Pharma’s most expensive question is the last one asked

Pharmaceutical firms are making an expensive mistake. They develop a drug, see it through clinical trials, and only then pause to ask whether it will secure reimbursement. But by then, capital has already been committed, and the drug has already been built. The most important – and most costly – decisions have already been made.

Policies like the United States' Most-Favoured Nation (MFN) and the EU's Joint Clinical Assessment (JCA) have compounded commercial pressures for firms. While it has long been true that regulatory approval alone is not enough to determine a drug's success, JCA and MFN have made it so that a drug's commercial viability is determined even earlier in development. These commercial requirements can’t be retrofitted onto an otherwise successful drug. They increasingly shape what evidence needs to be generated during development itself.

Firms that don't adapt their development process to meet this change will increasingly find themselves confronted with commercial failure after scientific success. But AI is changing the economics of drug development, removing the financial constraints that previously made it difficult to conduct rigorous commercial assessment as soon as it becomes clear that a drug is clinically viable.

The commercial stakes have increased

Failure to demonstrate cost-effectiveness and commercial value has very real consequences that have been playing out for some time. A recent study comparing HTA assessment rejections across seven OECD countries found a 12.9% rejection rate of otherwise successful drugs. Those weren't scientific failures. They were commercial failures: drugs that had already demonstrated efficacy, but which lacked the evidence, value proposition, or economic case needed for reimbursement. By the time those shortcomings emerged, they were too late to fix.

Now, MFN and JCA have raised the commercial bar even higher. MFN, for example, ties US prices more closely to those paid in other developed markets, meaning reimbursement outcomes in Europe now have much greater implications for US pricing and revenue (The White House). And, in Europe, the Joint Clinical Assessment raises the evidence bar higher still, requiring a major clinical evidence package to support reimbursement decisions across all 27 EU Member States (European Commission).

These requirements aren’t going to be satisfied by a dossier of evidence hastily pulled together at the last minute. Health economics and outcomes research (HEOR) needs to be brought in sooner – either after Phase II and before Phase III starts, or immediately once Phase III is finished. Rather than being treated like a set of conditions to be satisfied at the end of the process, it should be used to actively shape a drug around its commercial risks, strengthening its business case and improving its reimbursement prospects in the process.

AI is changing the economics of decision-making

Historically, this hasn't been practical. HTA-grade evidence synthesis and economic modelling could take up to five years and substantial investment. Unable to justify the spend on 20 candidate programmes, analysis has often been left until the end of development, taking place only once a drug was nearly finished.

AI has changed that. It isn't simply making HEOR faster, it's also shifting what's commercially viable to analyse. With AI, work that once took years – from lengthy literature analysis to economic modelling – can now be executed in weeks, if not days. When rigorous analysis costs a fraction of what it once did, questions that were previously too expensive to ask across an entire portfolio can now be answered before investment decisions are made.

It also presents an opportunity to design trials against the questions assessors will actually ask. Too many Phase III programmes are powered to satisfy a regulator and then arrive at HTA missing the comparator or the endpoint that payers care about; a gap no amount of post-hoc statistics can close.

Commercial analysis must shape development

This changes the purpose of HEOR entirely. By carrying out HTA-grade assessment before capital is committed, companies can identify which indications, comparators, and evidence packages are most likely to succeed commercially. Developers can shape the drug around the commercial reality of the markets they’re targeting, rather than discovering at submission that the economic case was never strong enough.

None of this means delegating decisions to a machine. AI is an incredible accelerator of analysis. It can parse literature quicker than any human, and gather together evidence in minutes when it would have taken a team of researchers months. But humans must keep the judgement calls, the framing of questions, and the final recommendations.

It could be said that the commercial evidence bar is rising at exactly the right time. Yes, there are more conditions to satisfy – but the advent of AI also makes it easier and more affordable to meet them.

Gathering and analysing this evidence is not the sort of thing you should be starting at the end of the development process, but as soon as it becomes clear that a drug is clinically viable: immediately after either Phase II or Phase III trial. Whether a drug can secure reimbursement is pharma's billion-dollar question, and the companies that succeed will be those that answer it sooner, rather than later.

About the author

Tim Reason is the founder of Estima Scientific, an HEOR consultancy that applies AI to accelerate evidence generation, synthesis, and technical analysis for pharmaceutical launches. Prior to founding Estima Scientific, Reason contributed to the development of NICE clinical guidelines at the Royal College of Physicians. He has co-authored papers in journals such as Value in Health and PharmacoEconomics, and regularly speaks at leading industry conferences, sharing insights into the future of AI-driven evidence generation and the evolving role of HEOR in pharmaceutical strategy.