The title snippet here is potentially misleading. From the paper:

> Households with at least one GLP-1 user reduce grocery spending by 5.3% within six months of adoption,

The reduction is only within those households using GLP-1 drugs, NOT across the US as a whole. Same for the other claims in the paper.

(That still suggests that these drugs are responsible for a 0.8% drop in total grocery spending in the US, which is remarkable.)