Vertebroplasty for Spinal Compression Fractures: What Sham-Controlled Trials Reveal

When vertebroplasty was tested against a convincing fake procedure, its pain advantage largely vanished. In the blinded VERTOS IV trial, injecting bone cement into a fractured vertebra reduced pain no more than a sham needle did over twelve months. The gap between glowing early reports and later rigorous trials traces almost entirely to one design feature: whether patients and assessors knew which treatment had been delivered.

What vertebroplasty is meant to do

A vertebral compression fracture happens when a spinal bone collapses, often because osteoporosis has weakened it. It can bring sudden, severe back pain and a measurable loss of height. Percutaneous vertebroplasty is a procedure in which a needle is guided into the fractured vertebra and liquid bone cement is injected to stabilize it. The intuition is simple and appealing: fill the crack, harden the bone, relieve the pain.

For much of the 2000s, that intuition seemed to be borne out. Case series and open-label studies reported that patients felt dramatically better soon after cement injection. An unblinded randomized trial comparing vertebroplasty with ordinary conservative care also favored the procedure, with patients in the cement arm reporting faster pain relief. On the strength of results like these, vertebroplasty spread widely.

The trouble is that back pain from a fresh fracture tends to improve on its own, and a person who has just undergone an active, high-tech procedure expects to feel better. Neither the natural course of the fracture nor the expectation effect has anything to do with the cement. To separate the drug from the theater, you need a comparison that looks and feels identical to the real thing but omits the active ingredient. That is what a sham arm provides.

The 2009 sham-controlled trials

In 2009 the New England Journal of Medicine published two trials that did exactly this. In the study led by Rachelle Buchbinder, 78 patients with painful osteoporotic fractures were randomized to vertebroplasty or a sham procedure in which the needle was advanced onto the vertebral periosteum but no cement was injected. The multicenter INVEST trial led by David Kallmes randomized 131 patients to vertebroplasty or a simulated procedure without cement. Both trials blinded patients to their assignment.

The results landed hard. Neither trial found a statistically significant advantage for real vertebroplasty over the sham at the primary follow-up points. Patients in both arms improved, and they improved by similar amounts. In INVEST the difference in pain trended toward the cement group without reaching significance. These were smaller studies, and critics argued about enrollment, patient selection, and the age of the fractures. The debate stayed open for nearly a decade.

VERTOS IV closes the gap

VERTOS IV, published in The BMJ in 2018, was designed to answer the remaining objections. It enrolled 180 patients across four Dutch community hospitals between 2011 and 2015, focusing on the population critics said would benefit most: people aged 50 and older with acute fractures, back pain lasting roughly six weeks or less, a pain score of at least 5 on a 10-point scale, and bone marrow edema on MRI confirming the fracture was recent and active.

The design was double-blind and sham-controlled. Every participant received local anesthetic at each pedicle so that the sensations of the procedure matched. The vertebroplasty group then received cement; the sham group underwent the same steps without it. Pain was tracked on a visual analogue scale at one day, one week, and one, three, six, and twelve months.

Both groups got substantially better. Mean pain fell by about 5.0 points in the vertebroplasty group and about 4.75 points in the sham group. The difference between them was 0.13 points, with a 95% confidence interval running from roughly -0.41 to 0.66, well inside the 1.5-point threshold the investigators had set for a clinically meaningful gap. The difference was not statistically significant at any of the six time points. The authors concluded that vertebroplasty did not produce significantly greater pain relief than a sham procedure over twelve months in this population.

Why blinding was the decisive variable

Line the trials up and the pattern is clean. The studies that compared vertebroplasty with no procedure, or that left patients aware of their assignment, tended to favor the cement. The studies that inserted a credible sham and kept patients blinded did not. The active ingredient did not change across trials. What changed was whether the comparison controlled for the natural improvement of the fracture and the powerful expectation that surrounds any needle-and-imaging intervention.

This is the core lesson for procedures in general. When an intervention treats a subjective symptom like pain, and when the act of undergoing treatment is itself dramatic, the placebo response can be large. An unblinded trial cannot tell you how much of the benefit is the treatment and how much is everything around it. A sham arm can. VERTOS IV did not overturn a well-established fact; it confirmed what the 2009 trials had suggested and answered the design criticisms that kept the question alive.

None of this means the fractures are imaginary or that patients should simply endure the pain. It means the specific claim that cement injection outperforms a convincing placebo did not survive rigorous testing. Guidelines and specialty societies continue to disagree about whether a carefully selected subgroup might still benefit, and research into patient selection continues. This article is educational and is not medical advice; decisions about treating a spinal fracture belong to a patient and their own clinician, who can weigh the individual situation.

What the evidence does establish is a way of reading procedure claims. A striking before-and-after story, or even a trial against no treatment, tells you far less than a study in which patients could not tell the real procedure from the fake one. When the fake one works nearly as well, the honest conclusion is that the mechanism you assumed may not be doing the work.