Making the Learning Health System A Reality, Part 1: Concepts & Barriers
This post is the first in a series exploring the concepts, barriers and goals central to widespread adoption of the Learning Health System.
What is an LHS?
The Learning Health System (LHS) is an idea, a concept and an inspiration, but it’s also something that can work in the world.
It’s fundamentally a very simple idea—that health delivery systems can become learning health systems when they acquire the ability to routinely study and improve themselves.
Now this has always been possible. Some of the methods that are used as part of a learning health system have been around a long time, but they have not been used routinely. Most often, they’ve been used to address crises or emergencies. With the LHS we’re talking about something very different, where we view the experience of every patient as an opportunity to learn and improve.
What makes this possible now is the power of information technology and the increasing fact that healthcare is being documented electronically. Every patient encounter leaves an electronic footprint with details about what happened, the patient’s condition and what the result was. As a result of their digital footprint, these data are available for analysis and the results of those analyses can be used to drive improvement.
Central Concepts and Living Examples
The LHS and the concept of a learning organization are ideas that have been around for a long time. Similarly, the idea of a virtuous cycle of observation, data collection, analysis, decision to act, implementation and assessment of the actions is an idea that has been with us for a long time. It’s the notion that we can create an infrastructure that allows cyclical improvement to happen routinely that has emerged over the past 8 or 9 years.
What is it that separates systems that act as learning health systems from the rest? Organizational culture.
One could argue that four health systems in the US have acquired most of the characteristics of an LHS: Kaiser Permanente, Geisinger, Mayo and Intermountain Health Care. Much of what differentiates them is organizational culture.
Over a period of time they have evolved a culture that is supportive of, and frankly expects as part of routine operation, the study and improvement cycles that are key to the LHS. All of them also got a head start over many other health systems in the deployment of integrated information technology, which is a key enabler. They have expanded their capability to continuously improve and frankly, it’s so wide spread and pervasive in these places that if you ask them why they do it, they’ll say this is who we are and this is what we do.
Technology and culture are two of the stated barriers to scaling the LHS, but key as we move forward is the notion of an infrastructure or platform that makes the LHS possible.
If you look at the Geisingers and Intermountains, the platform they use for continuous learning and improvement may not be called out as something independent of their other infrastructure. But, all the elements of the platform needed to enable learning cycles are in place.
Complementing these bright spots in the US health system is the emergence of learning networks: different organizations that have come together around a circumscribed set of diseases or a particular disease domain to collaborate in learning and improvement across organizational boundaries.
In some ways, these networks are more exciting if we’re thinking about how to disseminate the concept of an LHS, because these networks were engineered from the ground up to be inter-organizational. The integrated delivery systems, which have become learning health systems within their own boundaries, have done it in an idiosyncratic way -- and that’s not a criticism of them. It enables them to be extremely efficient in the way they do this, but the methods used cannot be externalized from those organizations and used anywhere else.
When you have to build an infrastructure that serves multiple organizations, you build it differently and that makes it scalable to other organizations. If you have a learning network with 10 organizations and you build an infrastructure that accommodates those 10 organizations, it’s pretty likely that if an 11th organization wants to join you won’t have to modify the infrastructure to accommodate the 11th.