EMA AI Paper, NICE RWE Pilot, CMS Value Model, ISPOR Digital Framework

MarcusWelcome to the Access Brief — your daily briefing on what's moving in HEOR, HTA, and market access. I'm Marcus, and it's great to have you with us today.

SaraAnd I'm Sara. Always good to be here — and I'll say, the AI reflection paper from EMA has me particularly interested today.

MarcusSame here. We're looking at EMA's new reflection paper on AI in drug development — why it signals a major shift in regulatory evidence. Then NICE's pilot for real-world evidence in early appraisals — what this means for speed and evidence quality. And CMS's new value-based purchasing model for hospital conditions — the implications for drug pricing in the US. Finally, ISPOR's new framework for digital therapeutics — how it's changing value assessment.

SaraThat second one about NICE's pilot is really intriguing — the budget impact question there is one the field hasn't fully worked out yet.

MarcusExactly. Let's get into it.

MarcusThe EMA has just published a reflection paper on the use of artificial intelligence in drug development. This is significant because it's the first time they've laid out their thinking on how AI might be used in generating evidence for regulatory submissions. The paper highlights areas like AI for clinical trial design, patient recruitment, and data analysis.

SaraWhat strikes me about this is the emphasis on data quality and transparency. The EMA is clear that while AI can accelerate development, it must not compromise the robustness of the evidence. They're calling for detailed documentation of AI algorithms and validation processes. That's a key point for HEOR teams because it means we'll need to adapt our evidence strategies to include AI-generated data.

MarcusExactly. And the paper also touches on the potential for AI to generate real-world evidence that could be used in HTA submissions. But here's the challenge: how do we ensure that AI-generated RWE is comparable to traditional evidence? The EMA is asking for more research on this, which could lead to new guidelines.

SaraThat's fair, though I think payers would see this differently. They might be concerned about the interpretability of AI models. If an algorithm is a black box, how can we trust the evidence it produces? The EMA's focus on transparency is a step in the right direction, but I wonder if that's the full picture though. We also need to consider how HTA bodies will validate AI-generated data.

MarcusThat's a valid point. And it connects to something I keep coming back to: the need for cross-disciplinary collaboration between regulatory, HEOR, and HTA teams. If we don't align early, we risk having AI evidence that's acceptable to regulators but not to payers. The EMA paper is a starting point, but the real work will be in creating common standards.

SaraRight, and from the payer side, we're already seeing some HTA bodies expressing caution. For instance, NICE has recently emphasized the need for human oversight in any AI-evidence submissions. So while the EMA is pushing innovation, payers are pushing for guardrails. That tension is going to be interesting to watch.

MarcusAgreed. And the paper also mentions the potential for AI to reduce development costs, which could lead to lower prices. But that's one of those stories where the reality is more complex. If AI development requires new expertise and tools, the costs might not decrease as much as hoped. We'll need to see how this plays out in practice.

SaraExactly. The part that gives me pause is the timeline. The EMA is saying this is a reflection paper, not binding guidance. So it's going to take time for these ideas to become concrete. Companies will need to invest in AI capabilities now, but the regulatory clarity is still evolving. That's a tricky balance.

SaraNow, NICE has announced a pilot program to incorporate real-world evidence into early technology appraisals. This is a big deal because NICE traditionally relies on randomized controlled trials for early evidence. The pilot will allow RWE to be used in the initial stages of appraisal, potentially speeding up assessments for innovative therapies.

MarcusThat's right. The pilot is focused on areas where RWE is most promising, such as in rare diseases or for expanding indications. NICE is hoping that RWE can fill evidence gaps when RCTs are not feasible or ethical. But I wonder if this will lead to a lower bar for evidence? RWE can be prone to bias, and NICE will need robust methods to ensure its validity.

SaraThat's one read — I'd frame it slightly differently. The pilot is actually about increasing the flexibility of evidence generation. NICE recognizes that for some therapies, RCTs take too long and may not capture real-world effectiveness. By using RWE early, they can make faster decisions, which is crucial for patients. But the key will be the methods they use to assess RWE quality. They're proposing a new framework for evaluating RWE in appraisals.

MarcusWhat strikes me about this is the potential for RWE to change the cost-effectiveness calculations. If RWE shows better real-world outcomes than RCTs, that could make therapies more cost-effective. But on the other hand, if RWE shows worse outcomes, it could be a barrier. How do you think this will impact the threshold for cost-effectiveness?

SaraThat's a critical question. NICE is maintaining that RWE will be used alongside other evidence, not as a replacement. But in practice, if RWE is used early, it might set the direction of the appraisal. And we know that RWE can sometimes show different results than RCTs. For example, in oncology, RWE often shows lower effectiveness due to real-world factors. That could lead to lower cost-effectiveness estimates, making it harder for drugs to get recommended.

MarcusExactly. And this connects to something I keep coming back to: the need for transparency in RWE. NICE will need to be very clear about how they're using RWE and what weight they're giving it. Otherwise, companies might invest in generating RWE that's not actually used in the appraisal. That would be a waste of resources.

