Forty three studies met the inclusion criteria (n=17,179, range 56 to 1,468). Investigators were blinded in 12 studies, intermediate test results were reported in 17. Twenty-seven of the 37 studies that used cytotoxicity assay as the reference standard, 27 reported sufficient details for the test to be reproduced.
Compared to cytotoxicity assay:
Well-type EIAs that detected toxins A and/or B (22 studies) had a mean sensitivity of 82% (95% CI 79% to 84%) and specificity of 97% (95% CI 97% to 98%).
Membrane-type EIAs that detected toxins A and/or B (24 studies) had a sensitivity of 72% (95% CI 69% to 74%) and specificity of 98% (95% CI 97% to 98%).
Well-type EIAs that detected glutamate dehydrogenase (GDH) (two studies) had a sensitivity of 93% (95% CI 88% to 97%) and specificity of 89% (95% CI 87% to 91%).
Membrane-type EIAs that detected GDH (eight studies) had a sensitivity of 90% (95% CI 87% to 92%) and specificity of 90% (95% CI 88% to 90%).
Real-time polymerase chain reaction (three studies) had a sensitivity of 92% (95% CI 83% to 97%) and specificity of 96% (95% CI 94% to 98%).
Compared to toxigenic culture:
Well-type EIAs that detected toxins A and/or B (seven studies) had a mean sensitivity of 66% (95% CI 61% to 71%) and specificity of 98% (95% CI 97% to 99%).
Membrane-type EIAs that detected toxins A and/or B (seven studies) had a sensitivity of 52% (95% CI 47% to 57%) and specificity of 98% (95% CI 97% to 99%)
ImmunoCard C. difficile (one study) had a sensitivity of 60% (95% CI 32% to 83%) and specificity of 76% (95% CI 68% to 83%).
Real-time polymerase chain reaction (one study) a sensitivity of 86% (95% CI 73% to 95%) and specificity of 97% (95% CI 94% to 99%).
Prevalence rates:
At a prevalence of 5%, PPVs ranged from 0.12 to 1.0 and NPVs from 0.97 to 1.0; at 50%, PPVs ranged from 0.71 to 1.0 and NPVs from 0.62 to 0.96.
Further results were reported.