Prediction of survival in patients with symptomatic spinal metastases: Comparison between the Tokuhashi score and expert oncologists

Abstract

Background: Existing scoring systems have suboptimal accuracy in prognosticating patients with spinal metastases. Currently, there is no superior method in predicting survival. This study aims to compare the accuracy of survival prediction by expert oncologists versus the revised Tokuhashi scores with actual survivals in a cohort of symptomatic spinal metastases patients.

Methods: All patients who underwent surgical treatment for metastatic spinal tumours in a tertiary hospital between January 2011 to December 2015 were reviewed. Each patient's data was reconstructed into an anonymised clinical scenario and presented independently to five blinded attending oncologists with at least three years' post fellowship experience. They were surveyed for survival prediction twice at no less than four weeks' interval apart; the test-retest reliability was examined. The agreement of their prognostication and modified Tokuhashi scores were compared with actual survivals.

Results: Fifty-five patients were included during the study period. The mean age at presentation was 61.1 years (range, 41 to 79), and mean actual survival was 21.6 months (range, 1 to 68). Cohen's kappa agreement with actual survival was higher by oncologists' estimation (0.52) than by revised Tokuhashi score prediction (0.31) (p = 0.018). Intra-class correlation showed high inter-reliability (0.71) between the five oncologists and a high test-retest reliability (0.69) between both rounds of the survey.

Conclusion: This study showed that expert oncologists provided more accurate survival prediction than revised Tokuhashi scores in patients with spinal metastases. Future studies are required to identify factors in their assessment that led to improved accuracy.