168. Factors associated with the failure to meet 90-day estimated survival after the surgical treatment of spinal metastases

Abstract

BACKGROUND CONTEXT Many prognostic scoring systems have been developed and provide estimated prognosis for surgical decision making in the treatment of vertebral metastatic disease. External validation studies have confirmed the diagnostic ability of these scoring systems to discriminate between various disease states and relative survival. However, analysis of the individual patient factors associated with misprognostication has not been completed and provides important insight into the failure of patients to meet an estimated prognosis. PURPOSE N/A STUDY DESIGN/SETTING N/A PATIENT SAMPLE N/A OUTCOME MEASURES N/A METHODS Preoperative prognostic scores were calculated for 90 patients undergoing operative treatment of metastatic spine disease using nine prevalent scoring systems including the Skeletal Oncology Research Group (SORG) Classic Algorithm, SORG Nomogram, original Tokuhashi, revised Tokuhashi, Tomita, original Bauer, modified Bauer, Katagiri, and van der Linden. The diagnostic accuracy of each scoring system at 90 days was evaluated using a receiver operating characteristic (ROC) model and the scoring system that demonstrated the highest accuracy was used to calculate an estimated 90-day survival for each individual patient. The actual survival of each patient was compared to estimated survival and the accuracy of each individual prognosis estimation was evaluated for correctness using a cutoff point of 0.5, which represents a 50% probability of survival at 90 days. The factors associated with incorrect estimated prognosis were then identified using a binary logistic regression model. RESULTS Among all scoring systems, the SORG nomogram demonstrated the highest accuracy at predicting 90-day survival with an AUC of 0.702. In 13 records, 90-day survival was overestimated (incorrectly classified records from probability 0.5-1.0) in which patients lived less than the estimated 90 days. Most misclassifications were optimistic, with a median estimated probability of 0.75 (total range 0.46-0.85). Both postoperative Karnofsky Performance Scale Index (KPS) and return to preoperative KPS level at 30 days post-operatively were found to be significant predictors of failure to meet 90-day predicted survival with an odds ratio of 51.3 (95% CI: 10.0-262.1, p: CONCLUSIONS Prognostic scoring systems provide clinicians with a preoperative estimation of a patients expected survival at various time points. It is shown here that failure to reach a functional status that allows for basic selfcare within 30 days of surgery, as evidenced by a KPS level greater than 40, is a significant predictor of failure to meet estimated survival. Clinicians should be aware of this and resources should be directed at maximizing physical rehabilitation and function specifically within the first 30 days after surgery, patients should be counseled accordingly, and further treatment decisions should also take this into account. FDA DEVICE/DRUG STATUS This abstract does not discuss or include any applicable devices or drugs.