277. What are the major drivers of outcomes in cervical deformity surgery?

Abstract

BACKGROUND CONTEXT Cervical deformity (CD) correction is becoming more challenging and complex. Understanding the factors that drive optimal outcomes has been understudied in CD corrective surgery. PURPOSE To weight baseline factors on impact upon outcomes following CD surgery. STUDY DESIGN/SETTING Retrospective review of a single-center database. PATIENT SAMPLE A total of 61 cervical patients. OUTCOME MEASURES Two outcomes were measured: ‘improved outcome’(IO):(1) radiographic improvement: ‘non-deformed’ Schwab PT/SVA, Ames cSVA/TS-CL, (2) clinical: MCID EQ5D, NDI, or improve in mJOA modifier, (3) complications/reop: no reop or major complications; ‘poor outcome’(PO):(1) radiographic deterioration: ‘moderate’ or ‘severely’ deformed Schwab SVA/PT, Ames cSVA/TS-CL, (2) clinical: not meeting MCID EQ5D, NDI worsening in mJOA modifier; (3) complications/reop: reoperation or complications. METHODS CD patients included: full baseline (BL) and 1-year (1Y) radiographic measures and HRQLs. Reoperation for infection excluded. Patients were categorized by IO, PO or not. Random forest assessed ratios of predictors for IO and PO. Categorical regression models predicted how BL regional deformity (Ames cSVA, TS-CL, horizontal gaze), BL global deformity (Schwab PI-LL, SVA, PT), regional/global change (BL to 1Y), BL disability (mJOA score) and BL pain/function impact outcomes. RESULTS A total of 61 patients were included (55.8yrs, 54.1% F). Surgical approach: 18.3% anterior, 51.7% posterior, 30% combined. Avg levels fused: 7.7. Mean operative time: 823min, EBL:1037ccs. At 1Y, 24.6% had an IO, 9.8% PO. Random forest analysis showed the top 5 individual factors associated with an IO: BL Maximum Kyphosis, Maximum Lordosis, C0-C2, L4 Pelvic Angle, and NSR Back Pain (80% radiographic, 20% clinical). Categorical IO regression model (R²=0.328,p=0.007): low BL regional deformity (β=-0.082), low BL global deformity (β=-0.099), global improve (β=0.532), regional improve (β=0.230), low BL disability (β=0.100), low BL NDI (β=0.024). Random forest demonstrated of the top 5 individual baseline factors associated with PO, 80% were radiographic: BL CL Apex, DJK angle, cervical lordosis, T1 slope and NSC Neck Pain. Categorical PO regression model (R²=0.306,p=0.012): high BL regional deformity (β=-0.108), high BL global deformity (β=-0.255), global decline (β=0.272), regional decline (β=0.443), baseline disability (β=-0.164), BL severe NDI (>69)(β=0.181). CONCLUSIONS Categorical weight demonstrated radiographic as the strongest predictor of both improved (global alignment) and poor outcome (regional deformity/deterioration). Radiographic factors carry the most weight in determining an improved or poor outcome, and can be ultimately utilized in preoperative planning and surgical decision-making in order to optimize outcomes. FDA DEVICE/DRUG STATUS This abstract does not discuss or include any applicable devices or drugs.