Shorter and sweeter: the 16-item version of the SRS questionnaire shows better structural validity than the 20-item version in young patients with spinal deformity

Purpose In patients with adult spinal deformity, it was previously shown that 16 of the non-management items of the SRS-instrument showed a better fit to the theoretical four-factor model (pain, function, self-image, mental health) than did all 20 items. Whether the same phenomenon is observed in data from younger (< 20y) patients, for whom the questionnaire was originally designed, is not currently known. Methods Confirmatory factor analysis was used to evaluate the factor structure of the 20 non-management items of the SRS-instrument completed by 3618 young patients with spinal deformity (75.5% female; mean age, 15.0 +/- 2.0 years) and of its equivalence across language versions (2713 English-speaking, 270 Spanish, 264 German, 223 Italian, and 148 French). The root mean square error of approximation (RMSEA) and comparative fit index (CFI) indicated model fit. Results Compared with the 20-item version, the 16-item solution significantly increased the fit (p < 0.001) across all language versions, to achieve good model fit (CFI = 0.96, RMSEA = 0.06). For both 16-item and 20-item models, equivalence across languages was not reached, with some items showing weaker item-loading for some languages, in particular German and French. Conclusion In patients with adolescent idiopathic scoliosis, the shorter 16-item version showed a better fit to the intended 4-factor structure of the SRS-instrument. The wording of some of the items, and/or their equivalence across language versions, may need to be addressed. Questionnaire completion can be a burden for patients; if a shorter, more structurally valid version is available, its use should be encouraged.

Automated summary

The study aimed to validate the shorter version of the SRS questionnaire in adolescents with spinal deformity and compare its equivalence across different language versions. The results showed that the 16-item version exhibited better model fit across all language versions, but further improvements are needed in terms of equivalence across languages.