The popular trope of the past decade (and then some) has been that pharmaceutical companies’ productivity is lagging – that R&D spending is becoming ever-less useful. This phenomenon has become so commonly accepted that it’s been coined “Eroom’s Law;” Moore’s Law in reverse. Productivity has sunk so law, the story goes, that as of 2000 the number of drugs approved per billion dollars spent in R&D is less than one. That is literally true. But a clever PhD candidate at the University of Chicago has developed what might be a more robust (but not necessarily more accurate) measure of productivity that tells a somewhat different story.
Typically, productivity is fairly easy to define. It’s simply a ratio of some outcome and input needed to get to that outcome. Economists often measure a country’s overall productivity as GDP produced per hour worked, for instance. Productivity in manufacturing might be measured by the number of products produced, or revenue generated, per factory or per employee. The simplest measure of pharmaceutical productivity has been the number of drugs approved (the output) per dollar of R&D spending (the input), and it’s been useful in illustrating the difficulty and complexity of the FDA’s drug approval process.
But as Tomas Phillipson, a University of Chicago health economist, and Kristopher Hult, the aforementioned clever PhD candidate, note in Forbes, the story of declining productivity is hard to reconcile with a massively increasing valuation in pharma.
This does make the conclusion of declining pharma productivity hard to accept. After all, declining productivity in an industry should send a message to investors that other investments may offer a better deal.
Hult’s paper (his dissertation to be precise) reframes this conundrum, redefining productivity in terms of (arguably) the outcome that matters for drugs – patient health. This has some intuitive appeal; after all, the real output for pharmaceutical firms isn’t the total number of drugs they put out, but how those drugs help the patients that take them. Phillipson and Hult put it best: “a single drug that cured all cancers would be immensely more valuable than 100 new molecules that treated acne.”
In his paper, Hult calculates what he calls the “health impact” of drugs – a measure that considers adherence , total number of users, and efficacy (measured in quality-adjusted life-years or QALYs). The advantage here isn’t only in more accurate measurement of productivity, but (again, as Phillipson and Hult note in Forbes) it also allows consideration of so-called “incremental innovation” as compares to “novel innovation.” (Note: The adherence measure may not be entirely accurate for very long-term therapies because the data source used to construct it is the 2-year Medical Expenditure Panel Survey. This may (though not necessarily) bias upward the adherence metric for therapies that are taken longer than 2-years.)
The latter includes new drugs based on active ingredients (technically, based on moieties) that have not been approved by the FDA in the past. But pharmaceutical innovation is more than just new drugs. Dosage changes, reformulations, dosing route changes, and new indications all represent innovations that can benefit patients tremendously. For instance, moving away from a pill form to a liquid or syringe may allow patients to take a drug that they couldn’t before. New formulations or combinations too can be very valuable – much of modern-day cancer treatment and HAART therapy for HIV/AIDS in the 90s was based on reformulations and combinations of drugs. This type of incremental innovation (often derided as “me-too” drugs) is an important part of the pharmaceutical space, and based on Hult’s analysis, makes up more than half of all non-generic prescriptions.
The results represent a complete about-face from Eroom’s Law. When productivity is measured by its health impact, pharma’s productivity appears to have increased by around 30 percent from 1980 to 2009. The naïve value of drugs (without accounting for R&D expenditures) is over 3 times greater than it was in 1980.
But everything isn’t roses and rainbows for pharma. Incremental innovations, responsible for a significant share of the increase, are projected to slow down in Hult’s model, primarily because of a slowdown in novel innovations (the former are dependent on the latter). Without more novel innovations to support incremental ones in the future, productivity (in terms of health impact) is projected to slow down to about 78 percent of what it was in 1980 (calculations based on Table 6 in the paper).
(Of course, the increase in productivity still doesn’t fully explain the massive spike in valuation – nor should it. The bulk of the increase in valuation came after 2010, well-beyond the study’s time-period. Indeed, the massive increase appears to coincide with the creation of the “breakthrough therapy” designation.)
Increasing pharmaceutical revenues probably isn’t something that government or society should concern itself with. So while Eroom’s law posed some problems, they weren’t directly tied to pharma’s profitability – rather, fewer new therapies available may impose negative consequences on patients. The decline in pharma’s productivity as measured by Hult, however, directly affects patients as it implies a lower health impact from medicines.
And that means the policies that might boost pharmaceutical productivity become a bit more convincing. Policies that include: getting the FDA to do a better job considering biomarker science in drug approval, improving the agency’s risk-tolerance to better match patient preferences, and perhaps de-risking drug development to get more drugs approved (and pulled off the market) faster, all now have much more justification.
By Yevgeniy Feyman