LAUSR

I wish to express my deep concerns over the uncritical claims and publication of this paper in Osteoporosis International. In my view, it reflects badly on the authors, the reviewers, and the editor who handled this submission and ultimately on the journal, as it makes claims for priority that are incorrect and does not adequately reference previous literature. This is especially galling as the earliest statistical shape modelling paper to assess fracture risk was published in this very journal! It claims to be the first to use 2D statistical shape modelling and statistical appearance modelling from DXA images to assess risk of osteoporotic hip fracture. In fact, these approaches have been used previously, and several studies have now been published from ourselves and from other groups. Statistical shape modelling was first known as active shape modelling and was based on a segmentation technique developed by Tim Cootes and colleagues in the University of Manchester [1, 2]. They later added appearance modelling [3]. We pioneered the application of these methods, first to radiographs and, later, to DXA images. In 2004, we published a study using radiographs and BMD to assess risk of fracture [4]. We proceeded to make measures of bone shape, structure, and density in an effort to improve fracture risk assessment [5]. As DXA images were improving in quality, we did repeatability studies comparing radiographs and DXA images [6] and started to apply the methods to osteoarthritis [7–11] as well as to other joints [12, 13]. Most importantly, we combined shape and appearance modelling to explore relationships with hip fracture [14, 15], the subject of this paper, and started publishing these studies 10 years ago! None of these papers is referenced, and it reads as though 2D shape modelling has been newly discovered; clearly, it has not. My question is why did neither the authors nor the reviewers pick up on this? It strongly suggests that a literature review was not performed properly and that reviewers were chosen who are peculiarly ignorant of this approach to morphological modelling. Unfortunately, this in now in print, but I feel it reflects badly on the authors. Perhaps an addendum needs to be added to the on-line version retracting the priority claims and adding a more comprehensive literature review.