{"subscriber":false,"subscribedOffers":{}} Decline In Economic Returns From New Drugs Raises Questions About Sustaining Innovations | Health Affairs

Research Article

Decline In Economic Returns From New Drugs Raises Questions About Sustaining Innovations

Affiliations
  1. Ernst R. Berndt is the Louis E. Seley Professor in Applied Economics at the Alfred P. Sloan School of Management, Massachusetts Institute of Technology, in Cambridge.
  2. Deanna Nass is a senior researcher at the IMS Institute for Healthcare Informatics, in Plymouth Meeting, Pennsylvania.
  3. Michael Kleinrock is director of research development for the IMS Institute for Healthcare Informatics.
  4. Murray Aitken ( [email protected] ) is executive director of the IMS Institute for Healthcare Informatics.
PUBLISHED:February 2015Free Accesshttps://doi.org/10.1377/hlthaff.2014.1029

Abstract

The sales and financial returns realized by pharmaceutical companies are a frequent topic of discussion and debate. In this study we analyzed the economic returns for four cohorts of new prescription drugs launched in the United States (in 1991–94, 1995–99, 2000–04, and 2005–09) and compared fluctuations in revenues with changing average research and development (R&D) and other costs to determine patterns in rewards for pharmaceutical innovation. We found that the average present values of lifetime net economic returns were positive and reached a peak with the 1995–99 and 2000–04 new drug cohorts. However, returns have fallen sharply since then, with those for the 2005–09 cohort being very slightly negative and, on average, failing to recoup research and development and other costs. If this level of diminished returns persists, we believe that the rewards for innovation will not be sufficient for pharmaceutical manufacturers to maintain the historical rates of investments needed to sustain biomedical innovation.

TOPICS

The drug development process in the United States has long been dominated by private companies. For these companies to sustain the development and marketing of innovative medicines, their financial returns must be sufficient to recoup research and development (R&D), manufacturing, marketing, and other operating costs. Specifically, when assessing whether to invest in the development of an innovative pharmaceutical, firms routinely evaluate the drug’s prospects for addressing a therapeutic need, receiving regulatory approval, and achieving commercial success. Regarding commercial success, firms evaluate whether expected lifetime sales will be sufficient to generate positive returns on investment beyond recouping R&D and operating costs.

In this study we compared average lifetime pharmaceutical revenues to average R&D and lifetime operating costs. This allowed us to determine the “net economic returns” for four cohorts of new drugs launched in the United States—in 1991–94, 1995–99, 2000–04, and 2005–09—and to examine the returns’ levels and patterns over time. The cohorts included large-molecule (biologic) and small-molecule (chemical) drugs. To provide a standardized basis for comparison over time, we converted all revenue and cost numbers to 2005 dollars.

A number of widely cited studies have focused on the average costs of bringing a new drug to market. 1–4 However, relatively little attention has been paid to lifetime expected revenues or the implicit net economic returns on R&D costs. Two frequently cited studies that quantified average pharmaceutical R&D returns are a 1998 study of drugs launched in 1980–84 5 and a 2002 analysis of drugs entering the US market in 1990–94. 6

Since those studies were published, a number of important developments have occurred in the pharmaceutical marketplace. These include growth in the availability of biologic drugs and targeted medicines; heightened scrutiny by payers in coverage decisions—involving, for example, the introduction of tiered formularies, utilization management, and step therapy; growth in the number of therapeutic areas that have a standard of care based on generic drugs, coupled with dramatic increases in the use of those drugs; and significant restructuring and consolidation of pharmacy purchasing organizations, such as pharmacy benefit managers, which administer drug benefits on behalf of payers, and wholesalers, which supply medicines to pharmacies, clinics, and hospitals.

Major regulatory and legislative efforts have also affected access to medicines. These include the Medicare Part D prescription drug benefit program, implemented in 2006, and the Biologics Price Competition and Innovation Act of 2009, 7 which created a regulatory pathway for the approval of biosimilar drugs.

The cumulative impact of these developments on net economic returns on pharmaceutical research and development is unknown. Yet it is of great importance to drug developers, providers, public and private payers, policy analysts, and public health.

