Peanut allergy
We included 51 studies [38, 54, 57, 58, 61, 80, 81, 92, 94, 103, 111, 114, 135-137, 139, 141, 148, 150, 154-156, 159, 180] on the accuracy of diagnostic tests for peanut allergy. For meta-analyses, 20 studies of SPT-peanut [38, 54, 57, 58, 80, 81, 92, 94, 103, 111, 136, 137, 139, 141, 148, 150, 155, 156, 159, 180] and 24 studies of sIgE-peanut [38, 44, 54, 61, 62, 67, 70, 75, 81, 90-92, 96, 112, 119, 136, 137, 141, 148, 150, 156, 159, 171, 180] met the inclusion criteria.
Studies for SPT showed a pooled sensitivity of 84% and specificity of 86% at a 4 mm median cut-off (Table 2 ). We could not detect differences in accuracy of SPT to peanut in younger age groups (Table 6 ). There were differences in data obtained in different geographical regions. Most notably there was a high specificity for SPT to peanut in Australian studies (97%) but not in Asian studies (Table 7 ). sIgE to peanut showed a pooled sensitivity of 81% and specificity 83% at a 4.3 kUA/L cut-off. In children ≤2 years of age, sIgE-peanut shows an increase in accuracy with better sensitivity and high specificity of 94%. We also observed specificity of 93% for studies from Western Europe and Australia.
Twenty-seven [31, 38, 44, 57, 58, 61, 62, 66, 67, 70, 75, 81, 90-93, 96, 101, 102, 112, 119, 139, 141, 148, 150, 154, 159] included studies employed CRD. When applying optimal cut-offs, CRD tests for peanut showed high specificity, 92% for Ara h 2-sIgE [31, 38, 44, 57, 58, 61, 62, 66, 67, 70, 75, 81, 90-93, 96, 101, 102, 112, 119, 139, 141, 148, 150, 154, 159], 93% for Ara h 3-sIgE [67, 70, 90-93, 112, 159] and 94% for Ara h 6-sIgE [31, 90, 91, 141]. In studies using ISAC [80, 81, 93, 99], the performance of Ara h 2 sIgE was less heterogeneous with a specificity of 93% using the 0.3 cut-off. The specificity of Ara h 2-sIgE increased for adult subjects. Ara h 2-sIgE was highly accurate in Northern Europe and Australia with specificity of 97% in both regions. The specificity was lower for North American subjects at 89% and was lower even for Asia subjects at 75%. Data on Ara h 8-sIgE [70, 90-92, 112, 119] and Ara h 9-sIgE [31, 70, 91, 93] was highly heterogeneous. In general, sensitivity for CRD in peanut allergy was lower than specificity. BAT to peanut was analysed in 4 studies [38, 141, 148, 150] with pooled sensitivity of 84% and high specificity of 90%. These studies were less heterogeneous and had a lower risk of bias compared to other index test studies of peanut allergy (Figure 3A).
The maximum sensitivity and maximum specificity were ≥90% for SPT to peanut, sIgE to peanut, Ara h 2-sIgE and BAT to peanut (Tables S5 and S6 ).