Starting around 16:25 UTC, we were alerted that the average request handle time had increase substantially and some connections to our API were timing out. An increase in CPU-bound database queries due to outdated PostgreSQL statistics caused the issue.