Psychometric Properties of the SF-12, Penn Shoulder Score, and Visual Analogue Scale-Pain for Rotator Cuff Retears

Abstract

Purpose: We assessed reliability, responsiveness, and sensitivity of the Short Form-12 (SF-12), Visual Analog Scale (VAS) Pain score, and Penn Shoulder Score (PSS) in patients with surgical repair of rotator cuff tear. Methods: The SF-12, PSS, and VAS-Pain were administered at a screening visit and at six follow-up visits over 1 year. Internal consistency, test-retest reliability, standardized effect sizes, and corrected item-total correlations were calculated and cumulative distribution function plots were constructed. Results: Cronbach’s alpha coefficients exceeded 0.70 for PSS Pain and Function subscales but not for PCS-12 or MCS-12 of the SF-12. The PSS Function, PCS-12, MCS-12, and VAS-Pain gave intraclass correlation coefficients (ICCs) between 0.60 and 0.80, while PSS Pain and Satisfaction subscales gave ICCs between 0.44 and 0.73. Standardized effect sizes between week 52 and baseline were between 1.5 and 3.5 for all scales, apart from the MCS-12 effect size of 0.005. Effect sizes between retear and no retear groups were between 0.3 and 0.7 for all scales, apart from MCS-12 effect size of 0.10. Cumulative distribution functions between retear and no retear groups revealed visual separation between changes in the PSS total score, PCS-12, and VAS Pain score, but no separation in the MCS-12. Conclusions: Cronbach’s alpha for PCS-12 and MCS-12 scales were below the generally accepted level of 0.70. The PCS-12, MCS-12, PSS total, PSS Function subscale, and VAS-Pain showed moderate test-retest reliability while PSS Pain and Satisfaction subscales showed fair test-retest reliability. All scales except the MCS-12 showed excellent responsiveness and discernible sensitivity.