Page 96 - Predicting survival in patients with spinal bone metastasesL
P. 96
CHAPTER VI
into account. This variable was associated with survival on multivariate analysis and could potentially be used to increase the accuracy of the predictive models.
Lastly, the recently published Bollen model was based on a retrospective analysis of 1043 patients treated with surgery and radiotherapy. Essentially a hybrid of the Tomita and Van der Linden models, it has quite good results on both the short and long end of the survival spectrum. Even though the primary tumor classification is reasonably accurate, it still requires further specification of the malignancies in order to increase the discriminative ability.
CONCLUSION
This study provides an insight on how the available SBM models perform on a large dataset. The results suggest that the Bollen model performs best when identifying patients with a short and patients with a long expected survival. With the predicted survival outcome, appropriate choices for treatment can be made by patients and their treating physicians.
94
