Sensitivity analysis
Sensitivity analyses for all significantly elevated risks using
penicillins as the exposure comparison group, gave AOR similar or
greater than that obtained for no antibiotics (primary exposure
comparison), although with less precision (Supplementary Table S10). The
study results were also not altered when we excluded the three
registries with a high proportion of excluded registrations due to
unknown antibiotic timing from the analyses.
The overall risk for non-genetic controls as a group was not raised
compared to genetic controls (AOR 1.01; 95%CI 0.69-1.46), confirming
absence of strong macrolide associations in this control group. With
regard to specific CA in the non-genetic control group, we found an
increased risk with macrolide exposure for teratogenic syndromes with
malformations (AOR 6.50; 95%CI 1.92-22.03) explained by its subcategory
maternal infections resulting in malformations (OR 6.44; 95%CI
1.89-21.92) (Supplementary Table S3). These are recognised infection
syndromes such as congenital rubella/CMV/toxoplasmosis, where antibiotic
use may be expected, but numbers are very small. Results from further
analysis after excluding these teratogenic syndromes from the controls
were similar to the original study results.
The study results were similar when the analyses were restricted to case
subgroups with isolated CA, and when genetic controls with signal
anomalies were excluded.