I broke my ankle three years ago playing tennis. Thankfully, I am back to the game, which I have been playing rather badly for more than half a century. The ensuing strict immobilization for a few months after my accident, I am a bit ashamed to say, gave me time to read and think about things, apart from doing nothing.
You Might Also Like
Explore This IssueJune 2011
The first thing I read after my accident was the Ottawa Ankle Rules.1 These rules are a set of clinical findings that tell you when you should get a radiograph for a suspected fracture in an ankle injury. They are almost 100% sensitive and about 40% specific for a fracture. In other words, when one follows these guidelines and takes a radiograph in an ankle injury, the radiographs almost always show the fracture. In addition, the Ottawa rules obviate about 40% of unnecessary radiographs. Many times, we high-nosed rheumatologists consider our orthopedic colleagues to be mere technicians. There is also the naughty joke: “A double-blind study is two orthopedic surgeons reading an EKG.” We should be ashamed. I am not aware of similar accurate and practical guidelines in our discipline.
Lessons from Blink
The first full book I read during my incarceration was Blink: The Power of Thinking Without Thinking by Malcolm Gladwell.2 The essay that gives the book its title is actually about how the human brain makes a decision. It is the compelling tale of how a kauros (a rare and particular type of Greek statue), after much deliberation and expert opinion, is purchased by the Getty Museum in California and how a certain super-expert sees it and, in a blink of the eye, declares that it is most probably a fake. The author stresses the blink is not due to intuition but rather a very quick processing of relevant information by an expert brain. I strongly suspect this is how we make diagnoses in medicine.
This book then delves into the amazing story of how a hospital administrator in Cook County Hospital in Chicago greatly improves patient care and the finances of his hospital by implementing specific guidelines with which an emergency-room physician should decide which patients with chest pain should be admitted to the coronary care unit. He accomplishes this by using an algorithm proposed by Dr. Lee Goldman about a decade ago, created with the help of a group of mathematicians and lots of patient data that Dr. Goldman put through many computer cycles.3 I learn that this algorithm, apparently not more than a witticism for many, was suddenly found to be very useful in the Cook County Emergency Room. If you have chest pain and one or more of the following—1) your chest pain is unstable angina; 2) you have hypotension; and 3) you have rales at lung bases that indicate heart failure—chances are rather high that you have a heart attack and should be put in the coronary care unit.
The ensuing strict immobilization for a few months after my accident, I am a bit ashamed to say, gave me time to read and think about things, apart from doing nothing.
I learned from Blink that 1) the human brain can make very important and precise decisions in an instant; 2) this is not intuition but classification based on lots of data very precisely processed in the most sophisticated computer thus far ever described—our brain; and 3) cardiologists also are to be envied along with the orthopedic surgeons.
And on Vasculitis…
Meanwhile, an e-mail came from a rheumatologist colleague, an expert on vasculitis. At the time I was (and still am) involved in a group of experts from either side of the Atlantic who set out to formulate more accurate classification criteria for systemic vasculitis. I admit that I was flattered. The e-mail said that the best set of vasculitis criteria thus far described was that for Behçet’s disease in 1990.4 I humbly wrote back that the starting plans for these criteria had been realized in my kitchen in Istanbul. However, it was only after some years had passed that I realized our criteria were only good after we excluded patients with other diseases. This was exactly what was written as a footnote in the table proposing the criteria. Might not the unwary say, “What good are any criteria if one excludes other diseases by some other means to start with?”
My colleague’s e-mail made the assertion that we had set out to do classification and not diagnostic criteria for vasculitis. The diagnostic criteria would come later. Even with a cast on, I had qualms about this. Why? Once, all the disease criteria in our discipline were diagnostic. These were the days of authoritative, eminence-based, ad hoc criteria. The prototype was the Jones criteria for rheumatic fever. Then we began to learn about sensitivity and specificity. The criteria began to be data based—the ACR criteria set for systemic vasculitis is an example. It soon became apparent, however, that these criteria were of limited use in an actual clinical setting. The response was ready: Such criteria were not for diagnosis of the individual patient but were mainly for research studies, enrolling well-defined groups of patients in drug trials or for other clinical and laboratory studies. Nobody questioned why we felt that we had to have an exact diagnosis each time we planned the care of an individual and we could be less precise when we did research. The sophisticated cliché became that criteria available were for classification only. Parallel to this assertion has been the cliché of hope that we would prepare a separate diagnostic criteria set. This is what our vasculitis expert group was and is trying to do.
To me, there is actually no difference in the mental or mathematic activity in the formulation of criteria for diagnosis, classification, or both. Diagnosis, to me, is nothing more than classification in the individual patient.5 The comparator groups, the odds ratios, and the pretest probabilities are there in either exercise, whether we do them by an electronic computer or by our brain, as reading Blink has taught me. Further, the assumption that we should always be more precise in making a diagnosis than when enrolling a patient into a clinical trial may not have as sound a basis as we generally assume.
