There’s been an interesting discussion about pharma R&D productivity and drug pricing on Twitter the last few days – here’s the starting point, from John Tucker. His thesis is that the hefty rate of inflation for medical services/hospitalization, where the data seem alarmingly solid, is one of the things driving the problem. That’s because it’s a surrogate for the expense of running clinical trials. Starting back in the mid- to late-1980s, the costs of medical services began to truly diverge from the general consumer price index, and it’s never looked back. Even the 2008 recession (which put a speed bump into the CPI data) didn’t slow down medical inflation one bit.
Tucker goes on to show that (at least since 2005) that while (1) drug approvals have generally been rising, (2) trials, on average, are not getting larger, (3) the success rates of these trials have not changed appreciably, and (4) the number of Phase 2/Phase 3 trials has not changed, either. What has changed is the cost of running them. We can debate the connection to drug pricing in general, but the effect of this on drug-company financials would seem to be real, and companies either soak these rising expenses up or pass them on.
Another round of discussion on this point starts here. This brings in David Grainger and Bernard Munos, and one point that ends up being made is that if cost-of-trials is indeed increasing, as it seems to be, then it would follow that we might want to be pickier about what we put into those trials. But that’s not what’s happening. Munos, in particular, has been using the example of Alzheimer’s for some years now, and it’s a tough case to refute. If you have a disease whose underlying clinical rationales are so insecure, and whose trials are so long and expensive, with historic failure rates approaching 100%, then don’t do it. Do something else with that time and money. Put some of it into figuring out more basic info about Alzheimer’s, if you want to, but don’t push things into trials under such conditions.
That leads to this exchange:
Strong words, but there can really be no doubt that (1) incentives matter! and (2) it is easy to put in perverse ones that don’t do what you think they’re doing. It’s like running your med-chem department by rewarding people based on the number of compounds they register per year. Many of us have seen that tried, to one degree or another, and the same flippin’ thing happens every single time: people game the system by shoving easy amides/sulfonamides/etc. into the hoppers. On the macroeconomic level, it’s akin to collectivizing the farms: dang, collapsed agriculture again. Dang, filled the compound collection with self-similar junk again.
And at the recommend-to-the-clinic level, anyone with experience in this business has seen perverse incentives at work. I recall sitting in meetings and thinking “Now, why are we recommending this program/lead compound for clinical development, exactly?“, followed by the dispiriting thought “Oh yeah. Look outside. The leaves are dropping. End-of-year goals.” Lots of junk has been shoved along in order to check boxes, because those boxes are tied to bonuses, promotions, and other immediately tangible desirable things. The harm, meanwhile, is spread out over the years and the responsibility is diffused. Clinical trial failures happen, y’know – and besides, we did everything by the book and everyone signed off on it. Whaddya want us to do, not send stuff to the clinic at all? Never going to have any successes that way.
It’s true, you’re not. But (general principle time again) just because you can screw things up in one direction doesn’t mean that you can’t screw them up in the opposite one. If you put nothing into the clinic, you will run out of money. If you put everything into the clinic, you will run out of money. And if your clinical programs are disproportionately great big huge long megatrials, you are taking an even bigger risk of running out of money – especially if those trials costs are ratcheting up every year. We all know this. Doing something about it is harder.