Scientific Programme
Year 2021
October 2021


Abstract Title
Predicting functional outcomes in conservatively managed proximal humerus fractures with radiographic evaluation



MOH Holdings Pte Ltd (MOHH)1, Tan Tock Seng Hospital2, Woodlands Health3

Background & Hypothesis

Treatment of proximal humerus fractures that are minimally displaced remains controversial with variable functional outcomes. This paper aims to evaluate the relevance of radiographic parameters in predicting functional outcomes and assess the intra- and inter-rater reliability of these radiographic measurements.


Radiograph images were assessed for 132 patients. Caput-collum-diaphyseal (CCD) angles, Y-scapular angles, and humeral head height (HHH) were obtained. The patients were split into varus and valgus groups based on the CCD angles and retroverted and anteverted groups based on Y-scapular angles. Functional outcome was measured by Oxford Shoulder Score (OSS), Constant Shoulder Score, and quick Disabilities of Arm, Shoulder and Hand score at 1 year follow-up.  Intra- and inter-rater reliability for 2 observers were measured with the intraclass correlation coefficients (ICCs). Receiver operator curve (ROC) analysis and logistic regression analysis defined the optimal value for abnormalities on radiographic evaluation as an outcome predictor.


The intra-rater and inter-rater reliability for the radiographic parameters were excellent for both observers. There was a significant relationship between OSS and final Y-scapular angle in the retroverted group in both univariable and multivariable analysis (coeff 0.058, 95% CI (0.008, 0.318), p=0.025) (adj coeff 0.056, 95% CI (0.008, 0.104), p=0.022) respectively. The optimum predictive angulation at the final Y-scapular was 25o for predicting poor functional outcome with an area under the ROC curve of 0.611 (95% CI, 0.460,0.761).

Discussion & Conclusion

This study showed that there was excellent reproducibility for radiographic measurements of proximal humerus fractures. Moreover, Y-scapular views are the most critical in predicting functional outcomes.