Vertebroplasty and the 'decline effect'

The favourable results of treatments can fade over time (the Decline Effect, see later post) for many reasons. Often the initial enthusiasm (bias) of the proponents is not replicated in later studies (see previous blog). But sometimes it can be shown that as the scientific validity of the studies improve, the effect decreases. This is because studies with better scientific methods will (by definition) have less bias (causes of error) and therefore the results will provide a better estimation of the truth. Vertebroplasty (injecting osteoporotic vertebral fractures with cement) is a recent (and controversial) case in point.

The decline in effectiveness of this procedure was documented in an editorial in The Spine Journal (The vertebroplasty affair: the mysterious case of the disappearing effect size). Initial reports of vertebroplasty, without controls (comparison patients) were very positive. Some comparative studies were also positive (but less so). There was still some apparent benefit when some non-placebo randomised clinical trials (RCTs) were done. But by the time the blinded placebo-controlled RCTs were done (here and here), the effect was gone.

The RCTs were scrutinised in a published report* by the North American Spine Society, and so they should be, as there is no such thing as a perfect study. And that is what science is all about: putting your findings out there for scrutiny. The authors addressed the criticisms, and even went to the extent of publishing a systematic review in the British Medical Journal, combining their data in order to answer the criticism about not having enough acute (fresh) fractures.

It is disappointing that many are dismissive of the RCTs, but are less critical of the unblinded, uncontrolled or non-placebo controlled studies. Unfortunately some people trust their eyes (seeing their patients getting better after giving the treatment) and make the logical error of post hoc ergo propter hoc (after this, therefore because of this). This human trait is a common logical error when assessing treatments for conditions that are self-limiting (get better anyway), like vertebral fractures.

My answer to the promoters of vertebroplasty is to show me the randomised trials that demonstrate a benefit. And if they finally produce one, we should scrutinise that paper, and not accept it on face value. To quote Rachelle Buchbinder, one of the lead authors of the two RCTs:

“Vertebroplasty appears to confer no benefit over placebo but poses some risk … The onus is on the proponents of the procedure to perform further high-quality randomized placebo-controlled trials”

And I didn’t even get time to report the complications associated with this procedure. If you are interested, click here for a glimpse.

*I realise that many of you may not have full access to these articles. If you don’t, I am sorry but I am not allowed to reproduce them for you.