Speakers

Prof Sir Martin Landray, FMedSci
Professor of Medicine & Epidemiology
University of Oxford
Chief Executive, Protas

Khair ElZarrad, PhD, MPH
Director, Office of Medical Policy
Center for Drug Evaluation and Research (CDER)
U.S. Food and Drug Administration (FDA)

Slides

Keywords

Pragmatic trials; Research; Guidance, Regulatory, Data Governance

Key Points

• The clinical trial enterprise needs modernization. Time, cost and failure to recruit trial participants are significant barriers that must be addressed. The rapidly evolving ecosystem now relies on real-world data, which is becoming increasingly more advanced, and clinical trial design is becoming more complex and innovative.
• The International Council for Harmonisation of Technical Requirements for Human Use (ICH) is a unique harmonization organization involving regulators and the pharmaceutical industry. Its objective is to improve efficiency of new drug development and registration processes, in addition to promoting public health, preventing duplication of clinical trial in humans and minimizing the use of animal testing without compromising safety and effectiveness.
• The E6 Good Clinical Practice clinical trial conduct principles was intended to be an improvement of the G8 clinical trial design principles. The working group is engaging with academic stakeholders to determine new approaches to enhance transparency. An extensive training program with use-cases focused on trial designs still needs to be developed in the application of GCP guidelines.
• The new structure aims to provide clarity and better readability. The scope is focused, and the language aims to facilitate innovation in trial design. It also aims to set a foundation for practical and feasible expectations around the responsibilities from the sponsor and investigator, encourage a fit-for-purpose approach and incorporate learning from innovative trial design. Annex-2 of E6(R3) will include additional considerations of GCP principles focused on decentralized elements, pragmatic elements and real-world data sources.
• The principles aim to promote good science and ethics, be clear and concise, be inclusively developed and be progressive and durable.
• While E6(R3) provides a foundation for GCP expectations, guidelines alone are not adequate. The field will also need to collaborate on implementation and capacity building, develop responsive and accessible training and avoid all-or-nothing approaches on design and technologies.
• Better guidelines are needed to promote better trials, which will promote better health. Clinical trials should incorporate quality in their scientific and operational design, conduct and analysis.
• The principles focus on the concepts that “Good Trials” produce a scientifically sound answer to a relevant question through the principles: Principle 01) Informative and relevant; Principle 02) Respectful of participants; Principle 03) Collaborative and transparent; Principle 04) Feasible for their context; and Principle 05) Efficient and well managed.
• The current 11 ICH principles are generally sound, and the structure of Introduction > Principles > Annexes support critical application. However, improvements can be made, and further modifications are needed to ensure that the annexes are treated as implementation guides and not rules.
• Current proposed modifications include grouping principles into overarching themes, restructuring each principle and explicitly cross-referencing principles in Annex-1. There are also concerns that the ICH document fails to include scientific issues encountered quite often, such as allocation concealment, randomization, loss-to-follow-up, blinded evaluation of endpoints and others.
• There are also some excessive details on data, records and computer systems, which are unlikely to stand the test of time. This could stifle innovation and quality when new and different technologies are more relevant. Unnecessary, irrelevant, or ultra-specific guidance can cause more harm than good.
• GCTC will publish its comments at goodtrials.org and the public is encouraged to submit feedback as the work continues to improve the guidance.

Discussion Themes

–How much work was needed to get to this stage where the revisions are up for public comment? The work started before COVID-19, and the idea was to meet every six months. We had an opportunity to meet once, and then everything became virtual. The amount of work that went into the gap analysis alone was several months because we needed to have a thorough understanding of where the community is coming from, especially in areas that we knew were seeing advancements in design. We wanted to make sure we were encouraging flexibility of design and use of technology, while still highlighting what people need to be careful with. We also included multiple meetings with our academic stakeholders. We need to be clearer and more concise, while making sure that the statement fits with the needs of researchers. Training is also going to be essential to move this forward.   

–How did the pandemic impact the updates to the guidance? COVID-19 made us think about things differently, because one couldn’t apply rules, one had to apply principles. You must apply what really matters regarding the critical quality factors – for example, you can’t disrupt critical health care. The principles were the only thing we could fall back on, so I think that’s why it’s so important to get the principles right and lean on the trainings in all of our department.

–How do you balance precision and fit-for-purpose quality checks with more situations in which investigators are conducting trials using data from entire systems of health care? We need to create a space where this is an open dialogue and these kinds of conversations can happen. I think collectively, we want to promote the utilization of health care in research. We see that as an important part of the future. It would be such a waste of valuable data, valuable approaches, and valuable infrastructure to not embed clinical trials effectively into health care systems.

