Can handheld dynamometry predict rotator cuff tear size? A study in 2100 consecutive patients.

Abstract

This study aimed to determine whether handheld dynamometry measurements could predict rotator cuff tear size in patients who required surgical treatment of their shoulder pathology. Handheld dynamometer readings were collected prior to surgery and analyzed retrospectively for 2100 consecutive patients. Post hoc, the cohort was divided into patients with rotator cuff tears (n = 1747) and those without rotator cuff tears (n = 353). The tear group was stratified into partial- vs. full-thickness tears and into 4 groups based on tear size area. Patients with partial-thickness tears had greater internal rotation (P = .03), external rotation (P < .001), and supraspinatus (P < .001) strength than patients with full-thickness tears. Patients with tears had lower supraspinatus strength than patients without tears (r = -0.82, P < .001). Patients with a larger tear size had lower values of external rotation (r = -1.46, P < .001) and supraspinatus (r = -1.18, P < .001) strength. A model involving internal rotation and supraspinatus strength could predict the presence of a tear with a sensitivity of 82% and specificity of 29%. The correct prediction rate was 73% overall (82% in tear group and 29% in no-tear group). The following formula was found to predict rotator cuff tear size, showing modest correlation with our raw data (r = 0.25, P < .001): Tear size = 482.8 + (3.9 × Internal rotation strength) + (1.6 × Adduction strength) - (7.2 × External rotation strength) - (2.0 × Supraspinatus strength). Handheld dynamometer readings could not reliably predict rotator cuff tear size, showing only modest correlation with our raw data. Handheld dynamometry readings could predict the presence of a tear, although tears in the intact cohort were overestimated (a specificity of 29% and negative predictive value of 25%).