METHODS: We searched nine databases from inception to 8 February 2018 for randomized controlled trials evaluating pharmacological interventions and clinical outcomes in adult bacterial meningitis. An updated search from 9 February to 9 March 2020 was performed, and no new studies met the inclusion criteria. Study quality was assessed using the revised Cochrane Risk of Bias Tool. The Grading of Recommendations Assessment, Development and Evaluation system was used for quality of evidences evaluation. Meta-analyses were conducted to estimate the risk ratio with 95% confidence interval for both direct and indirect comparisons on the primary outcomes of all-cause mortality, neurologic sequelae and any hearing loss. The study was registered in PROSPERO (CRD42018108062).

RESULTS: Nine RCTs were included in systematic review, involving 1,002 participants with a mean age ranging between 25.3 to 50.56 years. Six RCTs were finally included in the network-meta analysis. No significant difference between treatment was noted in meta-analysis. Network meta-analysis suggests that corticosteroids in combination with antibiotic therapy was more effective in reducing the risk of any hearing loss compared to mono antibiotic therapy (RR 0.64; 95%CI, 0.45 to 0.91, 4 RCTs, moderate certainty of evidence). Numerical lower risk of mortality and neurological complications was also shown for adjunctive corticosteroids in combination with antibiotic therapy versus mono antibiotic therapy (RR 0.65; 95%CI, 0.42 to 1.02, 6 RCTs, moderate certainty of evidence; RR 0.75; 95%CI, 0.47 to 1.18, 6 RCTs, moderate certainty of evidence). No differences were noted in the adverse events between different therapies. The overall certainty of evidence was moderate to very low for all primary outcomes examined.

METHODS: Data from a retrospective review of 13-year S.suis patient records in a tertiary hospital in Chiang Mai, Northern, Thailand was obtained. Univariate and multivariate logistic regressions were employed to develop a predictive model. The clinical risk score was constructed from the coefficients of significant predictors. Area under the receiver operator characteristic curve (AuROC) was identified to verify the model discriminative performance. Bootstrap technique with 1000-fold bootstrapping was used for internal validation.

KEY RESULTS: Among 133 patients, the incidence of hearing loss was 31.6% (n = 42). Significant predictors for S. suis hearing loss were meningitis, raw pork consumption, and vertigo. The predictive score ranged from 0-4 and correctly classified 81.95% patients as being at risk of S.suis hearing loss. The model showed good power of prediction (AuROC: 0.859; 95%CI 0.785-0.933) and calibration (AuROC: 0.860; 95%CI 0.716-0.953).