Dr. Jon Keevil, a cardiologist in Madison, Wisconsin, and, until very recently, an associate professor of medicine and radiology at the University of Wisconsin School of Medicine and Public Health, has a strong affinity for cardiac risk assessment and treatment decisions and, not incidentally, technology. He has developed a computer program that predicts individual risks of developing heart disease, and over the years what began as a labor of love in fellowship training, has grown into a library of clinically useful tools for shared decision-making by patients and physicians. Dr. David Rakel, Editor-in-Chief of PracticeUpdate Primary Care spoke with Dr. Keevil recently, discussing the decision-making tools and how computers may increase quality of care for patients.
Dr. Rakel: Dr. Keevil, welcome. Thank you for joining us at PracticeUpdate. I have to tell you that prescribing statins is a nice opportunity to use a medical decision tool. I use your HealthDecision tool (www.HealthDecision.org) to assess patients’ risk, and I am constantly surprised that most of the patients with whom I went through this process of shared decision-making actually didn’t need statins, number one, or, number two, didn’t want statins because you also include in your tool evidence of harm or side effects.
You are adding to your tool set, and so will you bring us up to date on where you are in the process? And then can you talk a little bit about the importance of bringing in the evidence of harm or side effects in the shared decision-making model?
Dr. Keevil: Sure. Well, recently, we built more tools, and other institutions started asking about them; so, I ultimately left my clinical practice this past July 1. I have retained a clinical adjunct associate professor appointment in the Department of Medicine within the University of Wisconsin, but I’m now the founder and CEO of HealthDecision, LLC, which is committed to building out more shared decision-making tools for electronic medical records.
I was working on these tools long before shared-decision making was the buzzword that it is now; I ran into that phrase about four years ago. As we started new versions in 2013, we began really thinking about what that model means and how to incorporate the best practices of shared decision-making. I think the model encourages being very clear about both benefits and harms of any medication or testing decision. There are numerous studies, as you know, showing that doctors tend to overestimate the benefits and underestimate the harms, particularly relative to procedures that they themselves perform. As a proceduralist myself, I am sure I was at fault for that as well. But, to have the best conversation, you must offer up both the good and the bad that can happen from any piece of it. An explanation I used with patients in practice about informed consent is that it’s unfortunate if something bad happens, but it’s even worse if it happens and you never knew it could happen.
Dr. Rakel: So, you’re doing a deep dive into how these tools with computer assistance can really help us improve the efficiency of care. Where do you think that line will be drawn between computers making decisions for us and the art of medicine of applying this tool to the context of someone’s life?
Dr. Keevil: Well, I love the question because it’s an issue that I've been thinking about a lot. The tools live in a computer, but they are very deeply rooted in the literature and the guidelines, which are efforts that experts created intentionally. Our tools do not try to make up new data, but we just work to find all the relevant data and present it. I think many decisions still need human intuition, and there are two intuitions in the room: the intuition of the physician who knows the science and has a sense of whether a particular choice is going to really match up with the patient or not; and then there’s the intuition of the patient, which is critical but has traditionally been left out of the conversation.
So, intuition is amazingly powerful, and I don’t think we should try to supplant that with decision tools. What I think the best tools do is get the busywork out of the way: they do the calculations, gather the data you need by zooming through the branching diagrams in the guidelines for you because they have enough data do that, and then the conversation starts. To me, that’s when intuition and the physician–patient relationship comes to bear. I am skeptical of whether computers will or even should ever take over that portion of it. I think there’s always a role for the person, the interactional communication, and intuition in the conversation.
Dr. Rakel: Yes, which is the joy of practice. So often in healthcare we think that the patient just needs what we know. But the beauty of these tools is that they can honor the needs of both the patient and the clinician, guided by the best science we have.
Dr. Keevil: I agree.
Dr. Rakel: We are only as good as the data that we put into these tools. For example, nutrition and psychosocial stress often aren’t included in these tools, but we know they are very big predictors of cardiovascular risk. Are you ever going to be able to include some of these other key ingredients to improve the accuracy?
Dr. Keevil: Yes, I’m with you on the imbalance. For example, we have pretty sophisticated knowledge about a statin’s impact on cholesterol numbers and then risk, and that’s strongly predictive. But for the lifestyle layer, we know it is also predictive, but we don’t have as good numeric data. I would love to integrate that less precise data into the tools as best we can, and one way to do it may be to use levels of data that are less sure, but may still be informative. This might lower the threshold of what we include or we don’t include; but I see that as an ongoing challenge for both the practitioner and the tools.
