A qualitative study of design stakeholders’ views of developing and implementing a registry-based learning health system

Why it matters

Supporting people living with chronic conditions in ways that fully respect their needs, values, capabilities and priorities is a challenge for health systems. At the same time, there is growing demand for better data to support improvement in care and to facilitate research.

Learning health systems are a promising approach to addressing these challenges by harnessing opportunities to record, collate and analyse routine patient data, and presenting them in a format that is accessible and useful to patients and clinicians alike.

One way a learning health system might work would involve patients recording daily information about their symptoms, medication side effects and health electronically, using an app or website. Combined with clinical information from their electronic health record, this detailed information could support communication between the patient and their clinician and facilitate joint decision-making about treatment and care. Additionally, this information, particularly when combined with data from many other patients, could enable learning across the health system, facilitating improvements in quality, safety and value.

However, real-life examples of learning health systems are rare, and few have been studied.

Our approach

We set out to examine the views of people involved in designing and implementing the US-based registry-supported care and learning system for cystic fibrosis (RCLS-CF).

The RCLS-CF exemplifies many of the promising features of learning health systems. It is based on a defined patient population (people with cystic fibrosis), and it combines data from a patient registry of clinical measures and patient-reported data. These are shown together on an electronic dashboard that patients and clinicians can view as a basis for shared review, planning, and decision-making. The data can also be repurposed for research and service improvement.

Our study was in two parts. First, we interviewed 19 people involved in designing and delivering the programme. These initial interviews aimed to understand their assumptions about its essential features and what would be required to make it work. Then we undertook 11 follow-up interviews to deepen understanding of the initial findings.

What we found

In both parts of the study, participants expressed strong consensus on the goals of the programme. It should use new, digitally enabled forms of data-sharing (both patient-generated and clinical) to try to change the dynamics of healthcare. Participants understood the ultimate aim as being the facilitation of better partnerships between patients, families, clinicians, researchers, and healthcare improvers.

In the follow-up interviews, participants suggested thinking of the clinical encounter as involving two experts: the clinician, an expert in biomedical knowledge, and the patient, an expert in their own experience of living with a life-limiting illness. It was proposed that this reconfigured relationship would result in new forms of partnership.

For the system to succeed, participants proposed that the data collected must be both clinically useful, and meaningful to patients and clinicians. They emphasised that co-design of the dashboard was both an ethical imperative and necessary to ensuring that the system would secure engagement.

They suggested that the system needed a technological infrastructure that would support easy data entry and joint decision-making. It also needed the right social conditions, including that patients and clinicians are willing to work together in new ways.

Our study has shown that accommodating the needs of patients and clinicians along with those of research and service improvement is not straightforward. Even when the values and vision are clearly communicated by design stakeholders, implementation can be challenged by social issues and ‘mundane’ technical problems.

Some patients may be keen to engage with a learning health system, while others may prefer a more passive relationship with their clinicians. Understanding how the system affects clinicians’ workflow is particularly important for its success. Clinicians would have to see that a system has benefits if they are to use it. However, the challenges in this programme did not arise from any fundamental disagreement about underlying objectives or values.