Clinical Innovators | Innovating Cardiology: Lessons for Health Care An Interview with Robert Califf, MD

By Amol Navathe, MD, PhD

With technology rapidly evolving, the push for real-world data has become, well, "real." The Patient-Centered Outcomes Research Institute (PCORI) was established by the Affordable Care Act to improve the evidence physicians use to make clinical decisions, but the challenges of integrating innovation are as numerous as the types of interventions to study.

Robert Califf, MD, vice chancellor of clinical and translational research, director of the Duke Translational Medicine Institute (DTMI), and professor of medicine in the Division of Cardiology at the Duke University Medical Center in Durham, North Carolina, has played an international role in leading the conversation about these issues. In this interview, Dr. Califf delves into this brave new world with a particular focus on cardiovascular medicine.

What are the priorities for innovation in cardiology?
In order to have the biggest impact on longevity and quality of life, we need to fully invert our hospital and health system-based perspective and start with the person living in the community. Our first priority is to help them avoid illness and disability, and even when that happens, we should strive to develop systems that enable self-monitoring to preempt cardiovascular events as much as possible. In that view, seeking care at a clinic happens when the patient feels the need for it or when evidence-based monitoring suggests that an in-person visit will have a benefit.

For the most part, hospital admission is acknowledgement of a failure of care and prevention. Of course, cardiologists should continue to perfect technology and delivery systems for hospital care—that is a special role that is lifesaving— but we should acknowledge that preventing the need for hospitalization is a critical goal. Accordingly, where we need innovation is in monitoring and deploying services close to the patient, close to where he or she lives and works.

A key driver will be the ability to self-monitor and send data to a remote health provider, identifying high-risk patients before the cardiac event occurs. And the notion of high risk will change from considering general schemes to prevent heart attacks on a population basis to "based on your monitor, we think you are high-risk for the next 24 hours."

This concept is especially interesting to me because I had the amazing privilege of assuming responsibility for a coronary care unit when we figured out blood clots cause MIs. It always bothered us that we couldn't prevent the event, but had to deal with it when the patient sought attention or came to the emergency department. This is not to downplay the importance of traditional delivery systems, but I see that as incremental and less transformative.

Should we expect delivery innovations to have more effect on population health than molecular therapeutics in the future?
It's hard to say whether molecular therapeutics or delivery innovation will be more impactful. Therapeutics and risk monitoring are two types of device technologies that will be critical. Finding a way to organize and maximize information is also very important. With our current state of knowledge about systems biology and behavior, we already know that a patient's consumer preferences may be more important to their health than the best biomarker panel (except in the rare cases of a few unusual genetic diseases). I'd say that it's really the integration of molecular biology with physiology, behavior, social interaction, and environment that will make the difference. In the past, we just didn't have the computing power to put these massive amounts of data together and analyze them, but now we do.

With the advent of comparative-effectiveness research and the enhanced focus on "real-world" impact, will the research paradigm move away from the traditional randomized controlled trial?
I predict the opposite. We, as a research enterprise, are going to go deeper into randomized trials, but future trials will look different in several ways. First, our ability to monitor entire populations will change. A challenge with our current approach is that we can optimize a clinic or a hospital population, but it is possible we miss that a drug is hurting those who aren't engaged.

To improve health, we are going to need a sensor for the entire region (city, county, or state). For example, if we have a billion dollars, do we put it into fancy shiny hospitals versus distributed clinics in neighborhoods?

I would argue the impact on health would be better with neighborhood clinics, and that the use of new personalized monitoring devices will tremendously expand the ability to enroll people in studies that have immediate importance for practice. In the past we couldn't really ask that type of question, but now with increasing technological capability we might be able to do so.

Second, we will be using more connected health. Once remote monitoring gains traction, we can have a model that is continuously running. We can look at inputs and outputs, when patients show up to clinics or hospitals, how often they go for a walk, if there are crimes in their neighborhoods, and so on. Then, if one parameter changes, we can measure what changes across the rest of the system in real time.

Through these kinds of data, rapid-cycle engineering will emerge, often using randomization when appropriate, but driven by continuous quality improvement. Finally, randomization is here to stay. When it comes to measuring whether intervention A affects outcome B, unless we expect huge effects, we absolutely need randomization.

I like to reference Google in illustrating this concept. Very early on, Google realized it had a new paradigm called "cookies" that was their independent variable while the dependent variable could be advertising revenue over time. Google studied how changing characteristics of a computer screen's appearance affected its advertising revenue, just as we study how medical interventions affect patient outcomes.

To do so, Google started to randomize interventions like bolding and underlining to evaluate the impact on advertising. Now they are the largest organization that does randomized clinical trials with thousands of "trials" going on at any given time. They found that small changes aggregated over large numbers of people have a profound effect on advertising revenue, and the value of ensuring a valid base of knowledge based on randomized evidence is profound.

They also studied maintenance of effect with multiple changes. In cardiology, we still don't know if after years of being on a beta-blocker post-MI whether your cardiologist should still be prescribing it to you. So Google is far better than medicine at creating a learning environment. We need to catch up!

Will the burden of proof, our p-value threshold, change as the trials become more real world?
We have demanded a high level of evidence for only a few, selected clinical decisions, and we ignored everything else. Only 12% of cardiology practice guidelines are truly based on robust evidence. Hopefully, PCORI will play a key role in helping to prioritize the evaluation of interventions using outcomes that are important to patients and families, addressing those with the highest level of evidence. For other issues, we can reduce the strictness of evidence to move practice forward, and I would be in favor of being a bit less obsessive about the alpha level of trials, as long as we agree on which trials need different levels of evidence.

How are these innovations in connected health and community-based clinical trials going to affect the practitioner?
First, the level of evidence to support practice will get much better because the cost of doing research is going to drop. A challenge is that clinicians everywhere are going to need to keep up, for which I think cardiology has been the exemplar thus far.

Second, technology in the cardiology domain is going to be awesome. Cardiologists should be participating in testing out technological innovations. Recently, I was talking with a surgeon who was confronted with a 92-year-old patient with aortic stenosis and a pulmonary artery systolic pressure of 108 mm Hg. He performed a percutaneous aortic valve replacement while talking to the patient. How amazing is that?

Third, cardiologists are really savvy at integrating technology, communication, evidence, and innovation. In this information-driven world, we are going to have the astounding opportunity and responsibility to help people monitor themselves and seek care at the appropriate time. Over time, this will move the paradigm from the hospital as the primary focus of operations (as if filling the hospital is the goal of the health system) to a distributed system with a virtual command center linking individuals, neighborhood services, doctors' offices, and hospitals, thereby deploying services to the right place at the right time as close to home and work as possible.

Amol S. Navathe, MD, PhD, is a physician at Brigham and Women's Hospital, a clinical fellow at Harvard Medical School, and adjunct faculty member at the Leonard Davis Institute of Health Economics of the Wharton School of Business, University of Pennsylvania. Dr. Navathe is also the co-Editor-in- Chief of Health Care: The Journal of Delivery Science and Innovation.


< Back to Listings