To assess the impact of these developments, we analyzed financial data for 466 novel active substances launched in the United States between 1991 and 2009, and we determined how revenues, costs, and returns have fluctuated over time. We defined a novel active substance as a molecular or biologic entity or combination product in which at least one element had not previously been approved by the Food and Drug Administration (FDA). Of the products in our study, 378 (81.1 percent) were small-molecule (chemical) drugs and 88 (18.9 percent) were large-molecule (biologic) drugs, inclusive of combination products in these groups. We considered launches only through 2009 because too few years of postlaunch data are currently available to allow reliable modeling of lifetime sales predictions for subsequent new drug launches.

Study Data And Methods

Data Source

For each novel active substance, we calculated the present value of lifetime net sales using quarterly US data from the period 1991–2012 from IMS Health Inc.’s MIDAS database. We adjusted for estimated rebates, discounts, and free goods not reflected in that data source. We also adjusted for general inflation, converting amounts to 2005 dollars using the US gross domestic product deflator and discounting to a present value using an annual discount rate that is consistent with recent literature. 3 We used an 11.0 percent discount rate for small-molecule drugs and an 11.5 percent rate for biologics.

All forms and strengths of a drug were included with its parent novel active substance, including subsequent formulations launched for any indication. To calculate worldwide net revenue, we multiplied annual US sales by standard values, based on recent stable trends for protected brand-name drugs (with separate trends for biologics and small molecules). Estimates on rebates, discounts, and free goods were taken from IMS Health research that compared manufacturers’ public filings and documents (such as Securities and Exchange Commission filings that reported net sales) with IMS Health data using invoices at manufacturers’ sale prices. 8

Methods

We defined a patent-protected drug’s lifetime as the time from launch through loss of patent exclusivity in the United States plus twenty-four months for small-molecule drugs and plus sixty months for biologics, to account for the introduction of generic and biosimilar competitors into the marketplace. To calculate the lifetime sales for drugs whose lifetime had not concluded by the end of 2012, we modeled subsequent sales, applying different methods for small molecules and biologics. All of the drugs that we studied had at least three years of actual sales performance data, and most had substantially more.

For small-molecule drugs, we modeled sales based on quintile rankings of each novel active substance’s sales at twelve, twenty-four, thirty-six, forty-eight, and sixty months after launch. For drugs where complete lifetime revenue data were not yet available, we aligned predictions with historical data based on 153 small-molecule drugs that had full lifetimes and with the average sales trajectories of similarly quintile-ranked substances in this set of products.

Because complete lifetime data did not yet exist to create such average curves for biologics, we projected their sales data using either the compounded annual growth rate or the average absolute growth over the period 2010–12. For products that showed accelerating compounded annual growth rates outside of historical norms, we used the average annual growth rate instead.

The end of life for biologics was determined by the presumed date of biosimilars’ entry into the market, based on consensus date estimates in the public domain regarding expected loss of patent exclusivity or other publicly available information. If the FDA begins approving biosimilars this year or next, the most recent cohort of biologics launched will have shorter times from launch to loss of exclusivity, compared to their historical counterparts. Assumptions of sales volume erosion for biosimilars that experience loss of exclusivity in the United States were based on experiences in the European Union, where a biosimilars pathway has existed since 2003.

Because the 2005–09 cohort of drugs and biologics relies on projections more than do the other cohorts, it is the most sensitive to forecasting methods. Nevertheless, all drugs in this recent cohort have at least three years of recorded market sales performance data, and 62 percent have more than five years of data.

For all four cohorts, we modeled expected lifetimes based on currently known line and patent extensions, but we did not account for possible future extensions. We therefore performed a sensitivity analysis by extending time from launch to loss of exclusivity for the 2005–09 cohort to average levels attained by molecules with complete lives in the 1991–94 and 1995–99 cohorts. This permitted us to set upper bounds on the possible impact of line extensions on lifetime revenues.

We employed published data on average operating expenses (cost of goods sold, 5,9,10 sales and general administrative expenses, 9–11 and depreciation and effective tax rates 6 ) separately for small-molecule drugs and biologics, 9 to derive net operating income after taxes. We then computed the present value of net operating income, all in 2005 dollars.

