 7.5% (IQR, −1.9% to 18.6%). Net prices decreased for 21 drugs (27%) and increased for 58 drugs (73%) (Table 1). Changes in list and net prices were positively correlated in weighted (r = 0.34; P = .002) (Figure 1) and unweighted (r = 0.39; P < .001) analyses.
The median number of commercially insured patients in our data set who used each drug was 4925 (IQR, 1281-16 266) in 2015 and 4078 (IQR, 772-19 907) in 2017. The sum of patients included in the median out-of-pocket spending estimates was 1.4 million in each year, although some patients may have used more than 1 drug in the cohort in a given year.
Median out-of-pocket spending for a 30-day drug supply was $29 (IQR, $26-$43) in 2015 and $30 (IQR, $26-$45) in 2017, with a median change from 2015 to 2017 of 3.5% (IQR, 1.4%-9.1%). Results were similar in unweighted analyses.
In 2017, 16.7% of patients in our cohort were enrolled in high-deductible health plans and 53.7% paid either a deductible or coinsurance (Table 2). Out-of-pocket spending was higher among patients in high-deductible health plans (median, $56 in 2017; IQR, $43-$65) than among those not in high-deductible health plans (median, $28; IQR, $25-$42) and among those who paid any deductible or coinsurance (median, $40 in 2017; IQR, $30-$54) compared with those with no deductible or coinsurance (median, $25; IQR, $23-$35). The change in out-of-pocket spending from 2015 to 2017 was higher among patients with any deductible or coinsurance (median, 15.0% vs −3.3%). In weighted analyses, there was no correlation between changes in out-of-pocket spending and changes in list price (r = 0.14; P = .22) or net price (r = 0.04; P = .71) (Figure 2).
Among patients in insurance plans that included deductibles or coinsurance, there was a moderate correlation between changes in out-of-pocket spending and list prices but no correlation with net prices (Figure 3; eFigures 2, 3, and 4 in the Supplement). The strongest correlations were among patients in high-deductible insurance plans (r = 0.43; P < .001) and those with any deductible or coinsurance (r = 0.38; P = .001) (Figure 3). Among patients who did not pay any drug deductibles or coinsurance (ie, copayments only), there was no correlation between changes in out-of-pocket spending and list (r = 0.09; P = .45) or net prices (r = 0.08; P = .48).
In unweighted analyses, we similarly found a moderate correlation between changes in list prices and out-of-pocket spending among patients with any coinsurance (r = 0.30; P = .008) and those with any deductible or coinsurance (r = 0.24; P = .03) (eTable 2 in the Supplement). There was no correlation between changes in net price and out-of-pocket spending. Results were similar when we excluded 21 drugs for which net price estimates exceeded list prices in at least 1 quarter (eTable 3 and eTable 4 in the Supplement).
Among a cohort of commercially insured patients using brand-name prescription drugs, approximately half paid fixed copayments and were insulated from increases in list prices. The other half of patients had prescription drug benefits that included deductibles or coinsurance and, in that cohort, out-of-pocket costs increased when manufacturers increased list prices. Changes in net drug prices accounting for manufacturer rebates were not correlated with changes in patient out-of-pocket spending, suggesting that increasing rebates offered by manufacturers to partially offset list price hikes are not being directly passed on to patients, even if they limit increases to total drug spending.
Earlier studies of the association between changes in drug prices and out-of-pocket costs have yielded mixed results.10,11 Our findings suggest that the association is influenced by insurance benefit design, with flat copayments protecting patients from increasing list prices but coinsurance and deductibles exposing patients to these price increases.
In a national survey, 84% of patients with private health insurance had a pharmacy benefit that included 3 or more drug tiers, with more expensive drugs in higher tiers requiring higher out-of-pocket costs.18 Traditionally, insurers set flat copayments for each tier, but insurers increasingly rely on coinsurance to limit spending on high-priced drugs.19 In 2019, 28% of commercially insured patients faced coinsurance on tier 2 drugs (typically preferred brand-name drugs) and 41% on tier 4 drugs (typically specialty drugs), with coinsurance rates ranging from 24% to 34% of a drug’s list price.18 In our study, nearly half of patients paid prescription drug coinsurance.
