Yanhua Zhou, Qiong Wu, Xiangzhi Xiao, Huashan Zhou, Yan Ouyang, Jue Hu, Zhen Wang, Wengao Zeng
Cumulative neurological complication burden encompassing components present at admission or documented later during hospitalization was associated with stepwise increases in in-hospital mortality among non-HIV TBM patients across the lifespan. Hydrocephalus, documented acute seizures, and elevated ICP were key mortality-associated factors. The score should be interpreted as a study-developed summary of hospitalization-period neurological severity rather than as a baseline prediction model, causal exposure, or externally validated prognostic score.
BACKGROUND: Tuberculous meningitis (TBM) is a severe form of extrapulmonary tuberculosis and is frequently complicated by neurological injury. Existing studies have usually evaluated individual complications or admission-based prognostic models, whereas cumulative neurological complication burden documented across a hospitalization remains poorly characterized in non-human immunodeficiency virus (HIV) cohorts spanning pediatric and adult patients. We evaluated whether neurological complications present at admission or documented later during the index hospitalization were associated with in-hospital mortality.
METHODS: We conducted a single-center retrospective cohort study of non-HIV TBM patients hospitalized between October 1, 2013, and January 1, 2024. Potentially eligible cases were screened using electronic health records and discharge-diagnosis systems and confirmed by chart review. TBM diagnostic certainty was retrospectively harmonized according to the Marais uniform research case definition. The primary five-component neurological complication burden score included hydrocephalus, documented acute seizures, cerebral infarction, cerebral hemorrhage, and elevated intracranial pressure (ICP). Components could be present at admission or first documented later during the index hospitalization. Among non-survivors, chart review confirmed that no burden-score neurological complication was first documented within the final 48 h before death. Logistic regression and exploratory model-performance analyses were performed using complete-case data.
RESULTS: Among 1,574 non-HIV TBM patients, including 211 children/adolescents (≤18 years) and 1,363 adults (>18 years), 187 (11.9%) died during hospitalization. Mortality increased across primary five-component burden categories: 8.8% for 0 components, 13.3% for 1 component, 20.3% for 2 components, and 34.5% for ≥3 components. In adjusted models, documented acute seizures [adjusted odds ratio [aOR] 4.10; 95% confidence interval [CI], 2.43-6.90], hydrocephalus (aOR 3.06; 95% CI, 1.92-4.88), and elevated ICP (aOR 1.48; 95% CI, 1.03-2.11; P = 0.032) were independently associated with mortality. Adding neurological burden to the base model improved discrimination modestly (area under the curve from 0.596 to 0.652; DeLong P = 0.002), with acceptable calibration; exploratory decision curve analysis showed generally small between-model differences in net benefit.
CONCLUSIONS: Cumulative neurological complication burden encompassing components present at admission or documented later during hospitalization was associated with stepwise increases in in-hospital mortality among non-HIV TBM patients across the lifespan. Hydrocephalus, documented acute seizures, and elevated ICP were key mortality-associated factors. The score should be interpreted as a study-developed summary of hospitalization-period neurological severity rather than as a baseline prediction model, causal exposure, or externally validated prognostic score.