To take both pre- and post-approval R&D costs into account, we used average R&D cost assumptions from published literature 2,3 and subtracted after-tax inflation-adjusted total R&D costs from the present value of net operating income. This allowed us to obtain lifetime net returns per novel active substance.

We evaluated revenues, costs, and profitability using a framework informed by microeconomic theory and traditional corporate accounting. In particular, we defined economic profits as revenues minus economic costs.

The economic costs of investments, such as those made in research and development, differ from accounting costs in that the former include the opportunity costs of investments—that is, what investors could have earned elsewhere had they not invested in pharmaceutical research and development. Revenues and costs were therefore converted to their present values, reflecting the opportunity cost of capital. Further details regarding our data sources, analytical procedures, and sensitivity analyses are provided in the online Appendix. 12

Limitations

Our study had several limitations. We did not observe the full lifetimes of about two-thirds of the novel active substances launched since 1991. Thus, we projected at least some of their post-2012 revenues and costs over the remainder of their estimated lifetimes based on historical data. This limitation particularly applies to the 2005–09 cohort, which was the most highly projected.

All efforts to model future product sales are open to risk. However, significant evidence exists that the sales uptake trajectories (sales values and their typical growth patterns over time) of novel active substances are determined early in their lifetimes (usually within the first six months after a product’s launch) and do not improve significantly over their initially established market share and sales trajectory. 13 This fact allowed us to more confidently draw projections for this cohort using a minimum of 3.0 years of actual data and an average of 5.5 years per novel active substance.

The actual data recorded and the model therefore reflect both demand-side changes that have occurred, such as the impact of tiered formularies and increases in competition within certain therapy areas, and the impact of these changes on the uptake of products after launch. We discuss this further in the online Appendix. 12

Another limitation was the lack of access to drug- or company-specific operating costs or R&D costs. We therefore used average cost data based on published studies. These are based in large part on information or public documents provided by a subset of manufacturers. Thus, our results on net lifetime returns should be viewed as averages and are unlikely to reflect the experience of individual companies or pharmaceutical products. The extent to which the set of companies that filed public documents represents the universe of pharmaceutical companies is unclear, particularly with regard to profits, losses, and rebates.

On the revenue side, IMS MIDAS data currently capture sales data in more than seventy countries, which collectively account for more than 88 percent of the world market. Nonetheless, there could be some impact of data set coverage on the global multiplier. Since almost all sales outside the countries covered in the data are of generic—not brand-name—products, however, the impact on sales of brand-name drugs and the global multiplier is minimal.

Our use of a constant global sales multiplier over time likely resulted in some overstatement of sales levels of brand-name drugs outside of the United States in the 1990s. It likely also resulted in understated growth rates for such sales outside the United States in later periods.

The ratios of US sales to sales outside of the United States were relatively steady over time across the small-molecule and biologics cohorts. However, ratios vary substantially by product and therapy area, and thus individual results could differ from actual average values.

Study Results

Present Values Of Lifetime Global Net Sales

The average present values of lifetime global net sales varied substantially across annual cohorts and between small-molecule drugs and biologics. To highlight major trends and differences, we aggregated net sales data for novel active substances into four multiyear cohorts: 1991–94, 1995–99, 2000–04, and 2005–09. We report average values both for all drugs and separately for small molecules and biologics.

For all novel active substances, the average present values of lifetime global net sales (hereafter, “lifetime global sales”) were less for drugs in the most recent 2005–09 launch cohort compared to those in previous cohorts ( Exhibit 1 ). There was a 43 percent reduction between the last two cohorts.

Exhibit 1 Average Present Value Of Lifetime Global Net Sales Of Novel Active Substances, By Launch Cohort, 1991–2009

Exhibit 1
SOURCE Authors’ analysis of 1991–2012 data from IMS Health Inc.’s MIDAS database. NOTE Average present value is the value discounted for the cost of capital and reflects the time value of money.

In addition, levels and time trends in lifetime global sales for biologics differed from those for small-molecule drugs. Lifetime global sales levels were larger for biologics than for small molecules in the first three cohorts (about 84 percent higher in the peak launch years for biologics, 1995–99). However, biologics’ sales then fell sharply, from a peak of $7.7 billion in 1995–99 to $2.7 billion in 2005–09. Biologics’ sales in the last cohort were about 7 percent less than the $2.9 billion garnered by small-molecule drugs. Hence, while initially higher, average levels of lifetime global sales for biologics have begun to approximate those for small-molecule drugs.

