How Distributed R&D Could Spark Entrepreneurship in Biopharma

This piece originally appeared in the Timmerman Report.

Remember the patent cliff and the general lack of new and innovative medicines in the industry pipeline? That was the big story of the past decade in biopharma. It caused a lot of searching for the next best way to organize R&D to improve productivity. One doesn’t hear that quite as often today. There are more innovative drugs both recently approved and moving forward through the pipelines of several biopharma.

The conversation these days has shifted toward drug pricing, and how the public is going to pay for some of these new, exciting drugs (the answer, in some cases, is maybe it can’t).

I don’t think the industry out of the woods yet. One of the main reasons drug prices have become such an issue is because even though there are new, innovative drugs, there aren’t enough of them. At the same time many of the drugs being approved are incrementally better but nevertheless being priced at a premium. And good reporting has made the public more aware of how many of our existing drugs are rising in price on a yearly basis. Especially in a time of little inflation, prices of most goods have not been going up at nearly the rate of pharmaceuticals.

Biopharma sits in a tough place. Analyses suggest the cost of developing a new drug has generally been doubling every nine years, which may be a by-product of some combination of the complexity of biology, our inability to predict which drugs will work, and the “better than the Beatles” problem. The question then is how to overcome these issues and increase the efficiency of developing new, innovative drugs. Without some kind of change, the industry is looking at a very difficult future in which price hikes run headlong into the wall of payers who finally say enough. Then what?

There’s a terrific short analysis by Gary Pisano (and if you haven’t read his book Science Business, it’s got some great insights for biopharma) describing a framework for classifying R&D strategies in terms of which element(s) of the business they affect: Architecture, People, Processes, and Portfolio. He describes how all biopharma are fighting the drug candidate attrition problem, and how different companies have adopted different R&D strategies to try and overcome that problem. For example, GSK created small, semi-autonomous business units to try and take advantage of a biotech-like structure and culture. This was a change in Architecture, according to Pisano’s taxonomy. Novartis used aspects of all four elements—locating research sites where talented People are, splitting off basic research as a change in Architecture, using opportunistic science to guide the Portfolio development, and setting up clear Process metrics for project advancement.

These approaches were implemented to try to meet the corporate goal of reducing attrition of projects through the pipeline. No one has figured out the best approach yet. On a practical level, given the relatively small number of projects even large biopharma can develop, the decades-long process of drug development, and the possibility that the best approaches may only increase success rates by a few percent, we may never be able to definitively proves what’s the best approach. The experiment, as we might say, is tremendously underpowered and uncontrolled, with continually changing environmental conditions.

Still, one clear pattern is that many big companies are continuing to cut back on internal R&D, leaning more and more on in-licensing products for further development. They are casting a wide net across academia and small biotechs. As Joy’s law states, “no matter who you are, the smartest people work for someone else.” The wide net goal isn’t incompatible with the goal of a reduced attrition rate, although in a way it’s a hedge. If a company can’t reduce attrition rate, it can increase the number of things it evaluates so that at the end of the day, the company still ends up with more products on the market.

My question, then, is what kinds of organizational R&D structures might also help in reaching this goal of casting a wide net? And I’ve come up with one idea that is kind of, sort of, inspired by some of the populist ideas floating around the Republican and Democratic primaries. What if, rather than organizing R&D sites around a specific site head with a vision and research philosophy, and with a hierarchical structure, we created sites that were allowed to self-assemble into teams pursuing many different visions and ideas about meeting a medical need: a kind of a distributed R&D approach?

How would we do this? Let’s start with the most fundamental thing. Let’s start with money.

Direct payment has proven to be a powerful tool in global health interventions. Randomized controlled trials have shown that giving money with no strings attached to people in the developing world leads to better outcomes than donations that are organized and filtered through Non-Governmental Charitable Organizations (NGOs). There are even bigger experiments going on now in direct payments and the related idea of basic income in the developed world (see some descriptions here and here). This made me think: what would it look like if every person in an R&D site was given control over an equal portion of the site’s R&D budget? No strings attached in terms of how it was spent. Well, other than that you can’t take the money and abscond to Costa Rica.

The roots of this idea come from both the actions of charities like GiveDirectly, as well as a recent proposal by researchers at Indiana University at Bloomington for a similar plan for academic research.

So: some nuts and bolts. The key starting point is the mission. I’ll suggest there should be a specific but generically worded goal for research at the site. For example, “develop ways to improve the lives of patients with lupus”. As has been described in management books galore, a good mission statement is the main thing employees need because they can always simply ask: “Is what I want to do contributing to the goal of improving the lives of patients with lupus?” If the answer is “No,” then that idea, no matter how exciting or innovative, doesn’t get done. There’s another great example from Dan Heath involving Southwest Airlines and Chicken Caesar Salad that you can jump to below if you want. I’ll wait.1

The second important point is people, but I’ll defer that for a few paragraphs, since selection criteria is going to make more sense in light of this proposed organizational structure.

Let’s say we have a lupus R&D site with 100 people and a $12M yearly research budget (not counting salaries and overhead). Let’s take $2M for basic, generic laboratory support—common equipment, general consumables, basic tissue, cell and animal materials. Split up the remaining $10M among the 100 people so each person is in direct control of $100K per year, as well as complete control of how she spends her time. The exact numbers aren’t that important by the way, I just wanted some nice round numbers to make the math easy. And when I say everyone I really do mean everyone. Each technician, each scientist, no matter how senior or decorated. See where the influence of populism is coming in?

