Highlander Health Institute Hosts Inaugural Summit

Introduction

Highlander Health, composed of Highlander Health Institute, a public interest platform and Highlander Health Partners, an investment arm, works to solve the most pressing challenges related to evidence generation. The Highlander Health Institute (“HHI” or the “Institute”) focuses on establishing clinician-led learning laboratories to tackle issues in evidence generation such as siloed data, scalable solutions, and the complexity of integration between research and care delivery. In furtherance of this mission, HHI and Lyda Hill Philanthropies recently partnered with leading institutions to address three key topics: Baylor Scott & White Research Institute is assessing the ability to use routinely captured patient data to populate clinical trial and registry datasets, Memorial Sloan Kettering Cancer Center is demonstrating a practical approach to decentralized clinical trials, and the Duke Margolis Institute for Health Policy is conducting an analysis of the issues, policies, and technologies related to developing a nationwide interconnected learning health care system. Our goal is to partner with a broad stakeholder community to develop learning labs that can help demonstrate new approaches to solving key issues.

HHI recently convened a small group of collaborators at the Highlander Health Institute (HHI) Inaugural Summit, a working session focused on the future of clinical evidence generation, to discuss the initial readouts from the ongoing workstreams and chart a course for what comes next. We believe it is important to convene the correct actors to (1) accurately define and simplify the problem statements related to scaling evidence generation that have morphed, evolved, and mushroomed over the past few decades and (2) rightsize our solution appetites. Accordingly, we are focused on practical clinical trials, longitudinal data collection, demand signal creation, evidence generation modernization and advanced tools, and clinical research and care integration. The Institute is committed to transparently sharing its findings, so that other parts of the healthcare system can learn with us and work towards a more holistic evidence generation approach.

A common question presented to the Institute notes that the healthcare ecosystem has faced the same problems for the last thirty years and asks how HHI differentiates its work from previous efforts. The Institute deliberately chose these projects because they represent purposeful building blocks for a single, all-purpose data infrastructure that can solve evidence generation and capacity challenges while not overburdening the health system and impacting the delivery of care. In addition to a first principles approach, HHI is taking a long view towards solutions development: we are focused on sustainability and scalability. That being said, we understand the importance of demonstrating progress and encouraging momentum. Our initial projects all focus on the short- and intermediate-term in order to maintain the velocity of change. Finally, technological leaps in the past decade have enabled intrepid problem solvers to address previously immovable siloes or intractable problems. 

Below, we present a brief overview of our three in-flight projects and summarize some of the rich discussion that took place in between sessions.

Baylor Scott & White Health Research Institute: Building the Infrastructure for High-Quality Real-World Data

The Baylor Scott & White Health Research Institute (BSWRI) has undertaken an effort to determine the feasibility of using real-world data (RWD) generated by its health system for regulatory decision-making. This project focuses on evaluating the data maturity of in-system electronic health data and identifying domains for improvement. BSWRI researchers began by conducting a baseline assessment of the ability to use data being routinely captured at BSWH to populate clinical trial and registry datasets. This involved extracting data from a central warehouse using structured and NLP queries and documenting the practicality of utilzing it to populate registry and clinical trial case report forms (CRFs). The project will now focus on identifying an approach to abstracting unstructured data while meeting data quality standards, better understanding data augmentation, and developing granular data quality assessment frameworks. 

BSW has participated in the transcatheter valvular device registry, TVT, for over a decade; they also led some of the related clinical trials. CRFs from TVT and one of the landmark clinical trials are being used to test the success of this approach. While the project is still underway, the BSWRI team has identified a number of early learnings, including the importance of mapping to a consistent terminology for CRF variables, differences between registry and clinical trial variables, the value of parsimony, and dependence on prospective data entry for some fields. Additionally, while Baylor Scott and White uses a single EHR system, consistency challenges arose when aggregating data from multiple sites due to variations in workflows, documentation, and custom flowsheets. Equally importantly, they have key insights into the team and processes required to build reusable infrastructure that supports this work at sophisticated U.S. health systems. 

It was acknowledged in the discussions that advances in artificial intelligence (AI), such as large language models, may be able to accelerate this work, but it is imperative to build the infrastructure necessary to deploy and monitor such transformative technology. Next steps include deploying BSRWRI’s approach at other health systems to demonstrate its portability, to continue to define the path to scale at BSW, and leverage their approach to data quality and efficiency assessment to serve as a measurement system for elements beyond just registry and trial population, such as documenting new processes and understanding the impact of AI and other technologies over time.

Memorial Sloan Kettering: Improving Clinical Trial Operations Infrastructure to Support Decentralized Studies

The traditional clinical trial operations paradigm dictates that each site involved in a clinical trial requires its own self-contained ecosystem. For example, each site should have local start-up processes, research teams, primary investigators, labs, and data entry approaches. However, a decentralized model of clinical trials suggests that some elements of trial execution can be supported by a central team, allowing the trial to operate in a hub-and-spoke model. Local sites would support critical clinical elements while a central team could provide the rest of the needed scaffolding. Interestingly, this represents  a decentralized model facilitated through centralization of certain capabilities, and indeed, this decentralized model showed great promise during the pandemic, when clinical trials needed to adapt to social distancing and shelter-in-place recommendations. 

