Why Doctors Keep Doing Treatments That Don't Work

Joseph M. Smith, PhD, MD; Gary Wolf

April 18, 2012

Editor's Note:

Joseph Smith, MD, PhD, Chief Medical and Science Officer for West Wireless Health Institute in La Jolla, California, was interviewed by Gary Wolf, a contributing editor at Wired magazine, at a panel discussion in San Diego called "Quantified Self and the Future of Personal Health." The panel also included Eric J. Topol, MD, Director of the Scripps Translational Science Institute and Chief Academic Officer for Scripps Health, and Larry Smarr, founding Director of the California Institute for Telecommunications and Information Technology. The following is a transcript of the discussion with Dr. Smith.

Gary Wolf: Joe, of the panel here, you are the most directly engaged in how healthcare is managed today, because when you talk about lowering costs, you talk about lowering the cost of the healthcare system that we have today. For instance, to lower costs you have to address people who are patients or potential patients and see people as consumers or participants in the healthcare system.

What do you see as the big wins in lowering costs, in terms of new knowledge? We can lower costs by making the paper move faster in the system and such, but what new knowledge is available to us through these systems? What diseases or treatments or systems in the body do you think will produce the biggest payoff in the next few years?

Joseph M. Smith, PhD, MD: It's a tough question. People smarter than myself have previously said that prediction is difficult, particularly when it involves the future. I wouldn't want to definitively predict specific events, but there are some obvious opportunities and what seems like an unavoidable trajectory toward them.

You have talked rather generously about evidence-based medicine. Most of medicine isn't evidence-based. The overwhelming majority is more "eminence-based," to steal from my colleague to the right [Eric Topol]. We do things because we have always done them. That is going to be less tenable, and you will be put under more and more scrutiny about "Why is that? Why is this happening to me?" or "Why, doctor, are you doing that as opposed to this?" You peel back the level that says, "Well, actually, there isn't any evidence to support that. That was merely my historical preference as opposed to my data-driven wisdom and decision-making." That will put pressure on what we do and will ask us to answer some of the questions about dominant practices that are founded largely by history.

Gary Wolf: I am going to put you on the spot: What dominant practices? Name a couple.

Dr. Smith: If you go to your doctor at the moment with lower back pain, there is a pretty good likelihood that you will get some imaging for that, and there are pretty good data that say that no subsequent decisions hinge on the observations made in that imaging, or that those decisions will happen at some incredibly low likelihood. But it goes much deeper than the instances of known waste. We do a lot of things, as Eric [Topol] pointed out, that are population-based when we fully know that 30%-40% of the people to whom we provide such therapies derive no benefit but experience all the costs and all the adverse consequences. All it takes is understanding the genetic determinants, the historical determinants, or the epigenetic determinants that say, "In you, this therapy won't work, so skip it." The opportunity to take potentially life-saving therapies and give them only to the 30%-50% of a cohort that deserves them, by virtue of having some positive impact, saves half of the expense.

Estimates of known waste are $700-$800 billion a year. The things we don't yet know are larger because we are doing things that are in the guidelines. But when you peel back a layer, those guidelines are derived largely from apocryphal suggestions in remote history, right? So, there is a tremendous opportunity, as we put pressure on the system, to justify why we do what we do.

Importantly, we have a system with a bandwidth limitation living at the doctor. We can't keep up with the onslaught of information. We can't keep up with the patients we have to see. We are not really good at even figuring out which of the patients we are responsible for need to be seen at a particular time. We realize that "maybe I shouldn't be making those decisions because I can't comprehend all the diseases that my patients have. They are presenting information I don't yet know how to interpret." Maybe we need to offload that to smart systems.

Every other technologically sophisticated endeavor in which humans have participated has had the opportunity to use massive information technology and smart algorithms. You can take your car in and the mechanic no longer says, "Let me see what's wrong with your car." No -- they plug in a chip and say, "Look at that. It turns out that one cylinder is off by a little bit. Let me fix that for you." Why can't we do that with ourselves? We have to do it to ourselves because we are bandwidth-limited at the moment, and so we have to move to that type of system. It offers a great opportunity for saving.