Can I put my type 2 diabetes into remission with food, and which food?

Both men are addressing an adult with established type 2 diabetes, which is unusually clean for a disagreement like this: they are not two experts talking past each other about different people. Two real splits remain.

First, medication: Fung's protocol is written for insulin-treated patients and titrates insulin down before the first fast, so it is a supervised clinical protocol from the start; Barnard's diet is written for a general diabetic audience and his supervision sentence sits on a page almost nobody reaches.

Second, evidence tier: Barnard's position rests on randomised trials with objective endpoints, Fung's on mechanism, on a critique of drug trials, and on a case series of three. That asymmetry is the most important thing a reader should see and it does not resolve the question, because Barnard's own 74-week trial lost significance on both weight and HbA1c, which is the durability problem Fung's argument is built to attack.

More than either headline suggests, and the agreements are the actionable part. Both tell you to stop drinking sugar and to cut refined starch: Barnard's diabetes guidance leads with cutting added sugars and sugar-sweetened beverages (bar-cut-added-sugar, Established), and Fung's first step is little or no refined carbohydrate (fun-cutsugar, Established). Both reject calorie counting and portion control as the mechanism.

Both treat type 2 diabetes as a condition that can improve rather than only progress. And both attach the same safety instruction to their own protocol, which is the single most useful thing in this block: Fung publishes it in capitals on his own site, Barnard said it in 2008. Neither disputes that someone on glucose-lowering medication must have a prescriber watching.

For Barnard: a randomised trial of a low-fat vegan diet in type 2 diabetes running two years or longer that holds a significant between-group HbA1c advantage, which is the comparison his own 74-week extension failed.

For Fung: a randomised controlled trial of intermittent fasting against an isocaloric comparator in type 2 diabetes with remission as a pre-specified endpoint, which would move his position from mechanism plus case series onto the tier Barnard's already occupies. Both sides would move on head-to-head evidence, which does not exist: no trial was found comparing a low-fat vegan diet against a fasting protocol in the same population.