Saturday, 5 May 2012

Does ultrasound make bones heal? No.


The acronyms are getting bigger. LIPUS (Low Intensity Pulsed Ultrasound) therapy is commonly used to heal fractures faster, or to get them to heal when they have not. It is a machine that straps on to the limb and is worn for minutes or hours each day, for a few weeks or months. Just like the techniques in my recent posts, it costs several thousand dollars and people assume that if it costs that much and is high-tech, it must be working. Lets cut to the chase.

Again, I am concentrating on the scientific evidence. Like many other treatments, there is abundant scientific evidence that it does amazing things to cell in a dish, and even mice and sheep, but the list of treatments that work well in lab animals and cells but have no effect in humans is too long to mention. To help me find the real (human) evidence, one of the companies has helpfully listed the evidence for me under a tab named “Here’s the proof” on their patient-targeting website, enticingly named “healmybone.com”.

There are two claims: that it heals bones that have not healed after a long time (non-union) and that it makes fresh fractures heal faster. Scientific clinical studies (randomised trials in humans) are limited, but they are there.

For the treatment of fresh (acute) fractures, it makes the bone heal faster on X-ray. We now that new bone formation around fractures is stimulated by movement. The ultrasound shakes up the bone, and consequently more bone is made around the fracture, so it kind of makes sense. However, in the landmark papers from the 90’s on tibia (leg) and radius (wrist) fractures, all of the fractures healed anyway. These two papers are similar. So similar that they have cut and pasted whole sections from one to the other, and the authors in both studies received benefits from the sale of the device, and several authors were employees of the company that made the device. And several of the authors appeared on both studies. Even the authors that wrote the review on this topic had conflicts of interest.

I don’t mind if the employees wrote the study, particularly as the two landmark studies were good studies, using concealed placebo devices, which is the best way to test these things. As long as there is verification from and independent study (where the authors are not receiving benefits form the manufacturer), everything is fine. However, a similar study from non-conflicted authors in Sweden found no statistically significant difference between the treatment groups (the placebo groups actually healed a little quicker).

Interestingly, there are no studies comparing LIPUS to placebo for the treatment of non-union (which surprised me, as I actually thought there was good evidence out there for the use of LIPUS in non-unions).

In short, this expensive treatment is only supported by studies that were sponsored by the manufacturer and written by employees and doctors who benefit from the sale of the device. Even in these studies, all of the fractures healed anyway. And there is no good evidence for using it in fractures that have not healed. My advice? Strap on a placebo, independent research shows that it might work a little better.

10 comments:

  1. Dr Skeptic
    Thanks for the blog, entertaining and enlightening - keep up the good work - a few comments regarding LIPUS
    (1) Regarding the Swedish study you allude to, it is important to look at why it may have failed to demonstrate any benefit - it was a good study, conducted independently, randomized, placebo controlled - but the study group selected was a poor choice - young, hyper-fit professional football players? You would expect them to already heal faster than normal, how could you possibly demonstrate they would heal even faster?! In fact, in that study both the LIPUS group and the placebo group heal faster than you would normally expect - the message? If you are young and fit you are already going to heal rapidly and reliably - don't waste money and resources trying to get back on the field faster - even if your employer is keen to get you back in the game...
    (2) The study group is all important, as are the outcome measures and the study design - each of these factors can very critically influence what your study may or may not show - it would be wise in this instance to select a study group where the frature is already at great risk of delayed healing or non union - choose a group like diabetics, smokers, the elderly, or a fracture pattern that is notoriously slow to heal like a segmental fracture or sub-trochanteric fractures - Steven Cook's paper demonstrates more rapid union with LIPUS in distal radius fractures when he specifically looked at smokers - just makes a lot more sense, you wouldn't test anti-hypertensives in young fit people either, you just won't show any benefit!
    (3) It is very important for new products to be developed in concert with surgeons and clinicians, and some surgeons have consulting agreements with our industry partners in order to help maintain the integrity of the studies that are conducted - most consulting agreements are not designed to directly benefit the surgeons involved through the increased sales of a given product; to suggest surgeons or physicians always derive significant additional benefits from the actual sales of implants or devices is somewhat misleading - very good studies can of course be conducted by surgeons with consulting agreements with industry (although admittedly the manufacturer may choose to suppress a study with negative results in favour of one reflecting positively on their product) - I know you have addressed this elsewhere on your blog, but in my opinion having a relatioinship with industry does not completely invalidate the results of a well-designed study, and with this I am sure you agree!

