The costs of developing new medicines and diagnostics reflect both the technical complexities of product development and costs related to regulatory approval, which requires clinical trials to establish product safety and efficacy. Although the relative contributions of these two components are difficult to distinguish empirically—and even conceptually—there is general consensus that increasing regulatory requirements have contributed to the rising costs of new product development in the United States. In considering the costs of new product development for diseases prevalent in low-income countries (LICs), we attempt to identify those costs that might be influenced by regulatory policy as opposed to the unavoidable costs resulting from the hard science of new product development.
R&D Costs for Drugs for Industrial Countries
The most detailed evidence on the cost of developing new drugs is from DiMasi, Hansen, and Grabowski (2003), who estimate the cost of bringing a compound to market at US$802 million in 2000 dollars. Their estimate is based on U.S. data from 10 major companies for a randomly selected sample of 68 compounds that entered human testing between 1983 and 1994 and reached approval between 1990 and 2001. The 68 compounds include 61 small molecule chemical entities, 4 recombinant proteins, 2 monoclonal antibodies, and 1 vaccine. Together, the 10 companies accounted for 42 percent of R&D by U.S. companies. The cost estimates are based on project-level data obtained from the companies for the period 1980-99. The sample was restricted to compounds that originated within these companies to avoid omitted costs of in-licensed products.
Earlier studies using similar data and methods found significantly lower R&D costs for drugs launched in the 1970s and 1980s (DiMasi and others 1991; Hansen, 1979). For the 1980s drug cohort, the estimate was US$359 million per NCE (U.S. Congress, Office of Technology Assessment 1993). Thus, the estimate for the 1990s drug cohort of US$802 million represents a significant increase over and above inflation.
Three main factors contribute to this high and rising cost of R&D. Understanding the contribution of each of these factors is important to understanding whether drug R&D costs might be lower in developing countries.
First, the inputs into pharmaceutical R&D are costly, including highly trained scientists, highly specialized capital equipment, expensive animal studies, and clinical trials involving thousands of human subjects that are often coordinated across multiple countries. Clinical trial out-of-pocket costs reflect expenditures on patients' medical treatment and monitoring, data collection, and analysis. In the study by DiMasi, Hansen, and Grobowski (2003), the average expected clinical cost, adjusted for the probability of entering each clinical phase, was US$60.6 million per compound entering human trials. In addition, the authors estimated that the out-of-pocket costs of drug discovery and preclinical development account for 30 percent of overall R&D costs, raising the total expected out-of-pocket cost to US$86.8 million per compound entering clinical trials. The average number of clinical trial patients per compound was 5,303, and the average cost per patient was US$23,500 before adjusting for the probability of entering each clinical phase.
Second, in the United States, the Food and Drug Administration (FDA) approves only roughly one in five compounds that enter human clinical trials.1 The costs incurred for the four out of five compounds that failed must be included as costs of bringing one new compound to market. Failures occur because of safety concerns, lack of significant efficacy, and poor economic prospects. Even though the new technologies of drug discovery should eventually improve predictive accuracy for both safety and efficacy, success rates were no better in the 1990s than in the 1980s (DiMasi, Hansen, and Grabowski 2003; DiMasi and others 1991). Adjusting for failure rates raises the total out-of-pocket cost from US$86.8 million to US$403 million per approved compound.
Third, the US$802 million total cost estimate includes the opportunity cost of capital over the roughly 12-year investment period. Using an 11 percent real (net of inflation) cost of capital, DiMasi, Hansen, and Grabowski (2003) estimate the total cost of capital at US$399 million. This figure represents the return that shareholders would have received had they invested in activities that yielded immediate returns rather than in the lengthy drug discovery process. If pharmaceutical R&D is financed by—and hence must compete for—private equity capital, shareholders must be compensated for this opportunity cost. Thus, the cost of capital is appropriately included as a cost of R&D if the R&D is undertaken in commercial firms and financed by equity capital. As discussed later, if not-for-profit organizations finance R&D, the opportunity cost of capital may be lower. If we assume financing by private equity, adding the US$399 million cost of capital to the US$403 million out-of-pocket cost yields US$802 million as the capitalized cost at launch, before taxes, per approved compound. The after-tax estimate is considerably lower because, like any business expense, R&D expenses are tax deductible, plus R&D tax credits may be available in certain circumstances. However, for purposes of comparing the costs of R&D to the revenues a commercial firm would require to cover these costs, if costs are measured net of tax, then revenues must also be measured net of tax, in which case adjusting for tax makes little difference. Hereafter we use the before-tax R&D cost estimates to facilitate comparison with other estimates of R&D costs. The before-tax estimates are also most relevant to not-for-profit firms and PPPs that are not subject to taxes.
