What to add to nothing? Use and avoidance of continuity corrections in meta‐analysis of sparse data | Synapse