Results
During the study period, 510 suspect meningitis cases examined
on arrival at health structures in the two regions were
included in the clinical trial and randomised for treatment,
495 (97.1%) of whom had a successful lumbar puncture. Of
these 495 CSF samples, 494 (99.8%) had either PCR or culture,
491 (99.2%) had direct microscopy (WBC count and
Gram stain) and 488 (98.6%) were tested using the Pastorex®kit (Table 1). Of the 494 samples tested by PCR and/or
culture, 352 were positive for N. meningitidis (of which
350 were serogroup A; data not shown). Figure 1 gives the
schematic diagram for the initial microscopy tests followed
by the tests performed for serogroup confirmation.
After exclusions due to contamination, or uninterpretable
or inconclusive results, the total number of samples
with positive or negative results for serogroup A, for
both the Pastorex® test and ‘gold standard’ (culture and/or
PCR), was 484 (Table 1). For direct microscopy followed
by Pastorex®, after excluding 207 samples with doubtful
microscopy results and 19 with discordant microscopy and
Pastorex® results, the number of samples remaining for supplementary
analysis was 263 (Figure 1).
Table 2 shows the results for the sensitivity, specificity,
PPV and NPV for the Pastorex® test alone and the Pastorex®
test with prior direct microscopy compared with the gold
standard. The sensitivity for the Pastorex® test alone was
88% (95% CI 85—91%). With prior microscopy, although results
were available for fewer samples, sensitivity improved to
97% (95% CI 94—99%). Our data suggest an excellent PPV for
the Pastorex® test alone (97%; 95% CI 95—98%). The NPV for
Pastorex® alone was 75% (95% CI 71—79%), which improved
to 88% (95% CI 85—92%) with prior direct microscopy.
The rate of misclassification (positive and negative) for
the Pastorex® test alone was 11% (95% CI 8—14%), falling to
5% (95% CI 2—8%) for samples with prior direct microscopy