The Validity of the SRS-22 Instrument in an Adult Spinal Deformity Population Compared With the Oswestry and SF-12: A Study of Response Distribution, Concurrent Validity, Internal Consistency, and Reliability

Abstract

In Brief Study Design. Prospective analysis of a consecutive cohort of adult spinal deformity patients queried over a 12-month period. Objectives. To assess the SRS-22 instrument compared with the SF-12 and Oswestry. Summary of Background Data. Very few reports in the literature have applied the SRS-22 to adult spinal deformity patients. Methods. Consecutive adult spinal deformity patients were applied the SRS-22, SF-12, and Oswestry. Four analyses were done: 1) floor/ceiling effect; 2) Pearson’s correlation coefficients between the SRS-22, SF-12, and Oswestry; 3) Cronbach’s alpha analysis for internal consistency within the SRS-22; and 4) test/retest. Results. Floor/ceiling range for the SRS-22 compared favorably with the SF-12 and Oswestry. The Pearson’s coefficients correlating the two questionnaires relative to the SRS-22 were > 0.7. The Cronbach’s alpha within each domain for the SRS-22 were > 0.7, except for pain (0.67). Test/retest correlation coefficients ranged from 0.84 to 0.95 for the subscales. Conclusions. The SRS-22 is a disease-specific instrument with the capacity to demonstrate change in health status more effectively than the SF-12 and in more domains than the Oswestry. The SRS-22 showed high criterion validity with the SF-12 and Oswestry based on Pearson’s coefficients. High Cronbach’s alpha scores suggested a high internal consistency within each domain of the SRS-22, except for pain (0.67). Test/retest reliability was excellent. A total of 228 adult spinal deformity patients were administered the SRS-22, Oswestry, and SF-12 outcomes instruments. The floor/ceiling range for the SRS-22 was superior to that for the SF-12 and Oswestry. The correlation coefficients showed high concurrent validity between the SRS-22 and the other two instruments. The Cronbach’s alpha analysis demonstrated > 0.7 for the domains of self-image, function, and mental health and 0.67 for pain. Test/retest for the 48 patients demonstrated excellent reliability for the five scales.