The design appears to be based on the notion that most of the confounders are washing out in the propensity to seek cardiac treatment, which was a precursor for recruitment into the study. I agree this doesn't mitigate all confounding concerns. Given the design where patients are measured at baseline and the endline event measure is measured later, I'd like to see a balance table at the baseline measures.

One possible pitch: I'm also not sure whether the circulating blood levels of this chemical are actually connected to consumption of the sweetener. You could imagine a world where two individuals who have identical consumption of the sweetener but different free-circulating levels of the chemical, the higher levels of the chemical could be an indication of other confounding health issues causing malabsorption.

I am not a biologist or doctor and don't know anything about sweeteners or any of the mechanisms that may or may not play a role here, just commenting on design.