The effects of oxytocin administration on social and routinized behaviors in autism: A preregistered systematic review and meta-analysis

Abstract

Oxytocin administration has demonstrated considerable promise for providing individualized support for autistic people. However, studies evaluating the effects of oxytocin administration on autistic characteristics have yielded inconsistent results. This systematic review and meta-analysis investigates the effect of oxytocin administration on social and routinized behaviors in autism using recently developed methods to accurately assess potential impact of effect size dependency and publication bias. Our frequentist meta-analysis yielded a significant summary estimate for the effect of oxytocin administration on social outcomes in autism (d = 0.22, p < 0.001). The summary effect size estimate for routinized behavior outcomes was not statistically significant (d = 0.14, p = 0.22), with follow up tests indicating that the effect size estimate was not either statistically equivalent (Z = -1.06, p = 0.2), assuming a smallest effect size of interest of 0.25. Frequentist and Bayesian assessments for publication bias, as well as results from Robust Bayesian meta-analysis of oxytocin effects on social outcomes in autism, indicated that the summary effect size might be inflated due to publication bias. Future studies should aim to reduce bias by preregistering analysis plans and to increase precision with larger sample sizes.