If commercial firms facing a commercial cost of capital and with no in-kind contributions (see the next section) undertake R&D, they might be able to save roughly 10 to 20 percent of their costs by conducting trials in developing countries and possibly more if they adhere to the countries' regulatory requirements, which may permit fewer and shorter trials than are normal for the FDA. Whether firms could realize those potential savings may be a matter of judgment depending on perceived liability risks. If commercial firms conduct R&D for LIC diseases in not-for-profit spinoffs, they may realize tax advantages and a lower cost of capital, which would further reduce their cost below the US$802 million estimate.
R&D Costs for Drugs for Developing Countries
Recent studies by two PPPs that focus on new product development for diseases in developing countries yield much lower cost estimates for drugs in their portfolios than those in the previous section. The GATB and the MMV estimate the costs of R&D at US$150 million (MMV 2002) and US$178 million (midpoint of the range of US$115 million to US$240 million) (GATB 2001, 101) or less than a quarter of DiMasi, Hansen, and Grabowski's (2003) estimate of US$802 million. The reasons for these large differences are instructive.
First, the GATB and MMV estimates reflect only out-of-pocket costs, with no allowance made for the opportunity cost of capital. Nevertheless, the estimates of out-of-pocket cost are less than half of the US$403 million out-of-pocket cost estimated by DiMasi, Hansen, and Grabowski (2003). This difference in out-of-pocket costs primarily reflects two factors: (a) fewer clinical trials and, hence, fewer patients in trials—namely, 1,368 patients per drug for the GATB compared with 5,303 in the DiMasi, Hansen, and Grabowski (2003) study—and (b) lower costs per patient of US$1,000 to US$3,000 for the GATB for trials run in developing countries compared with US$23,500 per patient in the DiMasi, Hansen, and Grabowski (2003) study.
Some drugs for LICs may require fewer trials, fewer patients, or both per trial because of differences in drug types and trial objectives and different regulatory requirements. For example, some of the drugs in the two PPPs' portfolios are modifications of existing drugs for which some data have been established. R&D costs for LIC drugs may also be lower to the extent that these drugs are tested for fewer indications, with less within-sample stratification by patient subgroup and less need to test for drug interactions. Clinical effects for infectious and parasitic diseases may also be greater than for chronic diseases, which permit smaller trial sizes.2 The lower trial cost per patient for LIC drugs partly reflects the lower costs of conducting trials in developing countries, with much lower costs of medical care and personnel than in the United States. The trial duration may also be shorter because the target diseases are acute rather than chronic. To the extent that the lower out-of-pocket clinical costs in the GATB and MMV studies reflect fewer patients in trials and lower cost per patient, such savings could, in principle, apply to LIC drugs regardless of whether these drugs are developed by not-for-profit or commercial enterprises.3
Another factor contributing to the lower out-of-pocket costs reported by the MMV and the GATB is that these PPPs benefit from in-kind contributions of personnel, technologies, and other resources supplied by their industry and academic partners. The MMV estimates these in-kind contributions as equivalent to its own incurred costs. Thus, if these in-kind contributions are included, the full social cost for developing LIC drugs increases to US$250 million to US$300 million per compound, or only 25 to 35 percent less than the DiMasi, Hansen, and Grabowski (2003) estimate of US$403 million. However, as long as such in-kind contributions are available without charge to PPPs, the actual budget cost to PPP funders is only US$150 million to US$178 million, or less than half DiMasi, Hansen, and Grabowski's (2003) estimate.4
The second major determinant of R&D costs is failure rates. The GATB and MMV estimates show overall drug failure rates similar to those in DiMasi, Hansen, and Grabowski's (2003) study. Indeed, there is no obvious reason to expect significant differences in failure rates if LIC drugs face similar scientific challenges and are reviewed by the FDA or the European Medicines Evaluation Agency applying the same safety, efficacy, and risk-benefit tradeoff standards as are applied to drugs for the industrial countries. However, if the regulatory review of LIC drugs uses risk-benefit tradeoffs that reflect conditions in developing countries, then success rates might be higher, implying a lower budget cost per approved compound for LIC drugs.
