The Cost of Research: Why Medicines Are Expensive

Ask why a new medicine is expensive and two stories usually collide. In one, a successful product must pay for a decade of research, complex trials, and the many candidates that failed. In the other, a patent-protected company charges what a health system will bear, even when manufacture is cheap and public science supplied the foundation. Both stories contain evidence. Neither can explain a price by itself. The cost of creating a medicine and the price attached to a pack, vial, or infusion belong to connected systems, but they are different quantities.

Research and development begins long before a clinical trial. Laboratories investigate disease biology, identify targets, design or screen candidates, refine their properties, develop tests, and learn how to manufacture experimental material. Most ideas stop. Candidates that reach people move through dose-finding and exploratory studies before large confirmatory trials. Teams must run sites, recruit participants, make and ship trial product, monitor safety, manage data, and satisfy regulatory and ethical standards across countries. Specialist facilities and long follow-up can make a programme especially demanding. Advanced cell and gene therapies add individualised production, testing, and logistics.

Failure is a real economic cost. Money spent on a candidate that proves unsafe or ineffective cannot be recovered from sales of that candidate. A portfolio that yields one approval must therefore fund the projects that ended along the way if private investors are to keep supplying capital. Clinical success also varies widely. One analysis estimated that 13.8 percent of candidates entering phase 1 eventually achieved approval, with lower or higher probabilities across disease areas [1]. The estimate is sensitive to database and method, and it does not mean that each approved drug literally carried the same number of failures. It shows why averages across a portfolio differ from the direct spending on one winner.

The phrase cost of capital adds another layer. A pound spent on research today is tied up for years before an uncertain return. Economic studies often capitalise that spending by adding the return that investors could have expected from an alternative use of the money. They may also adjust for failure by assigning unsuccessful projects to successful ones. Both choices have a rationale, and both inflate the estimate above the cash paid directly for the approved product. The result is an economic opportunity-cost estimate, not an audited invoice. Readers need to know which kind of number they are being shown.

Published estimates vary accordingly. A 2021 systematic review found large differences in methods, data access, included stages, treatment of failure, and assumed cost of capital. Estimates could not be compared responsibly without unpacking those choices, and many studies relied on confidential company data that independent researchers could not verify [2]. The review also found that no approach captured every relevant factor well. This is why a confident claim that a new medicine costs a single universal sum to develop should invite questions about the sample, accounting boundary, success rate, time period, and source of the data.

One influential industry-based study estimated a capitalised pre-approval cost of about US$2.6 billion in 2013 dollars for a new prescription compound. It drew on confidential data supplied by ten firms and included failed projects and an assumed cost of capital [3]. Critics have challenged the representativeness and transparency of that sample. A later study using public company filings for 63 medicines approved in the United States from 2009 to 2018 estimated a median capitalised R&D investment of US$985 million per new medicine, including failures. Those 63 were a nonrandom subset of 355 approvals, with data more often available for smaller firms and certain product groups. Its estimated mean was US$1.336 billion, and the 95 percent confidence interval around the median ran from US$684 million to US$1.229 billion [4]. Public data improved reproducibility while still requiring assumptions to allocate company spending to individual products.

These are estimates of an average or median under specified methods. They do not establish how much was spent on a particular medicine, and they cannot dictate its price. The cost of an unsuccessful programme is unknowable at the start, while a successful medicine may serve a few thousand people or hundreds of millions. A uniform mark-up over historical R&D would reward expensive inefficiency and struggle to allocate shared science. The World Health Organization advises countries against relying primarily on cost-plus pricing for medicines, citing weak transparency and the absence of an agreed framework for which costs count [5].

Manufacturing matters more for some products than others. Once a small-molecule tablet has a mature process, the active ingredient and packaging may cost little relative to the branded price. A sterile biologic requires living production systems, purification, contamination control, cold storage, and careful testing. Vaccines combine biological manufacture with high-volume quality assurance and distribution. A personalised cell therapy may require a patient's cells to be collected, engineered, tested, and returned through a time-critical chain. Raw materials, energy, skilled labour, quality failures, regulatory compliance, and resilient capacity all have costs. Shortages show what happens when a low price fails to sustain enough reliable suppliers.

Clinical trials can be a large part of development spending, especially when outcomes are uncommon, follow-up is long, or many centres are needed to recruit enough people. Trial design also affects social cost. A poorly chosen endpoint or comparator can consume money and participant effort without answering the decision that regulators, clinicians, or payers face. Regulators charge fees and require extensive submissions, inspections, safety systems, and post-market work. Those requirements add expense because society wants credible evidence and consistent product quality. Regulatory cost is therefore part of the bargain, while the existence of regulation cannot explain why two countries may pay very different prices for the same medicine.

Patents and regulatory exclusivities connect the investment story to pricing power. A UK patent can last up to 20 years from filing, which usually occurs well before launch. A supplementary protection certificate can provide up to five further years of protection for an authorised medicinal active ingredient, partly compensating for time used in testing and review [6]. Separate data and market protections can delay reliance by competitors on the originator's regulatory dossier. These time-limited rights are intended to make risky investment attractive by allowing a period without direct generic or biosimilar competition.

Exclusivity does not set a price. It changes the bargaining conditions under which one is set. A company with the only effective treatment for a serious disease has more leverage than a company entering a class with several close substitutes. The number of eligible patients, severity of illness, clinical advantage, competitor prices, payer budgets, and willingness to refuse or restrict coverage all matter. A review of high US prescription-drug prices concluded that market exclusivity and limited negotiating power were central, and found no evidence that product prices tracked R&D costs [7]. The American market is unusually permissive, but the underlying relationship between scarce alternatives and pricing power applies more broadly.

