The question is what average; some people apparently view "in the US" as implying US population-level averages (which it does not explicitly imply), whereas authors report the average within adopting households, which for this study's data source, all happen to be in the US
>>a123b4+(OP)
If the claim was just that grocery spending is down 5.3% across the country they wouldnt have said average, the title would just be "Ozempic reduced grocery spending by 5.3% in the US"