Research Permissions–Angels on the Head of a Pin, or the Key Issue to Decipher?
NEJM today has a pair of sentinel papers on the issue of research permissions for comparing therapies already in routine use in practice. Ruth Faden, Nancy Kass and colleagues from Hopkins, building on the ir Learning Health System work featured in the Hastings Center Reports, argue that in a proper context specific informed consent is not needed. Scott Kim and Franklin Miller from NIH argue that consent is needed, but that current circumstances call for a very different approach that integrates information needs in clinical care with knowledge acquisition through randomization. In their approach the consent would be very short and consistent with a fabric of information already given to people about what is known about the care they are receiving and the choices being made by the patient and provider.
The exciting element of the pair of articles is that the agreement is much greater than disagreement. Both groups of ethicists agree that the current approach is not optimal. In discussions about this issue, many colleagues have pointed out to me how seldom we really inform people about the uncertainties of current clinical recommendations. Over 85% of decisions in cardiovascular disease are not based on high quality evidence and other specialties have even a lower rate.
I hope these articles will move people to a better place that might include:
1. much better information in practice about what we really know (evidence based medicine) vs our opinions (opinion based medicine) and a better characterization of the uncertainty
2. notification and consent that doesn’t treat randomization as a major event, but rather as a tool to help inform us all about what we should do to optimize health
A colleague recently said: “When you get offered the next procedure where the evidence is unclear, maybe you should be asked: “Would you rather that I guess whether this procedure is good for you, not tell you that the doctor next door would do something different and do the procedure with a promise that we will learn nothing that will give the next patient a better chance? Or would you rather enter a study in which we let you know about our uncertainty and learn about the procedure so that all future patients can benefit?”.
A Tipping Point?
In the past 48 hours I have been involved in many meetings and discussions about learning, health decisions and healthcare. Ray Gibbons, a former Duke fellow who went on to greatness at Mayo Clinic and was President of the American Heart Association, pointed out that we could save a million lives if we just implemented systems that made sure effective therapies were given. On many conference calls and discussions, patient advocates and researchers were realizing that the evolving texture of health system data would give us the ability to intelligently provide useful information about choices to patients in the near future. Liberating us from the ignorance that comes from the isolation of practitioners as they try to learn from their own limited experience will be a major advance in medical history. Even as we struggle trying to put data into cumbersome systems like EPIC we are beginning the see the fruits of our labor as patterns emerge in data that guide better decision making. The invigoration of administrators and clinicians, now armed with information, is leading to fascinating discussions about the boundaries (or lack thereof) between quality improvement and research involving human “subjects” (also known as participants!).
The outcome seems inevitable– much better decisions made by patients and clinicians armed with better and better data. The question is: when will we reach the tipping point and what can we do to reach that point more quickly?
The Human Being as the Target for Study
AS we are reviewing our efforts to use modern technologies to advance our knowledge of: 1) human biology; 2) the impact of new and old interventions on human biology, it has become increasingly clear that we are not configuring our enterprise optimally. In most of our country’s best academic centers we have expertise in gene sequencing, genomic technologies, imaging, pharmacology and informatics. And in pharma and biotech, the success rate remains miserable and surprises continue to occur in late phases. Yet, we just haven’t organized our resources to optimize knowledge acquisition. Academic centers are not well organized and pharma continues to use cookie cutter phase I designs.
The recent work of Geoff Ginsburg and multiple colleagues at Duke is demonstrating that taking a new look at old problems (aspirin dosing and infection detection) provides a different view of health and disease than we had expected. Mike Snyder’s work at Stanford shows the same when intensive measurements are applied to an individual. Cancer continues to evolve to a view that pathways may be as important, or more important, than organ specificity when it comes to therapy and prognostic stratification.
Is it time to reconfigure? How do we do it without recapitulating the issues that led to the dissolution of the old GCRC system?
What is a Research Site?
We are accustomed to thinking of a research site as a clinic or a unit in a hospital. In the past 48 hours I have had the chance to speak with leaders of the PCORnet based consortia for “research sites” that include the city of New York, the city of Chicago and seven states in the midwest. I believe that we’re going to need to think of a new nomenclature in an era in which clinics and hospitals are owned by integrated health systems which roll up into regional consortia. The entire infrastructure support will be different if this approach succeeds.
Building a National Network–Early Phases
We’re having a very interesting time meeting the leaders of 11 Clinical Data Research Networks (CDRNs) and 18 Patient Powered Research Networks (PPRNs) selected through peer review for PCORnet. The amount of talent, technological sophistication and desire to make a difference is impressive. The CDRNs will provide massive involvement of integrated health systems and their patients across the spectrum of health and disease, while the PPRNs will provide a deep set of data and commitment from patients and families with particular diseases and/or medical problems. The intersection of these 2 types of networks into the “network of networks” will be fascinating.
While the potential is obvious, the complexity and lack of coordination is the major challenge. So much of our clinical research infrastructure is either isolated from resources that could enhance the effort, and the approach of building a clinical trial, getting a result, taking down the trial system and starting over with the next one is insane!
This is our chance to move the clinical research enterprise to a new level, and much of the fun of it will be the combustible mix of people with so much talent who have not worked together before. The product is likely to be something that we could not have predicted.
