Provided you are using a random effects model, the estimates are adjusted to account for the heterogeneity so yes you can continue.

You can/should investigate the reasons for the heterogeneity. This can be qualitative (risk of bias, methodological differences, different follow up duration, inclusion criteria, outcome differences, etc) or quantitative (meta-regression).

I see, that's good to hear. Thank you very much for your reply!

You have statistically significant and relatively large (in magnitude) heterogeneity. Your emphasis should be on explaining that heterogeneity using subgroup analysis and meta-regression to test moderator variables that might explain the heterogeneity, at least some of it. You should pay little if any attention to the overall effect.

Thank you for your insight!