LAUSR

BACKGROUND Hospital length of stay (LOS) is an important cost driver in neurosurgery. Broader surgical literature has demonstrated patient-related factors, including comorbidities, and procedure-related factors, such surgeon experience, may be associated with LOS. Since value optimization strategies may be targeted towards either domain, this study investigated the contributions of patient- and procedure-related factors in predicting prolonged intensive care unit (ICU) LOS (iLOS) and total hospital LOS (tLOS). METHODS Data for adult patients undergoing brain tumor surgery (2017-2019) was collected. Bivariate analyses for iLOS and tLOS were performed using the Mann-Whitney U test and Fisher's exact test. Variables associated with either outcome with p<0.10 were included in patient-only, procedure-only, and patient+procedure factor multivariate linear regression models. Model discrimination was quantified using c-statistics. RESULTS Our 654 patients had a mean age of 57.54 years (standard deviation 14.34 years). For iLOS, the patient-only model significantly outperformed the procedure-only model (p<0.0001) and performed similarly to the patient+procedure model (p=0.50). Other than tumor diagnosis, mFI-5 score was the only factor associated with iLOS (p<0.001) and tLOS (p<0.001) on multivariate analysis. When predicting prolonged tLOS, the patient-only model significantly outperformed the procedure-only model (p<0.0001), and performed similarly to patient+procedure models (p=0.49). CONCLUSIONS Patient-specific factors are the main drivers of prolonged iLOS and tLOS among brain tumor patients. Frailty was significantly associated with both iLOS and tLOS on multivariate analysis. Efforts to improve care value should focus on strategies to optimize patient status, such as pre-habilitation and enhanced recovery after surgery.