The 3% from the meta analysis above was not for a single "best case scenario fitness regime". It was a number calculated by pooling data from ~28 studies with varied methodologies. A significant portion of those studies employed dietary changes alone (no fitness regime).