    ReplyDelete
    Replies
    1. Doc Bruce, thanks for your comment.
      I agree that industry sponsorship does not invalidate the results, but the evidence points to it being a factor that might influence the results.
      And your points about the individual studies mentioned are valid. This is the kind of critical thinking I am trying to invoke in the public and our medical colleagues.
      You also mentioned smoking as a negative predictor for fracture healing. I think that is an area worthy of an upcoming blog.
      Stay tuned.

      Delete
  2. Dr. Skeptic,
    I would just like to ask if LIPUS is available now in the market and where can we buy this thing? And how much does this cost and where to purchased this ? And how effective is this?


    ReplyDelete
    Replies
    1. As explained, there are no good tests of LIPUS for fracture healing, so I would not be recommending, buying or using it myself.

      Delete
  3. "For the treatment of fresh (acute) fractures, it makes the bone heal faster on X-ray."
    "Even in these studies, all of the fractures healed anyway."

    The last quote feels funny to me. Cannot quite put my finger on it.

    ReplyDelete
    Replies
    1. Thanks. Promoters of treatments look for angles at which their product looks good. A purely clinical scientific question, on the other hand, looks at the most patient-relevant angle. This is why these studies often look at how the device (ultrasound, BMP, whatever) affects the appearance of the X-ray in the weeks following the fracture, rather than looking at the patient's long term function or pain, or whether or not the bone heals (much more important questions).
      Their argument will be that the changes seen on X-ray may reflect improvements in patient-relevant outcomes. Then why not measure those outcomes in the first place?

      Delete
  4. LIPUS for fresh fractures is a luxury - if you believe in the science. Emami didn't have great parameters but many of the other fresh fracture studies did.

    AAOS Definition of a nonunion: "When a broken bone fails to heal". Providing a placebo control in the nonunion application is unethical and . By definition, that bone wouldn't have healed on its own.

    Nonunions:
    http://www.ncbi.nlm.nih.gov/pubmed/22424956

    This study wasn't funded by the manufacturer. These physicians took it upon themselves to do the study in the interest of preventing unnecessary nonunion surgeries. 89% chance of success is pretty good - and, as you know, rivals that of surgery. There are no risks associated with treatment.

    Will you not believe the nonunion data without a placebo control, knowing the ethical considerations?

    ReplyDelete
    Replies
    1. Firstly, I believe the data (that 88% of fractures un-united at 6 months healed). But that data does not tell me what would have happened if the device had not been used, because there is no control group. As stated in the study, union was correlated with stable fixation. Most fractures with stable fixation and less than 10mm gap without union at 6 months will heal.
      I disagree with your statement that "by definition, that bone wouldn't have healed on its own". Non union is usually failure of progression at a minimum of 6 months. Some argue it should be longer. Either way, failure to heal at a certain time does not state that it will not heal with more time. You must be very careful using the term "by definition"; your definition states that the bone HAS NOT healed, not that it WILL NOT heal - two very different things.
      I treat many non-unions. Stable fixation without union at 6 months doesn't worry me that much.
      As far as the ethics go, why is a placebo control unethical? That is how we prove effectiveness. Treatment without proof of effectiveness is unethical. See my blog post on this: http://doctorskeptic.blogspot.com.au/2012/07/why-placebo-surgery-is-ethical-and.html

      Delete
    2. Stable fixation at 6 months without union - is there no concern for hardware failure?

      Here is a study comparing delayed unions, LIPUS vs Placebo. I would love to hear your thoughts on the data:

      http://www.biomedcentral.com/1471-2474/11/229

      Delete
    3. Thanks Sam, an interesting study.
      My observations are that the primary outcome should have been clinical, and the analysis simple. This might stem from the study design, analysis and writing being done by a trials consultant (Jon E Block) who was being funded by the manufacturer.
      That there was no clinical difference reported (like pain, function or need for further surgery) means that it is unlikely that there was such a difference. Those outcomes would have been much easier to determine, and with less error, than the imaging outcomes they chose to report.
      Regarding hardware failure, I am less concerned with hardware failure than the risks of unnecessary revision surgery.

      Delete