Stop (em)bracing burst fractures?

Jailbreaking spine fracture braces. The image is CC by Parée.
Jailbreaking spine fracture braces. The image is CC by Parée.

Continuing on the theme of #2MuchMed (see BMJ’s campaign) I figured it may be worth highlighting two recent studies on stable compression fractures of the spine. Both failed to show any benefit with standard bracing treatment. As some patients experience the brace as claustrophobic and uncomfortable (from my own patients’ experience), I’m thrilled to have this option available.

Bailey et al.

The study from Bailey et al. is the larger one with 96 randomized patients. The inclusion criteria were:

  • An AO-A3 burst fracture (a rather severe form)
  • 16 to 60 years old
  • Levels Th10-L3
  • Kyphotic deformity < 35 °
  • No neurological symptoms

There were a few additional exclusion criteria, but in summary I would refer to this cohort as one that I would definitely considered bracing.

There were 2 groups, one that received a rigid brace for 10 weeks and one that did not. The primary outcome was functional and quality of life at 3 months, the patients were though regularly followed for 2 years.  The baseline characteristics indicated comparable groups.

The results indicated no difference in any of the scores, they had illustrative graphs that showed the two groups perfectly aligned. Important to notice is that the power calculation was correctly performed as a non-inferiority study, i.e. has the actual ability to tell if there is little support for difference between the groups.

My thoughts on the study

  • They had some missing data and used last observation carried forward or baseline carried forward. They should have tested their model through multiple imputation by chained equations (MICE), currently considered best practice.
  • Scores and VAS-scales have a forced bounded characteristic and the t-test can therefore not be applied. They should have used either robust or bootstrapped confidence estimates.
  • The average may not be the best measure – we are often OK with over-treating if we avoid a few disasters, e.g. patients with intractable pain. Unfortunately this would require an even larger cohort, something probably impossible considering that the authors recruited their cohort during a  7 year period from three Canadian spine centers.
  • The ClinicalTrials registration reported for the study was submitted after completion.

Kim et al.

Complementing Bailey et al.’s study is Kim et al.’s study where they looked at patients aged above 60. Here they also included a soft-brace group and used a smaller total sample of only 60 patients (20 per group). It is though nice to see that the results align well between the groups in the graphs.

My thoughts on the study

  • I doubt that the authors did a proper non-inferiority power calculation. They seem to discuss non-inferiority in the statistics but it seems unlikely given the numbers that they truly designed a the study in a non-inferiority fashion and even more annoyingly they fail to report the confidence interval in the results. They report a large number of p-values but there are no confidence intervals reported.
  • The statistical thoughts regarding Bailey et al.’s study also apply to this one. Note though that they didn’t do any attempts at adressing the missing data in this study.
  • Including thoracal vertebrae 7-9 are most likely only diluting any results. In my experience they are frequently not relevant findings in this elderly population.
  • The ClinicalTrials registration reported for the study was submitted after completion.

Final thoughts

In summary I find the two studies well aligned and despite some methodological thoughts I feel that we can question the use of bracing. My personal opinion is that we should avoid active treatment if we don’t know. Although there are some exceptions not treating is just not an option, e.g. a smashed proximal tibia fracture in a young patient is probably better treated. Unfortunately drawing this line is difficult and our loss aversion certainly complicates things.

I’m disappointed that both studies seem to have registered their ClinicalTrials.gov after completion – the point of registering a study is so that the study findings aren’t influencing the analyses. If I claim pain to be my main outcome after the study, I cannot change to SF-36 just because “that was more interesting”.  These two journals are top orthopaedic journals and should look into their review process.

References

Flattr this!

This entry was posted in Orthopaedic surgery, Research. Bookmark the permalink.

Leave a Reply