Development and validation of the Home time and Overall survival after Metastatic spine tumor surgery Estimator (HOME score)

Abstract Background This study reports the development and validation of the Home time and Overall survival after Metastatic spine tumor surgery Estimator (HOME score). Methods A population cohort study was conducted, including 2348 adults with spine metastases treated with surgery in the 2005 to 2020 Ontario Cancer registry. HOME score predictions were the likelihood of post-surgery home time of 3-months or less, and overall survival at 6 months, 1 year, and 1.5 years after surgery. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC) for home time predictions, and the concordance index (C-index) for survival. Variable importance was quantified using standardized coefficients. Results Mean age was 62.4 years (SD: 12.6) and the most common primary cancer was lung (N = 513, 21.9%). Patients treated between 2005 and 2018 were allocated to training, and those treated in 2019-2020 were used as a hold-out test cohort. The final model included 17 items for home time prediction, and 24 items for survival prediction including demographic, comorbid, cancer, and presentation features. Performance of the home time model (AUC: 0.70), and survival model (C-index 0.70, 6-month AUC: 0.73, 1-year AUC: 0.75, 1.5-year AUC: 0.76) was stable across training and testing. Primary cancer origin and history of congestive heart failure (CHF) ranked highest among features impacting outcome predictions. Conclusions The HOME score (https://shakilh.shinyapps.io/home_app/) accurately predicts home time and survival for patients with spinal metastases undergoing surgery. Key factors influencing predictions were primary origin, and history of CHF. This represents a significant advancement to patient centered preoperative risk stratification.