–What are the next steps? We encourage people to review the guidance and provide feedback. One important note is that we specifically moved away from a checklist approach. The guidance should not remove the need to think. Even when we have a guideline, we want to lead investigators to think based on the principles of this guidance.

Tags

#pctGR, @Collaboratory1

Speaker

Eric Perakslis, PhD
Chief Science & Digital Officer
Duke Clinical Research Institute
Professor, Department of Population Health Sciences
Chief Research Technology Strategist
Duke University School of Medicine

Slides

Keywords

Artificial intelligence; Machine learning; Gamification; Healthcare

Key Points

• Once people start learning about what technology can do right and wrong, the immediate response is often to start minimizing the risks of technology. However, Dr. Perakslis advocates that “instead of spending time and energy solving small problems, we should be solving big problems with bold solutions.”

• The key to quality AI and utilizing AI in a beneficial way relates to the quality of the data. It’s important for organizations to recognize this and upgrade their data in order to use these technologies more effectively and efficiently.

• “Serious games” are games designed for a purpose beyond pure entertainment. They use motivation, game design, competition, curiosity, collaboration, challenges, media, and learning to accomplish something. An example of when serious games have been useful is the creation of video games to treat depression and insomnia. However, gamification could have serious negative effects, and the benefits of these games in health care would relate to how well or how poorly they’re gamified, based on data, execution, and use cases. It’s important that people apply good judgement when utilizing these new technologies, especially in fields of science and medicine.

• Serious games pique the interest of people and can increase their likelihood of continuing a behavior. This can be useful in the context of clinical trials. For example, if the method of patient reported outcomes is gamified, patients will be more likely to continue reporting their outcomes on a regular basis rather than giving up or forgetting after the first time, which may increase the quality and quantity of data for the trial overall.

• In an interactive video game, people interact with NPCs (non-player characters). Those NPCs learn from interacting with human intelligence. This concept could be an interesting way to think about consent. In some ways, the idealized informed consent process would be one where the patients could ask the questions on their mind and get the answers, rather than only checking boxes on a plethora of situations that feel foreign to them. There are still risks associated with this process, but gamifying the game could ultimately create a more individualized approach to health care that prioritize the needs of each patient.

• Today’s toughest health care problems could be easily gamified so that people can access technologies that allow them to ask and answer questions. Implementing this is critical and doable according to Dr. Perakslis. The key is to have a deep understanding and control over the data being used. The smallest change in data can completely change the results when AI is involved.

Discussion Themes

-It seems like the speed of AI use is accelerating and our awareness isn’t keeping up. What is going to keep AI in the boundaries for health hand research right now? People need to be really focused. Tech companies that are promoting AI and rolling out these new tools quickly are going to be in the cross hairs of the aftermath when those tools are inevitably misused. It’s important to be thoughtful in picking partners to work with and making sure you ask the right questions before proceeding. The bleeding edge of these technologies is human talent, so if you’re struggling to keep up in this space, hire smart people. It’s also important to focus on the basics (cleaning up your data, etc.) before jumping into these new technologies to ensure the technologies work in your favor.

-Do you think AI within the context of health and research will be replacing people? If there’s one thing that’s absolutely true it’s that these new technologies can produce an unprecedented amount content, which is good and bad. For people, there are some jobs that will inevitably be replaced by these new technologies. However, these new technologies are more likely to replace people who can’t use the new technologies. The important thing is to learn and adapt to the changes.

Tags

#pctGR, @Collaboratory1

Speaker

Carl T. Bergstrom, PhD
Professor, Department of Biology
University of Washington

Slides

Keywords

Research, Misinformation, Media, Social Media

Key Points

• Misinformation abounds in medicine and about medicine. As researchers, we are producers and consumers of the medical literature.

• It is important to know and understand the information landscape into which your research is going out. What are the debates and controversies in the press and social media? If you’re working on a paper about vaccine safety, you have to understand the conversation about vaccine safety. How will it be framed in the public discourse? This will help you release information responsibly.

• It is important to avoid hype. Be very explicit about what findings do and do not provide.

• Issue responsible press releases. Researchers should consider themselves personally responsible for press releases. Researchers need to understand how preprints are received. It is important to understand that there are different standards depending on the topic. If you’re not prepared to see your manuscript appear in a leading medical journal, you should think carefully before posting it as a preprint.