Well, you might say, okay everyone has control of some money—what do you mean they’re also in control of their time? I mean exactly that. This is going to be the flattest of flat structures. It starts off with nobody reporting to anyone else, nobody directing anyone’s actions. In year 1 the site essentially holds a startup competition to figure out what is going to be done over the next year. Maybe this period lasts three months, and during that time coordinators and facilitators, who are outside this structure, help people at the site learn anything they want to know about lupus. They might bring in Key Opinion Leaders (KOLs) and patients, or connect people to patients’ organizations, or to individuals or groups in other business areas like mobile health or cloud computing. Another potential group of outside experts would be financial analysts to ensure any ideas have a reasonable chance of actually making a profit at some point.

Everyone is tasked with coming up with research ideas and trying to figure out what it would take to make that idea happen. This is where teams begin to assemble. For example, someone comes up with an idea that requires informatics expertise and familiarity with high throughput cell screening. If she doesn’t have all the necessary skills, she goes out and pitches her ideas to people who do. On the other hand, there may be some lone wolves who decide to go it alone and figure out what they can do with their $100K. Great! One of the main goals is to have a wide variety of search directions. This was one of the general insights of the Indiana researchers I mentioned above: that we want to make the  funding mechanism more like internet search.

As teams assemble, they also need to figure out their funding needs. An important point here is that money and time are two separate commodities, and a person can choose to donate both, either, or neither to any given nascent project.

At the end of the three months, there’s a pitch day in which teams describe what they’d like to do and what, if anything, they still need. The last outstanding pots of cash and unallocated hours are divvied up and every successful proposal gets started. After that, there’s full transparency—regular reporting by every team back to the site, every experiment updated in public notebooks, and every advance and failure celebrated—as they should be. Only celebrating advances sends the wrong message about risk. Also, being transparent about failures means the same mistakes won’t keep getting made.

While the year progresses, if projects are failing or flailing, people are always free to reapportion their time and dollars. When the next year rolls around, successful projects can continue, building out or changing their capabilities as necessary, and new projects can start up.

I think this kind of arrangement could have a huge number of benefits. Going back to the original goal, this kind of distributed project creation and thinking increases the breadth of the net being cast. People will be much more engaged: they will be the driving forces behind the projects, not their bosses, or their bosses’ bosses, or even some layer higher than that. This system would be tapping into the internal motivating forces of pride, autonomy, purpose and learning. All of these incentives will, I think, drive people to be as creative as they can possibly be. By creating a truly flat structure, this system would also leverage diversity of opinion and experience.

Let me delve a little more into the personal development aspect. Participating in this would give everyone at the site experience in team building, pitching, acting as a venture capitalist, creating and managing realistic budgets and goals, communication, presenting, and emotional intelligence. In a word, entrepreneurship. How many times do we hear people in the industry bemoaning the lack of an entrepreneurial spirit? How often are we told the US is actually suffering a decrease in the number of startups (as one measure of entrepreneurship)? Well, here’s one way to instill and nurture that in a lot of people at once.

In fact, I think a positive and expected side effect of this would be spin-out companies. Let’s say a team comes up with an idea that reaches a go/no-go decision point and for whatever reason is stopped internally. If the team truly still believes in the idea, let them take it outside to pitch to external funders. The parent company might retain some minor stake and/or a right to license, and would encourage this team to try to make it. I’m reminded of something I’ve heard LinkedIn does, which is they make a point of telling their employees that LinkedIn is there to help prepare them for their next job. Paradoxically, I think that kind of attitude would lead to more retention and greater loyalty. Among those who stay, the parent biopharma will now have a very broadly trained group of people who can evaluate all kinds of in-licensing opportunities and have fresh skills in doing so.

I’ll take a step back here and admit that with respect to this proposal I’m a wild-eyed optimist. I don’t doubt that there’s a good chance this kind of setup would implode, leading to a Lord of the Flies meets Survivor kind of crazy crackup. And that’s why I’ll return to the point about people. This clearly isn’t going to work for a lot of researchers. Maybe not most of them. Figuring out how to assemble the site, whom to hire, what experiences to combine, what personality types and in what proportions—these are all going to be key decisions influencing whether this works. Probably the key traits that would be high on the list, though, would be humility, curiosity, open-mindedness, intelligence (emotional and otherwise), and diversity. Real diversity. Not “Oh, we’ve got one PhD guy from Harvard and also a PhD guy from Stanford, so that’s diverse.” Gender, ethnicity, life experiences, fields of research across and beyond biology—all of these would be important parameters to try and capture when assembling the site.

Okay. I admit. This kind of system will never happen in any of the established biopharma, of course. It’s too many changes piled together at once and I have yet to meet an executive who would be willing to take that great a risk with a site’s budget, and for very good and rational reasons. But it’s fun to think about, and if we’re going to change the R&D enterprise, it’s likely there will have to be experiments, large and small, because the low hanging fruit is gone and nobody’s yet invented the perfect ladder.

1:  From here: “Dan Heath: Southwest’s strategy to be the low-fare airline in its markets is a sticky idea, primarily because it’s meaningfully simple. It’s certainly not a story, and not particularly emotional.

Herb Kelleher, Southwest’s founder, once posed an interesting scenario to demonstrate the power of Southwest’s simple message.

Here’s an example, he said: Tracy from marketing comes into your office and presents survey results indicating that passengers on the long flight from Houston to Las Vegas might enjoy something more substantial to eat than peanuts. She thinks a chicken Caesar salad would be popular. What do you say?

You ask her if adding that chicken salad will help make us the low-fare airline. If not, we’re not serving any chicken salad.”