Development and internal validation of a nomogram for predicting neurobrucellosis in hospitalized patients with...

To develop and validate a nomogram based on routinely available clinical indicators for individualized prediction of the risk of neurobrucellosis (NB), thereby providing decision support for early clinical diagnosis and timely treatment while avoiding overtreatment.

A single-center retrospective study was conducted including 407 patients diagnosed with brucellosis and hospitalized at the General Hospital of Ningxia Medical University between January 1, 2020 and September 1, 2025. Demographic characteristics, comorbidities, clinical symptoms, physical signs, and laboratory findings at the time of first hospitalization were collected. The occurrence of NB served as the outcome variable. Univariate logistic regression analysis was first performed to screen potential predictors, and variables with P < 0.05 were subsequently included in a multivariate logistic regression model to identify independent risk factors and construct a nomogram prediction model. The dataset was randomly divided into a training cohort and a validation cohort at a ratio of 7:3 for model development and internal validation. Model discrimination was evaluated using the receiver operating characteristic (ROC) curve and the area under the curve (AUC). Calibration performance was assessed using calibration curves, and clinical utility was evaluated using decision curve analysis (DCA).

A total of 407 patients with brucellosis were included, comprising 275 males (67.6%) and 132 females (32.4%). The incidence of NB was 10.6% (43/407). Compared with non-neurobrucellosis (non-NB) patients, those with NB were younger and had a lower incidence of bone and joint pain. Additionally, NB patients exhibited higher levels of hemoglobin and albumin, but lower levels of fibrinogen, high-sensitivity C-reactive protein, and erythrocyte sedimentation rate (all P < 0.05). Univariate logistic regression analysis indicated that age, bone and joint pain, hemoglobin, absolute lymphocyte count (ALC), alanine aminotransferase (ALT), albumin, fibrinogen (FIB), and erythrocyte sedimentation rate (ESR) were associated with the occurrence of NB. Multivariate logistic regression analysis identified bone and joint pain (OR = 0.23, 95% CI: 0.11-0.49), ALC (OR = 1.09, 95% CI: 1.03-1.17), ALT (OR = 0.98, 95% CI: 0.97-0.99), and albumin (OR = 1.13, 95% CI: 1.06-1.20) as independent predictors of NB. The nomogram constructed based on these four variables demonstrated good discrimination and calibration in both the training and validation cohorts. Decision curve analysis showed that the model provided greater net clinical benefit within a reasonable range of threshold probabilities compared with the "treat-all" or "treat-none" strategies.

The nomogram model developed in this study demonstrated good predictive performance and potential clinical applicability. It may serve as a useful tool for early identification of high-risk patients with neurobrucellosis among individuals with brucellosis, thereby facilitating timely and individualized clinical management.