News

Health systems face a growing challenge. How to provide the high-quality care to a population that is ageing and suffers from increasing rates of chronic disease. Even new innovations (such as machine learning and precision medicine) that may improve the quality of care, present a problem for health systems. These innovations generate massive amounts of information that then need to be incorporated with the vast quantities of patient data already generated through routine care and medical research to inform the best care for the patient.  

At present, health systems are struggling with this latter challenge. On average, medical research takes up to 17 years to be translated into healthcare guidelines. Even then, only 60% of healthcare is provided inline with those guidelines. The rest is either of little to no value to the patient (30%) or outright harmful (10%). These percentages have stubbornly persisted within health systems, despite numerous efforts at reform, for decades.

The health system of the future: a Learning Health System

To change such recalcitrant systems, the US Institute of Medicine (IoM, now the National Academy of Medicine) proposed the adoption of the Learning Health System (LHS) model. LHSs were envisioned as the health system of the future. One that was capable of seamlessly producing, capturing, and learning from clinical data produced during routine care (e.g., diagnostic data, pathology reports, error reporting, clinical notes), medical research, and new innovations for continuous quality improvement. The continuous nature of an LHS meant that it was a journey, rather than a destination.

Since this futuristic vision of health systems was put forward nearly two decades ago, it has received a lot of attention with development and adoption occurring at different levels of health systems. These efforts have in turn generated a wealth of literature.

In their recent scoping review of LHS literature titled Learning health systems: A review of key topic areas and bibliometric trends, PCHSS researchers Jeffrey Braithwaite, Yvonne Zurynski and colleagues examined the current state of LHS literature to identify the key topic areas and trends in publications on LHSs.

Lots of theory, little evaluation

Building off other recent systematic reviews, the authors examined papers published between 2016 and 2020. They found that a substantial body of theoretical literature describing LHS designs and likely barriers and enablers to implementation has emerged. However, empirical and evaluative studies of implemented LHS were noticeably absent. This suggests that many authors are still focused on the conceptual portion of the LHS journey, and that many systems have yet to move onto the adoption and evaluation stages.

Another interesting insight is that most of the literature is focused on the data and technological aspects of an LHS, rather than on the organisational and cultural arenas, even though the later is critical to changing health systems. In addition, challenges posed to data access through research ethics, particularly around the use and re-use of data for clinical quality improvement, was often discussed. Easy access to all types of data (e.g., research, patient information, genomics) is a corner stone of an LHS. However, current ethical frameworks create obstacles to integrating data into learning and care. Unfortunately, changing organisational culture and research ethics are complex problems that are likely to continue to inhibit the development of LHS.

What are the next steps?

As we continue on the LHS journey, it is clear that more emphasis is needed on empirical studies and evaluation of developing LHSs. It is also essential to begin to address the barriers that inhibit the wider adoption of LHSs.

As the authors explain:

“The LHS literature warrants a shift in focus to move the field from the conceptual LHS to take‐up and adoption, and from technical processes to emphasising the complexity of human factors.”