Method
Sixty-four stroke patients (40 male; 24 female, mean age: 59.2⁑10.4) were included in the study. All subjects underwent a clinical examination that included spasticity and Brunnstrom stage; and a radiologic examination. The diagnosis of CRPS was based on clinical criteria and three-phase scintigraphies. The degree of shoulder subluxation was assessed by the distance between inferomedial point of acromion and the center of humeral head and glenoid fossa.

Results
CRPS after stroke occurred in 34 patients (53%). There were significant differences in Brunnstrom stage and shoulder subluxation between the CRPS groups and the non CRPS groups. Among radiographic measurements, vertical distance ratio (involved/uninvolved) and oblique distance ratio indicated a strong correlation with CRPS (p＜0.01). Among clinical variables and radiographic measurements, oblique ratio was the most valuable determinant of the risk for CRPS.

Conclusion
Shoulder subluxation shows a significant correlation with CRPS after stroke. We suggest oblique ratio as a useful measurement of shoulder subluxation to estimate the risk of CRPS. (J Korean Acad Rehab Med 2003; 27: 21-26)