The abstract is at http://cad.sagepub.com/content/60/3/471.abstract (there's a paid link for the full study to the right.)

I'd like to see error bars, or at least a clear indication of the assumptions that went into the model.