Performance-based drug pricing and the "Wanamaker problem"

Even in the era of personalized medicine and targeted therapies, many patients may not benefit from drug treatments.  Luke Timmerman’s recent Xconomy article on “big data” referred to this as a “Wanamaker problem” in healthcare – i.e., only half the money payers spend on drugs benefits patients... but which half?  He argues that by analyzing real-world outcomes to identify likely responders, payers could avoid paying for drugs for the rest. "Big data" may be the flavor of the month (in healthcare and elsewhere), but I think it's unlikely to make a dent in drug utilization.  Luckily, though, there is a real solution at hand – performance-based pricing (PBP).  PBP may be less sexy and tweetable than "big data", but it's a near-term, sustainable fix for wasteful drug spending.

A major problem with payers using “big data” to define reimbursement is that this approach is simply not likely to pass clinical muster.  The scientific bar to validate predictors of drug effectiveness is high, and many published efficacy markers for anti-cancer drugs and other agents haven't withstood deeper analysis, even when they've been based on large data sets.  Even if efficacy "signatures" are identified, payers will likely face widespread opposition (and be on the wrong side of the science) if they impose unvalidated response predictors on clinicians and patients.

Under PBP, payers and manufacturers agree that the price of a drug will be higher if it shows results above a certain threshold, and lower (perhaps even free) if it doesn’t.  PBP has some clear advantages over "big data" approaches in oncology and other areas:

  1. PBP can be implemented today.  PBP builds on existing science, rather than trying to create new algorithms or metrics.  In cancer, for example, progression-free survival (PFS) is a clinically meaningful surrogate for efficacy for already approved drugs that could form the basis of a PBP agreement.  (Importantly, this has nothing to do with the debate over whether or not PFS is a suitable approval endpoint for new drugs; for PBP, we're talking about drugs that have already passed regulatory muster in terms of clinical efficacy.)
  2. PBP puts payers on the same side as physicians and patients.  Under PBP, physicians would have payers' blessing to continue to use existing and evolving data and their own scientific and clinical judgment to try to match the “best” therapy to the right patients.
  3. PBP further incentivizes advances in personalized medicine.  Payers currently restrict access to new therapies when they are not convinced there is sufficient impact on the population at-large, even if this means missing the opportunity for some patients to reap dramatic benefits.  Pharma companies don't want access to be restricted, but are often conflicted about the relative business cost and benefit of defining efficacy markers post-approval.  Under PBP, manufacturers and payers would have a shared interest in permitting patients to access new drugs while proactively defining how to identify likely responders.  This could drive a shift in thinking toward focusing on "exceptional responders" rather than average ones, even after a drug is approved, and in oncology, it could also be a key enabler of combination therapies, as I’ve discussed previously.

Importantly, although there are some examples of PBP in the U.S. (summarized here) and the UK (see here and here), it doesn’t seem to have taken off.  (If you know of additional, more recent cases or analyses, please tell me.)  Besides the argument that "it's new and looks complicated", one challenge for PBP is that it may be hard to generalize across disease areas.  Oncology and diabetes work because there are accepted nearer-term surrogate endpoints (PFS, HbA1c), as would acute treatments like hospital-administered anti-bacterials.  But for many of payers' big-ticket expenditures (e.g., chronic ailments like rheumatoid arthritis and chronic pain), PBP requires a validated efficacy metric that may not always exist.  (Again, comments and examples welcome.)

Even in oncology and other suitable areas, however, the biggest challenge to PBP may come from pharma, who perceive that they would lose money if revenues are actually tied to performance (gasp!).  But manufacturers would only lose money relate to what they can make using "traditional" commercial and market access approaches, and I'd argue these "traditional" days are over or ending quickly.  Payer restrictions are coming to oncology and other specialty care areas, and pharma can either wait for payers to define the reimbursable populations for them (possibly using "big data") or work with payers to implement PBP and help shape their own destiny.

PBP could help payers solve the “Wanamaker problem” in drug utilization, and it’s likely to be a much more beneficial, near-term applicable and durable approach than the hypothetical one offered by “big data”.  In oncology in particular, it’s probably worth a try sooner rather than later.