These averages, however, mask substantial volatility, both within and between large- and small-molecule drugs. A common measure of relative volatility is the coefficient of variation, or the ratio of the standard deviation to the arithmetic mean. In five of the nine years in the 1990s (1992–95 and 1999), only one new biologic was launched. Thus, in those years no variability was observed.

We therefore focused on relative volatility for large and small molecules during the period 2000–09. During this time, the relative volatility for all novel active substances was greater than 1.00 (1.63 within the 2000–04 cohort and 1.59 within the 2005–09 cohort). Relative volatility increased for biologics (1.50 in 2000–04 and 1.56 in 2005–09) and was even higher for small-molecule drugs, although it decreased slightly during the period (1.69 in 2000–04 and 1.60 in 2005–09).

The vast majority of novel active substances achieved relatively small lifetime sales ( Exhibit 2 ). Seventy-five percent had lifetime sales of less than $4.5 billion, and 50 percent achieved sales of less than $1.5 billion. On the revenue side, therefore, not only have average lifetime global sales fallen for all novel active substances launched in the most recent cohort, but there also was considerable volatility in those sales.

Exhibit 2 Distribution Of Present Value Of Lifetime Global Net Sales Per Novel Active Substance, By Launch Cohort, 1991–2009

Exhibit 2
SOURCE Authors’ analysis of 1991–2012 data from IMS Health Inc.’s MIDAS database.

Cost Components In Derivation Of After-Tax Net Income

The magnitudes of the various cost components (pretax operating costs including cost of goods sold and sales, general, and administrative costs; after-tax R&D costs, including pre- and postapproval expenditures; and taxes) applied in the derivation of after-tax net income (“economic profits”) are displayed in Exhibit 3 for the four multiyear cohorts and compared to lifetime global net sales for all of the novel active substances in the study. Across the first three cohorts, average total costs per novel active substance rose steadily, from $3.0 billion for the 1991–94 cohort to $4.5 billion for the 2000–04 cohort. For all novel active substances, R&D expenditures accounted for 18–23 percent of all costs. The shares for both cost of goods sold and sales, general, and administrative costs were 31–33 percent, and the shares for taxes were 14–16 percent.

Exhibit 3 Average Present Value Of Lifetime Global Net Sales And Total Costs Of Novel Active Substances, By Launch Cohort, 1991–2009

Exhibit 3
SOURCE Authors’ analysis of 1991–2012 data from IMS Health Inc.’s MIDAS database. NOTES Revenue is lifetime global net sales. R&D is research and development.

In the 2005–09 cohort, however, costs returned to the $3.0 billion level, in line with declining revenues. The R&D share notably increased from 18–23 percent to 34 percent. This reflected the fact that R&D costs are not variable in the short term, despite declining revenues. In contrast, the cost of goods sold and sales, general, and administrative costs both fell to 27 percent, and taxes fell to less than 12 percent.

After-tax R&D costs accounted for $1,032 million and $699 million of the costs for small molecules launched in the 2000s and 1990s, respectively. The comparable figures for biologics were $972 million and $658 million (for details, see the Appendix). 12 The higher overall costs in the more recent cohorts reflected the rising costs of research and development. However, total costs as a percentage of total revenues (88.3 percent) for the 2000–04 cohort fell to a level between that of the 1991–1994 and 1995–1999 cohorts (90.6 percent and 84.4 percent, respectively). In the 2005–09 cohort, expenses exceeded revenues (104.0 percent) for the first time.

Small-molecule drugs and biologics both experienced marked reductions in costs and revenues in the 2005–09 cohort. The R&D cost share grew for both types of drugs, but only for small molecules did costs exceed revenues.

Calculation Of Economic Profits

By subtracting the average present value of lifetime total costs from the average present value of lifetime global net sales ( Exhibit 3 ), we derived average lifetime after-tax net returns (“economic profits”). The results of these calculations are presented in Exhibit 4 by launch cohort and separately for small molecules, biologics, and all novel active substances.

