Improving public health – in real time: U of T experts on the need for Learning Health Systems

There are growing calls to seize on the disruption caused by COVID-19 to create better, stronger health systems – not least to deal with the coming wave of disease that will be diagnosed later due to missed care during the pandemic. That, in turn, has accelerated interest in intelligent health systems that make better use of data to make improvements in real time. 

Researchers at the University of Toronto’s Dalla Lana School of Public Health want to build on the momentum by calling for Learning Health Systems (LHS) that support rapid learning and continuous improvement.

In a recent white paper, Patrick Feng of the Institute of Health Policy, Management and Evaluation (IHPME) and Dalla Lana researchers Ross Upshur, Erica Di Ruggiero, Robert Reid and Andrew Pinto discuss how such a system could be acheived.

Upshur and Feng recently spoke with writer Heidi Singer about the potential role of public health in Learning Health Systems, and why now is the time to think about including it in the nascent movement.

What is a Learning Health System?

Feng: It’s about gathering data on what you’re doing and quickly feeding that back to processes and people, so that you’re continuously improving. You can look at it as rapid feedback – understanding how an intervention is going, learning from the experience and making adjustments to continually improve.

The Learning Health System concept developed in the U.S. in the early 2000s. We have some groups in Canada that are very interested in it, but it’s very aspirational. In public health, the idea is to use data more effectively and work with others to have timely and co-ordinated responses to health problems. As we think about how to build Learning Health Systems, we should also be thinking about how to create Learning Public Health Systems – and build the two together.

Upshur: It’s an opportunity to mobilize high-quality research into health practices. Traditionally, the health-care and public health sectors have not worked closely together. The LHS is an opportunity to change that. And now, because of COVID-19, we know how to talk to each other a bit better. As well, if we wish to truly grapple with the social determinants of health and advance health equity, there is a need for intersectoral collaboration that is robust and enduring.

Feng: In our white paper, we point to the collaboration between health-care systems and public health during the pandemic. We’ve seen a lot more sharing of data because they had to try to co-ordinate the response to this very complex threat. Instead of patient data living only in the health-care organizations and public health and social services data in their silos, we’ve seen sharing of data in near real time. For example, to make mass vaccination clinics successful, public health had to work with doctors, hospitals, municipalities and others to get clinics staffed in areas of most need.

How could public health and health systems work together in a Learning Health System?

Feng: Take diabetes. Public health has information about diabetes prevalence at the population level. They share data with family health teams to tell you which patients are at highest risk. Doctors might put a clinic in those communities and public health would identify structural issues to address – such as could this community use another park, more walking trails or access to fresh food? The problem could be tackled at both the individual and population levels.

Aren’t health systems already always learning and adjusting based on new research and opportunities?

Upshur: We squander a lot of information in health. Every encounter every day should be harnessed to learn what works, what doesn’t work, how we can improve. Instead, we collect a lot of data that goes absolutely nowhere. In hospitals and clinics, most of the data is just collected and doesn’t get acted upon. One patient might have nine different medication lists in their record, which nobody looks at.

The big barrier is that we don’t have access to data in real time. In Ontario, ICES is extremely useful, but its data is usually a year behind. Physicians’ offices can’t harness their data in real time. Public health units have abundant data and considerable expertise in analysis and interpretation at the local level which are invaluable. All of these assets need to be aligned to improve population health.

Feng: Data are important, but it’s also about culture change. Right now, evaluations of interventions are more the exception than the norm. We need to make continuous learning a core part of the health system. Every time we try something new, we should build in feedback on how it worked.

What are some examples of Learning Health Systems being developed now?

Upshur: There are efforts to close this research-to-practice gap. At DLSPH, post-doctoral fellows are being embedded into Ontario Health Teams to make research less free-standing from the delivery of health services. Research is often seen as extraneous to the day-to-day work of health systems. Learning Health Systems would change that.

The evolution and growth of learning health systems is seen as a priority by the Canadian Institutes of Health Research, which means there will be an opportunity for research in this area. The Institute for Better Health, part of Trillium Health Partners, is an example of a health-care institution premising its research on the principles of a Learning Health System. Several DLSPH faculty, including Dr. Robert Reid, who contributed to this paper, as well as Professors Laura Rosella, Walter Wodchis and Kerry Kuluski are involved in this initiative.

Feng: Healthy Barrie is an example of it. DLSPH researchers are working with primary providers, hospitals and public health to tackle chronic health problems in this small Ontario city by sharing health records and other data, and then using the information to customize treatment approaches to populations.

How do Learning Health Systems square with the unlimited potential for change in the era of big data?

Upshur: There will likely be convergence between big data, machine learning, artificial intelligence approaches and Learning Health Systems. This will require investment in critical data infrastructure to enable linking and sharing data across the health and social services sectors. It will also require a transformation in the governance of health information. That’s somewhere in the not-too-distant future, I hope.