Article Text

AB0886 A classification tree for axial spondyloarthritis in individuals with inflammatory back pain from TH
  1. F. Melchor1,
  2. I. Pelaez-Ballestas2,
  3. E. Gutierrez-Peña3,
  4. A. Lopez4,
  5. R. Flores-Camacho2,
  6. E. Navarro-Zarza2,
  7. R. Burgos-Vargas2
  1. 1Clinical Epidemiology, Instituto Nacional de Salud Publica
  2. 2Rheumatology, Hospital General De Mexico
  3. 3Estatistics, Instituto Investigaciones en Matematicas, UNAM
  4. 4Radiology, Star Medica Lomas Verdes, Mexico, Mexico


Objectives To determine the diagnostic value of inflammatory back pain (IBP) in a classification tree model for axial spondyloarthritis (SpA) in individuals from the community.

Methods A probabilistic sample of 121 individuals suspected of having IBP (Berlin’s criteria) collected during an epidemiological survey of 4,049 individuals assessed for non-traumatic back pain (n=467; COPCORD methodology) and then IBP were included in this study. Their assessment included clinical history, HLA-B27 testing, sacroiliac joint radiographic and magnetic resonance (MRI) imaging studies. Analysis of the data followed the Classification and Regression Tree (CART) methodology. Axial SpA was the dependant variable and 21 demographic and SpA related variables were considered independent. The decision tree was validated in a group of 25 individuals.

Results Mean age of the group was 41.3 (±11.1) years; 60% were females. The best classification tree model is shown in Figure 1. IBP positive individuals with three or more SpA caharacteristics should be classified as axial SpA. Those with less than three SpA features, but HLA-B27 positive should be also classified as axial SpA. Finally, those with less than three SpA features and HLA-B27 negative, but sacroiliac joint edema on MR should be considered axial SpA as well. This tree erroneously classified 5 individuals (4%). Its sensitivity and specificity were 84.62% and 96.30% and its +LR was 73.33%. A second classification tree (based on Rudawaleit’s one) disclosed the following partioning decisions: Regardless SpA features, individuals with IBP and either sacroiliac joint edema on MR or HLA-B27 should be considered as axial SpA. Classification errors were 6 (5%) and their sensitivity, specificity, andn +LR were 92.3%, 95.37%, andy 19.9%, respectively.

Conclusions The results of our study suggest that in community studies, the classification of individuals as axial SpA should follow the route IBP, SpA features, and then HLA-B27 followed by MR if necessary

Funding: CONACYT-Salud 2007-C01-69439. Abbott Laboratories de México: MRI funding

Disclosure of Interest None Declared

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.