Exhibit 4 Average Lifetime After-Tax Net Returns Of Novel Active Substances, By Launch Cohort, 1991–2009

Exhibit 4
SOURCE Authors’ analysis of 1991–2012 data from IMS Health Inc.’s MIDAS database.

The average lifetime profitability of newly launched novel active substances appears to have experienced a “golden age” during the late 1990s, with average lifetime economic profits at $725 million for the 1995–99 cohort. Since then, average economic profits have fallen steadily and sharply, and they became negative (−$111 million) for the 2005–09 cohort.

This pattern holds true for both small-molecule drugs and biologics. In both cases, profits peaked for the 1995–99 cohort and fell for the 2000–04 cohort, by 28 percent for small molecules and 45 percent for biologics. For the 2005–09 cohort, the average net return for small molecules was negative, at −$186 million, while that for biologics was barely positive, at $93 million.

For all four cohorts, the average profits were larger for biologics than for small-molecule drugs, although the gap between them has been declining. For the 1991–94 cohort, this difference—or premium in average returns of biologics over small-molecule drugs—was about $745 million. It more than doubled for the 1995–99 cohort but declined thereafter. Hence, the premium seen in earlier periods essentially vanished for the most recent cohort.

We conducted a sensitivity analysis across five variables—operating expenses, the global sales multiplier, tax rates, erosion scenarios, and the opportunity cost of capital—to understand the impact on returns if our base-case assumptions were altered. Overall, changes made to expenses, the global sales multiplier, and the cost of capital had the greatest impact on returns (for details, see the Appendix). 12

We also performed an analysis on the 2005–09 cohort, setting bounds on the likely upper limit of revenues resulting from line and patent extensions that have not yet occurred. Extending the protected life of small-molecule drugs in the most recent cohort to align with that observed in the 1990s resulted in a 10.3 percent increase in the present value of lifetime global net sales. The returns for the 2005–09 cohort remained negative, averaging −$25.7 million, but they were close to zero. The methods and results of this sensitivity analysis are discussed further in the Appendix. 12

Discussion

Many industry observers predicted that the profitability of the pharmaceutical industry would be reduced significantly as a result of the expiration of patents on “blockbuster” drugs during the past decade (known as the “patent cliff” era). However, there has been little discussion of the extent to which patent-protected and newly launched brand-name novel active substances also experienced declines in profitability relative to earlier cohorts.

Our research shows that the present values of lifetime global net sales for new novel active substances have declined. This has occurred even as the drugs’ relative sales volatility has remained persistently high and has increased for biologics.

To our knowledge, previous research has not focused on the risks associated with highly volatile sales revenues among novel active substances, but instead has pointed to risks incurred during drug development. 6 Additionally, there is some evidence that R&D costs have continued to escalate, increasing risk further. The most recent data from Joseph DiMasi and coauthors suggest that R&D costs may be 70–75 percent higher for compounds that first entered clinical trials from 1995 to 2007 than the costs we used, 14 which were based on products that entered clinical testing from 1990 to 2003. The recent increases appear to be driven by lower clinical approval success rates and higher attrition.

Perhaps we should not be surprised by the finding of declining economic profits. It would be highly inaccurate to characterize the pharmaceutical industry as being perfectly competitive, given intellectual property protection, marketing exclusivity, and other barriers to entry. However, even imperfect competition can result in positive economic profits’ being competed away.

Economic theory teaches that over a long period of time, in the absence of barriers to market entry or exit, competition ensures that economic profits are close to zero. That is, on average, firms’ earnings equal the opportunity costs of capital. In pharmaceuticals, however, patents that protect intellectual property and marketing exclusivity provisions that are provided by statute are intended to encourage investments in innovation through the profits accumulated during a finite period of market exclusivity. As a result, economic profits can be positive—that is, earnings on investments can be greater than the opportunity costs of capital—over sustained periods. However, to the extent that competition exists among brand-name therapeutic alternatives, and generic or biosimilar alternatives become increasingly important, the positive impacts on sustained corporate profitability are mitigated, and consumers and taxpayers benefit from downward pressure on prices.

