Fractured Evidence: Spine Repair Debate Heats Up

Hundreds of thousands have benefited from vertebroplasty, advocates insist. They say the minimally-invasive procedure has freed them from hospital beds and dependence on intravenous narcotics.

Spine physicians swear that inserting a large-gauge needle into fractured vertebrae and injecting a cement compound to stabilize the bone hastens healing and helps relieve the often-crippling pain of compression fractures brought on by osteoporosis or metastatic disease.

But two recently-published, randomized controlled trials -- the gold standard of evidence-based medicine -- say otherwise. As far as disability and pain relief were concerned, they found that vertebroplasty for osteoporotic vertebral compression fractures was no better than a sham procedure.

Publication of the results triggered an outraged backlash from radiologists, for whom vertebroplasty is a bread-and-butter operation. The entire specialist community lambasted the studies -- statistically and methodologically.

But experts in evidence-based medicine argue that when profits are on the line, it's easy to be persuaded that studies are flawed.

So the debate rages, with radiologists citing case after case of success, arguing that patients with the worst fractures will have no treatment alternatives if the nation's third-party payers -- Medicare and the insurance companies -- refuse to pay for the procedure anymore.

The Studies

Both randomized, controlled studies were published last August in the New England Journal of Medicine. They involved a combined total of about 200 patients. Each found that vertebroplasty did not yield significantly better results in terms of disability or short-term pain relief than sham procedures for patients with this type of vertebral fracture.

In an e-mail to MedPage Today, the author of one study, Rachelle Buchbinder, PhD, of Monash University in Australia, suggested that the research showed the procedure was ineffective.

"Based upon the results of both trials I don't think that this treatment should be offered in routine care," she declared.

But the leader of the other trial, David Kallmes, MD, of the Mayo Clinic in Rochester, Minn., offered a different interpretation. He said the studies have been largely misunderstood by the trials' critics.

"I've been practicing for 15 years and had strong confidence that the procedure was effective, and that's exactly what we showed," he said in a telephone interview.

Mean baseline pain intensity on a 10-point scale was 6.9 in the vertebroplasty group and 7.2 in the sham-operated group in his trial, which had 131 patients. One month later, these scores had declined to 3.9 and 4.6, respectively.

Kallmes said this degree of pain reduction with vertebroplasty "exactly reproduces our prior experience." He vehemently denied that the findings were "discordant" with prior experience, as two statements from the Society of Interventional Radiology (SIR) put it.

"I don't know what they're talking about," Kallmes said. "It is concordant."

He said the real surprise was the effectiveness of the sham procedure, which should be the focus of follow-up investigations.

It involved inserting needles into the spinal column and injecting short-acting painkillers such as lidocaine, as was also done with vertebroplasty prior to inserting needles into the fractured vertebrae and injecting the cement.

The Flaws

Critics of the two studies cite a laundry list of complaints: too few patients, failure to meet enrollment, inclusion of patients with milder degrees of pain and disability than those usually treated.

In a November commentary from SIR, issued in conjunction with NEJM's publication of letters critical of the studies, J. Kevin McGraw, MD, of Riverside Radiology and Interventional Associates in Columbus, Ohio, highlighted the fact that Kallmes' study originally called for 250 patients.

Only 131 enrolled, and the vast majority -- 1,682 of 1,813 screened -- were excluded, "introducing significant selection bias into the study," he complained.

McGraw also pointed out that Kallmes' group didn't use screening MRI to ensure that a fracture was the cause of the patient's pain.

In addition, he did some additional statistical noodling and found that if one additional patient had reported a favorable response in the vertebroplasty group, the P-value would be 0.04, rather than a nonsignificant 0.06. Likewise, if one more patient had an unfavorable response in control group, the association's P-value would become significant.

Finally, in the crossover part of the trial, McGraw highlighted that 12% of patients in the vertebroplasty arm elected the sham procedure, while 43% of those who got the sham went for the real thing.

"The tremendous crossover rate speaks for some obvious benefit of vertebroplasty over sham and is worthy of a future adequately powered analysis to evaluate," McGraw wrote.

