Correlation and responsiveness of global health, upper extremity-specific, and shoulder-specific functional outcome measures following reverse total shoulder arthroplasty for proximal humerus fracture

Purpose

Reverse total shoulder arthroplasty (rTSA) is effective and increasingly utilized for the management of proximal humerus fracture (PHF). However, the optimal patient-reported outcome metrics (PROMs) for the evaluation of patient outcomes after this surgery are unclear. We investigated the correlation among global, upper extremity-specific, and shoulder-specific PROMs in patients undergoing rTSA for PHF as well as the responsiveness of these PROMs as assessed by floor and ceiling effects. We hypothesized that patients’ post-operative outcome would be best reflected by a combination of these metrics.

Methods

Thirty patients with a history of rTSA for ipsilateral PHF filled out the following outcomes questionnaires at a minimum of 3 years post-op: EQ-5D, EQ-5D VAS, PROMIS physical function, DASH, SSV, SPADI, and ASES. Correlation between metrics was assessed using the Spearman correlation coefficient. Responsiveness was assessed by comparing the proportion of patients reaching floor or ceiling values using McNemar’s test.

Results

Global health metrics (EQ-5D and PROMIS physical function) were strongly correlated with the upper extremity-specific metric (DASH). Shoulder-specific outcomes (SPADI, ASES, and ASES) were moderately correlated with both the global metrics and DASH. There was no significant difference between PROMs with regards to floor and ceiling effects.

Conclusions

The DASH score has been shown to be valid and responsive for shoulder interventions, and our data demonstrate that it correlates strongly with overall quality of life. Shoulder-specific metrics are valid and responsive for shoulder interventions but correlate less with global quality of life. An optimal PROM strategy in rTSA for PHF might involve both DASH and a shoulder-specific score. Based on our assessment of floor and ceiling effects, none of these metrics should be excluded for poor responsiveness.