METHODS: We searched nine international biomedical databases for published, in-progress, and unpublished evidence. Studies were independently screened by two reviewers against predefined eligibility criteria and critically appraised using established instruments. Our primary outcomes of interest were symptom, medication, and combined symptom and medication scores. Secondary outcomes of interest included cost-effectiveness and safety. Data were descriptively summarized and then quantitatively synthesized using random-effects meta-analyses.

RESULTS: We identified 5960 studies of which 160 studies satisfied our eligibility criteria. There was a substantial body of evidence demonstrating significant reductions in standardized mean differences (SMD) of symptom (SMD -0.53, 95% CI -0.63, -0.42), medication (SMD -0.37, 95% CI -0.49, -0.26), and combined symptom and medication (SMD -0.49, 95% CI -0.69, -0.30) scores while on treatment that were robust to prespecified sensitivity analyses. There was in comparison a more modest body of evidence on effectiveness post-discontinuation of AIT, suggesting a benefit in relation to symptom scores.