We gladly welcome the discussion in the commentary by Li et al. Reference Li and Chen[1]. While we appreciate the commentators’ interest in our work Reference Andalib, Emamhadi, Yousefzadeh-Chabok, Shakouri, Høilund-Carlsen and Vafaee[2], we would like to respond to the issues raised by them as follows.

It is clear that studies based on the same data source could be a potential source of bias in meta-analyses affecting both the pooled effect size and its standard error. Taking this into account, a sensitivity analysis was conducted to ascertain the robustness of the pooled results and related conclusion. Three separate meta-analyses were applied each time including data from only one of the Danish studies. As a result, negligible differences in OR and 95% CI were observed in the separate meta-analyses, compared with those reported when including all the three Danish studies.

The notion is attractive that every meta-analysis may be affected by potential confounders when using crude model; nonetheless, it should be noted that data source is chosen based on its availability in the literature and that is why crude model is the most applicable. In our meta-analysis, we emphasized that maternal depression was a joint adjusted confounder. However, there were a few eligible studies for inclusion and controlling all the potential confounders was impossible.

It is worth mentioning that our aim was not the assessment of trimester in the meta-analysis, but to demonstrate whether an association exists between maternal SSRI exposure and ASD offspring. Moreover, such trimester assessment was not applicable inasmuch as the required data was not available in all the included studies and more importantly, such subgroup meta-analysis with a few number of included studies does not provide sufficient power of test. We hope that the issue can be addressed in a future study when sufficient data is available.