Statistical Methods
Data were analysed using STATA version 12. Odds ratios (OR), crude (COR) and adjusted (AOR), were produced by logistic regression. We performed analysis for macrolides as a class, then for each of the five subclasses of macrolides. Registries without any exposures for a specific macrolide subclass were excluded from the analysis of that subgroup. Each investigation of an association between a macrolide and a case CA subgroup made use of the four comparison groups i.e. the two control groups (non-genetic controls and genetic controls) and the two exposure comparison groups i.e. no antibiotics exposure (primary exposure comparison) and exposed to other medications (secondary exposure comparison). We adjusted for the following confounders: year of birth in three time periods (1995 – 2000; 2001 – 2006; 2007 – 2012), EUROCAT registry, and maternal age in categories (<20; 20-24; 25-29; 30-34; 35-39 and 40+).
We divided our analyses into signal testing analyses regarding the five subgroups of CA shown in Table 1, and exploratory analyses regarding the non-macrolide antibiotic signal CA subgroups. For the latter, all results are given in supplementary files, and only statistically significant results with at least 3 exposed cases are given in the main tables.
We considered a result to be robust if AOR were statistically significant in two or more control and exposure comparison groups. Only results involving the primary exposure comparison group are presented (results involving secondary exposure comparison are found in the Supplementary Tables S4b-9b).
As our primary controls (non-genetic controls) are malformed babies, there is a possibility that some CA in this control group could be associated with the macrolides under investigation (teratogen non-specificity bias), leading to reduced OR. 36,37 As a sensitivity analysis, we therefore investigated the association between non-genetic control anomalies and macrolides using the secondary controls (genetic controls) as the comparison group. These results are given in the Supplementary Table S3.