Recently, I visited an internist, my primary care physician (PCP). I have been visiting this PCP quarterly for a decade. That’s a bit excessive, but I’m (1) very interested in this stuff, (2) my PCP is a very engaging guy who unknowingly provides some insights for these columns and (3) I’m a senior who takes the time and has the money. I’m monitoring my decay because I can, and no informed consent approved by an institutional review board is necessary.
I do, on the other hand, initial a bunch of statements with each visit, signing off to absolve everyone of liability. I’m not sure, as with agreeing to internet policies by multiple vendors, what I am agreeing with, but I do it. I have asked other patients if they read the sign-offs. Thus far, no one has told me they’ve studied the material. My investment advisers ask for similar signings. I’ll admit, I’ve not read all the great books either, many of which I find art of the absurd. I digress. A good thing about my medical visits: I’m not asked to check a box acknowledging that “I am not a robot.” Perhaps next year.
There is much debate about the value of chemistry. Medical examinations are traditionally said to be physical exams, and that’s what they once involved. The parameters measured were in the domain of physics: temperature, pressure, heart rate, height, weight. These could be seen or determined with centuries-old devices. Problems were described the same way. I’m hot, broke my arm, cut my finger, head hurts, been shot, have a rash and the like.
These are real concerns, but today most of the diagnostic data are chemical. So, I go to the doctor for “a chemical” and have nearly 50 parameters I can access in my electronic health records and plot within the same day following a fasting blood draw. For the last five years I’ve had access to trend information, and I really like having that available, both for the physics and the chemistry. I’ve been able to use these numbers for behavioral adjustments and titrate them to a better place with exercise and fewer pieces of pie.
What’s not to like? I’m told that few patients ever look at these records. Fewer still spend any time considering what they mean. I’d prefer to receive an indication of the uncertainty of the measurements, but no luck there. I do receive a reference range expected for an undefined population with no detail for gender, age or co-morbidities. I likewise have no idea what the measurements cost. Most of them appear to be of good quality these days, with a pleasing internal consistency over years. They are reportedly the basis of 60 to 70 percent of diagnostic conclusions. That sounds important.
For a reality check, there have been multiple reports of the many diagnostic errors in American healthcare delivery. Combined with other medical errors, the consequences are reported to be rather dire, with the loss of several hundred thousand lives per year, albeit not so many out of three million annual deaths in the United States—but it could be 5 to 10 percent.
Nevertheless, this is a far larger number than we lose from more politically visible sources such as guns, automobiles or opiate overdoses. Medicine is moving decade by decade to be more information-rich and constrained by the resulting increases in cost from advances in both the physics (imaging) and the chemistry. The tools in the black bag of 1960 are not enough for the today’s best result. There are now thousands of potential tests, many of which are not fully validated in either the laboratory or clinical sense. Which to order? Medical schools spend rather little time teaching in-vitro diagnostics (IVD) vs. other things. Change is slow. Throw genomics into the mix and we have more than a single medical decider can digest, given so little time per patient. More is feasible, but not yet affordable.
For example, we are not yet at the stage of dosing based on data for an individual vs. a population. That will take clinical pharmacology participation in patient care. The informatics load already begs for a more consultative or team approach to diagnostic decisions. How will these professionals be rewarded? Can artificial intelligence play a role? Dangers lurk in machine learning. For example, IBM Watson has recently been accused of giving “unsafe and incorrect” advice for treating cancer patients. The AI hyperbole is premature. Dr. Watson perhaps should have asked Sherlock Holmes for his thoughts. The data supporting AI must be good. Data scientists, oddly, don’t know much about where the data comes from. Much of it is very sloppy and biased, uncorrectable by statistics.
Today, CMS (Center for Medicare and Medicaid Services) concludes that laboratory medicine costs too much. They want to take hundreds of millions in cash out of it under the Protecting Access to Medicare Act (PAMA). Medicine long ago became a menu of reimbursed actions, not achieved results. Thus, now we have the new Clinical Laboratory Fee Schedule. It makes sense to have a schedule vs. chaos, but it also makes sense to look for those unintended consequences. Among them will be a further erosion in IVD innovation both in academic clinical labs and the commercial test developers. There are so many places where IVD measurements are made that beg for a more flexible costing approach. While hundreds of millions of dollars are few vs. the approaching four trillion in total healthcare costs, it is a lot when it comes to staffing labs with the talent to help avoid all those medical errors noted above.
With very costly therapies making progress for smaller patient populations, the adverse consequences of invalid measurements will only grow. The wrong medicine, in the wrong patient at the wrong dose at the wrong time. can be very hard to fix. Reducing diagnostic reimbursement could perhaps increase overall healthcare cost. With my recent cataract surgery, I was asked four times if we were to work on the left eye or the right eye that day. Check lists can help, but pharmacology questions are harder. Some say we test too much, because most tests suggest nothing. Others say we test too little, because we missed something.
Very little of the possible clinical chemistry universe is affordably measurable today. When drug concentrations are determined for narrow therapeutic index drugs, typically it is at a single point in time. For most drugs, the circumstances vary a great deal with polypharmacy, organ dysfunction and diet. Nothing is determined to confirm drug concentration rate-of-change beyond the original clinical trials. This pretty much remains unaffordable and complicated by blood sample logistics and reimbursement. It can be much better, but resistance to change in medicine exceeds even that of fundamentalist religion. Sometimes we go out on the fringes a bit. I encourage that and like it out there. Let’s look at a fun example.
A few weeks ago, I was alerted by the Wall Street Journal of a U.S. Army study on the relationship between sleep duration, caffeine consumption schedule and alertness to achieving good results for “three different psychomotor vigilance task alertness statistics.” Bayesian statistics were put to work to develop a software tool to forecast the “alertness of an ‘average’ individual as a function of sleep/wake schedule and caffeine consumption.” This sounds like pharmacodynamics to me. But I’m not the average coffee drinker and would be a customer for a biosensor strip to monitor caffeine in finger prick blood collected in a nanotainer. I’d do this several mornings a week. Some things are perhaps best not done, although a grad student can be found to do it in trade for a Ph.D. To learn more about the app, visit https://2b-alert-web.bhsai.org and references therein.
Speaking of listening to molecules, the 2000 Nobel Prize winner Arvid Carlsson passed away this summer. Arvid, a neuropharmacologist, listened to dopamine in the 1950s, a substance correctly thought to be an intermediate on the way to another molecule norepinephrine, the subject of the 1970 Medicine and Physiology Prize to fellow Swede Ulf von Euler. I was fortunate to meet both gentlemen. They truly set in motion much of the effort in my academic laboratory and a couple of instrument firms I helped begin. Dopamine turned out to be far more important than the 1950s thinking suggested. Upon receiving the prize in late 2000, Dr. Carlsson noted that his team had the “advantage of being ignorant and not so much burdened by dogma.”
May your dopamine be not too much, not too little, but just right. Listen to your dopamine talk! It says a lot about politics, shopping, mood, drug abuse and longevity. Dopamine is up, stimulating economic growth and stock market returns.
[Notes: A helpful review of the status quo with medical diagnostic errors can be found here: Michael Laposata, Journal of Applied Laboratory Medicine, July 2018, pp. 128-134. More on the caffeine work can be found at https://onlinelibrary.wiley.com/doi/epdf/10.1111/jsr.12725. Obituaries for the late great Arvid Carlsson are readily available through your favored search engine.]