Predictive model for major complications 2 years after corrective spine surgery for adult spinal deformity
ASD surgery improves a patient’s health-related quality of life, but it has a high complication rate. The aim of this study was to create a predictive model for complications after surgical treatment for adult spinal deformity (ASD), using spinal alignment, demographic data, and surgical invasiveness.
This study included 195 surgically treated ASD patients who were > 50 years old and had 2-year follow-up from multicenter database. Variables which included age, gender, BMI, BMD, frailty, fusion level, UIV and LIV, primary or revision surgery, pedicle subtraction osteotomy, spinal alignment, Schwab-SRS type, surgical time, and blood loss were recorded and analyzed at least 2 years after surgery. Decision-making trees for 2-year postoperative complications were constructed and validated by a 7:3 data split for training and testing. External validation was performed for 25 ASD patients who had surgery at a different hospital.
Complications developed in 48% of the training samples. Almost half of the complications developed in late post-op period, and implant-related complications were the most common complication at 2 years after surgery. Univariate analyses showed that BMD, frailty, PSO, LIV, PI-LL, and EBL were risk factors for complications. Multivariate analysis showed that low BMD, PI-LL > 30°, and frailty were independent risk factors for complications. In the testing samples, our predictive model was 92% accurate with an area under the receiver operating characteristic curve of 0.963 and 84% accurate in the external validation.
A successful model was developed for predicting surgical complications. Our model could inform physicians about the risk of complications in ASD patients in the 2-year postoperative period.
KeywordsAdult spinal deformity Complication Surgery Risk stratification Predictive model
This study was approved by the appropriate institutional review board.
Compliance with ethical standards
Conflict of interest
The authors report no conflict of interest.
- 24.Abbott D (2014) Applied predictive analytics: principles and techniques for the professional data analyst, 1st edn. Wiley, IndianapolisGoogle Scholar