Several days ago, Paul Graham, co-founder of noted Silicon Valley accelerator Y-Combinator (YC), wrote an exceptional post, “Black Swan Farming,” observing how crazy difficult it is to predict success in the startup space, and noting that just two companies – Airbnb and Dropbox – account for about 75% of the total value created by all YC-associated companies.
Yesterday, Dave McClure (the white-hot seed-stage Silicon Valley investor, familiar to readers of this column – see this discussion of his small bets style in connection with digital health) responded in a post titled (what else?) “Screw the Black Swans” that his investment model (at 500 Startups) is slightly different.
While most VCs are looking for the big score, McClure said, he’s deliberately seeking singles and doubles, which he basically expects will result in a similar expected value for his portfolio but reduce the chances of getting shut-out. He anticipates and is hoping for a greater number of successes (albeit more modest ones) than achieved by other VCs.
This will be a familiar dialog not only to investors but also to those in biopharma (who perhaps should be thought of as investors as well), as they continuously need to decide whether to go for a risky potential blockbuster or more of a sure-thing that ostensibly may be associated with a smaller market.
I’ve been fascinated with this exact question for a while (see here and here), and I’ve always looked at the problem a bit differently than McClure – which, if I’m right, may actually be good news for him.
In the current issue of The New Yorker, surgeon Atul Gawande provocatively suggests that medicine needs to become more like The Cheesecake Factory – more standardized, better quality control, with a touch of room for slight customization and innovation.
The basic premise, of course, isn’t new, and seems closely aligned with what I’ve heard articulated from a range of policy experts (such as Arnold Milstein) and management experts (such as Clayton Christensen, specifically in his book The Innovator’s Prescription).
The core of the argument is this: the traditional idea that your doctor is an expert who knows what’s best for you is likely wrong, and is both dangerous and costly. Instead, for most conditions, there are a clear set of guidelines, perhaps even algorithms, that should guide care, and by not following these pathways, patients are subjected to what amounts to arbitrary, whimsical care that in many cases is unnecessary and sometimes even harmful – and often with the best of intentions.
According to this view, the goal of medicine should be to standardize where possible, to the point where something like 90% of all care can be managed by algorithms – ideally, according to many, not requiring a physician’s involvement at all (most care would be administered by lower-cost providers). A small number of physicians still would be required for the difficult cases – and to develop new algorithms.
The gap between model or potential solutions and solutions that work in the real world – the translational gap — is arguably the greatest challenge we have in healthcare, and is something seen in both medical science and in digital health.
Translational Gap in Medical Science
The single most important lesson I learned from my many years as a bench scientist was how fragile most data are, whether presented by a colleague at lab meeting or (especially) if published by a leading academic in a high-profile journal. It was not uncommon to watch colleagues spend months or even years trying to build upon an exciting reported finding, only to eventually discover the underlying result was not reproducible.
This turns out to be a problem not only for other university researchers, but also for industry scientists who are trying to translate promising scientific findings into actual treatments for patients; obviously, if the underlying science doesn’t hold up, there isn’t anything to translate. Innovative analyses by John Ioannidis, now at Stanford, and more recently by scientists from Bayer and Amgen, have highlighted the surprisingly prevalence of this problem.
In a piece just posted at TheAtlantic.com, I discuss what I see as the next great quest in applied science: the assembly of a unified health database, a “big data” project that would collect in one searchable repository all the parameters that measure or could conceivably reflect human well-being.
I don’t expect the insights gained from these data will obsolete physicians, but rather empower them (as well as patients and other stakeholders) and make them better, informing their clinical judgment without supplanting their empathy.
I also discuss how many companies and academic researchers are focusing their efforts on defined subsets of the information challenge, generally at the intersection of data domains. I observe that one notable exception seems to be big pharma, as many large drug companies seem to have decided that hefty big data analytics is a service to be outsourced, rather than a core competency to be built. I then ask whether this is savvy judgment or a profound miscalculation, and suggest that if you were going to create the health solutions provider of the future, arguably your first move would be to recruit a cutting-edge analytics team.
The question of core competencies is more than just semantics – it is perhaps the most important strategic question facing biopharma companies as they peer into a frightening and uncertain future.
Among the most frustrating dilemmas facing patients – and physicians – is when doctors are unable to assign a specific diagnosis. Just having a name for a condition can be remarkably reassuring to patients (and families), providing at least a basic framework, a set of expectations, and perhaps most importantly, an explanation for what the patient is experiencing.
Sara Wheeler, writing in the New York Times in 1999, poignantly described the experience of traveling through “the land of no diagnosis.” Ten years later, the NYT featured a story called “What’s Wrong with Summer Stiers,” about another patient without a diagnosis – and about a fascinating initiative at the NIH, the “Undiagnosed Disease Program” – specifically created to meet this need.
The reality of today’s funding environment for digital health entrepreneurs is that it’s traditional tech investors who have the lion’s share of the money, while most long-time healthcare investors are on the ropes, contending with fleeing LPs and at least the perception of disappointing returns.
While it’s great news that some tech funds seem interested in dipping their toes into the healthcare space, it’s concerning that the investors with the most resources are not necessarily the ones who understand healthcare the best.
Tech investors, in general, are not always comfortable with physicians, and seem much more at home with engineers and developers. These investors also tend to gravitate to businesses selling directly to consumers rather than dealing with the sordid complexities of our current healthcare system.
