Thirty-four studies (13,134 patients) were included. Some studies assessed more than one type of EMA test.
Most studies did not provide sufficient details to determine whether ascertainment bias had been avoided (i.e. whether the EMA or tTG tests were used to identify coeliac patients for inclusion). The authors stated that most control groups consisted of patients in whom coeliac disease was suspected.
The sensitivities of both the tTG antibody and EMA tests ranged from 70 to 100%. The specificity of the tTG test ranged from 91 to 100%, and that of the EMA test ranged from 90 to 100%. The pooled sensitivity for the EMA test (34 studies) was 93% (95% CI: 92.1, 93.8) and the pooled specificity was 99.7% (95% CI: 99.5, 99.8). The estimates were similar when analysed separately for studies that used monkey oesophagus and human umbilical cord, and for adults alone. The pooled sensitivity for all tTG studies (42 studies) was 92.8% (95% CI: 91.9, 93.6) and the pooled specificity was 98.1% (95% CI: 97.8, 98.4). Recombinant tissue tTG (rhtTG) (19 studies) showed slightly higher pooled sensitivity and specificity than guinea pig tTG (gptTG) (23 studies): the pooled sensitivity was 93.8 (95% CI: 92.8, 94.7) for rhtTG and 90.4% (95% CI: 88.8, 91.9) for gptTG; the pooled specificity was 98.7% (95% CI: 98.5, 98.9) for rhtTG and 92.4% for gptTG (95% CI: 90.8, 93.8).
Head-to-head comparisons of EMA and tTG showed that EMA had higher sensitivity in 48% of studies and higher specificity in 62% of studies, while tTG had higher sensitivity in 28% of studies and higher specificity in 17%; in all other studies the estimates were equal. When only rhtTG was considered, EMA showed better sensitivity in 28% of studies and better specificity in 56%, while rhtTG showed better sensitivity in 44% of studies and higher specificity in 22%; in all other studies the estimates were equal.