A New Year with Great Opportunity for Researchers
I believe that 2014 will be the year in which a transformation happens in the 2 big spheres of clinical research: intensive human biology and clinical studies involving cohorts and populations. A major shift to a new data fabric of electronic health records, registries and data derived directly from people/patients from personal devices will begin to supplant traditional stand alone data systems. This change will trigger a reformation of clinical epidemiology and physiology as we apply big data methods to reclassify disease and understand biology, and it will be applied to populations. The population component will lead to a fusion of observational studies, clinical trials and epidemiology to a common discipline applying different methods on top of a new information infrastructure. The launching of the PCORnet (www.pcornet.org) will be one major example, but other efforts will soon become apparent. These changes will not happen over the course of one year, but I believe that 2014 will be the tipping point.
I had the chance to hear about Duke Medicine’s plans to contribute to downtown Durham. The city has been through many phases, but what is happening now is quite amazing. An increasing number of our research and operations people are moving into renovated or new buildings in the old “tobacco and textile” town.
The trajectory is clear: a mixed model of business, academia, biotech, cultural and great food. It should be an economic and cultural engine for the region. The DCRI’s Durham Centre is a key part of the plan and the building is fantastic.
One question is whether we can use this engine to also improve the regional ambience in terms of education, health and participation in cultural advances.
Another Reminder that Evidence is Needed for Practice!
Today Steve Kimmel from Penn presented the results of the COAG Trial, the NIH funded study of pharmacogenetic testing with warfarin administration. Simultaneously in the NEJM, 2 other studies appeared, one of which was also presented at the American Heart Association Meetings. Interesting lessons are learned from these trials on many fronts.
The COAG Trial was more like a well-controlled Phase 2 trial. Patients slated for warfarin initiation were randomized to genotyping plus a sophisticated algorithm based on known clinical modifiers of warfarin metabolism or to the sophisticated algorithm without genotyping. The goal was to tease out the contribution of genotyping.
A British/Swedish Trial compared genotyping with a simple clinical algorithm to â€śusual careâ€ť for which a loading and relatively fixed dose of warfarin was recommended. The goal was to improve upon usual care.
A European Trial compared genotyping with a clinical algorithm alone. This trial, however, used other types of Vitamin K antagonists rather than warfarin.
The primary outcome for all 3 trials was â€śtime in therapeutic range,â€ť an interesting, but complex construct. None of the trials had adequate statistical power to measure differences in major bleeding, death or recurrent thromboembolic events.
Bottom lines from my perspective:
1) Genotyping adds nothing significant to a good clinical algorithm.
2) A good algorithm is better than â€świlly-nillyâ€ť usual care.
3) Before wide-scale implementation of expensive approaches to clinical care, comparative effectiveness trials are a good idea.
4) Even a diagnostic decision algorithm can do harm (see result of COAG in African American patients)
5) If we go to broad genotyping and databasing as a part of routine clinical care, I would use the information in white patients if it was free.
Cholesterol, Statins and Evidence: Revenge of the Nerds
Here at the American Heart Association, there is a lot of discussion about statins and LDL cholesterol management. Several interesting themes:
1. For many of the most important questions, we just don’t have the answers. Rather amazing that after all the profits of industry and all of the public health prominence, we don’t know whether treating to an LDL target is best and we don’t know what to do with people over age 75. We need a better way of developing priorities so the right questions get answered by our research engine.
2. Many more people should be treated with statins. While this recommendation has raised concerns, especially by critics who typically recommend against medical therapies, it seems reasonable based on available evidence. Exactly where to demarcate the eligible patients remains “squishy,” but in general if reducing death, stroke, heart attack and heart failure is our goal, we need more statin use, not less.
3. The risk algorithm in the new guidelines may be suboptimal. The controversy here is fascinating, but it all emphasizes the need to really understand the operating characteristics of decision support algorithms. Maybe people will finally grasp the importance of accurate predictions and careful assessment of the operating characteristics of decision support algorithms. What was previously thought to be esoteric academic work is really important!
We’ll see how all of this settles out, but the current debate is probably healthy and will lead to a better system. As we prepare for the PCORInet system of research, it seems like we could answer a lot of these questions very quickly!
Should America emulate other successful healthcare systems or do it our own way?
Over the past several weeks, I’ve had substantial interactions with colleagues from Japan and Scandinavia. Life expectancy and functionality are clearly superior for the average person in both Japan and Scandinavia compared with the US. In the US if you’re highly educated and wealthy, you’re right up there with our peers.
In Japan the healthcare system is not modernized and depends on a patchwork of “mom and pop” primary care in the community and highly focused specialty practices that are hospital-based. While many records are electronic in the hospital based system, they are not linked in a way that provides fluid data. The prescription system is tied to physician payments in many cases and not a modernized e-prescribing system.
In Scandinavia everyone has a health record number, everyone is organized and the data systems are first rate. The system is grinding towards evidence generation and evidence based practice in the same system. Yet, the longevity is close to Japan and not better.
Of course, all this tells us that lifestyle and culture are much more important for overall health than the traditional health delivery system. In America, it is mostly an issue of disparities and differences are built into our culture. How do we deal with this? “Obamacare” is bringing coverage to many more people. We need a way to make the health care system work in tandem with cultural change in the community that leads to better health.