Memorial Sloan Kettering (MSK) has created a comprehensive project to understand best practices, using existing opportunities to operationalize decentralized trial capabilities. MSK has broken down the project into three workstreams: (1) activation, budgeting, and contracting, (2) screening and matching, and (3) data management and operations. The short-term focus of the first workstream involves partnering with MSK Cancer Alliance sites on existing strategic projects,leveraging clinical research collaboration and master institutional review board (IRB) reliance agreements. This workstream will ultimately establish a network of vetted spoke sites that align with a streamlined feasibility model, aiming to transition towards single IRB and prime agreement structures to reduce the average time to activation. The screening and matching workstream intends to transform operations from manual screening of site-identified patients to incorporating a trial matching solution into a spoke site’s scheduling and electronic health record (EHR) systems. This will enable automated screening of patients for open trials and reduce the manual work. The final workstream aims to move away from manual data entry by utilizing an Electronic Health Records to Electronic Data Capture (EHR2EDC) system that will improve the timeliness of data entry, reduce the burden of data entry, and ultimately improve the quality of the resulting data. 

A key feature of the project is that it is a test environment - a learning lab - intended to evaluate and document the net contributions of new technologies and workflows as they are reimagined. Baseline measurement sets the floor, and new approaches must be net positive to be incorporated. Many of these have been discussed in the past, but this represents a unique moment in time where they may truly prove valuable. A transparent measurement environment supports the rapid and practical adoption of new solutions.

Duke Margolis Institute for Health Policy: A Policy Framework to Catalyze High-Quality Source Data Collection, Curation, and Linkage

The Duke Margolis Institute for Health Policy (DMI) conducted a landscape analysis to better understand relevant issues, policies, technologies, methods, and implications for care delivery related to source data collection, curation, linkage, and use for evidence generation. Following the conclusion of the analysis, DMI and HHI believe that policies to advance fit-for-purpose longitudinal clinical data for patient care and quality improvement can provide sustainable support for learning networks across healthcare organizations and a scalable, reusable, pragmatic clinical research infrastructure to improve health outcomes and reduce costs.

DMI presented the results of a research project that identified 41 use cases related to developing sustainable real-world data generation. The projects spanned multiple source data types (EHRs, claims, registries, etc.) and involved a variety of therapeutic areas. The DMI team also presented several recommendations focused on (1) improving data quality at the points of collection and curation, (2) fostering interoperability and optimizing a limited number of fit-for-purpose data elements, and (3) supporting a holistic patient view and enabling long-term follow-up.

The full research report will be published shortly and will serve as the basis for future research and policy assessments.The policy framework introduced in the research report proposes developing pilot or demonstration projects to explore improvements for specific clinical care interventions in health systems. Interventions would be paired with technological investments and regulatory alignment to foster pragmatic clinical research to synergize with learning health systems. Future research would in turn inform needed policy reforms that can collectively orient stakeholders to address the systemic siloed nature and inefficiencies of the current evidence generation system.

Discussion

One of the most impactful parts of the day was the opportunity for rich discussion among project participants and observers. HHI envisions a model for scaled evidence generation, achieving clinical research insights in half the time and at half the cost. There was broad agreement that data, operations, and policy represent core components in the path to accelerating evidence generation. There was also strong support for the initial strategy of continuing to build learning labs in partnership with health systems as platforms for testing new workflows and honing in on outcomes that truly move the needle. Stakeholders also acknowledged that true progress that will reimagine evidence generation writ large could take decades, but we can incrementally build the infrastructure, alignment, and use cases to enable change over time.

A few participants noted the importance of broadening HHI’s scope to focus on use cases involving clinical care, patients, and payers. In particular, there is a strong demand signal for use cases on the evolution of the Centers for Medicare and Medicaid Services (CMS). Explicitly defining and creating projects with an eye towards areas such as coverage with evidence development may provide a foothold for CMS policy generation and accelerate change.

One repeated challenge from many voices was to keenly focus on “good enough” from a data infrastructure perspective. As a whole, we tend to advocate for an entirely comprehensive data collection and curation infrastructure with every possible outcome measure and data element collected and validated. The reality, however, is that only a fraction of the collected data is needed to conduct high-impact research. The collection of extra or unused data places a significant burden on providers and does not necessarily yield better evidence. Similarly, a conversation needs to be had on the optimal length of longitudinality in a dataset - are we may be overthinking how much is required.

Additionally, an interesting paradox arose when discussing care delivery. There has been significant advancement in technology and infrastructure that has improved the actual delivery of medical care; however, it has corresponded with the patient’s perspective of care becoming more complicated, more opaque, and harder to access. We must keep the patient at the forefront of our actions and ensure we are not losing sight of what matters.

The discussion sessions also left us with a few open questions to explore in future convenings and projects. First, particularly as it relates to AI, how do we ensure that tech advances are ready for prime time and increased scrutiny? In other words, how do we ensure the validation of tools that help us build and scale data infrastructure. Second, what lens are we viewing the solutions through? Using a regulatory decision-making lens was lauded, but are there other lenses that may enable better operationalization of repeated, varied, and scaled re-use of data? Finally, is the time right to consider new approaches to privacy and consent?

We are thrilled by our early progress and believe that our Inaugural Highlander Health Institute Summit enabled a dialogue and spurred conversation that will continue to drive progress. We will continue to provide updates on our projects as they advance, and we look forward to involving others in the conversation.

We thank all of the Summit participants and Lyda Hill Philanthropies for their partnership.