Prediction of Oswestry Disability Index (ODI) using PROMIS-29 in a national sample of lumbar spine surgery patients.

Abstract

The primary purpose was to examine the measurement properties of the PROMIS-29 to better understand its use in patients undergoing spine surgery. A secondary objective was to calculate a predictive equation between PROMIS-29 and ODI, to allow clinicians and researchers to determine a predicted ODI score based on PROMIS short form scores. 719 patients with PROMIS v2.0 and ODI responses were queried from the quality outcomes database. Validity was assessed using coefficient omega, ceiling/floor effects, and confirmatory factor analysis. Multivariable regression predicting ODI scores from PROMIS-29 domains was used to create a predictive equation. Predicted ODI scores were plotted against ODI scores to determine how well PROMIS-29 domains predicted ODI. Results showed good reliability and validity of PROMIS-29 in patients undergoing lumbar spine surgery: convergent and discriminant validity, low floor/ceiling effects, and unidimensional domains. The conversion equation used 6 PROMIS-29 domains (ODI% = 37.847- 1.475*[PFraw] + 1.842*[PAINraw] + 0.557*[SDraw] - 0.642*[SRraw] + 0.478*[PIraw] + 0.295*[DEPraw]). Correlation between the predicted and actual ODI scores was R = 0.88, R2 = 0.78, suggesting that the equation predicted ODI scores that are strongly correlated with actual ODI scores. Good measurement properties support the use of PROMIS-29 in spine surgery patients. Findings suggest accurate ODI scores can be derived from PROMIS-29 domains. Clinicians who want to move from ODI to PROMIS-29 can use this equation to obtain estimated ODI scores when only collecting PROMIS-29. These results support the idea that PROMIS-29 domains have the potential to replace disease-specific traditional PROMs.