What It Means to Diagnose in Rheumatology
Why do we need diagnostic or classification criteria in rheumatology? The immediate answer is that we do not have histologic, laboratory, or radiologic leads to accurately define many of our diseases. Equally important is the question of why we want to diagnose or classify in the first place. According to Webster’s Dictionary, the verb diagnose is originally Greek, and the best synonym is discern. It also has the connotation of “understanding the nature of.” The verb classify, on the other hand, is newer; its first use is in the English language was at the end of 18th century and simply means to put into classes. As such, it implies less precision and perhaps less of an anxiety “to know the nature of.” So, when we do research we like to classify, and when confronted with a patient we prefer to diagnose. In other words, with a patient, we are not content with a classification only, we also want to attach an “omniscience” to what we classify. We, as rheumatologists, painfully know we do not know the exact nature of many of the ills we recognize and manage—rather successfully in recent times, at that. Perhaps we should be more open about this, first to ourselves, then to our patients, and finally to the public.
To me, there is actually no difference in the mental or mathematic activity in the formulation of criteria for diagnosis, classification, or both. Diagnosis, to me, is nothing more than classification in the individual patient.
Let us now take this line of thought into a recent undertaking in our discipline, the new ACR/European League Against Rheumatism (EULAR) rheumatoid arthritis (RA) criteria set. Between you and me, we still do not know what causes, propagates, and, in a few patients, even heals RA. Most probably it is more than one disease. However, in recent years we have been quite successful in our management of what we diagnose or classify as RA.
We also have the well-grounded premise that the earlier we start treating our RA patients, the better the outcome. The nature of the beast is that early in the game, many of our patients with recent onset of inflammatory arthritis do not permit themselves into being classified as RA according to the “one before” RA criteria set. In a Machiavellian way and with omniscience, we diagnose them as having RA, prescribe them methotrexate early in the course, and they do better. We also have some evidence that a similar early prescription of biologics would make them even better.
Experts on both sides of the Atlantic set out to make up new classification criteria based on how early RA had been being recognized in the best hands. Cleverly, they chose to define early RA as the decision to start methotrexate in a patient when other causes for inflammatory arthritis could be excluded. This exclusion, mind you, reminds us of the little footnote I alluded to earlier, under the table in the Behçet criteria. Going back to the ACR/EULAR criteria, the related manuscripts did not answer the nagging question of how many patients in daily practice with psoriasis and early onset arthritis are diagnosed as psoriatic arthritis when they present, only to find a year later that they have RA, or how many young women with synovitis of the small joints of the hand, Raynaud’s phenomenon, and a positive FANA are initially diagnosed with systemic lupus erythematosus and again turn out to have erosive RA at the end of a year’s follow-up. Nevertheless, the ACR/EULAR exercise tells me that there is strong agreement between the better (OK, the best!) rheumatologists on either side of the Atlantic as to when and how they start methotrexate in a patient with early onset inflammatory arthritis. If I start to do a trial for a new biologic or draw blood for research purposes, I will, in all probability, use these criteria. Then what is the issue? The issue, the incongruity, is to call these criteria RA criteria. They are, in fact, “When and if to start methotrexate in a patient with early onset inflammatory arthritis?” criteria. Why not call a spade a spade?
And why not, when confronted with a patient with an early onset of arthritis, simply say, “I am not certain whether you have rheumatoid arthritis, the bad disease you tell me your unfortunate next-door neighbor has, with which you tell me she is crippled and jobless. At this stage I do not have a sound and firm diagnosis. However, based on scientific evidence, if I prescribe methotrexate and perhaps other medications for you now, if you indeed do have RA, the chances are very good that you will benefit from what is prescribed. Meanwhile I will follow you closely, especially within the next few months, not only for the effects and side effects of what I prescribe but also for whether my initial classification—the diagnosis—of your condition was correct.” I am afraid it is as much our overestimation of our science as our underestimation of our patients’ cognitive abilities that keep us from such direct and honest dialogue. Perhaps we do not need to have a broken ankle to realize this.
Editor’s Note: This article was originally published as part of a “Liber Amicorum” presented to Professor Ben Dijkmans of Amsterdam on the occasion of his retirement.
Dr. Yazici is professor and chair of the department of medicine, division of rheumatology at the University of Istanbul in Istanbul, Turkey.
- Stiell IG, Greenberg GH, McKnight RD, Nair RC, McDowell I, Worthington JR. A study to develop clinical decision rules for the use of radiography in acute ankle injuries. Ann Emerg Med. 1992;21:384-390.
- Gladwell M. Blink: The Power of Thinking Without Thinking (First ed.). New York: Little, Brown and Company; 2005.
- Goldman L, Cook EF, Johnson PA, Brand DA, Rouan GW, Lee TH. Prediction of the need for intensive care in patients who come to the emergency departments with acute chest pain. N Engl J Med. 1996;334:1498-1504.
- International Study Group for Behçet’s Disease. Criteria for diagnosis of Behçet’s disease. Lancet. 1990;335:1078-1080.
- Yazici H. Diagnostic versus classification criteria—a continuum. Bull NYU Hosp Jt Dis. 2009;67:206-208.