SaraRight, and from the payer side, we're concerned about the reproducibility of RWE. If different datasets lead to different results, how can we be confident in the evidence? NICE is proposing to use standardized RWE datasets, but that's easier said than done. The field hasn't fully worked out the methodology for RWE in early appraisals, and this pilot will be a test case.

MarcusCMS has announced a new value-based purchasing model for hospital-acquired conditions. This model will tie Medicare payments to hospitals based on their performance in reducing these conditions, such as infections or falls. For drug companies, this could mean new opportunities for therapies that prevent these conditions, as hospitals might be more willing to adopt them to improve their performance.

SaraThat's an interesting angle. But what I'm thinking about is the budget impact. If hospitals are rewarded for reducing these conditions, they might be more likely to cover preventive therapies. But the model is still in the proposal stage, and we don't know the specifics. For example, how will CMS measure the outcomes? And what thresholds will hospitals need to meet? Those details will determine how much this actually moves the needle.

MarcusExactly. And the model is part of CMS's broader shift toward value-based care. But hospital-acquired conditions are a specific focus. The idea is that by preventing these conditions, hospitals can reduce costs and improve patient outcomes. For drug companies, this could open up new markets for drugs that prevent infections or falls. But the challenge will be demonstrating that the drug actually leads to the desired outcome in a real-world setting.

SaraThat's one read — I'd frame it slightly differently. The model is more about creating incentives for hospitals to invest in prevention, which could include drugs, but also other interventions. And the budget impact for payers might be mixed. On one hand, if hospitals reduce these conditions, Medicare could save money. On the other hand, if the model leads to more use of preventive therapies, it could increase costs. It's a classic value equation: are the savings greater than the costs?

MarcusWhat strikes me about this is the potential for new evidence requirements. To be included in this model, hospitals might need to show that a drug is effective in reducing the condition. That could mean more real-world evidence studies. And for companies, this could be an opportunity to generate data that supports both the hospital model and traditional HTA submissions.

SaraThat's a good point. And it connects to something I keep coming back to: the need for data that speaks to multiple stakeholders. If a drug can demonstrate value in a hospital setting, it might be easier to get coverage from both hospitals and payers. But the challenge is that hospital data can be messy, and it might not be comparable across institutions. CMS will need to standardize the data collection to make it meaningful.

SaraFinally, ISPOR has released a new framework for assessing the value of digital therapeutics in chronic disease management. This is important because digital therapeutics are a growing class of interventions, but there's no standardized way to assess their value. The framework provides guidance on how to measure outcomes, cost-effectiveness, and budget impact.

MarcusThat's a much-needed development. Digital therapeutics are often software-based, so traditional RCTs might not capture their real-world impact. The ISPOR framework emphasizes the use of real-world evidence and patient-reported outcomes. But I wonder if this will be enough to convince payers. Digital therapeutics can be expensive to develop and deploy, and the evidence might not be as robust as for drugs.

SaraThat's fair, though I think payers would see this differently. The framework is actually designed to address payers' concerns by providing a structured approach. It includes guidance on how to handle issues like interoperability with other systems and long-term adherence. These are key pain points for payers. And by standardizing the assessment, it could make it easier to compare digital therapeutics to traditional treatments.

MarcusWhat strikes me about this is the potential for this framework to influence HTA bodies. If ISPOR, which is a respected organization, endorses this approach, it might lead to more consistent assessments across different countries. But the challenge will be adapting HTA methodologies to account for the unique aspects of digital therapeutics, such as their iterative nature and the need for ongoing data collection.

SaraExactly. And this connects to something I keep coming back to: the importance of defining the comparator. For digital therapeutics, what are they being compared to? Usual care? Other digital interventions? The framework provides guidance on this, but it's still a challenge. And from a budget impact perspective, digital therapeutics might have lower upfront costs but higher long-term costs if they require ongoing support. The framework helps to break down these costs.

MarcusThat's a good point. And I think companies developing digital therapeutics will need to generate evidence that fits this framework. But the framework is also flexible enough to accommodate new types of evidence. For example, it allows for the use of data from digital health records. This could make it easier to generate the necessary evidence. But how do you see this affecting the adoption of digital therapeutics by payers?

SaraI think it will help by providing a common language. Right now, each payer might have their own way of assessing digital therapeutics. With a standardized framework, they can compare more easily. And the framework includes guidance on value-based contracting, which could make it easier for companies to negotiate reimbursement. But the proof will be in the pudding. We need to see if HTA bodies actually adopt this framework and if it leads to more positive coverage decisions.

SaraA lot to think about today. I'll be watching how NICE's RWE pilot plays out in the first few appraisals.

MarcusSame — and for me the thread running through today is the increasing need for innovative evidence generation methods to keep pace with new technologies and payment models. Worth sitting with.

SaraThanks so much for listening — really glad you're here with us.

MarcusWe'll be back tomorrow. Show notes and transcripts at outcomes-analytica.no. See you then.

SaraThanks for listening — see you tomorrow.

MarcusBack tomorrow on Access Brief. Show notes at outcomes-analytica.no.