Summary of included studies
The systematic search identified 1,494 unique records. 149 studies [31-180] were included after application of our predefined eligibility criteria. Figure 1 llustrates the PRISmeta-analyses flowchart for the study screening and selection process. This resulted in 24,489 subjects included in the analysis. The current SR includes representative data for thirty-two countries and all continents. The data principally originate from Europe (54.3%), Asia (19.9%) and America (13.9%). Only 13.4% of eligible data are derived from multicentre studies.
The studies evaluated were predominantly prospective (63.8%) including consecutive (51.0%) and randomized (6.7%) studies. Within the retrospective studies (34.2%), most were performed consecutively (30.2%). The studies included are mostly cross-sectional (59.1%) or cohort studies (35.6%). Only 5.4% are case controls. Most studies include subjects under 18 years of age (79.2%), most of which were in infants or children ≤12 years of age (60.4%). Studies exclusive on adults represent only 7.4%. The included studies where largely performed in an allergy or paediatric clinic setting (94.6%). All included studies used OFC as a reference standard in a proportion of patients, most of them in over 70% of the subjects included. Only 7.4% of studies explicitly stated OFCs were done in all subjects. Overall, 63.8% used open OFCs and 21.5% double blind placebo-controlled food challenges (DBPCFC). The full summary of characteristics of the included studies is given in Table 1 . Further information for individual studies is compiled in Table S1 .
Nineteen different index tests were identified, most commonly sIgE (128 studies), component resolved diagnosis (CRD) (87 studies) and SPT (79 studies). Additional identified tests were SPP (15 studies), basophil activation test (BAT) (13 studies), mast cell activation test (meta-analysesT) (2 studies) [38, 149] and bead-based epitope assay (BBEA) [159]. Table S2 lists the identified index test studies. Note that studies may have more than one diagnostic test reported. Evaluable data was available for 21 foods: foremost peanut (34.2%) followed by HE (25.5%), CM (18.1%), tree nuts (12.1%), wheat (10.7%), sesame (6.0%), soy and shellfish (4.7%). All foods tested in the studies are listed in Tables S3 . The group of most frequently studied foods with three or more included publications per index test were peanut, CM, baked HE, extensively heated HE, raw HE, sesame, soy, walnut, hazelnut, cashew, almond, wheat, and shrimp.Table 2 summarizes results for each food where meta-analyses was possible.