In the last issue of Curta on Call, I wrote about what appeared to be the start of a trend towards integration of HEOR groups into Medical Affairs or Market Access teams among major pharmaceutical companies. To my surprise, that article went viral (as much as that means in our world). I also received dozens of comments, some within minutes of posting. Since then, I have heard from other colleagues who have told me that their HEOR groups have experienced the same fate. There doesn’t appear to be anywhere to hide from this brave new world.*
It’s hard not to think a bit more about a subject when it gets that much attention. In this issue, I will share some reflections on what it might mean for our work.
But first, the smoking gun.
Last year, McKinsey and Company published, “A vision for medical affairs 2030: Five priorities for patient impact”(1). Part manifesto, part instruction manual, this document appears to be the blueprint for what we are seeing happen to pharma HEOR groups in 2024. A textbook example of McKinsey hubris, it starts by congratulating itself for the foresight of its earlier paper predicting the rise of Medical Affairs. After that, the report goes on to offer several thoughtful insights about many of the forces impacting pharma and drug development, and how companies should adapt to remain competitive.** It soon becomes clear where HEOR fits into this picture, but what struck me was something else they say over and over: the winners will move faster. And I’ll tell you this—we all better get ready.
*The exception might be small biotech and pharma groups. The HEOR folks that I know in those firms have told me that their jobs are secure (at least until a big pharma buys them!).
**Pharma-based HEOR experts: Take a proton pump inhibitor and give it a read.
Accelerate
That word comes up again and again in the document. Let’s face it, accelerate is the force majeure driving pharma today. First, use machine learning to mine massive genomic and proteomic databases for faster identification of “druggable” targets. Next, create entire new “tech enabled” clinical trial infrastructures to get drugs through trials faster.† Finally, after FDA approval, use artificial intelligence to mine real world data, synthesize evidence, and in McKinsey-speak, “achieve next-level patient impact.”‡ Speaking of AI, we may not even have to rely on ever-so-slow humans to write scientific papers summarizing the findings (2).
† Even pharmacy chains are building out clinical trials capabilities.
‡US President Harry Truman once famously said, “Give me a one-handed Economist. All my economists say, ‘on ONE hand…’, then ‘but on the other…”
Culture Clash
So how will the forced marriage of HEOR and Medical Affairs play out? It has the potential to be a rocky relationship. From the Medical Affairs perspective, the stereotypical HEOR person tends to be viewed as “academic” in a negative sense: pedantic, favoring complex, dense analyses simply for rigor’s sake, and having a hard time reaching conclusions (see the “further research is required” sections of almost every HEOR paper). And worst of all… slow. Certainly, the C-suites in most pharmaceutical firms have little time or patience for the type of extended pro/con discussions that made at least one US president want to “remove a hand” from every economist that worked for him.
From the HEOR perspective, many of us have had a “where did THAT come from?” reaction when reviewing estimates from colleagues in Medical Affairs. Overly optimistic forecasts about disease burden and eligible patients. Downplaying uncertainty about incremental benefits compared to existing treatments. Perhaps there is a kernel of truth to this typecast, but over many, many years of interactions with Medical Affairs teams, I have come to appreciate the forces behind their worldview. Yes, they often create data-light estimates, but those estimates are often made very early in product development when several or even dozens of potential products are being reviewed for investment. Guesstimates about market size, disease prevalence, or anticipated benefits that are tossed in during these frenzied meetings tend to stick when the winning product moves to the next phase of the development cycle.
So HEOR colleagues, let’s be honest, we play a supporting role in a system where the ultimate metric is how much the product actually gets used. And it’s worth noting where the accountability lies if products don’t make their sales projections.
Living with Uncertainty
Stereotypes aside, I strongly believe there is a huge opportunity for HEOR professionals who have been moved to Medical Affairs (and for the consultancies that support them). Our strong quantitative backgrounds, and interdisciplinary skills in epidemiology, statistics and economics can be an important complement to these teams, particularly in the very early phases of product development discussions where we typically have been absent. But to be useful, we are going to have to move faster, and perhaps more importantly, live with the great amount of uncertainty that comes with rapid decision making. As an example, population funnels for budget impact models contain a lot of useful information, and often could avoid both overoptimistic or overly pessimistic market projections for developing products—let’s call them Type I and Type II Market Access errors.*** Can we generate them in hours, in minutes? As perhaps the simplest of applications of an algorithm, I don’t see why not. Yes, sometimes they will be wrong, but will they be closer to the truth than a human guesstimate with all the expected cognitive biases?
