Tuberculous meningitis: a comparison of scoring systems for diagnosis.

OBJECTIVES: Tuberculous meningitis (TBM) is a major clinical and public health problem, both for diagnosis and management. We compare two established scoring systems, Thwaites and the Lancet consensus scoring system for the diagnosis of TB and compare the clinical outcome in a tertiary care setting.

METHODS: We analyzed 306 patients with central nervous system (CNS) infection over a 5-year period and classified them based on the unit's diagnosis, the Thwaites classification as well as the newer Lancet consensus scoring system. Patients with discordant results-reasons for discordance as well as differences in outcome were also analyzed.

RESULTS: Among the 306 patients, the final diagnosis of the treating physician was TBM in 84.6% (260/306), acute CNS infections in 9.5% (29/306), pyogenic meningitis in 4.2% (13/306) and aseptic meningitis in 1.3% (4/306). Among these 306 patients, 284 (92.8%) were classified as "TBM" by the Thwaites" score and the rest as "Pyogenic". The Lancet score on these patients classified 29 cases (9.5%) as 'Definite-TBM', 43 cases (14.1%) as "Probable-TBM", 186 cases (60.8%) as "Possible-TBM" and the rest as "Non TBM". There was moderate agreement between the unit diagnosis and Thwaites classification (Kappa statistic = 0.53), as well as the Lancet scoring systems. There is only moderate agreement between the Thwaites classification as well as the Lancet scoring systems. It was noted that 32/ 284 (11%) of patients who were classified as TBM by the Thwaites system were classified as "Non TBM" by the Lancet score and 6/258 (2%) of those who were diagnosed as possible, probable or definite TB were classified as Non TB by the Thwaites score. However, patients who had discordant results between these scores were not different from those who had concordant results when treatment was initiated based on expert clinical evaluation in the tertiary care setting.

CONCLUSION: There was only moderate agreement between the Thwaites' score and the Lancet consensus scoring systems. There is need to prospectively evaluate the cost effectiveness of simple but more effective rapid diagnostic alogrithm in the diagnosis of TB, particularly in a setting without CT and MRI facilities.