Study setting | Study design | Inclusion criteria | Exclusion criteria | Sample size | Level of evidence (OCEBM) | Key results | ||
Berkley et al 6 | Rural district hospital, Kenya | Cross sectional | All paediatric patients older than 60 days admitted during 13-month study period. | 4582 patients (91 bacterial meningitis cases) | 2 |
Neck stiffness
LR+ 13.3, LR− 0.74 | ||
Lehmann et al 4 | Rural hospital, Papua New Guinea | Cross sectional | Children aged 1–59 months admitted during normal working hours with suspected meningitis. | CSF not collected for analysis. | 697 | 2 |
Neck stiffness
Sensitivity 53%, specificity 81% PPV 39% [NPV 80%] | |
Offringa et al 11 | Urban hospital, Netherlands | Cross sectional | All children between 3 months and 6 years of age who presented with first episode of seizure with fever at two urban hospitals during 2-year period. | 309 (23 cases) | 2 |
Nuchal rigidity
OR 128 if ‘definite’ OR 2.1 if ‘dubious’ Combined history and examination features to define ‘major features of meningitis’ LR+=∞ LR−=0.3 PPV 100% NPV 98% | ||
Curtis et al 2 | Global | Systematic review of prospective studies | Studies that described history and examination features of children with meningitis confirmed on LP compared with an LP-negative comparison group. Data collected prospectively and amenable to calculation of accuracy estimates. Not pretreated with antibiotics. | Not prospective. No comparison group. No data or mixed data. Pretreated with antibiotics. | 10 studies | 1 |
Meningeal signs (any one of NR, KS or BS) Sensitivity 64%, specificity 89% LR+ 4.5 (2.4–8.3), LR− 0.41 (0.3–0.57) Neck stiffness Sensitivity 51%, specificity 89% LR+ 4 (2.6–6.3), LR− 0.56 (0.43–0.72) |
Kernig’s sign
Sensitivity 53%, specificity 85% LR+ 3.5 (2.1–5.7), LR− 0.56 (0.41–0.75) Brudzinski’s sign Sensitivity 66%, specificity 74% LR+ 2.5 (1.8–3.6), LR− 0.46 (0.31–0.68) |
Bilavsky et al 7 | Urban hospital, Israel | Prospective cohort study | All children aged 3 months to 17 years who were diagnosed with bacterial meningitis in a 4-year period. Control group were children who were suspected to have bacterial meningitis but ruled out on LP. | Antibiotic therapy before arrival to ED. | 86 (40 in study group) | 2 |
Nuchal rigidity
Sensitivity 64.9%, specificity 53.5% LR+ 1.39, LR− 0.66 Brudzinski’s sign Sensitivity 52.6%, specificity 77.5% | LR+ 2.34, LR− 0.61 Kernig’s sign Sensitivity 51.3%, specificity 95% LR+ 10.27, LR− 0.51 |
Amarilyo et al 5 | Urban tertiary hospitals, Israel | Cross sectional | Included patients with clinically suspected meningitis between the ages of 2 months and 16 years. | Severe chronic disease, immune deficiency or any neurological condition, patients in whom LP was not performed. | 108 (58 cases of meningitis, 6 confirmed bacterial) | 2 |
Nuchal rigidity
Sensitivity 65%, specificity 67% PPV 0.8 [NPV 0.62] Brudzinski’s sign Sensitivity 51%, specificity 80% | PPV 0.81 [NPV 0.58] Kernig’s sign Sensitivity 27%, specificity 0.87 PPV 0.77 [NPV 0.51] Prevalence 53.7% |
Predictive values in square brackets were not reported by the original study but have been calculated by the authors for the purposes of this review. 95% CIs are given in brackets.
CSF, cerebrospinal fluid; ED, emergency department; LP, lumbar puncture; LR+, positive likelihood ratio; LR−, negative likelihood ratio; NPV, negative predictive value; OCEBM, Oxford Centre for Evidence-Based Medicine; PPV, positive predictive value.
See appendix for search strategy used to identify studies for inclusion (online supplementary file 1).