Prognosticating outcomes and survival for patients with lumbar spinal metastases: Results of a bayesian regression...

To assess the impact of surgical intervention on survival, ambulatory capacity, complications and readmissions following treatment for lumbar metastases.

We identified all adult patients treated for lumbar metastases between 2005-2017. To limit the potential for inherent bias to influence determinations, we used principal component analysis to identify confounders to be included in multivariable testing. Multivariable logistic regression was performed, followed by Bayesian analysis to generate conservative estimates of effect size and 95% confidence intervals (CI). In a sensitivity test, analyses were repeated in a population where patients who died before they could initiate treatment were excluded.

In the period under study, we identified 571 patients who met inclusion criteria. Twenty-one percent of the cohort received a surgical intervention. Bayes regression indicated surgical intervention was independently associated with decreased mortality at 6-months (odds ratio [OR] 0.49; 95% CI 0.34, 0.68) and 1-year (OR 0.63; 95% CI 0.51, 0.76), along with lower odds of being non-ambulatory at 6-months following presentation (OR 0.29; 95% CI 0.18, 0.45). Surgery was also associated with increased odds of complications (OR 1.60; 95% CI 1.24, 2.06) and readmissions (OR 1.37; 95% CI 1.09, 1.72). Numerous clinical characteristics were found to be associated with the outcomes of interest including serum albumin, lung metastases and vertebral body collapse.

Given the favorable outcomes associated with the incorporation of surgery as a component of treatment, we believe that such interventions may be considered part of the treatment approach in patients with lumbar metastases.