Finally, the third major contributor to R&D costs is the opportunity cost of capital, which accounts for US$399 million, or almost half of DiMasi, Hansen, and Grabowski's (2003) US$802 million cost per compound. The GATB and MMV estimates do not include the cost of capital. Whether the cost of capital should be included in estimating the cost of R&D for LIC drugs depends on the circumstances and the perspective. If LIC drugs are to be developed by commercial firms that must generate a competitive return for their shareholders, then the cost estimates appropriately include a cost of capital at roughly 11 percent, as in the DiMasi, Hansen, and Grabowski (2003) study. However, if LIC drugs are developed by PPPs or other not-for-profit institutions with financing from philanthropic or governmental agencies, the opportunity cost of capital may be lower if these funders typically do not require a rate of return on their investment to compensate them for the forgone alternative uses of the funds during the investment period. For example, government investments sometimes assume a social opportunity cost of capital of about 5 percent. Using a 5 percent cost of capital for financing from philanthropic or governmental agencies implies a roughly 50 percent markup over out-of-pocket R&D costs to reflect the cost of capital rather than the roughly 100 percent estimated byDiMasi, Hansen, and Grabowski (2003), assuming the same time flow of investments.
Applying this markup to the US$150 million to US$178 million estimated out-of-pocket R&D cost for the MMV and the GATB yields a total capitalized R&D cost of roughly US$250 million for LIC drugs if they are developed by PPPs with foundation or government funding, assuming that in-kind contributions are at current levels and that trials are conducted in developing countries. Alternatively, these funders might choose to use a zero cost of capital, reflecting the importance that they attach to developing new medicines to treat currently untreatable diseases and to replace existing drugs that are increasingly ineffective because of resistance. In that case, the appropriate capitalization cost is zero, and the out-of-pocket costs of US$150 million to US$178 million are the full R&D costs per new compound for LIC diseases.
Economics of Vaccine Discovery and Development
In discussions of the economics of vaccine development, comparing the findings with those obtained for pharmaceuticals is useful. Note, however, that the two product categories are different in many fundamental and practical aspects. Pharmaceuticals are used to treat an existing clinical condition with the ultimate aim of reversing the course of disease. By contrast, vaccines are used to prevent a future threat. In addition, pharmaceuticals may be administered over a prolonged time frame and, in many chronic conditions, may be taken from the time of diagnosis for the rest of the patient's life, whereas most vaccines are administered once or a few times.
The costs of vaccine production consist of the traditional components of discovery, process development, scale-up, and manufacturing, as well as the costs pertaining to regulatory requirements, liability, and post licensing studies (Andre 2002; Grabowski 1997). Furthermore, the economic framework for disease prevention (Kou 2002) raises many questions that are less clear than calculating the cost of treatment of a specific pathological condition in an individual or setting priorities for government budgets. Finally, the financing of vaccine purchasing and immunization programs has traditionally been separated from the totality of health care financing. Although this practice may have appeared to be advantageous at some point globally or in individual countries, the current outcome is less than satisfying in that the financing of vaccines is fragmented (Institute of Medicine 2004) and competes at a less favorable level with other budgetary priorities.
Costs of Vaccine R&D
The decision to develop a new vaccine is usually based on medical need, scientific feasibility, and market conditions. Because most currently available vaccines have been developed over relatively long periods and multiple organizations have been involved in their discovery, our cost estimates are based on historical data and on many assumptions that are probably changing rapidly (Agre and Rapkin 2003; Barrett and Parker 2003; Emanuel and others 2004; McMillan and Conlon 2004). The cost elements are similar to those for pharmaceutical R&D except for the specific regulatory procedures for vaccines, such as the completion of plant construction before phase 3 trials.
