A new scoring system to predict tuberculosis among South-Tunisian patients with lymphocytic meningitis

Background

Death and poor outcome due to Tuberculous meningitis (TBM) is greatly influenced by the delayed treatment initiation, which often occurs in lymphocytic meningitis (LM). This study aimed to propose an easy-to-use clinical prediction score that can accurately diagnose TBM among LM patients.

Methods

It was a prospective cohort study including all patients with LM hospitalized at the infectious diseases department in Southern Tunisia from 2008 to 2022 were included.

Results

Among 290 LM patients, 105 cases (36.2%) had TBM. By multivariate analysis, age ≥60 years, rural origin, symptom duration ≥5days, thrill, deteriorated general conditions, a hospital stay ≥ 10 days, complicated forms, blood white cells count <4000/mm3, CSF/blood glucose ratio<0.5, hydrocephalus and arachnoiditis were independent factors of TBM. Individual diagnosis indexes from 1 to 3 points were attributed to these factors to produce an overall score ranging from 0 to 16 points. At a cut-off of 7, the predictive score had a receiver operating characteristic (ROC) area of 0.94, a sensitivity, a specificity and a diagnosis accuracy of respectively 81%, 91.4%, and 87.5%.

Conclusions

This original study proposed a novel scoring system that can reliably identify patients with TBM at hospital admission and could be easily used in clinical practice.