Abstract
OBJECTIVE: To Estimate the probability of treatment success 1 year after a total shoulder arthroplasty by developing a model based on preoperative clinical factors. METHOD: Between June 2003 and December 2006, 140 patients undergoing shoulder operations were assessed for age, gender, current rheumatoid arthritis, Short Form (SF) 36 physical and mental sum scores, previous shoulder operations, the Disabilities of Arm, Shoulder and Hand (DASH) symptom and function scores, the Shoulder Pain and Disability Index (SPADI), and insurance status. One year after the operation a Constant score of 80 or more out of 100 indicated successful treatment. Patient variables were analyzed with a logistic regression model augmented in a stepwise manner and bootstrapped 100 times. Variables selected at least 33 times were incorporated into a final model and the Area under the Receiver Operating Characteristics Curve (aROC) was calculated. RESULTS: There were 47/140 (33.6%) successful treatments. The probability of success was reduced in patients with previous shoulder operations (Odds Ratio [O.R.] 0.17, 95% Confidence Interval (95%CI) 0.04-0.85; P=0.03) and older than 75 years (O.R. 0.21, 95%CI 0.05-0.77; P=0.02). The probability of success increased in patients with a higher SF 36 mental sum score (O.R. 1.03, 95%CI 0.96-1.09, P=0.42) and a higher DASH function score (O.R. 1.05, 95%CI 1.02-1.07, P=0.001). The aROC was 0.79 (0.70-0.88) indicating that the model has a high predictive capacity. CONCLUSION: Once validated this model based on four preoperative clinical factors offers a prediction of whether a patient will respond to treatment 1 year after total shoulder arthroplasty.
Simmen, B R