Mike's Blog

While we’re not sure what they teach in civics class today (do they even still teach civics?), there was a time when a major topic was how an idea became a bill and subsequent law. In the biomedical field, most can outline the steps from concept to grant to published work. The true ‘last mile,’ whether it’s drug development leading to FDA approval or myriad other types of health care innovations that reach our patients, is a bit murkier. But if we’re talking about healthcare innovation and its eventual impact on patients (as opposed to an individual investigator’s career, journal editors, or institutional reputation), that ‘last mile,’ what happens after the peer-reviewed literature has been ‘disseminated’ is the most critical aspect of our work and needs to involve many more players beyond the lofty, ivory tower biomedical research community, particularly the diverse collection of healthcare systems. If we expect our healthcare systems to evolve along with our scientific and medical research advances, then those systems must be capable of learning, in addition to executing.

When thinking about learning health systems, there can be a real temptation to conclude that any 21st century health system that’s even somewhat research-oriented is a learning health system. After all, don’t all of us generate new knowledge, consider that new knowledge in the context of the larger scientific and healthcare contexts, implement new knowledge for the better health of those we serve, lather, rinse, and repeat? That is indeed what many of us do, and though these are core functions of a Learning Health System (LHS), the excellent presentations and discussions at the LHS Collaborative Workshop on May 22, 2024 (click here for more information) made it clear that a true LHS does not represent “business as usual” nor should it.

We don’t want to steal any thunder from what are sure to be outstanding outputs from the workshop. So, we’ll highlight a few topics that emerged in various ways throughout the day but that might be difficult to capture in a white paper.

The idea of “buy-in” as an indispensable ingredient in the LHS recipe emerged early in the workshop. If one wants to build and maintain a functioning LHS, there’s buy-in required at multiple levels. It needs to start at the top of the organizational unit, whether that’s a discrete medical center or an entire health system. If the C-suite doesn’t see the value proposition and agree to the concept as a whole, to the particulars that will be required to make the LHS work, and to the support (person-hours, dollars, space, etc.) needed to enable success, the best LHS plans in the world won’t get off the ground or fly very far. Of course, there needs to be buy-in from the people who actually deliver healthcare in all of the forms and settings that will be touched by the LHS. Beyond the C-suite and the boots-on-the-ground healthcare workers, though, an LHS can interface with lots of other units within a health system and will need their buy-in as well – the CISO/CMIO/CRIO, the COO, the pharmacies, the directors for any affiliate hospitals/care settings, and potentially many others as well as their customers, the patients.

The topic of developing and training the future architects and leaders of Learning Health Systems also got some airtime. If an LHS is not “business as usual” then developing the future leaders in this kind of translational science and delineating career paths for them is perhaps not “training as usual.” This is an area that will likely require some special creativity. Many current examples of LHSs have been very capably led and developed by accomplished senior clinician-scientists who have successfully worked together with their institutions and health systems over a period of time to craft the vision that we are coming to recognize as an LHS. But how do we train people to do this from a much earlier point in their careers, and how do they get the necessary support to do it well? Involvement of earlier career professionals as LHS leaders will likely become increasingly important as the pace of healthcare evolution continues to increase (ahem… AI-enabled LHSs).

Finally, there were some really interesting examples that were offered in both the talks and in the breakout sessions for what might broadly be called “design features” of LHSs. Early in the day, Leora Horwitz discussed how A/B testing has been used incredibly effectively in the NYU Langone health system to increase rates of specific screening exam completion by optimizing the wording of communications in a way that was almost invisible to providers and that did not impact their workflows at all. In the short term, increasing screening rates certainly will be perceived as value by the C-suite since reimbursements rates will increase, but the ultimate value to patients and the public will be the downstream impact of that increased screening has on subsequent disease (which may actually, negatively impact future healthcare system reimbursements). We will need appropriate metrics in place to monitor the overall population health improvements.

In conversations later in the day, though, it also became clear that invisible is not necessarily the perfect initial implementation for all LHS interventions, as shown by demonstrated need to prompt providers to adhere to evidence-based resuscitation fluid selection following publication of the SALT-ED and SMART trials. While many successful implementations may not be the stuff of Nobel-worthy efforts (reworking fluid order menus for example), these straightforward tweaks offer significant opportunities to truly improve health outcomes, but require a team with a diverse skillset.

If all of this sounds like exciting opportunities disguised as some potentially daunting challenges that require nontraditional teams taking nontraditional approaches… well, yeah, that’s right. And isn’t that what the CTSAs are best at?

We cannot solve our problems with the same thinking we used when we created them.

- Albert Einstein