Correlation and conversion between the QuickDASH, Constant Score, and Oxford Shoulder Score in patients with conservatively treated proximal humerus fractures.

BACKGROUND

There is significant heterogeneity of outcome measures used in the research of proximal humerus fractures (PHFs). Current evidence regarding the correlation and conversion between the various outcome measures is sparse. This study aims to study the correlation and conversion between the QuickDASH, Constant Score (CS), and Oxford Shoulder Score (OSS) in conservatively treated PHFs.

METHODS

A prospective cohort study of patients (n = 136) with conservatively treated PHFs between August 2017 and April 2020 was conducted. Patients had a minimum follow-up period of 1 year. The 3 outcome measures (QuickDASH, CS, and OSS) were collected at 4 time points-6 weeks, 3 months, 6 months, and 1 year after injury. Changes in scores across time and correlation between each pair of outcome measures were calculated. A linear regression model was used to derive conversion equations which were then internally validated.

RESULTS

A significant strong negative correlation was observed between the OSS and QuickDASH (coefficient: -0.746; P < .001), a significant moderate negative correlation was observed between the CS and QuickDASH (coefficient: -0.581; P < .001), and a significant moderate positive correlation was observed between the CS and OSS (coefficient: 0.697; P < .001). The 6 derived regression equations showed low mean differences between predicted and actual values (ranging from -1.21 to 2.51). The correlation between actual and predicted values was moderate to strong, ranging from a coefficient of 0.57 in the conversion from the CS to QuickDASH to 0.74 in the conversion from the CS to OSS and OSS to CS.

CONCLUSION

In a cohort of patients with conservatively managed PHFs, moderate to strong correlations were seen in pairwise comparisons of the OSS, QuickDASH, and CS. With linear regression analyses, 6 regression equations were derived to estimate one score from another. On internal validation, there was good agreement between the means of the predicted and actual scores but high within-individual variability. These formulae can help to compare studies with heterogeneous outcome measures and facilitate meta-analyses. However, these equations should not be used to predict one score from another in an individual due to variability when converting individual scores.