By Dr Tom Foley, Dr Fergus Fairmichael.
Dr Michael McGinnis is a physician, epidemiologist, and long-time contributor to national and international health programs and policy. An elected Member of the Institute of Medicine (IOM), he has, since 2005 also served as IOM Senior Scholar and Executive Director of the IOM Roundtable on Value & Science-Driven Health Care. He founded and stewards the IOM’s Learning Health System Initiative, and, in prior posts, also served as founding leader for the Robert Wood Johnson Foundation’s (RWJF) Health Group, the World Bank/European Commission’s Task Force for Health Reconstruction in Bosnia, and, in the U.S. government, the Office of Research Integrity, the Nutrition Policy Board, and the Office of Disease Prevention and Health Promotion.
Evolution of the concept of the learning health system
Initially, the institute of medicine began to explore evidence based medicine and there was a small group of organisations who felt that there was often no evidence base for many clinical decisions. The question was initially about whether there should there be a body like NICE in US. There was a quick realisation that there was a need to build a demand function, with interest from public, clinicians and congress. It became clear that to be able to build this demand in an effective fashion there was a need for all of these varied stakeholders to come together. The IOM therefore pulled together a round table on evidence based medicine. From this, it was obvious that in order to strengthen our evidence base, we needed to move beyond the hugely expensive five year RCT. To continue in this way would be ignoring a generation of advancements in research and digital infrastructure. This led to the development of the concept of the continuous learning health system that could eventually generate evidence in real time. The Affordable Care Act and the creation of The Patient Centred Outcomes Research Institute (PCORI) helped to develop the infrastructure to try to work on this challenge and our focus then shifted to building a continuously learning system.
The first step was laying out a vision for the learning health system; “a system in which science incentives and culture are aligned for continuous learning”. This was followed by identifying the building blocks to do this, with clinical data at the centre of it. Each element of this was explored in a series of workshops, with invited experts to create a planning committee that represented a broad group of stakeholders and built the case to create a series of publications (http://www.nap.edu/catalog/13301/the-learning-health-system-series).
The IOM is normally known for consensus study recommendations, here, the aim was to work faster than that and to develop virtual consensus studies without the recommendations, although these are often implicit. After six years, a consensus committee was formed, which prepared “Better care at lower cost”. This document does have recommendations.
There was also a decision to add an action dimension to the proceedings and a series of collaborative affinity groups were formed. At every step of this, we have tried to expand the sphere of involvement. There have been 6 collaborative groups with involvement from vendors and regulators, large healthcare institutions, etc. Each affinity group has 3 requirements:
1. It addresses an important issue
2. That multiple stakeholders are involved
3. National Academy of Sciences can offer some advantage to its progress. If the project could be done better elsewhere, then it should be
The notion of a continuous learning health system is never going to be achieved without patient involvement in particular, because of the concerns regarding issues of privacy and use of data. Therefore, there was a need to directly involve patient groups and we are trying to build a network of patient advisory councils. The IOM will aim to steward this group until it can support itself.
Work across all of our collaborative affinity groups focus on three areas:
1. Science or evidence
2. Value or systems improvement
• Current shortfalls relate to a digital infrastructure that has capacity or potential for seamless interoperability to generate evidence in real time. At present interoperability is diminutive.
• Another issue is changing the notion of clinical research from one which is a guild to one where everyone is invested and every interaction is an opportunity for knowledge generation. This has both cultural and practical considerations but is about changing the clinical research paradigm in fundamental ways.
• The third issue is regulating the use of clinical data, current regulation is inhospitable to the continuous learning system.
2. Systems improvement challenges
• The fragmented, fractured nature of healthcare delivery, along with the current economic perceptions, creates a difficult challenge to treat this as a system. This requires digital infrastructure to be triple purposed for, delivering care, quality improvement and also knowledge generation. There is no reason that the required infrastructure cannot be built.
• In order to engage the fragmented system and multipurpose the infrastructure there is a need for board level engagement. We are in the process of forming a national network of CEOs with the aim of making a reality of what is practically possible.
• Another element of the systems piece is transparency, key to the notion of a continuous learning system is open access to information.
• Without a supportive culture, an institutional commitment to learning activities and a recognition of the patient as the driver of evidence generation, the system will not succeed.
We are unlikely to check many boxes off quickly. This concept is for a continuous learning system and this would be a never-ending process with continuous opportunities for improvement. The real fly in the ointment is the disjuncture between a myopic view of opportunity and the reality of long-term challenges. However, we have made more progress in the field than many would have expected.
In 5 years there may be progress on interoperability and transparency, which is one of the biggest problems. There is currently no incentive for providers to show the prices that they are charging or the results that they are achieving.
Interoperability is tough, it is not just a technical problem, but also a problem of will. Part of our job is to explore open source platforms, where everyone opens up their APIs. The IOM is not the government and cannot force this upon people but what it can do is to get everyone in the room to answer the technical question as to whether this open source platform would stimulate the kind of access and innovation that has been promised. Government could require that all machines bought with public money can talk to each other. The timeframe for interoperability is unclear but we are obliged to take a run at it.
Meaningful measurement of outcomes would be sensible in the learning health system. We can start right now with a small set of indices that ought to be easily attainable at every level in the short term, e.g. oriented around 2 dozen standardised clinical process and outcome measure. It is possible to develop core measures that deal with population health, patient safety, evidence based medicine and that deal with cost issues, not necessarily at the individual or service level but with enough granularity to give some comparatives across institutions.
What could be easily achievable would be to implement 3-5 measures for the most important conditions, heart disease, stroke, diabetes, heart conditions asthma etc., which could be standardised across the country. This could prove to be a very useful first step. However, it is not solely about individual diseases and perhaps more importantly we should also be looking towards outcomes that reflect the overall function of the system.