Dynamic nomogram for predicting long-term survival in patients with brain abscess.

Abstract

Background: Brain abscess (BA) is a serious condition that causes significant mortality and morbidity. While various prognostic factors have been studied, there is limited research on long-term survival predictions. The present study aimed to identify predictors of long-term survival in BA patients and develop a dynamic nomogram for individualized prognostication. Additionally, the secondary objective was to develop and validate a dynamic nomogram for predicting long-term survival in BA patients.

Methods: A retrospective cohort study was conducted on BA patients diagnosed at a tertiary care hospital in Southern Thailand. Demographic, clinical, laboratory, and imaging finding were analyzed. Cox regression was used to identify independent prognostic factors. A dynamic nomogram was developed and validated using Harrell's concordance index (C-index), calibration plots, and cumulative case/dynamic control survival receiver operating characteristic (ROC) curves.

Results: A total of 205 patients were included, with a mean follow-up of 41.66 months. The 1-year, 2-year, and 5-year survival probabilities were 0.77, 0.73, and 0.69, respectively. Independent predictors of long-term survival included age, Karnofsky performance status, hemoculture results, preoperative coagulopathy, neutrophil-to-lymphocyte ratio, bandemia, and occipital BA. The dynamic nomogram revealed strong predictive performance, with a C-index of 0.855 for apparent validation and 0.701 for validation with testing data. Calibration plots and ROC analysis further supported its reliability.

Conclusions: This study presents a validated dynamic nomogram for predicting long-term survival in BA patients. The model provides an interactive tool for individualized risk assessment and facilitating clinical decision-making. Future research should focus on external validation and refinement of the model for broader applicability.