As for the Buchbinder trial, McGraw said it was convoluted by selection bias, since two-thirds of patients came from a single center and their procedures were performed by a single radiologist. The commentary also criticizes the volume of cement injected into vertebrae as lower than normal.

Similarly, the North American Spine Society issued a critique of patient selection criteria and outcome measures, and questioned whether the sham treatment was actually an active therapy. The statement suggested that dry needling might be a more appropriate control.

One of the letters published in NEJM also criticized the protocol requirement that patients undergo four weeks of medical therapy prior to enrollment in the trial. During that time, some fractures would have already healed, "resulting in a study on healed fractures," a group of Australian physicians wrote.

SIR president Brian F. Stainken, MD, took issue with the fact that patients with the most pain -- typically older, osteoporotic women -- weren't represented.

McGraw said these patients would be the least likely to agree to be in a randomized trial with a 50% chance of receiving the sham treatment.

"Most people in severe pain won't enroll," Stainken said. "At some level, research design has to take reality into consideration."

"The population with low-grade pain, that's the population these papers focused on," he continued. "It's not clear what the right solution is for this group. But I think the contribution of these [two NEJM studies] may be toward that."

On the Defense

Kallmes, however, responded that there was no difference in baseline pain scores between patients entering the study and those considered eligible but refusing to participate.

He also told MedPage Today that he and colleagues "enrolled patients that are very similar to those treated around the world."

"These studies were by far, by far, the best studies ever done," Kallmes declared.

"I have full confidence that if we had shown something different, that is, if we had found the procedure was more effective than placebo, people would have embraced it and said they were great studies," he continued.

"They would have said, 'Look at it, they were prospective, randomized, blinded, near 100% follow-up.' They would have been held up as the paragon for how to do studies in the future -- if we had reinforced people's preconceived notions."

Richard Deyo, MD, MPH, professor of evidence-based medicine at Oregon Health & Science University in Portland and deputy editor of Spine, told MedPage Today that the studies are the best evidence to date regarding the effectiveness of vertebroplasty in these patients.

"No study is perfect, and these are not perfect, but I do think they're the best we have," Deyo said.

Deyo said there was some validity to complaints that the studies enrolled too few patients, although both trials were adequately powered to detect a difference in pain reduction.

"If the benefit of this treatment were as enormous as many of the advocates argue," he said, "then it would take a much smaller study to demonstrate a huge benefit."

On the other hand, the studies were too small to evaluate the effects among various patient subgroups.

Kallmes also criticized the societies' press releases, charging they were written by some "who haven't read the studies carefully."

"The societies are in a great position," he said. "They have thought leaders that they can partner with to move the science forward. Moving the science forward is not done by sending out press releases."

Radiologists' Concerns

SIR's reaction may not be surprising, given that radiologists are so vested in vertebroplasty. Medicare will pay physicians from $522 to $554 for a procedure on one vertebra performed in a hospital or outpatient surgery clinic -- or more than $2,000 if it's conducted in the physician's office.

The number of annual procedures varies from office to office, and radiologists provide varying estimates. McGraw said he performs about 150 vertebroplasties annually, and Stainken said the procedure accounts for about 20% of radiologists' procedures, although that estimate "may be a little high."

One of the radiologists' main concerns is that insurance companies will use the trials to justify ending coverage of vertebroplasty for osteoporotic spinal fractures.

But insurers started playing that card long before the NEJM trials were published.

In a 2008 report, the Technology Evaluation Center of the Blue Cross and Blue Shield Association (BCBSA) concluded that neither vertebroplasty nor the related kyphoplasty -- which restores compression-fractured vertebrae to their normal size with a balloon before the cement injection -- had been demonstrated to be any better at improving net health outcomes than medical treatments.

That same year, Wellpoint, the insurance giant with 35 million members, announced plans to classify both procedures "investigational." That prompted a letter from SIR imploring the company to reconsider.

Another spine physician, Christopher Bono, MD, of Brigham & Women's Hospital in Boston, told MedPage Today that Aetna was also reconsidering coverage for vertebroplasty following the NEJM publications.