• As a research community, we need to think about auditing citations because they are used as evidence for claims we make but many are falling short of that aim. One step journals can take is checking to see if cited papers have been retracted. More broadly, citations in medical journals often do not reflect the material cited. 10-20% of the time the claim is not a fair representation of what is being cited. We need to find a way to tackle this.

• Design stand-alone data visualizations because one thing that happens in a social media world is that the graphs and visualizations very quickly escape their containers. A paper is posted, it has a data figure, model or schematic, and it gets excerpted in the media or on individual’s social media feeds and the context is left behind. All of the necessary context should be included in the figure itself so people cannot accidentally snip it away from the relevant context.

• When we are communicating about our science, it is important to engage with traditional media and take the time to get to know reporters who are interested in the science. Engaging with social media  is important, particularly as false information starts to spread, it can make its way into traditional media. One of the best ways to stop false information on social media is having the author step in and clarify or correct misinformation. This is a very powerful and effective thing to do.

• We all have to prepare for generative AI. The landscape is changing so fast compared to other technologies. We already have the devices and platforms where generative AI is displayed. AI generated work has been submitted to journals and other publications. Publishers will need to rethink how to identify who is saying things and what are their credentials to make sure the ideas in the scientific literature are coming from humans who are trustworthy.

Learn more

Read Misinformation in and about science.

Discussion Themes

-Are you an optimist for the future? I am an optimist about science. We can innovate and change our norms and institutions to tackle problems. I am very pessimistic about big tech and social media, who are motivated by clicks and ads with no oversight. The data of what is happening with social media is not available unless you are collaborating with social media.

-How can we prioritize where to focus the resources to address misinformation? One thing that is essential is working within marginalized communities to understand the effects of misinformation on and in these communities and to partner with the communities most effected by misinformation to better understand where are the biggest harms being caused.

Tags

#pctGR, @Collaboratory1

Speaker

Christopher B. Granger, MD
Donald F. Fortin, M.D. Distinguished Professor of Medicine
Professor of Nursing
Duke University

Neha J. Pagidipati, MD MPH
Associate Professor of Medicine,
Duke University School of Medicine

Slides

Keywords

COORDINATE-Diabetes; Cardiovascular disease; Intervention; Cluster-randomized trial; Implementation; Prescription

Key Points

• High-intensity statins, ACEi/ARBs, and SGLT2i and GLP-1RA have been shown to improve recovery and maintenance outcomes for patients with diabetes and atherosclerotic cardiovascular disease (ASCVD). However, these therapies are highly underused in routine clinical practice.

• The objective of COORDINATE-Diabetes trial was to improve the implementation and adoption of these therapies by testing the impact of a clinic-level, multifaceted intervention on the prescription of 3 key groups of evidence-based therapies. This was a cluster-randomized trial of 42 cardiology clinics across the U.S., in which participants had type 2 diabetes mellitus and ASCVD. The cardiology clinics were randomized to one of two groups. The first group consisted of a usual care arm, where they received basic guidelines for treating diabetes or ASCVD. The second group of clinics implemented a multifaceted intervention at the provider level. The intervention included assessment of local practices and barriers, development of strategies to overcome those barriers, and audit and feedback of quality metrics, with the goal of improving the way clinics approach treatment of these diseases.

• About 6-12 months after enrollment, patients were assessed for the primary outcome, which was the prescription of all 3 recommended therapies. These therapies included an anti-hyperglycemic agent with evidence for cardiovascular benefit, ACEi/ARB/ARNI, and high-intensity statin.

• In the usual care arm sites, 14.5% of the patients ended up being prescribed all 3 therapies, and 37.9% of patients were prescribed all 3 therapies across the intervention sites. The 23.4% absolute difference highlights the important benefits of utilizing the provider education and interdisciplinary care pathway tools included in the intervention being tested in this trial.

• An important aspect of the intervention was regular phone calls and check-ins to patients prescribed the therapies to ensure they were taking the prescriptions. Results of the study showed a positive correlation between prescription of the therapies and the patients actually using the medications.

• Key study limitations were the need for remote delivery due to the COVID-19 pandemic and the lack of total representation of the broader U.S. or international population across the selected sites and patients.

• Ultimately, this trial concluded that a coordinated, multifaceted intervention increased prescription of 3 groups of evidence-based therapies in adults with T2D and ASCVD. The results of this trial highlight the under-use of evidence-based therapies in clinical practice and the lack of high-quality data on how to improve this gap. It is essential to scale this intervention across cardiology practices in order to improve the quality of care being delivered more broadly to ensure the implementation of these trial results.