Many tech investors are also — understandably — drawn to the power of data, and the possibility of analytics, a sensible affinity but one that at times can translate into an excessively reductive view of medicine that fails to capture the maddening but very real ambiguity of medical science, and especially of clinical practice.
(Note: the following commentary was co-authored with Tory Wolff, a founding partner of Recon Strategy, a healthcare strategy consulting firm in Boston; Tory and I gratefully acknowledge the insightful feedback provided by Jay Chyung of Recon Strategy.)
Medicine has been notoriously slow to embrace the electronic medical record (EMR), but, spurred by tax incentives and the prospect of cost and outcomes accountability, the use of electronic medical records (EMRs) is finally catching on.
There are a large number of EMR vendors, who offer systems that are either the traditional client server model (where the medical center hosts the system) or a product which can be delivered via Software as a Service (SaaS) architecture, similar to what salesforce.com did for customer relationship management (CRM).
Historically, the lack of extensive standards have allowed hospital idiosyncrasies to be hard-coded into systems. Any one company’s EMR system isn’t particularly compatible with the EMR system from another company, resulting in – or, more fairly, perpetuating – the Tower of Babel that effectively exists as medical practices often lack the ability to share basic information easily with one another.
There’s widespread recognition that information exchange must improve – the challenge is how to get there.
One much-discussed approach are health information exchanges (HIE’s), defined by the Department of Health and Human Services as “Efforts to rapidly build capacity for exchanging health information across the health care system both within and across states.”
With some public funding and local contributions, public HIE’s can point to some successes (the Indiana Health Information Exchange, IHIE, is a leading example, as described here). The Direct Project – a national effort to coordinate health information exchange spearheaded by the Office of the National Coordinator for Health IT – also seems to be making progress. But the public HIEs are a long way from providing robust, rich and sustainable data exchange.
As an ever increasing amount of money seems determined to chase an ever greater number of questionable ideas, it’s perhaps not surprising that inquiring minds want to know: (1) Are we really in a tech bubble? (2) If so, when will it pop? (3) What should I do in the meantime?
I’m not sure about Question 1: I’ve heard some distinguished valley wags insist we’re not in a tech bubble, and that current valuations are justified, but I also know many technology journalists feel certain the end is neigh, and view the bubble as an established fact of life – see here and here. The surge of newly-minted MBAs streaming to start-ups has been called out as a likely warning sign of the upcoming apocalypse as well.
I have the humility to avoid Question 2: as Gregory Zuckerman reviews in The Greatest Trade Ever, even if you’re convinced you’re in a bubble, and you’re right, the real challenge is figuring out when to get out. Isaac Newton discovered this the hard way in the South Sea Bubble, leading him to declare, “I can calculate the motions of heavenly bodies but not the madness of people.”
I do have a thought about Question 3, however – what to do: reconsider digital health — serious digital health.
Here’s why: Instagram and similar apps are delightful, but hardly essential; most imitators and start-ups inspired by their success are neither. It doesn’t strain credulity to imagine investors in these sorts of companies waking up one day and experiencing their own Seinfeld moment, as it occurs to them they’ve created a portfolio built around nothing.
Reviewing “The Myth of The Paperless Office” for the New Yorker in 2002, Malcolm Gladwell argued that if the computer had come first, and paper didn’t exist, someone would have had to invent it. Paper, it turns out, is a lot more useful than we typically appreciate.
It occurred to me that perhaps the same might be said of another product we seem to take for granted in the digital age – medicines. (Disclosure: I work at a company that makes them.)
Medicines – you know, those little white pills that everyone loves to critique – are in many cases remarkably effective solutions to very difficult problems; it’s actually kind of amazing how useful some of these products can be. What an incredibly powerful idea – addressing a difficult and complex health problem with a simple pill you can pop before breakfast.
I read a tweet recently asserting that physicians may soon prescribe health apps as an alternative to medications; my initial reaction: good luck with that one. It’s certainly easy enough to envision how magical thinking about the power of health apps will soon be replaced by disappointment as app developers realize something drug makers have known for years: it’s hard to improve health, and it can be very difficult to get patients to stick with a treatment long enough to make a difference.
Durably improving health is really, really hard.
I’ve discussed this in the context of drug discovery, which must contend with the ever-more-apparent reality that biology is incredibly complex, and science remarkably fragile. I’ve discussed this in the context of patient behavior, focusing on the need to address what Sarah Cairns-Smith and I have termed the “behavior gap.”
Here, I’d like to focus on a third challenge: measuring and improving the quality of patient care.
I’ve previously highlighted the challenges faced by Peter Pronovost of Johns Hopkins in getting physicians to adhere to basic checklists, or to regularly do something as simple and as useful as washing hands, topics that have been discussed extensively and in a compelling fashion by Atul Gawande and others.
Several recent reports further highlight just how difficult it can be not only to improve quality but also to measure it.
Consider the recent JAMA article (abstract only) by Lindenauer et al. analyzing why the mortality rate of pneumonia seems to have dropped so dramatically from 2003-2009. Originally, this had been attributed to a combination of quality initiatives (including a focus on processes of care) and clinical advances. The new research, however, suggests a much more prosaic explanation: a change in the way hospitals assign diagnostic codes to patients; thus, while rates for hospitalization due to a primary diagnosis of pneumonia decreased by 27%, the rates for hospitalization for sepsis with a secondary diagnosis of pneumonia increased by 178%, as Sarrazin and Rosenthal highlight in an accompanying editorial (public access not available).