Dr. Rakel: At least that’s a nice opportunity for shared decision-making because in primary care often we know of some of those psychosocial stressors, and how well the patient eats; so, it’s a nice opportunity to combine the insight from relationship with the data.
Dr. Keevil: Absolutely.
Dr. Rakel: What are you most excited about in creating these tools? What gets you up in the morning?
Dr. Keevil: I think the piece that is the biggest challenge, but the biggest excitement, is where computers really do have the potential to create value and add to the quality of care. So many institutions have moved to electronic health records, and there are portions of the decision process that computers and tools can do well and thus amplify our ability to work through these decisions with patients. As I was finishing my practice, I was often in clinic as a subspecialist, and I’d see 8 or 12 patients, and then, just at UW, a hundred patients using the tools at the same time. So, personally, I can have a larger impact in the world of patient care with these kinds of tools and the sort of leverage that they bring. That’s where I get excited by really being able to make a bigger impact on care.
Dr. Rakel: You should be excited and proud because, with the high rates of burnout, particularly in primary care, tools like this, which help improve the efficiency and ease of practice and reduce the clutter, are going to be valuable in helping us return the joy to our work.
What are the most important ingredients of those tools and how will they be most useful for primary care docs in practice?
Dr. Keevil: Well, I think it’s the stuff that is hard to get at. You know, it’s the stuff that you can’t hold in your head well or that needs multiple steps. I’ve been on guideline committees and we might work for a year, and there are 20 people on the committee, and there’s all sorts of discussion about what goes where and what words are you going to use in this sentence in this particular combination. And you feel restricted because you have to write it down. So, there can be these big battles about how someone interprets those data and how someone else has interpreted them in a different way. Ultimately, you print it up, put it in a PDF, put it on the hospital website, and then it can get lost.
I think the other cause of excitement is that the guidelines themselves can get brought into these tools and can thus be brought to the workflow of the docs. To me, the most useful tools are the ones that bring the numbers and the calculations to life and put them in front of the clinicians quickly and easily, and then clinicians can get on with the business of educating and working with the patients.
Dr. Rakel: How often do you update these tools since the literature changes so quickly?
Dr. Keevil: Here’s the thing. Literature changes quickly, but the guidelines don’t move all that fast. So, for example, the guideline for cholesterol from 2001 had its next major update in 2013; so, it gave us a few years to get ready. We have people responsible for each tool and watching the literature, so we keep them up to date. We have a newer lung cancer tool, and we just released a mammogram tool this week; they are current, but they have to be living documents. By example, I think UpToDate is a wonderful model—it’s a textbook, but its living. There’s a person responsible for every chapter and staying current with it. It’s important to be that way, and I think a lot of tools that are done well are kept current that way too. The Mayo Clinic makes some great tools, and Option Grid is a different type single-page tool out of Dartmouth, and they’re both actively involved with those tools and keep them updated. It has to be an ongoing process.
Dr. Rakel: Thank you. Anything else you’d like to say on the topic?
Dr. Keevil: I can share another getting-me-up-in-the-morning experience. We did a survey, which I sent to you, too, and had 109 clinicians reply. And the most enjoyable responses were clinicians who found that the tools actually gave them more credibility in the eyes of the patient. More than just them coming up with a number, the tool amplified trust by having a number that calculated in front of them; so, that’s been really satisfying as well.
Dr. Rakel: It’s also really helpful when the electronic health record combines with the tool to pull all the data so you don’t have to insert it manually. That was so helpful. Do you see that growing?
Dr. Keevil: I do think that’s a really important piece. There was a Cochrane Review from 2011, with about 115 studies of decision aids showing that the aids raised patient knowledge, and they reduced a 1979 measure of decisional conflict; so, that’s been replicated many times. But very few of those tools got into regular use in part because of the challenges that happen in practice; if you don’t have tools right there, integrated in your workflow, it’s very hard for tools to gain acceptance.
Dr. Rakel: What are three or four web resources that our readers can go to for some of the most useful shared decision-making tools?
Dr. Rakel: Thank you for helping us help our patients, Dr. Keevil!