The UK makes that bargaining visible through health technology assessment. The Medicines and Healthcare products Regulatory Agency decides whether a product's quality, safety, and efficacy support authorisation. NICE separately considers comparative clinical effectiveness and value for NHS resources, often estimating cost per quality-adjusted life year gained. Confidential discounts or commercial agreements can bring the effective NHS price below the list price. A medicine may be licensed yet not recommended, or recommended for a narrower group, because regulatory benefit-risk and health-system value answer different questions [8]. Price is negotiated in the shadow of the payer's ability to say no, delay, limit use, or request a discount.

Public funding complicates any claim that the private developer paid the whole scientific bill. Governments and charities support universities, research institutes, clinical networks, patient cohorts, and enabling technologies. An analysis traced publications and US National Institutes of Health grants associated with all 210 new drugs approved by the FDA from 2010 through 2016. It identified more than US$100 billion in relevant NIH funding, over 90 percent associated with basic research on biological targets rather than the products themselves [9]. The method captures a broad knowledge base and cannot assign that entire sum to the eventual drugs. Its point is that public and private investment are complementary parts of the same innovation system.

Industry often takes on costly translation that public laboratories are not structured to complete: optimising a candidate, scaling manufacture, running global trials, assembling the regulatory dossier, and accepting the risk that the asset fails. Companies may also license discoveries from publicly funded institutions or acquire biotechnology firms after early evidence reduces uncertainty. The boundary shifts from product to product. Recognising public contribution does not erase private risk. Recognising private expenditure does not grant a moral entitlement to any price. The debate concerns how rewards and obligations should be divided when knowledge has mixed parentage.

Marketing and administration further widen the gap between development cost and company expenditure. Promotion can inform clinicians about a new option, support training in correct use, and help a product reach eligible patients. It can also expand demand, steer prescribing toward a branded medicine, and consume resources unrelated to generating evidence. A review of US medical marketing found that annual spending across prescription drugs, disease awareness, health services, and laboratory testing rose from US$17.7 billion in 1997 to US$29.9 billion in 2016, with marketing to health professionals the largest component [10]. Those figures are American and historical; they illustrate that commercialising medicine is a separate activity from discovering it.

Accounting comparisons between a company's total R&D and selling, general, and administrative expenditure should be handled cautiously. Categories differ across firms and can contain costs that do not fit everyday meanings of research or marketing. Buying another company may acquire a pipeline without appearing as internal R&D. Share buybacks and dividends reward capital holders rather than produce medicines, although investors price those expected returns when deciding where to place money. Taxes, rebates, wholesaler margins, dispensing fees, and confidential discounts also separate a manufacturer's list price from its net revenue and from what a patient or health service ultimately pays.

Competition can change the equation dramatically. When patents and exclusivities expire, generic manufacturers can rely on evidence that their version contains the same active substance and performs equivalently, avoiding repetition of the original efficacy trials. Biosimilars require a more extensive comparability exercise because biological products are complex, while they also avoid rebuilding the entire clinical dossier. Entry can lower prices, though the size and speed of the fall depend on the number of competitors, tender design, manufacturing difficulty, prescribing behaviour, and litigation. Old medicines can still become expensive when supply is concentrated and competition weak, demonstrating again that low historical R&D does not guarantee a low market price.

High prices feed back into research choices. Expected revenue attracts investment toward diseases and products with a plausible paying market. Antibiotics present the opposite problem: stewardship rightly limits use of new drugs, weakening sales even when society urgently needs a reserve against resistance. Neglected diseases concentrated in low-income countries may offer too little commercial return. Orphan incentives can stimulate work in rare conditions, while small populations can also support exceptionally high per-patient prices. Grants, prizes, advance purchase commitments, public-private partnerships, subscription payments, and patent pools are attempts to align research with health need when ordinary sales incentives misfire.

Access is where the abstractions become consequential. A medicine that exists but cannot be afforded, supplied, or delivered does not produce its potential health benefit. High-income health systems can negotiate and prioritise, though every expensive purchase carries an opportunity cost. Lower-income countries face tighter budgets, weaker bargaining power, diagnostic and delivery constraints, and sometimes later market entry. WHO's fair-pricing work treats affordability and a sustainable supply as joint goals [11]. Driving a price below reliable production cost can create shortages; tolerating every monopoly price can exclude patients and destabilise health budgets.

Better policy begins with refusing a false choice between innovation and access. Transparent product-level R&D data would make cost claims more testable. Trial registration and results reporting can reduce waste. Comparative evidence can reward meaningful improvement over duplication. Strong purchasing, generic and biosimilar competition, and licensing mechanisms can moderate prices. Public or philanthropic funding can target neglected needs, with affordability conditions considered when contracts are written. Each tool has trade-offs, and a country that pushes prices down may benefit from research financed by revenues elsewhere. That international collective-action problem is real, though it does not turn any particular price into a scientifically determined number.

Medicines are expensive for several reasons that operate at different levels. Discovery, failure, trials, manufacture, quality systems, and surveillance consume real resources. Patents and regulatory exclusivity create a temporary shelter from competition so that investors may recover risk and earn returns. Within that shelter, clinical value, scarcity, payer power, strategy, and politics shape the price. Public science helps create the opportunity, and marketing helps create the market. The honest answer therefore resists both slogans. Research cost helps explain why society built a reward system. It does not, by itself, explain the size of the bill.