Impact of radiologic variables on item responses of ODI, SRS22 and SF-36. in adult spinal deformity patients: differential item functioning (DIF) analysis results from a multi-center database

Abstract

To determine if responses given to each question of the Scoliosis Research Society-22 (SRS22), Oswestry disability index (ODI) and Short Form-36 (SF-36) questionnaires are influenced by the radiological parameters. Patients enrolled in a multi-centre prospectively collected adult spinal deformity database who had complete SRS22, ODI and SF-36 data at baseline and at one-year follow-up were analysed. The presence of a differential item function of each question within each score in relation to radiological parameters was analysed using a mixed Rasch model with the radiological threshold value(s) determined. Of those patients analysed (n = 1745; 1406 female, average age 51.0 ± 19.8years), 944 were surgically and 801 were non-surgically treated. For the SRS22, questions (Q) 3, 5 and 18 were sensitive to almost all radiological parameters and the overall score was found sensitive to the Cobb angle. For the ODI, Q3, 6, 9 and 10 were not sensitive to any radiologic parameters whereas Q4 and 5 were sensitive to most. In contrast, only 3 of the SF-36 items were sensitive to radiological parameters. 78% of the SRS-22, 60% of the ODI and 8% of the questions in the SF-36 are sensitive to radiological parameters. Sagittal imbalance is independently associated with a poor overall outcome, but affects mental status and function more than pain and self-image. The assembly of questions responsive to radiological parameters may be useful in establishing a connection between changes in radiologic parameters and HRQL.