Table 1 Methodology of included studies (summary).

Reference (country)	Case selection	Control selection	Exposure measurement	Confounding factors considered	Sample size	Age
Greco et al22 (Italy)	Children with CD in Campania region, Italy	Healthy children from the same region.	Interview by one of the authors	Age and area of residence, age at gluten introduction, father's occupation	201 cases, 1949 controls	Cases, mean 2.14 years, SD, 2.6; controls, 2.34, SD, 2.93
Peters et al4 (Germany)	From a CD incidence study and appeals at a national meeting.	Identified through a population registry	Questionnaire	Age, sex, number of inhabitants in area, FH of CD, age at gluten introduction	143 cases, 137 controls	Mean age 6.4 years (median 6.2)
Falth‐Magnusson et al23 (Sweden)	Children with CD in participating centres	Healthy children living in the same region selected.	Questionnaire	Age, area of residence	72 cases, 264 controls	Median age 3.1 years (range 1.4–5.1)
Ivarsson et al3 (Sweden)	Children on CD database	Selected through national population register	Questionnaire	Age, sex, area of residence, other infant feeding practices	491 cases, 781 controls	Age range of participants was 0–14.9 years
Auricchio et al5 (Italy)	Cases in participating centres who had healthy siblings	Siblings of cases without symptoms of CD	Interview	Not specifically stated but by using sibling controls, some genetic, environmental and socioeconomic factors might have been controlled for	216 cases, 289 controls	Median age at diagnosis 15 months (range 6 months to 14 years).
Ascher et al6 (Sweden)	Siblings of known CD patients found to have HLA genotype DQA1*0501‐DQB1*02 and CD on screening	Siblings with the HLA genotype but who did not have CD on small intestinal biopsy	Interview	Controlled for HLA genotype and socioeconomic factors but not for age	8 cases, 73 controls	Median age 7.9 years for cases and 7.4 years for controls