The Problem of Medicine that Doesn't Work

From the Respectful Insolence blog, written by a pseudonymous doctor, an essay on how hard it is to stop a medical procedure that demonstrably doesn't work.

The procedure in question is vertebroplasty, which is the treatment of a cracked vertebra by the surgical injection of acrylic cement into the spine. Cracked vertebrae are common in older people with osteoporosis, and they can be extremely painful and take years to fully heal. So when, in the 1990s, this procedure seemed to offer real help to many patients, it took off. There were always hints that it might not be all it seemed, for example, the amount of cement used and the way it was applied seemed to have no effect on how well it worked. Plus any doctor who works with back pain knows how fundamentally mysterious it is, coming and going without much explanation.

Now the New England Journal of Medicine has published two studies rigorously comparing the procedure to a placebo and finding no difference. From the abstract of one of the studies:

Vertebroplasty did not result in a significant advantage in any measured outcome at any time point. There were significant reductions in overall pain in both study groups at each follow-up assessment.

As the RI blogger points out, the placebo effect from bogus surgery is often the most powerful kind of placebo. An added argument here is that the author of one of these studies was one of the pioneers of the procedure, who began the trials with the conviction that he would be vindicated:

Dr. Kallmes, who helped develop vertebroplasty and has been performing it for 15 years, said his team was "shocked at the results." Shock is understandable. As had been pointed out in both papers, vertebroplasty had become the standard of care for the treatment of vertebral compression fractures. Medicare and insurance companies had, on the basis of unblinded studies, begun to reimburse for the procedure.

All credit to Dr. Kallmes, who has performed that very difficult feat of admitting in public that he was wrong, not to mention cutting himself off from a big source of revenue. (Memo to cynics: yes, people really are capable of taking scientific positions that are against their own personal and professional interests.) Science is supposed to work that way, and, what do you know, it has.

Now comes the hard part, which is getting the medical system to respond to the new information. Dr. Kallmes has shown the way, but will others follow? One of the study authors suggests that maybe the procedure should be continued even if it doesn't work, because the placebo effect is so strong, which causes RI to remark, "A $3000 placebo?" Here is his conclusion:

Here's my prediction. These studies will not be enough to change practice, at least not in the short term. I wish I could say otherwise, but I can't. A procedure as well embedded into the standard of care and as profitable as vertebroplasty is won't disappear overnight or without a fight. Less mercenary, vertebral compression fractures are a serious problem for which other treatment options (bed rest, pain killers, and back braces while the fractures heal) are both slow and not palatable to patients or physicians, who want immediate results. However, unlike CAM [alternative medicine] practitioners, eventually physicians will yield to the weight of the negative evidence. There may be a few more studies, and it's even possible that there may be found a subgroup of patients who actually do benefit from vertebroplasty, but eventually the procedure will either be abandoned or scaled back to patients who might actually benefit from it. The process may be far messier than we would like. It may take far longer than we would like. It may even take a turnover to a new generation of physicians for the process to be complete. But, make no mistake, science will eventually win out.

If it doesn't, then we are no better than the homeopaths.