He said that he believed the firms would not end coverage entirely, but policies would be more restrictive.

"My gut sense is that payers are going to be much more selective in who they will pay for [regarding vertebroplasty]," Bono said. "They will ask for certain documentation and many more criteria than they did in the past."

McGraw worries that seniors will be "denied coverage when they are in the twilight of their lives and could possibly have long-standing suffering" that "could lead to their demise."

Stainken said there's "clear, unambiguous data that prolonged bed rest is the beginning of the cycle of decline for these patients. The key is to be able to do everything we can to ... keep these patients ambulatory and avoid that whole scenario."

Change is Hard

Deyo said money may partly explain the reluctance of physicians who perform vertebroplasties to scale back their use of the procedure.

"There are people who are making a living doing this, and you don't easily change what you do for a living," Deyo said.

The man credited with coining the term "evidence-based medicine" agreed.

"If you are making money from a procedure, it is very easy to persuade yourself that new evidence that the procedure is ineffective is in some way flawed or limited to allow you to continue to make money on the procedure," Gordon Guyatt, MD, of McMaster University in Hamilton, Ontario, told MedPage Today.

"When we have opinions and beliefs, we are very resistant to new evidence."

Kallmes, too, said there are psychological reasons for not accepting the results: "People just don't want to be convinced. They don't want to change their preconceived notions."

"They have this anecdotal experience," he added, "but I can tell them anecdotes of miraculous results with the placebo. So if they do this same study themselves, they may find the same thing I found."

Still, McGraw and Bono insisted the studies were too flawed to serve as the evidence base for clinical practice.

"Before we make these two studies the Holy Grail of evidentiary medicine with regard to vertebroplasty, we need to have larger trials," McGraw said.

Bono said evidence-based medicine, properly implemented, has three components: reviewing and applying the best data, "but also incorporating surgeon experience and patient preference."

"If you are just using the first, and I'm a strong advocate for using data and literature and references, I think you're doing a disservice," he said. "And then if you are misinterpreting the data or twisting the data or slanting the data, and eliminating the other two, you are really abusing the word evidence-based medicine."

Outcomes

So, will vertebroplasty go the way of other procedures that seemed to work clinically, but bombed in trials, such as knee arthroscopy for osteoarthritis?

SIR is awaiting the results of VERTOS II, a 200-patient Dutch trial comparing vertebroplasty with conservative therapy in patients with painful, osteoporotic vertebral compression fractures. And Stainken called for large trials that will address several of the methodology issues raised with the NEJM reports.

With regard to the potential placebo effect, Kallmes said he has nearly completed a 20-patient, open-label study of a procedure like that used as the sham in the randomized trial -- a spinal injection of short-acting painkillers.

Also in the works is a comparative trial of vertebroplasty and balloon kyphoplasty, he said.

Stainken said he has heard from colleagues who said they had slowed down or stopped doing the procedures after the studies were published, "which is not unreasonable -- to think it through and understand the situation."

However, most have resumed performing the procedure, and demand for it continues, he added.

Aman Patel, MD, an associate professor of radiology and neurosurgery at Mount Sinai School of Medicine in New York City, said he and his colleagues have not changed their practice since learning of the studies' findings.

However, he has started telling patients about the findings from the NEJM studies so they can be fully informed of the existing evidence about the risks and benefits of vertebroplasty.

"I firmly believe this procedure benefits some, if not many, patients," Patel said.

McGraw said he tells patients about the trials, but he also tells patients who he thinks would benefit from vertebroplasty that he doesn't believe the findings apply to them.

He said one patient had initially declined the procedure after talking it over. "That patient called me up a week later to proceed with vertebroplasty," he said.

Even if the findings don't change the way spine doctors use vertebroplasty in the short term, Deyo suspects the results will lead to fewer procedures in the long run.

Guyatt and Deyo predicted that eventually, a critical mass of evidence would be assembled -- assuming future randomized studies replicate these results -- that would convince even the most steadfast adherents.

Also, the NEJM studies could have an immediate effect in promoting more research by increasing the level of doubt about the procedure, Deyo said: "I don't think we have the final word here."

Comment