PROMIS-29 Validity and Conversion Equation to Neck Disability Index (NDI) using a National Sample of Cervical Spine Surgery Patients

Abstract

Abstract INTRODUCTION Patient Reported Outcomes Measures (PROMs) have become a vital tool for evaluating success of spine surgery. The purpose was to examine the validity of the PROMIS-29 to better understand it's use for cervical spine surgery patients. A secondary objective was to calculate a conversion equation between PROMIS-29 and NDI, to allow researchers and clinicians to determine a predicted NDI. METHODS We conducted a retrospective analysis of prospectively collected national registry data, the Quality Outcomes Database (QOD), which is designed to evaluate risk-adjusted outcomes for the most commonly performed spinal surgical procedures. We queried the QOD registry for patients with PROMIS-29 and NDI responses. Of 619 patients, 41% were female, 87% Caucasian, and 13% had revision surgery. The mean age was 58 yr (SD = 12). Validity of the PROMIS-29 was assessed using Cronbach's a, ceiling/floor effects, and principle axis factoring. Multivariable regression predicting NDI scores from PROMIS-29 domains used data from all three timepoints. Predicted NDI scores, derived from PROMIS-29 domains, were plotted against NDI scores to determine how well PROMIS-29 domains predicted NDI. A conversion equation was created from PROMIS regression coefficients. RESULTS Results showed good reliability (Cronbach's a >0.80) and validity of PROMIS-29 in patients undergoing cervical surgery: convergent (r > 0.65) and discriminant (r = 0.49-0.63) validity, low/high floor/ceiling effects (=2%), and unidimensional domains (factor loading >0.64). The conversion equation used 6 PROMIS-29 domains (NDIpercent = 7.928-0.527*[PFraw] + 1.570*[PAINraw] +0.995*[SDraw] + 1.101[PIraw] + 0.392[Fatigueraw] 0.409 [SRraw]). Correlations between predicted and actual NDI scores at each timepoint were: r = 0.84 at baseline, r = 0.76 at 3 mo, and r = 0.84 at 12 mo, suggesting that the equation predicted NDI-scores that are strongly correlated with actual NDI-scores. CONCLUSION Good reliability and validity support the use of PROMIS-29 in cervical surgery patients. Findings suggest accurate NDI score can be derived from PROMIS-29 domains. These results support the idea that PROMIS-29 domains have the potential to replace disease-specific traditional PROMs like NDI.