Robert M. Califf, MD

Today, the Hasting Center Reports published a pair of articles with 2 introductions and 7 commentaries.  The papers addressed a new way of thinking about research as we move into the era of “learning health systems”.  In the past, research and practice have been seen as existing in 2 worlds with a “bright line” between them.  Practice is characterized by a commitment of a practitioner to act in the best interest of the patient, whereas in research the patient is a subject, and the researcher is committed to the protocol with the “subject” as an object of research.

I will be discussing aspects of this remarkable effort over the next few days.  You can read the press release here: http://www.eurekalert.org/pub_releases/2013-01/jhm-eat011113.php

I’m having a fascinating time at a meeting we’re hosting that brings together academia, FDA, industry, CMS and others to discuss the care of patients with atrial fibrillation.  Mike Cuffe, one of our Duke products (internal medicine and cardiology), is here in his job as head of the physician practice at Columbia HCA, one of the largest hospital corporations in the world.  Now that the election is over, we’re absorbing the likely impact of the accountable care act.

The system is already strained to the extreme and millions of uninsured people will have insurance a year from now.  In my view, this is a good thing, but we’re not prepared.  There is an amazingly rapid implosion of private practice as we have known it, as independent practices are being bought or docs are signed up to long term contracts–essentially doctors are rapidly becoming employees of large systems. In our discussion, it seems that while there are many benefits, 2 trends are occurring that need to be taken into account as we design research systems: 1) doctors are working fewer hours as they become employees; 2) the systems are requiring efficiency (basically translated into fewer minutes allowed per patient).

Now factor in our new knowledge that we need to combine research and practice into a learning health system in which we learn from our cumulative experience.

The logical conclusion is that we need a major system redesign.  The economics and technology will not tolerate the current way of doing things, either in research or practice.  No one knows what the new system will look like.   Interesting times.

I have attended the first of a 2 day meeting on “large, simple trials” at the Institute of Medicine today.  There is a growing consensus among major funders of research and leadership at FDA that we need to streamline clinical trials and accelerate the pace of moving to electronic health records for the collection of much of our clinical trials data.  The “learning health system” in which we constantly use accumulated data from practice to learn about best approaches to diagnosis and treatment is the ultimate goal.  The case for this change in approaches was made by federal agency leaders, patient advocates and there was support from the medical products industry, both devices and drugs.   Tomorrow we will be addressing the obstacles that must be overcome to make the proposed new learning health system work.

My faculty career began as a fledgling Coronary Care Unit Director, making rounds and caring for people before we knew what caused heart attacks.  We would put them to bed, give morphine, lasix, and nitrates and study the “natural history” of the disease, marked by a high mortality rate and great uncertainty.  Shortly after I took over (on call 12 of 14 nights covering both Duke and Durham Regional Hospitals for $35,000 in salary), the miraculous discovery about blood clots causing heart attacks occurred and my career was launched as we went through an amazing phase of discovery and research that revolutionized the treatment of the leading cause of death in our society.

This weekend, I had the privilege of making a rounding effort in the CCU.  It made me realize that while the career opportunity and scientific thrills were amazing, the real value was the experience of being in a place where one could simultaneously work hard to help fellow human beings in serious trouble and interact with learners coming through the system created at Duke.  The caring and human element of dealing with critically ill people and their families and the amazing sensitivity of hospital staff to the complexity of the environment, taking the time to shepherd trainees through the experience while also dealing with understandably upset family members is reassuring about the real meaning of being human.  It was especially fun to realize that many of the nurses on the weekend shift have  been there for more than 20 years, having gone through many experiences themselves, but coming back to the fundamental opportunity to make a difference to the critically ill.  While critical care is frightening and overwhelming, it has an amazing way of erasing the veneer of pretense and formality as we confront the reality of life threatening illness and all the commonality it exposes.

The problems we faced in rounding this weekend were the same in many ways, but in 1980 we would have been overwhelmed with the number of 80-100 year old people otherwise in good functional status except for their cardiac status (and associated co-morbidities) and at the technological fortress that has been built at major medical centers today.

I leave the weekend feeling energized and more grounded about why I do what I do… and appreciate the many people who grapple every day with the complexity of our health system in an effort to improve the condition of their fellow human beings.

I had the chance last night to attend “Friends of Nursing”, an annual event hosted by the Duke Health System. The brainchild of grateful patients and Mary Ann Peter, this event honors the work of the real heroes of healthcare–the nurses who care for patients every day. Mary Ann was a nursing/hospital administrator when I was starting out as CCU director at Duke. Between she and Deborah Roth, I was enabled to carry out many of my ideas with their guidance.