As noted earlier, estimates indicate that an NCE costs US$403 million to US$802 million in 2000 dollars (DiMasi, Hansen, and Grabowski 2003). Clarke (2002) estimates that a vaccine costs approximately US$700 million by the time the product is marketed, including not only the actual costs of products, but also such items as the cost of failures and the cost of funds (Grabowski 1997). In addition, the size of phase three clinical trials has recently escalated along with costs.
The Institute of Medicine (2004) estimates that total expenditure on vaccine R&D in 1995 was US$1.4 billion. The large pharmaceutical companies accounted for approximately 50 percent of the total (Mercer Management Consulting 1995). However, the current situation is more complex for vaccine research than for drug R&D. In 2004, only five major multinational companies were investing in vaccine R&D and production (Institute of Medicine 2004). In addition, a multitude of smaller, new biotechnology organizations in both developed and developing countries are pursuing multiple vaccine targets that are of considerable value (Nossal 2004). Since September 11, 2001, U.S. government funding for microbial threats that can be used as agents of terror has increased: Project Bioshield is devoting more than US$5 billion during the next 10 years to discovering and producing vaccines and other therapeutics (Herrera 2004). These initiatives may have spillover benefits for vaccines and therapeutics for developing countries.
Another barrier, in addition to complexity and costs that may directly or indirectly affect investment in vaccine R&D, is the condition of the vaccine market. Even though experts anticipate healthy growth in the total global vaccine market from approximately US$6 billion in 2004 to US$20 billion in 2009, the number of large private pharmaceutical companies involved in vaccine research is down to five (Mercer Management Consulting 2002). As a recent Institute of Medicine report (2004) demonstrates, other significant barriers also stand in the way of a well-functioning vaccine research and production system. These barriers include the difficulties of entering the field and of financing research, plus in the United States they include the government's role in determining pricing in relation to the government's purchase of a significant proportion of vaccines. Similar situations arise in other countries. All lead to an under appreciation of the value of vaccines and reduce the incentives for investment in future vaccine products.
As noted earlier, whether the cost of R&D for drugs or vaccines intended for use in developing countries is less than for products targeted to high-income markets is questionable. Certainly, developing vaccines for LICs requires investment from both industrial and developing countries and participation by scientists from both industrial and developing countries. In the case of vaccines, discovery similar to pharmaceuticals is a costly process. Therefore, a research infrastructure has to be supported in academic institutions and private sector and government laboratories for new ideas to emerge and to be tested. The capabilities needed to discover a new HIV or malaria vaccine are different and far more complex than those used to manufacture traditional vaccines such as whole-cell pertussis. Indeed, the technological know-how needed to discover new vaccines is embedded in the advancing edge of science.
Alternative mechanisms of financing and managing the development of new vaccines for the developing world must be identified and may require governmental, international, and philanthropic funding. Appropriate new institutions or alliances could evolve from the multiple PPPs now being pursued. The case has repeatedly been made for a massive infusion of funds and global coordination if vaccines against great killer diseases such as HIV/AIDS are to be developed (Klausner and others 2003).
Effect and Cost of Vaccination Programs
The major societal and health effect of vaccines are realized mainly when immunization programs reach a significant proportion of individuals in a society (Mahmoud 2004). The effect of vaccines in interrupting or preventing the transmission of infectious agents depends on two concepts: inducing resistance in healthy individuals before exposure and extending the umbrella of prevention to the majority of the target population to achieve herd immunity (Anderson and May 1990). When deciding to mount a vaccination program, health professionals face scientific, public health, and financial considerations. The ultimate outcome is a cost structure that has to compete against well-established budgetary constraints and comparisons. The subject of the cost-effectiveness of vaccination programs has been examined at multiple levels and in many settings (Miller and Hinman 1999). The overall conclusion derived from most quantitative techniques—for example, cost-benefit analysis, cost-effectiveness analysis, and cost utility and decision analysis—indicates that vaccination was one of the most effective health measures of the 20th century (CDC 1999).