Two weeks ago I had the honor of participating in the 2016 Second Learning Health Summit held in Alexandria, VA. This was a meeting of over 100 thought leaders from industry, healthcare provider organizations, and consulting companies who got together to brainstorm on the actionable next steps to persuade organizations to adopt a learning health paradigm.
Over two days we listened to engaging speakers and broke into working groups to come up with ideas to tackle problems surrounding learning health adoption. This summit followed in the footsteps of the first meeting held in 2012 where core values were defined and endorsed by over 100 organizations (including ThotWave!)
What is a Learning Health System?
Although Learning Health has been defined in many ways, I think the following ‘per-view’ level as shared by Dr. Charles Friedman, Chair of the Department of Learning Health Sciences at the University of Michigan Medical School, takes the best approach:
- Macro view: A learning health system strategically connects data and insights across the pharmaceutical industry, universities, care providers, government agencies, non-profit organizations, and patients
- Micro view: A learning health system is one where the analytic life cycle is realized such that insight from analysts is used in a virtuous cycle to inform the practice of care
- Checklist view: A learning health system is one where organizations do the ‘X things’ with knowledge, such as share data externally or curate information in a warehouse
But take note—a Learning Health System concept is not a Utopian ideal of ‘everyone getting along’ in leveraging the creation and use of knowledge. The Learning Health values describe a series of organizational skills, abilities, and traits that allow them to provide optimal patient care while maintaining their capacity to compete well using data.
For example, patient reported outcomes by definition represent information sourced from outside the four walls of a care organization. Health systems must develop a strategy that ingests this data to improve patient outcomes and create a safety net that may alert of deteriorating patient conditions. Drug safety surveillance is another activity that requires Learning Health capacity to safeguard patients. Due to both the sample size required and the cost involved, it is not feasible to design and deploy all possible multifactorial clinical trial designs to examine the variety of safety concerns with a given medication, procedure, or device. The best way to do this is post-approval when a baseline safety threshold has been reached. Organizations that can combine data on outcomes allow the detection of additional adverse events in an observational manner.
Learning Health System Challenges
There was productive, frank discussion regarding why Learning Health activities fail to gain momentum. A good deal of the debate among participants was in asking ourselves how we could overcome these blockages:
- The data streams spewing from electronic systems spawn from either frontline patient care or billing. Few aspects of data capture and reporting systems are designed with a longitudinal analysis in mind.
- There persists a significant challenge with understanding the role of Institutional Review Boards (IRBs) and the rules of engagement for asking questions of data. There is incredible variety in how IRBs approve and enable data-drive research. People are often under the misimpression that they cannot legally ask questions of data given HIPAA rules.
- Data provenance, or the lack of it, is an enormous challenge. It is not enough to have different organizations all agree to use the same data standards. They must also agree upon similar definitions—at least for feeding aggregate studies—of what different data concepts mean. For example, how do you define a primary care visit? Do lab-only nursing visits count? What about telehealth modalities? As of now, all too often the data received is subject to the opinion (or ignorance) of the analyst who pulls it.
- Both health services researchers and healthcare operations analysts need to work in concert to make any findings actionable i.e. do that whole learning thing. The abbreviation ‘K2P’, or Knowledge to Practice was mentioned as too often interpreted as Knowledge to Papers when a research investigation is involved. Publishing is not the final act of a Learning Health project.
Pharma and Biotech Plays an Important Role
By attending, I gained more appreciation for the stake that Pharma & Biotech have in seeing Learning Health happen. Richard Rudick from Biogen spoke and described how they had spent much money in recent years to help position the concept at partner organizations. He noted that the drive isn’t just altruism but also a realization that drug development is changing and the industry needs these data from healthcare providers to make that happen.
In fact, organizations that don’t feel like they have the resources needed to develop the analytics required may find partnering with Industry players lucrative as they are willing to pay for real patient data. This does, however, have to be done in a manner consistent with privacy concerns and the desires of patients. But even an ‘opt-in’ research data network pitched with the right ad campaign (and Pharma knows how to spend money on marketing) could give them the information they desire. Laura Crawford from Eli Lilly noted, “we want to compete on value, but we need data to do that.”
We need methodology, not just strategy
Towards the end of the meeting, there was a lot of discussion about how to take the inspiration surrounding Learning Health and turn it into something actionable. At some points, the discussion got a bit heated as ideas and styles conflicted.
As a leader at ThotWave, it is clear to me that our mission to help nurture the development of data champions is not just directly relevant to the goals of furthering Learning Health but is an essential tactic to make the concept rise above just aspiration. One of the things that we have done is to define in excruciating detail the competencies needed to fuel the analytics lifecycle, which is the engine of the virtuous circle of Learning Health. It is hard to build a knowledge engine in your organization without having a roadmap of what it looks like, and that is exactly what our competency model seeks to offer.
We have used this model for all of our healthcare client engagements, and now we are packaging it within a program for provider organizations and are seeking feedback. If you are part of a provider organization, we would love for you to participate in our beta program and lend your thoughts. Please sign up!