More importantly, I learned from day to day work on really sick people how irrelevant theory can be relative to the importance of steadfast caring at the bedside. It was particularly exciting for Wanda Bride to get a new award: the Legend of Duke Nursing award. Wanda truly is a legend, having taught multiple generations of nurses and physicians how to provide first rate clinical care in an environment that also appreciates the value of research. Wanda literally is responsible for special skills among hundreds of Duke cardiologists now practicing all over the world. 39 years is a long time, but she has made the best of every minute.

I continue to hope that before I give it up that we’ll have a living example of the “learning health system” in which the kind of dedication to the well being of patients exhibited by the 850 nurses in attendance at this meeting is coupled with an integrated approach to clinical care in the context of learning through data collection and analysis.

We have a complex data set in which we have combined electronic health records, social and environmental information into a composite dataset that represents an entire county. This should allow us to make a prediction about each individual, each neighborhood and the entire county for future health outcomes.

Funded by the Bristol Myers Squibb Foundation and the Centers for Medicare and Medicaid Services, we’ve now put the data into an analysis data set. The first topic of interest is Type 2 Diabetes. What is the risk of death, major macrovascular events (MI , stroke, heart failure, unstable symptoms requiring hospitalization, etc.), and what are the anticipated costs? Armed with these predictive models, we’re hopeful that individuals, neighborhoods and health systems will work together to focus on the most effective strategies to improve outcomes and reduce costs.

But the next step in our work is to use the best possible analyses to devise the best predictive models. We’ve invited analysts from different backgrounds to work together on this project. The nice thing is that we can “hold back” an entire year’s worth of data with thousands of events so that the different potential predictive models can be tested in an independent data set.

In the end, I hope we can develop a community approach involving the DCRI, Department of Statistical Sciences, “big data people” on campus, our colleagues at NC State and perhaps industry collaborators such as SAS to develop a systematic approach to ongoing analysis to predict a variety of health outcomes from the this rich data source, including the many additional counties we may add on.

Now that the election is decided, it seems that there is a window to operationalize the Accountable Care Act in a way that would improve the overall health of the US. As we have developed our CMMI Innovations Challenge Grant and our BMS Foundation Effort to implement the Durham Health Innovations Project, it has become clear that much of the “progress” from the view of individual clinics, hospitals and health systems is really segmentation of populations in a way that may make one venue look good, while having little impact, or even a detrimental impact, on overall public health. We now have a data platform in which the electronic health records along with other data are embedded into geospatial maps to provide a medical, social and environmental snapshot relevant to entire counties, neighborhoods and individuals. We will be rolling this out over the next few weeks and months to a variety of audiences to get input into how to make best use of what has been developed.

Yesterday was an interesting day for late-breaking clinical trials. The FREEDOM Trial showed a clear superiority of bypass surgery over PCI in multivessel disease patients with diabetes. Not a complete surprise, but the clarity of the results was way beyond my expectations.

The TRILOGY Trial, led by Magnus Ohman, showed that differences between clopidogrel and prasugrel with regard to inhibition of platelet aggregation did not result in differences in clinical outcomes. The excellent lab work by Paul Gurbel’s group showed that this biomarker is not ready for clinical use and may never be sufficient. Time after time, we are forced to face the deficiencies of biomarkers and the residual uncertainties after testing.

The stunner was the TACT Trial, with DCRI as the data center and Dan Mark doing the quality of life. This trial was initially funded despite the objections of many in the scientific community who felt that the funding could have been spent much better on basic science. Many mainstream researchers and ethicists saw the trial as not ethical, since they could not imagine a benefit from the treatment. There has been a firestorm for the duration of the trial. However, when all the dust settled and the analysis was done, there was an 18% reduction in CV events. Detractors have come out in force to attempt to discredit the trial, but this NIH funded and monitored study was done in a rigorous fashion, despite the efforts of many to undermine its conduct. As with all trials it was imperfect and a sensible interpretation would seem to call for more evidence as the observed result was unexpected. None of the trial leaders are calling for immediate deployment of chelation therapy as a mainstream treatment.

However, I am quite disappointed at the tendency for leaders to avoid taking risk and to accept an honestly done study for what it has to offer. I hope medical journals and thought leaders will develop a more open mind. How many more trials like these 3 will show a counterintuitive result before we begin to accept that proper RCTs are needed to validate or refute beliefs about therapeutic benefit? What if there truly is an 18% event reduction with chelation? Who should be more ashamed: the investigators who did the study or doctors and hospitals that have profited from PCI while not participating in the RCTs that generate the evidence about the role of PCI in patient care?