Learn more

Read about the COORDINATE-Diabetes trial. 

Discussion Themes

-What are the challenges of implementing the implementation intervention of this trial? What is the trade-off between simple approaches versus multi-faceted approaches? Simple approaches are scalable and less expensive, but the benefits are less impressive. In this program, we were trying to change a more complex behavior with a higher activation energy, which is not only to prescribe one therapy, but to take a very complex patient population and prescribe multiple therapies, which inherently requires more than just a prompt here or there. One of the barriers we faced was a lack of education and familiarity with the therapies, which takes more than just a prompt to overcome. In this trial, we tried to land somewhere in the middle of the trade-off between simple and multi-faceted.

-Is there a reason for the disproportionate number of participants between the usual care and intervention groups? Ideally all participants would be identified and enrolled prior to starting the trial. This was not the case for COORDINATE-Diabetes primarily due to time constraints. They randomized the clinics, then enrolled the patients. This can be a limitation of cluster-randomized trials, as the knowledge of which group they are randomized to could impact patient enrollment.

-How can you sustain this intervention across all components of the process? That might be an important jumping point for the second leg of COORDINATE-Diabetes to answer questions such as: How do you get payers and Medicare to sustain these interventions? How do you elevate the pharmacists to get embedded in the clinics and remain engaged? How do you incentivize patients to see this through?

Tags

#pctGR, @Collaboratory1

Speaker

Ankeet S. Bhatt, MD, MBA, ScM
Associate Physician, Kaiser Permanente San Francisco Medical Center
Research Scientist, Kaiser Permanente Northern California Division of Research

Slides

Keywords

Heart Failure, Implementation Science, IMPLEMENT-HF

Key Points

• There are 4 drugs that modify 5 pathways, renin-angiotensin inhibition, neprilysin inhibition, SNS inhibition, aldosterone inhibition, and SGLT2 inhibition, that are the mainstays for treatment of heart failure with reduced ejection fraction (HFrEF).

• In modeling exercises, optimal implementation of these therapies, such as an ARNI, a beta blocker, an MRA, and an SGLT2 inhibitor, compared to conventional therapy (an ACEi or ARB plus a beta blocker) is estimated to afford more than 6 years of survival in a typical 55-year-old patient. This is compelling evidence that patients with HFrEF should be treated with this combination of therapies, but less than 5% of patients on are optimal guideline therapies for this condition.

• There is renewed interest in implementation science in cardiometabolic care to better understand optimal pathways for delivering these therapies in patients with HFrEF.

• The daily workflow consisted of a daily electronic health record (EHR) query that identified patients who had heart failure with reduced ejection fraction, who were admitted for any cause within sites in the healthcare system. Patients went through a screening by a study physician for eligibility. The physician-pharmacist team developed recommendations using an algorithm that was supported by the ACC, AHA clinical practice guidelines. They made only one suggestion a day for how to optimize the patient’s care, such as initiation of one therapy or an up titration of an existent therapy. The recommendations were communicated in the EHR as a note and the primary team received a page notification. The recommendations were at the discretion of the primary provider.

• There were substantial improvements in the intervention group of the pilot study and improved guideline medical therapy that indicated that the virtual nudge strategy might be beneficial and scalable to multiple entities within an integrated healthcare system.

• For the IMPLEMENT-HF Pivotal Study, 200 hospitalized patients were enrolled at 3 hospital sites in one integrated health care system, where patients either received the virtual care team intervention or usual care. The study team assessed the change in medical therapy from hospital admission to hospital discharge by assigning a guideline directed in-hospital medical therapy score.

• As compared to usual care, a virtual care team-guided strategy improved medical therapy optimization in hospitalized HFrEF patients across multiple hospitals in an integrated healthcare system. Benefits were consistent across most subgroups, including hospitalizations for non-HF indications and de-novo HF presentations.

Learn more

Read about the IMPLEMENT-HF study in JACC.

Discussion Themes

–How diverse were the settings where this was tested? The study team was interested in testing in a setting outside of a large academic hospital. Salem Hospital is affiliated with the health care system but is a community-based hospital. We need to test in systems that might have an enhanced usual care arm. Would this provide incremental value?

-What is going on at Mass General? It hasn’t been implemented in usual care yet but we hope that it will happen. There are a lot of programs that are aimed at this type of intervention at Kaiser but not one with a virtual intervention. It is something we are looking at seriously to see if we can run a small-scale version of this to see if it would be effective.

Tags

#pctGR, @Collaboratory1