Would the model here be pertinent?
I think that if you had the sample size but not the SE or SD, you could use partial-pooling to inform on the missing SD and thereby permit studies with that info missing to still contribute to the meta-analysis, no?