Objective: Metastases in the spinal column are a common manifestation of advanced cancer disease. Severe pain, pathologic fracture and neurologic deficit due to spinal metastases need adequate treatment. Considering oncologic aspects as well as quality of life, treatment decision should also include prediction of the survival period. In this study we analysed the scoring systems of Bauer, Bauer modified, Tokuhashi, Tokuhashi revised, Tomita, van der Linden and Sioutos, as well as the parameters they consist of, for their predictive value.

Methods: Two-hundred and fifty four patients with confirmed spinal metastases were investigated retrospectively (treatment 1998–2006; 62 received surgery, 189 only conservative therapy). The following factors were analysed: primary tumor, general condition (Kar-nofsky Performance Scale), neurological deficit, number of spinal and extraspinal bone metastases, visceral metastases, pathologic fracture. Survival period was calculated from date of diagnosis of the spinal metastases until date of death or last follow up (minimum follow-up: 12 months). For statistical analysis univariate and stepwise multivariate Cox regression analyses were performed.

Results: Median overall survival for all patients was 10.6 months. The following factors showed significant influence on survival in multivariate analysis: primary tumor (p< 0.0001), status of visceral metastases (p< 0.0001), and systemic therapy (p< 0.0001). Cox regression proved all scores significant in metric analysis. Distinguishing between the prognostic subgroups, only Bauer and Bauer modified showed significant results for this classification into good, moderate and poor prognosis. The other systems failed to distinguish significantly between good and moderate prognosis.

Conclusion: In our collective, Bauer and Bauer modified score prove to be the most reliable systems for predicting survival. We therefore want to propose the Bauer modified score (consisting of only four positive prognostic factors, excluding pathologic fracture) as valid for predicting survival and practicable for clinical use.