Intended for healthcare professionals

CCBYNC Open access

Rapid response to:

Research

Effectiveness of vertebroplasty using individual patient data from two randomised placebo controlled trials: meta-analysis

BMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d3952 (Published 12 July 2011) Cite this as: BMJ 2011;343:d3952

Rapid Response:

What information is in an EQ-5D comparison?

The comparisons of generic health status is in this article based on
EQ-5D, a standardised instrument for measuring health often used in
economical evaluations. This measure is, however, statistically
problematic, and failing to find an advantage of vertebroplasty over
placebo by testing mean values, is not necessarily an indication of
vertebroplasty not having an effect.

This may seem paradoxical, but it depends on EQ-5D being constructed
to have a bi- or tri-modal distribution (depending on whether subjects in
full health are included or not). The individual pairwise differences from
randomisation to follow up will therefore have a non-Gaussian multi-modal
distribution with even further modes.

The mean value from such a distribution is, in contrast to the mean
value from a Gaussian distribution, not representative for the entire
population of studied patients. A comparison of mean values says therefore
very little about actual treatment effects. Major treatment effects could
be present among some patients without this affecting the mean value. A
more sophisticated analysis of the subpopulations represented by the
different modes would have to be performed.

Competing interests: No competing interests

02 September 2011
Jonas Ranstam
Biostatistician
Philippe Wagner
Dept of Orthopedics, Skane University Hospital, Lund, Sweden