Sorry, I think there is a misunderstanding. I'm talking about the studies being analysed. They must provide both effect signs, or they cannot be included in the analysis. Therefore, it's impossible to meta analyse studies reporting positive effects only, as is commonly donewith other methods?