Cow’s milk allergy
We included 27 studies on accuracy of tests to support the diagnosis of CM allergy. Eleven studies for SPT to CM [37, 49, 52, 73, 84, 94, 95, 106, 134, 156-158, 167, 180], 5 for SPP using fresh CM [37, 49, 157, 167, 180] and 3 for SPT to casein [37, 49, 52, 73] met the inclusion criteria for meta-analyses. SPT to CM and SPT to casein showed sensitivities of 52% and 64% and specificities of 80% and 87% respectively. SPP showed a high sensitivity of 90% with specificity of 80% at the 4 mm cut-off (Table 2 ). CM-sIgE [37, 51, 52, 73, 94, 95, 97, 106, 134, 143, 152, 156, 157, 167, 178, 180] showed pooled sensitivity of 82% with a high specificity of 92% at the 3.5 kUA/L cut-off. CRD also showed high specificity: casein 93% [32, 37, 51, 52, 73, 127, 134, 167] and alpha-lactalbumin 92% [51, 52, 73, 167] with sensitivities of 67% and 58% at 1.8 kUA/L and 1.7 kUA/L cut-offs, respectively. Studies of CRD were less heterogeneous compared to SPT and sIgE to CM (Figure 2C). For CM, maximum sensitivity was ≥ 90% for SPP and maximum specificity was ≥ 90% for SPT-casein, sIgE-CM and CRD (Tables S5 and S6 ). We could not determine changes in accuracy for index tests on CM for various age groups or geographical regions. There was insufficient data on baked CM for meta-analyses.