We found that, for patients whose prescription drug plans included deductibles or coinsurance, increases in a drug’s list price were correlated with higher out-of-pocket costs. This design is problematic because list prices for brand-name drugs, which are unregulated in the US, have long been increasing much faster than inflation.3 List prices increased for nearly all drugs in our cohort, with a median inflation-adjusted increase of 16.7% over 2 years. Some of these list price increases have been offset by increasing rebates provided to insurers and pharmacy benefit managers, such that 1 in 4 drugs in our cohort had a decrease in net price from 2015 to 2017. Although increasing rebates may limit the effect of list price hikes on total drug spending, we did not find any correlation between changes in a drug’s net price and out-of-pocket spending, suggesting that rebates are not being reflected in lower point-of-care prices for patients.4,7
Tying patient out-of-pocket costs directly to the price of a drug can encourage patients to use less costly medications, and completely detaching patient expenses from drug prices could ultimately lead to higher premiums.20,21 However, the parallel unfettered increases in list prices for brand-name drugs and greater reliance by insurers on deductibles and coinsurance have placed important financial burdens on patients using high-cost brand-name medications. Higher out-of-pocket costs have been associated with decreased medication adherence and, in some cases, worse clinical outcomes in a wide range of clinical conditions, including cancer,22,23 diabetes,24,25 and cardiovascular disease.26-28 Policy interventions that lower out-of-pocket spending can increase medication adherence, improve clinical outcomes, and reduce health disparities.14,29,30
As a result, policy makers should consider how to protect patients from large increases in out-of-pocket spending resulting from year-over-year increases in list prices for prescription drugs. For example, Congress could cap annual increases on list prices or penalize manufacturers for increases above inflation through mandatory rebates in Medicare Part D, similar to the model already used in Medicaid.31,32 However, there are concerns that caps or penalties on price increases could encourage manufacturers to set higher prices for drugs at launch,32 and if such rules were imposed only on Medicare Part D, there may be no substantial benefit for the half of Americans who are covered by private employer–based health insurance plans.8
Policy makers could also address the increasing gap between list and net prices by eliminating confidential manufacturer rebates to insurers and pharmacy benefit managers or mandating that such rebates be passed directly to patients at the point of sale. A rule finalized by the Trump Administration in November 2020 would have eliminated drug rebates in Medicare Part D, although the rule faces substantial legal hurdles and its implementation has been delayed.33,34 Lawmakers have proposed more general bans on rebates to public and private payers unless the rebates are directly passed on to patients at the point of sale.35 Some private insurers have already begun implementing programs to pass rebates on directly to patients.36 However, limiting confidential rebates or requiring increased transparency could undermine the bargaining power of large purchasers, leading to an overall increase in drug prices and spending.37-39 Thus, a ban on rebates might reduce out-of-pocket spending by patients who use certain high-cost medications but lead to higher premiums among all patients.
This study has limitations. We examined only 79 drugs owing to strict inclusion criteria to ensure accurate measurement of prices, so our findings may not be representative of all brand-name drugs. However, the cohort includes many top-selling drugs, with the entire cohort accounting for $67 billion in spending in 2017, which represents nearly 15% of total US prescription drug spending.40 Drugs with multiple formulations or for which monthly price varies by dose are likely to have similar changes over time, although prices and subsequent out-of-pocket spending per patient may be more variable. Because we measured out-of-pocket spending from pharmacy claims, we were unable to include clinician-administered drugs, which account for a growing proportion of prescription drug spending.41
Another limitation is that our findings are applicable only to commercially insured patients, which include approximately half of Americans8; the association between increasing list prices and increasing out-of-pocket spending may be more pronounced among patients covered by Medicare, as the standard Part D benefit includes coinsurance (25% up until a catastrophic limit, and thereafter 5%).4,7 Commercially insured patients may use manufacturer coupons to offset some out-of-pocket spending, and coupon use is not captured in claims data. Coupon use has increased over the past several years,42,43 so not being able to account for coupons may lead to an overestimate of out-of-pocket spending for some drugs.
In addition, individual rebates to payers are confidential. The estimates of net prices from SSR Health are based on manufacturer revenue and may underestimate postrebate prescription drug spending by excluding fees and profits among other elements of the pharmaceutical supply chain, including wholesalers, pharmacies, and pharmacy benefit managers.44,45 SSR Health may underestimate Medicaid rebates, leading to a corresponding overestimate in estimated non-Medicaid rebates,46 although such an overestimate would be unlikely to substantially affect change in net prices over time.
Although some commercially insured patients who pay only prescription drug copayments are insulated from increases in prescription drug prices, more than half of patients pay deductibles or coinsurance and may experience increases in out-of-pocket spending when manufacturers increase list prices. Among these patients, we found no evidence that increasing manufacturer rebates directly offset increases in out-of-pocket expenses. Policy makers seeking to limit price increases among brand-name drugs should consider how any proposed policies affect patient out-of-pocket spending.
Accepted for Publication: March 13, 2021.
Published: May 4, 2021. doi:10.1001/jamanetworkopen.2021.8816
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Rome BN et al. JAMA Network Open.
Corresponding Author: Benjamin N. Rome, MD, Program On Regulation, Therapeutics, And Law, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital, 1620 Tremont St, Ste 3030, Boston, MA 02120 (email@example.com).
Author Contributions: Dr Rome had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Rome.
Critical revision of the manuscript for important intellectual content: Feldman, Desai, Kesselheim.
Statistical analysis: Rome, Desai.
Obtained funding: Kesselheim.
Conflict of Interest Disclosures: Dr Feldman reported receiving personal fees from Alosa Health, Aetion, and Blue Cross Blue Shield of Massachusetts outside the submitted work. Dr Desai reported receiving grants from Bayer, Novartis, and Vertex outside the submitted work. No other disclosures were reported.
Funding/Support: This study was funded by a grant from Arnold Ventures to Brigham and Women’s Hospital. Drs Rome and Kesselheim are supported by a grant from Anthem Public Policy Institute. Dr Feldman receives funding from the National Institutes of Health.
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Gregory Brill, MS (Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women's Hospital), assisted with extracting out-of-pocket spending data from MarketScan; his compensation was included in the study funding. The following Program On Regulation, Therapeutics, and Law colleagues provided uncompensated feedback about earlier versions of the figures and tables: Rachel E. Barenie, PharmD JD, MPH; Sheng Liu, JD, MSc; Brooke Raunig, JD, RN; Véronique Raimond, PhD, MSc; Leah Rand, PhD, MA; Victor L. van de Wiele, LLB, LLM; Rebecca E. Wolitz, JD, MPhil; and Steven Woloshin, MD, MS.
Create a personal account or sign in to: