Healthcare Startups – Why Now and So What?

It’s Screen Shot 2015-10-09 at 8.03.52 AM8:15 on Friday evening.  I’m almost through editing the job description for a user interface engineer after sending off an introductory slide deck to a potential client.  Today I met with a business development prospect, held calls with a potential advisor, a potential client, and finally made those changes to the website.  There’s not time to write this but when will there be?

I’m part of a growing trend of academics, programmers, and clinicians taking the startup path to try to make healthcare a better place. In fact, record breaking amounts of venture funding are pouring into healthcare with 2014 seeing $4.13 billion in digital health venture funding and 2015 showing no signs of slowing.  Established tech companies not typically associated with healthcare including Apple, Samsung, Google, and IBM are getting in on the act with substantial investments.  It seems that nearly every hospital and insurer is launching its own incubator or innovation fund.  

The real question is why, after decades of lagging behind nearly all other industries in the adoption and use of information technology, does healthcare seem to suddenly be such a hotbed of activity?  

The answer: data matters like never before in healthcare.   

Why data?

For as long as there have been 3rd party insurers, the vast majority of care in the US has operated under a perverse incentive system.  And while every clinician I know is dedicated to providing the best care possible, today’s healthcare systems are far more likely to invest in technologies that fetch handsome reimbursement returns.  That’s why a robot to perform surgery that fetches tens of thousands of dollars per procedure (regardless of the lack of evidence of improved outcomes) has trumped investment in systems designed to keep people out of the hospital.

If you’re interested in building systems that improve care, and said systems consume resources beyond those available from your typical governmental or philanthropic grant, you either work in another country or you take a job in one the few US systems with appropriately aligned incentives.  And for the past 8 years that’s exactly what I’ve done, chasing interesting work in government, in the non-profit sector, and overseas.

However, a US healthcare market driven by value demands efficiency + effectiveness.  Efficiency + effectiveness requires measurement and improvement.  In this new world, data is currency.  It points out inefficiencies, speeds and streamlines effective communication, and highlights opportunity.  In fact, the ability of healthcare organizations to use their data most effectively – from the C-suite to the visiting nurse – will soon determine whether or not that system survives.   

In order to measure, data silos will be opened by their owners in search of new insights and best practices to tackle the complex challenges of human health.  Solutions that are most capable of learning relevant patterns and automating the best practice tend to incorporate information technology.  

Why now?

The answer has everything to do with a gradual and long overdue shift in healthcare payment incentives from rewarding volume to rewarding value.  The credit goes to changes in public policy via the American Recovery and Reinvestment Act (ARRA), the Affordable Care Act, new risk-sharing contracts offered by insurers to providers, and growing demand by large employers that buy health insurance.  This shift is a prerequisite for our health system to make progress in addressing its ranking of 1st in cost and 37th in quality worldwide.  And organizations will fail or thrive based, in large part, on their ability to learn from and react to data.  

Enter the startups and increased attention by retail IT companies.

Of course, today’s new entrants are not the first to seek truth in healthcare data.  Health services researchers, epidemiologists, and quality improvement specialists have filled the stacks of medical libraries with not only important discoveries but often times proven formulae for improvement.

For the last 10 years these were my “customers.”  They posed the questions of the greatest clinical importance and I built software tools designed to let them get at the truth as quickly and easily as possible.  They published papers citing newly unlocked clinical discoveries.  I published papers demonstrating the ability of machine learning and natural language processing technologies to improve healthcare’s use of data.  No, we weren’t curing cancer, replacing doctors, or anything nearly as ambitious as some of today’s ‘big data’ marketing suggests.  But we were figuring out who actually had certain diseases, what was being done, and whether or not it was working – fundamental information that can be difficult and sometimes impossible to unlock using traditional business intelligence approaches.  

I believed and argued at conferences that the adoption of such technologies by clinics, hospitals, payors, and providers could make a real difference.  Only, they didn’t.  Not in a meaningful way.  Because there is a greater incentive to not know the answers when it only pays to treat, not prevent.  But for the first time in my career, that’s starting to change.  Today those most interested and knowledgeable about improving healthcare can finally take up the mantle of change with a new and most important champion – the bottom line.  

Of course, startups are far from the only way to have impact in this new regime.  In fact, if you’re currently suffering from a moderate to severe case of startup fatigue, I can’t say that I blame you.  It’s hard to grab a coffee at Voltage (or your local equivalent) without stumbling across the pitch of an app hoping to add order to the chaos that is healthcare.  And one need only ask “and how will you get the data?” to encounter a sort of naive optimism that has long been beaten out of those on the inside.  

However, if improving healthcare is truly your interest, there is another angle to consider.   Behind every healthcare startup, no matter how impractical the idea, is a group of usually bright, ambitious, and well-meaning individuals looking to help.  There may be value in listening to their ideas and asking the tough questions.  This army of the unindoctrinated, not yet bloodied by battles for IT’s attention or empty promises of “after the EPIC install” just might become your army. You may learn that, with your navigational support, their concept has a chance and realize a new opportunity for yourself to contribute to their team.  Frankly, it doesn’t matter which banner you carry, as long as your enemy is the estimated 400,000 deaths a year attributed to completely preventable medical mistakes.

It’s 6:23 p.m. Tues, and I’m editing this one last time (I hope).  The next few minutes of editing will come at the expense of writing up the outline of a proposal that will bring together the CMOs and CFOs of a health plan and a hospital around a joint program.  Claims, EMR, and free text notes will be combined to predict people in their shared population that are likely to benefit from additional support.  The project does not aim to sell more drugs or procedures but to demonstrate the value their organizations can offer to clinicians operating in an accountable care model (i.e., value based).  

If value-based financial reimbursement becomes a permanent fixture in our health system, such collaborations will become the norm and not the exception.  Why? Because “value” in healthcare can only be achieved when healthcare’s various stakeholders work together to keep people healthy throughout their continuum of care.  Data will be the engine driving these partnerships forward and new companies may be formed to meet emerging needs.

In time, healthcare will realize the same levels of innovation, efficiency, and capability that the US high tech industry is globally famous for.  And for a nice change of pace, billions of dollars will be exchanged not in the pursuit of better ad placement or market timing, but in turning data into better human health.  

Leonard D’Avolio Ph.D., is the CEO and co-founder of Cyft, Inc., assistant professor at Harvard Medical School, an advisor to Ariadne Labs and the Helmsley Charitable Trust Foundation.  He can be followed on twitter @ldavolio and his writings and bio appear at http://scholar.harvard.edu/len.

Categories: Uncategorized

Tagged as:

7 replies »

  1. Thanks and couldn’t agree more. What isn’t often discussed is the price of not learning from data. What are our moral obligations to understand what we are doing, to whom we are doing it, and is it working? Because without even looking at our data, we cannot begin to answer these questions. That’s the state of most of US healthcare.

    Reimbursement based on value is beginning to force a reconsideration but it’s early days.

  2. Dr. D’Avolio (Leonard)

    Thank you for the post. I agree with your premise- the ACA has elevated the value of data from intellectual curiosity (research driven activity) to the business of medicine. And it makes for exciting times as some of the brightest, most entrepreneurial people are now engaged.

    My worry is that the data vaults are secured by some of the tightest security in the world. I am not referring to encryption or other technologies but rather than byzantine legal and security procedures most hospitals deploy. It seems any risk, no matter how small, is viewed with trepidation. This is exacerbated by overwhelmed IT services that look to finish last year’s roadmap before entertaining any new projects. The result, valuable data remains locked away. Improved outcomes and competitiveness aren’t realize. Research, whether for intellectual curiosity or revenue (reportedly Harvard has made 100’s of millions by integrating medical research into their clinical culture) becomes nearly impossible. Ironically, I’ve spent almost a year trying to bring a small analytics engine into my hospital and have all but given up.

    Hospitals, the owners of this data, need to get on board. As the saying goes, no risk, no reward.


  3. Hi John, thanks for the provocative questions.

    #1: How do you differentiate with so much activity?
    Everyone with a product faces competition. The first question is whether you really have something of value to offer. If after careful consideration you are convinced that the answer is yes, it comes down to messaging. In this large, silo’ed world of healthcare with its competing economic interests, who exactly is your buyer and which of their top three problems do you solve? If you can’t answer that you’ll need to get real creative about where your funding will come from because unless you’re very lucky, it won’t be from customers.

    #2 regulation:
    It’s tough to take a side on regulation devoid of specifics. On the one hand, we’re only talking about healthcare startups and digital health because of regulation. Prior to the American Recovery and Reinvestment Act, adoption of EMRs hovered in the teens and now we’re seeing numbers over 80%. On the other, the government’s near overnight creation of EMR oligarchs meant that existing players with 30 year old client-server architectures now rule the EMR world. I advocated for policies that incentivized data sharing, standard formats, and the ability of users to customize decision support in JAMA before Meaningful Use regulations were announced (http://bit.ly/1GEzRKm). We got something more like the type of clinical metrics healthcare is used to (e.g., demonstrate that you can track vaccination).

    I think what we’re seeing today in MU3 is the pendulum swinging back toward flexibility, end user control, data sharing. Now that the government’s goals of adoption have been met, these seem like reasonable adjustments to existing incentives and would spawn more innovation as data and interfaces are unlocked.

    If instead you’re referring to regulation of EMRs as safety devices, I would become a bit more cautious. If the march toward pay for value continues forward, safety will emerge as a highest priority item for the C-suite of all care provider organizations. To survive, they will look for ways to optimize their systems. Then the ability of their systems to be adapted / suited to improve care becomes a competitive advantage. Vendors will be forced to either create all needed functionalities for all of medicine’s sub-domains (not happening) or to allow those with the best solutions to integrate.

    A health system financially incentivized to keep patients healthy should require far less regulation than we’re used to.

  4. Thanks Leonard,

    I think I fundamentally agree with your key takeaways.

    Particularly on the “So What?”

    A couple of follow up questions.

    #1: With so much interest in health startups, how do you differentiate yourself from the competition? Do you worry about less capable “me-toos” distracting potential investors and clients?

    #2: There’s been a lot of talk about the need for regulation floating around recently.

    Would regulation help bring order to the health tech world? Or would it impose the barriers to innovation that people in the technology industry like to talk about?

    Or is health tech different?

  5. As a physician turned digital health entrepreneur myself (SeamlessMD), this post resonates a lot with me. Thanks for sharing your story.

    I had a similar experience when I chose to leave medicine to build a startup. I asked myself: did I want to build this innovation from inside or build it outside and bring it in? I wanted to have the most impact possible, and I wanted to build something that would scale. So I chose the latter.

    One of advantages startups have is that they must scale – or they die. If I developed an innovation inside the healthcare system, it could sit on a shelf or never leaves our own four walls – it wouldn’t matter, I’d still have my job.

    But as a startup, if we don’t achieve scalable impact, if our product doesn’t help hospitals everywhere – our startup won’t last. That urgency is important, and is a key reason why startups are an important partner for bringing meaningful healthcare innovation to scale.

  6. Where does a JASON Public API for access to personal data fit into this picture?