Article Text

Download PDFPDF
Reproducibility and sensitivity to change of four scoring methods for the radiological assessment of osteoarthritis of the hand
  1. Emmanuel Maheu1,
  2. Christian Cadet2,
  3. Sylvie Gueneugues3,
  4. Philippe Ravaud3,
  5. Maxime Dougados4
  1. 1Rheumatology Department, Hôpital Saint Antoine, Paris, France
  2. 24 Place Martin Nadaud, Paris, France
  3. 3Université Diderot, Paris VII, Unité d’épidémiologie clinique et de Biostatistiques, Hôpital Bichat, Paris, France
  4. 4Medicine Faculty, Paris-Descartes University; Rheumatology B Department, Cochin Hospital, 27 rue du Faubourg Saint Jacques, Paris, France
  1. Correspondence to:
    Dr E Maheu
    4, Bd Beaumarchais, 75011 Paris, France;emaheu{at}


Background: Osteoarthritis (OA) of the hand could be a relevant model to study the progression of OA in structure-modification trials. Various methods are proposed to assess hand OA and its progression radiologically.

Objective: To compare intra-reader and inter-reader precision and sensitivity to change of four radiological scoring methods proposed in hand OA.

Methods: 2 trained readers scored separately 105 pairs of radiographs (baseline; year 1), selected from patients enrolled in a randomised controlled trial, for inter-reader reliability and sensitivity to change. They scored twice 60 pairs among the 105 for cross-sectional and longitudinal intra-reader reliability. Radiological hand OA assessment used: global, Kellgren–Lawrence (KL), Kallman and Verbruggen scoring methods. Inter- and intra-reader reliabilities were studied using intraclass coefficient (ICC) and the Bland–Altman method. Sensitivity to change was compared by calculating the standardised response means.

Results: Transversal intra-reader reproducibility ICCs ranged from 0.922 to 0.999. Verbruggen ranked the highest, followed by the KL and Kallman methods. Inter-rater reliability was higher for the Verbruggen scores, followed by the KL, global and Kallman scores (ICC 0.706–0.999). Longitudinal intra-reader reliability (baseline; year 1) was better using the Kallman and KL (ICC 0.986 and 0.990), followed by the Verbruggen (0.941) or global methods (0.939). Standardised response means ranged from 0.24 (KL) to 0.29 (Kallman).

Conclusion: All four methods compared well with respect to reliabilities. However, the Verbruggen and Kallman methods performed better. The method most sensitive to change was the Kallman method, followed by Verbruggen and global scores. This study also suggests that structural changes could be detected in hand OA over a 1-year period.

  • FIHOA, Functional Index for Hand OA
  • ICC, intraclass coefficient of correlation
  • JS, joint space
  • KL, Kellgren–Lawrence
  • M12, month 12
  • OA, osteoarthritis
  • SDD, smallest detectable difference
  • SRM, standardised response mean

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.


  • Published Online First 9 November 2006

  • CC performed the readings and contributed significantly to the protocol development, data analysis and manuscript revision. SG checked the database and performed the statistical analysis. PR participated significantly in devising the protocol of the radiographic study, planned and monitored the statistical analysis. MD participated significantly in devising the protocol of the study and in the analysis of data. He contributed to the manuscript writing and its revision. EM wrote the protocol of the trial and that of this study with MD and PR, performed the readings with CC, contributed to data analysis and wrote the manuscript.

  • Competing interests: None.