How does your team currently handle stale statistics in published articles?