I am Australia's most experienced vertebroplasty expert, having treated more than 2,000 patients. I have also spoken to every IR operator (all 4) in Buchbinder et al.

Buchbinder et al is a poor quality study for several reasons. Firstly, patient selection criteria are inappropriate. The duration of back pain in the study is less than 12 months - it should be less than 6 weeks (as the soon to be published VertosII study).
Vertebroplasty is usually applied (in Australia) to the exquisitely painful subgroup of acute osteoporotic fractures when the pain cannot be adequately controlled by oral medication. These elderly folks have fractures less than 6 weeks old (most often less than 4 weeks old) and find themselves incapacitated and often hospitalized unable to cope. They cannot sleep, find the pain unbearable and can talk of suicide. They may have bowel or cerebral complications from the narcotic medications. Traditionally they had up to 6 weeks in hospital. The advent of vertebroplasty has dramatically changed their outlook with early discharge, often the day after the procedure. The technique can only be truly effective on recent, un-united fractures. In this setting it provides internal fixation of the fracture fragments and ameliorates pain. Acute fracture fixation is well recognized to reduce acute fracture pain elsewhere in the skeleton.

This study has inappropriate patient selection with back pain for up to one year. These fractures unite in 6 weeks - to inject them at 6 months defies logic and, of course, can only have placebo effect.
There are very few, if any, inpatients in the study group. These are the patient group (ave age 80) with acute severe fracture pain unable to bear the pain at home. In Sydney these patients have a vertebroplasty and go home  within a week, often the following day. The internal fixation has dramatically controlled fracture pain. This group is not studied in Buchbinder et al.

Buchbinder et al could not recruit (aimed to recruit 200) and required 4.4 years in Melbourne, a city of 4 million people to recruit 78 patients. Two of the four hospitals (The Alfred and Monash) dropped out having done 5 vertebroplasty trial patients between them. They were concerned by the ethics of the study. Only 23 patients in Buchbinder et al had pain less than 6 weeks - too few for statistical analysis. 68% of patients were performed in one centre (Cabrini Hospital) by one radiologist who had no input into patient selection. These patients were selected by the principle author who therefore had control over the outcome of the study.
64% of patients refused enrolment (selection bias) and were not followed. This alone makes the study meaningless and makes a mockery of the hallowed concept of sham surgical studies.

What amazes me, a physician who has performed 2,000 vertebroplasties almost exclusively on subacute osteoporotic fractures is the ease with which this second rate study has been accepted. It has inappropriate patient selection, terrible recruitment and selection bias with the majority not followed.

This is not surprising with Buchbinder et al where the first four authors are physicians who have never performed a single vertebroplasty. The craft group (IR) who practice the procedure were effectively locked out of patient selection.

If the authors apply the same selection criteria to appendicectomy (right iliac fossa pain for up to one year) then they can collectively prove that appendicectomy also does not work better than a sham.