The mFI-5 and Postoperative Outcomes in Brain Tumor Patients: A Bayesian Approach to Quantifying Uncertainty

OBJECTIVE: To the best of our knowledge, prior research has not investigated the uncertainty in the relationship between patient frailty and postoperative outcomes after brain tumor surgery. The present study used Bayesian methods to quantify the statistical uncertainty between the 5-factor modified frailty index (mFI-5) and postoperative outcomes for patients undergoing brain tumor resection. METHODS: The present study used retrospective data collected from patients undergoing brain tumor resection during a 2-year period (2017-2019). Posterior probability distributions were used to estimate the means of model parameters that are most likely given the priors and the data. Additionally, 95% credible intervals (CrIs) were constructed for each parameter estimate. RESULTS: Our patient cohort included 2519 patients with a mean age of 55.27 years. Our multivariate analysis demonstrated that each 1-point increase in the mFI-5 score was associated with an 18.76% (95% CrI, 14.35%-23.36%) increase in hospital length of stay and a 9.37% (CrI, 6.82%-12.07%) increase in hospital charges. We also noted an association between an increasing mFI-5 score and greater odds of a postoperative complication (odds ratio [OR], 1.58; CrI, 1.34-1.87) and a nonroutine discharge (OR, 1.54; CrI, 1.34-1.80). However, no meaningful statistical association was found between the mFI-5 score and 90-day hospital readmission (OR, 1.16; CrI, 0.98-1.36) or between the mFI-5 score and 90-day mortality (OR, 1.12; CrI, 0.83-1.50). CONCLUSIONS: Although mFI-5 scores might be able to effectively predict short-term outcomes such as length of stay, our results demonstrate no meaningful association between mFI-5 scores and 90-day readmission or 90-day mortality. Our study highlights the need for rigorously quantifying statistical uncertainty to safely risk-stratify neurosurgical patients.