“To be useful, we are going to have to move faster, and perhaps more importantly, live with the great amount of uncertainty.”
Why will this be helpful? So many times, I have been brought into a discussion about HEOR support for a “breakthrough” product that is well into phase II or III testing where my first reaction was: “Huh?” Perhaps as a general internist, I knew too much (and of course have biases that come with that specialty), but my thoughts quickly turn to, “How is this going to play out in an analysis? (awful or really awful)”. On the other end, I wonder about what I don’t see: a genuinely useful product that never made it to an HEOR specialist’s desk because inaccurate back of the envelope math put it far down the pecking order.
To their great credit, HEOR groups in pharma have become much more adept at moving their functions to earlier phases of product development and even in discussions around in-licensing. That said, I often wonder if this time gift of foresight is wasted when massive intellectual efforts are spent trying to address uncertainty that simply isn’t addressable at that point in development. For example, I have seen many teams (including ours) spending weeks debating the best functional form to fit a progression free survival (PFS) curve from a clinical trial of cancer drug, with the goal of estimating survival for a cost-utility analysis. All of this ends up lost when the end users—clinicians and HTA groups—discount the result for the simple fact that PFS is a regulatory endpoint with limited clinical meaning. Perhaps we can get out of this mindset and use the information instead to tell our Medical Affairs and Market Access colleagues what is needed next to solidify the evidence chain and gain payer and clinical acceptance. In our parlance, this is the optimal time to apply Value of Information theory—provided that we can explain what that means in a clear and compelling way (3).
Plain Language
This brings me to my final point: to move faster within the new team structures, it is going to be critical to find simpler ways to answer questions and to convey our findings clearly and without jargon. Perhaps some enterprising person in our field can create a Google Translate app for HEOR-speak. In the meantime, let’s think about ways to use our analyses to simplify decisions from the point of view of the end user. For example, rather than relying on tornado plots or acceptability curves to convey uncertainty, provide a numerical scale (example: 1-very little to 5-a lot), then help the user understand the driving factors, then point to ways to address the uncertainty in ways that will be meaningful to end users (e.g., a pragmatic trial, real world evidence study).
***In this example, a Type I or false positive, is the rejection of the null hypothesis that the product isn’t really going to be successful when it is actually true, and a type II error, or a false negative, is the failure to reject a null hypothesis that a product won’t be successful when that is actually false.
In sum, HEOR has the opportunity to provide their teams with information that can identify potentially useful products, deemphasize those with marginal expected benefits, and help guide evidence generation in ways that can improve market access. As both a clinician and a person who sits on Pharmacy and Therapeutic committees, I’ve always appreciated pharma pitches that combine honest enthusiasm with transparent and balanced assessments of benefits and harms. We can and should play an important role in that process… if we can deliver our information at the speed that it’s needed.
–Scott Ramsey, MD, PhD
Senior Partner and Chief Medical Officer, Curta
REFERENCES
- McKinsey. Available at : https://www.mckinsey.com/industries/life-sciences/our-insights/a-vision-for-medical-affairs-2030-five-priorities-for-patient-impact. Accessed Sept. 9, 2024.
- AI writing scientific papers Available at: https://sakana.ai/ai-scientist/#example-papers-generated-by-the-ai-scientist. Accessed Sept. 9, 2024.
- We have shown that this is possible, albeit still needing some work to help facilitate decisions. See Bennette CS, Veenstra DL, Basu A, Baker LH, Ramsey SD, Carlson JJ. Development and Evaluation of an Approach to Using Value of Information Analyses for Real-Time Prioritization Decisions Within SWOG, a Large Cancer Clinical Trials Cooperative Group. Med Decis Making. 2016;36(5):641-51. doi: 10.1177/0272989X16636847.