Indeed, as we have seen, the economic profits of newly launched novel active substances have been driven close to zero, so that returns on investment end up being very close to the opportunity costs of capital. This drop in returns may be due to a number of factors, including the specific attributes and attractiveness of the products recently supplied to the market as well as the result of new demand forces in the market. For example, on the demand side, downward pressures on price have occurred because of consolidation among payers, wholesalers, and pharmacy benefit managers; increased experience with cost containment strategies such as multi-tier formularies; and a greater focus on incremental value in coverage decisions.

As a result, in recent years the forces of market competition that operate on both the supply and the demand sides have essentially eroded the economic profitability of newly launched brand-name pharmaceuticals. Payers and consumers benefit from such downstream pressure on prices in the short term, but sustainable levels of returns are needed to maintain continued innovation. Notwithstanding the recent introduction of such blockbusters as Sovaldi (sofosbuvir) to treat adults with chronic hepatitis C, it remains to be seen whether novel active substances launched since 2009 will again provide robust economic profits, on average.

Policy Implications

Total R&D expenditures by the pharmaceutical industry have been growing consistently since the early 1990s, but they appear to have plateaued in 2008. 15 Recent data also suggest declining venture capital deals and other investment in the pharmaceutical and biotechnology sectors since 2011. 16 Additionally, biotech’s share of venture capital investments has declined since 2007, and private equity investments in biotech have fallen since 2010. 16,17

Recent investments in biologics have delivered higher returns than those in small-molecule drugs. However, biologics’ returns are also declining, and new policies that might affect the sales or the costs sides of the equation will likely reduce biologics’ profitability.

For example, the FDA’s recent establishment of a regulatory pathway for biosimilar approval is likely to result in more rapid reductions in brand-name drug sales at the end of patent life than has been the case historically. The situation with biosimilars is likely to parallel (but not necessarily mimic) what has been the case when less expensive small-molecule generic drugs come on the market and when payers use a range of tools, including preferred formularies, to favor generics over more expensive originator versions. Future studies will be needed to examine this impact further.

The low levels of profit for recent launches plus the high level of sales volatility observed in this study indicate that policy changes intended to ensure an acceptable level of returns may be necessary. For instance, recent efforts to speed drug approvals through the awarding of breakthrough therapy designations for drugs that address life-threatening or other serious conditions may partially offset declining returns, thereby helping spur continued innovation, investment, and patient access.

Because of the importance of sales and reimbursements outside of the United States to total lifetime revenues, sustainable global reimbursement policies are also critical. As all countries seek to secure affordable medicines for their populations, some policy initiatives have restricted access or reimbursement. An example is the use of health technology assessments in Europe and elsewhere.

Health care costs have risen in the United States and globally, but the ability of innovative medicines to prevent costly complications of disease should not be overlooked. And while total pharmaceutical costs to the US health care system have risen, they have done so more slowly than overall health spending, frequently offsetting costs in other areas. For example, in 2012 the Congressional Budget Office began incorporating medical savings associated with the increased use of medicines among Medicare beneficiaries into its scoring estimates for federal health spending. 18 This and other recent evidence 19–23 suggests that better use of medicines can lead to lower health costs overall, and raises questions about the potential impacts on investment if drugs become the particular focus of additional cost containment measures.

Conclusion

Our study demonstrates that although returns on drug development peaked in the late 1990s and early 2000s, they have declined to their lowest levels in two decades, as a result of a combination of declining growth in demand and increasing R&D costs. If such levels persist, the negative returns would likely not be sufficient to sustain medical innovation over the long term. Recent public attention focusing on highly successful launches (such as that for sofosbuvir) needs to be weighed against less publicized unsuccessful launches and declining economic returns. This study calls into question pharmaceutical companies’ ability to maintain their historical high levels of innovation to advance medical science, and it asks whether policy changes will be required to provide needed incentives for investment or to mitigate risk.

ACKNOWLEDGMENTS

This research was supported in part by Pharmaceutical Research and Manufacturers of America, which provided funding for the data analysis undertaken by the IMS Institute for Healthcare Informatics. Ernst Berndt’s research was not funded. The authors acknowledge the contributions of Lauren Caskey, research manager at the IMS Institute for Healthcare Informatics, and Silvia Valkova, engagement manager at IMS Government Solutions, for their analyses, which laid the foundation for the content of this article.

NOTES

Loading Comments...