Health-related quality of life measures in adult spinal deformity: can we replace the SRS-22 with PROMIS?

Purpose To determine the validity and responsiveness of PROMIS metrics versus the SRS-22r questionnaire in adult spinal deformity (ASD). Methods Surgical ASD patients undergoing >= 4 levels fused with complete baseline PROMIS and SRS-22r data were included. Internal consistency (Cronbach's alpha) and test-retest reliability [intraclass correlation coefficient (ICC)] were compared. Cronbach's alpha and ICC values >= 0.70 were predefined as satisfactory. Convergent validity was evaluated via Spearman's correlations. Responsiveness was assessed via paired samples t tests with Cohen's d to assess measure of effect (baseline to 3 months). Results One hundred and ten pts are included. Mean baseline SRS-22r score was 2.62 +/- 0.67 (domains = Function: 2.6, Pain: 2.5, Self-image: 2.2, Mental Health: 3.0). Mean PROMIS domains = Physical Function (PF): 12.4, Pain Intensity (PI): 91.7, Pain Interference (Int): 55.9. Cronbach's alpha, and ICC were not satisfactory for any SRS-22 and PROMIS domains. PROMIS-Int reliability was low for all SRS-22 domains (0.037-0.225). Convergent validity demonstrated strong correlation via Spearman's rho between PROMIS-PI and overall SRS-22r (- 0.61), SRS-22 Function (- 0.781), and SRS-22 Pain (- 0.735). PROMIS-PF had strong correlation with SRS-22 Function (0.643), while PROMIS-Int had moderate correlation with SRS-22 Pain (- 0.507). Effect size via Cohen's d showed that PROMIS had superior responsiveness across all domains except for self-image. Conclusions PROMIS is a valid measure compared to SRS-22r in terms of convergent validity, and has greater measure of effect in terms of responsiveness, but failed in reliability and internal consistency. Surgeons should consider the lack of reliability and internal consistency (despite validity and responsiveness) of the PROMIS to SRS-22r before replacing the traditional questionnaire with the computer-adaptive testing.