A Brief History…

Today, Glen Reynolds of Instapundit linked to a rather heated denunciation of Evidence-Based Medicine. The term, vigorously debated in medicine, may not be familiar to those not part of the industry. The definition involves two major issues. One is the medical literature and what is called a randomized trial. This involves a new, or occasionally a standard, form of treatment. The question to be answered is whether the tested form of treatment is better than the control form, which may be no treatment at all. The treating physicians, or institutions, and the patients being treated, ideally, should not know whether the treatment being given is the test version or the control. Obviously, this is easier to do with pills. For ethical reasons, it is difficult to do with surgery although a very few such trials have occurred. Arthroscopy of joints involves very small incisions and a few sham operation trials have been conducted to test the effect of arthroscopic surgery. Those studies are very controversial. Another alternative is the use of randomized trials comparing surgery against non-surgical treatment. The problem here is that it is obvious to everybody concerned, who got the surgery and who didn’t. One such study compared surgery on the medial meniscus of the knee (The rubbery cushion in the knee that is subject to tears) to simple exercise therapy. The results? According to the outcome scores arthroscopic partial medial meniscectomy combined with exercise did not lead to greater improvement than exercise alone.

This brings up the concept of “Outcomes Research.” The difference between outcomes research and the standard clinical research we have done for 100 years is in what is measured. Most journal articles on clinical research list the results as mortality (death) or morbidity (poor health) or cure, if cancer is the topic. In studies of non-fatal conditions, such as low back pain or knee arthritis, results can be misleading. Dartmouth Medical School has studied benign prostatic hypertrophy (known as BPH) for 30 years. The results of surgery, removal of the prostate, cannot be assessed by measuring mortality rate as the mortality of that operation is very low. The cure rate is also misleading because removing the prostate “cures” the condition but often leaves a series of complications, major and minor, in its wake. Outcomes research uses survey methods to determine how the treatment affects the patient’s quality of life. One such is called the “SF 36.” If a man is cured of BPH but dribbles urine all day, the improvement from the pre-treatment condition may be minor or even negative. Similar studies have been done with low back pain. The reader will note that many of these studies come from countries with government funded health care programs. Some of that is because it is easier to follow patients for a long time in such systems because of a uniform record and a single source of data. In a distributed system like the US, there may be difficulty tracking subsequent care, a major consideration for this type of study. We want to know how this patent is doing five years later. In the US people change employer and/or insurance carrier every three years on average. Regrettably, insurance companies usually do not share data.

Take the example of spine fusion for back pain, a common procedure. For many years, the literature on such procedures looked like this. A small number of patients, followed only in retrospect and with no control for possible bias. Surgeons like to operate and they rarely report bad results. A series of 67 patients reported by a malpractice lawyer might look very different.

The Patient Outcomes Research Team approach (or PORT) looks like this. Other studies have shown that spine fusion done for back pain alone has a 95% failure rate when residual pain is the metric. An awful lot of spinal fusion surgery is done in this country every year, billions of dollars worth. Is it all useless ? There are “Guidelines” for what will produce results worth the cost and risk. How are these derived ?

Guidelines are of several types. Some are established by the government. How do they decide what will be included ? The best guidelines are based on randomized trials. Those are few. Many are based on the PORT method where common conditions are studied over years using every system of data collection availabe. Prospective trials, which are randomized, are the best in surgical cases, where it is obvious who got what treatment, but they are difficult to do. Patients may refuse to be included because they, or their doctor, are convinced one type of treatment is best. This is where fear of socialized medicine is most concentrated. They fear that the guidelines are based on cost, not efficacy. I might add that insurance companies often resist outcomes research because they fear that optimal treatment may be more expensive than what is commonly done now. There is always a lot of fear when changes come.

The least useful guidelines, and the most common, are called “Consensus Guidelines.” These are derived from committee meetings in which a group of experts concludes what the best treatment should be. Most of the time, the experts are using a lifetime of experience and a thorough knowledge of the medical literature to come to their conclusions. Bias, however, is not excluded by this process and the guidelines are often muddied with second guessing and reluctance to challenge colleagues who may be out of date. If all doctors kept up to date on medical progress, such guidelines would probably be unnecessary. As it is, they are better than nothing. Evidence Based Medicine, then, consists of trying to use “best practices” when they can be identified. In many instances, the art of medicine still remains the better indicator of what should be done. Doctors need to listen to their patients and they may find that explanation will cure something that surgery would only aggravate. I see this every week when I make hospital rounds with medical students. The students have much less information than the older resident physicans but they have time and interest and patients may respond to that when the science has failed. It is important to know the difference.

God grant me the serenity
to accept the things I cannot change;
courage to change the things I can;
and wisdom to know the difference.

This entry was posted on Tuesday, December 18th, 2007 at 9:28 am and is filed under health reform, medical history, medicine, science. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.