A pattern recognition approach to the development of a classification system for upper-limb musculoskeletal disorders of workers

Dorcas E. Beaton, Claire Bombardier, Donald C. Cole, Sheilah Hogg-Johnson, Dwayne Van Eerd, Peter C. Amadio, Rachelle Buchbinder, Simon Carette, Bradley A. Evanoff, Carol Kennedy, Christine B. Novak, Glenn Pransky, Eira Viikari-Juntura

Research output: Contribution to journalArticlepeer-review

6 Scopus citations


Objectives: Workers' musculoskeletal disorders are often pain-based and elude specific diagnoses; yet diagnosis or classification is the cornerstone to researching and managing these disorders. Clinicians are skilled in pattern recognition and use it in their daily practice. The purpose of this study was to use the clinical reasoning of experienced cclinicians to recognize patterns of signs and symptoms and thus create a classification system. Methods: Two hundred and forty-two workers consented to a standardized physical assessment and to completing a questionnaire. Each physical assessment finding was dichotomized (normal versus abnormal), and the results were graphically displayed on body diagrams. At two different workshops, groups of experienced researchers or clinicians were led through an exercise of pattern recognition (clustering and naming of clusters) to arrive at a classification system. Interobserver reliability was assessed (8 observers, 40 workers), and the classification system was revised to improve reliability. Results: The initial classification system had good face validity but low interobserver reliability (kappa <0.3). Revisions were made that resulted in a proposed triaxial classification system. The signs and symptoms axes quantified the areas in the involved upper limbs. The proposed third axis described the likelihood of a specific clinical diagnosis being made and the degree of certainty. The interobserver reliability improved to -0.70. Conclusions: This triaxial classification system for musculoskeletal disorders is based on clinically observable findings. Further testing and application in other populations is required. This classification system could be useful for both clinicians and epidemiologists.

Original languageEnglish (US)
Pages (from-to)131-139
Number of pages9
JournalScandinavian Journal of Work, Environment and Health
Issue number2
StatePublished - Apr 2007


  • Nosology
  • Repetitive strain injury
  • Reproducibility of results

ASJC Scopus subject areas

  • Public Health, Environmental and Occupational Health


Dive into the research topics of 'A pattern recognition approach to the development of a classification system for upper-limb musculoskeletal disorders of workers'. Together they form a unique fingerprint.

Cite this