Identifying High-risk Patients in Adult Spinal Deformity Surgery: Harnessing Predictive Analytics to Uncover Key Drivers of Medical Complications

Introduction: A lack of consensus among high-risk ASD committees persists regarding which specific patient characteristics and surgical parameters should trigger preoperative discussions, complicating efforts to standardize risk assessment and care pathways.

Methods: ASD patients meeting operative criteria—scoliosis >20º, SVA>5 cm, PT >25º, or TK>60º—were included if baseline and 2-year radiographic and HRQL data were available. Surgical risk was evaluated based on the occurrence of surgical complications. High medical risk was defined as the presence of major medical complications or revision surgery, along with failure to achieve the MCID for the ODI. A conditional inference tree (CIT) machine learning model was employed to identify threshold cutoffs for continuous variables, while multivariable logistic regression models were used to develop predictive scores for both medical and surgical risks.

Results: A total of 464 ASD patients met inclusion criteria (mean age: 60.5 years, 80% female, BMI: 27.1 kg/m², CCI: 1.7, mASD-FI: 6.8). For patients with high medical risk, factors such as age >70, baseline ODI >56, baseline Frailty Index >5, CCI >3, prior revision surgery, and histories of heart, liver, or lung disease, as well as baseline SVA >15 cm, C7PL >7 cm, and PI-LL >25, were predictive of poor outcomes, with an AUC of 97% and accuracy of 98%. As the number of risk factors increased, patients demonstrated significantly higher odds of major medical complications prior to discharge, medical complications within 90 days, complications within 2 years, overall complications, medical complications, and reoperations. Similarly, as medical risk factors increased, the odds of achieving BCO in ODI (at 6M, 1Y, 2Y) and SRS-22r total scores (at 1Y, 2Y) decreased, whereas the likelihood of experiencing WCO in ODI (at 6M, 1Y, 2Y) and SRS-22r (at 1Y, 2Y) total scores increased.

Conclusion : Identifying patient-specific and surgical factors linked to a heightened risk of major medical complications and suboptimal clinical outcomes allows surgeons to more effectively stratify patients who could benefit from multidisciplinary care. This targeted strategy promotes tailored perioperative optimization, enhancing patient safety and driving improved surgical outcomes.