INTRODUCTION Recent years have seen the emergence and increasing use of patient-reported outcomes in clinical studies of treatment effectiveness, and it has become clear that individual outcomes can be quite heterogeneous. When consenting a patient for surgery, it is important to be able to offer an evidence-based, individualised prediction regarding the likely outcome. This study used a comprehensive set of data collected over 12 years in an in-house registry to develop a parsimonious model to predict the multidimensional outcome of patients undergoing surgery for degenerative spinal pathology.
METHODS Data from 8374 patients (mean age 63.9 (14.9-96.3) yrs, 53.4% female) were used for model development, predicting the 12-month scores for the Core Outcome Measures Index (COMI) and its subdomain scores. The data were split 80:20 into a training and test set. The top predictors were selected by applying recursive feature elimination based on a Lasso cross validation model retaining the top 15 predictors (out of 172) per outcome, allowing the retention of a practical number of 20 (out of 39) input variables to be used as a clinical decision-support system (CDSS). Based on the 111 top predictors (of the 20 variables), Ridge cross validation models were trained, validated, and tested for each outcome dimension.
RESULTS Preoperative back and leg pain, nationality, the number of previous spine surgeries, age, type of intervention, preoperative quality-of-life, body-mass index, number of affected levels, Charlson comorbidity, and ASA score, were among the strongest outcome predictors in most models. The R-squared of the models on the validation/test sets averaged 0.16/0.13. Models based on all 39 input variables performed only slightly better in terms of R-squared (0.17/0.14) underlining the good performance of the CDSS based on 20 input variables. A preliminary online tool was programmed to present the predicted outcomes for individual patients, based on their presenting characteristics.
DISCUSSION The prediction models provide reliable estimates to enable a bespoke prediction of the outcome of surgery for individual patients with varying degenerative pathologies and baseline features. The models form the basis of a simple, freely-available online prognostic tool developed to improve access to and usability of prognostic information in clinical practice. This should serve to facilitate decision-making and assist in managing patient expectations.