Why did Peter discontinue the ketogenic diet? And what's his dietary strategy for 2018? (AMA #1)

On his representative day: woke up, rode the Peloton for 45 minutes, made coffee, worked, saw patients — and did not eat until he returned to his apartment at 2:00 PM. His first meal was the 'Peter Kaufman' high-fat shake. This is a practical example of informal time-restricted feeding maintained through habit rather than rigid scheduling.

Attia uses this rule as his single most portable dietary principle post-keto. He describes it in the context of what he tells his children about food — 'that didn't grow on a tree, that didn't you know come out of you know like that's fake food right' — framing it as a food literacy lesson, not just a personal rule. The rule allows for potatoes, rice, lamb, salmon, fruit, and most restaurant food while excluding crackers, protein bars, packaged snacks, and cereals.

Attia's specific example: on his representative day, peak post-dinner glucose was 117 mg/dL about an hour after a dinner of smoked salmon, Greek salad, lamb with lemon potatoes. By bedtime, glucose was 98 mg/dL. Peak nighttime glucose was 95 mg/dL — occurring during sleep, not traceable to food. He explicitly notes: 'I almost always see my highest glucoses at night they're almost always I think in response to cortisol.' A peak nighttime glucose of 95 makes him 'really happy' — suggesting his personal target for overnight metabolic health.

The shake is named 'the Peter Kaufman' — presumably a friend or patient, though the name origin is not explained in the transcript. Attia notes his co-host was 'salivating' for it. Despite the episode being about transitioning away from keto, this shake is actually quite low-carb: the only carb sources are almond butter (modest), frozen strawberries (minimal), and whatever is in the protein powder. The heavy cream contribution makes this calorically dense and fat-forward even outside of formal ketosis.

Attia's representative day dinner: smoked salmon appetizer, a 'big ass Greek salad in a bowl the size of my head,' lamb entree with lemon potatoes. No dessert. He notes he 'definitely like potatoes and I definitely like rice' and will eat them 'with at least one of my meals' most days of the week. This is a meaningful shift from his ketogenic period when he could not eat potatoes or rice without exiting ketosis. He frames himself as 'kind of a boring eater' — the variety is across protein source, not cuisine style.

Attia explicitly regrets not taking muscle biopsies during his three-year ketogenic period — he believes those biopsies could have provided unique insight into the cellular adaptations happening in muscle tissue during long-term nutritional ketosis at high training volumes. He notes he was 'about as good a responder as you're gonna see' but acknowledges many people do not respond nearly as well — suggesting his 18-month recovery timeline may represent a best-case scenario rather than a typical one.

Attia frames this as a question of intellectual humility: 'no offense to the keto community out there who I'm probably pissing off at the moment but I'm highly put off by the this view that like ketosis is for everybody.' He explicitly acknowledges he is one of the best responders he has encountered across all the patients and colleagues he has seen on ketogenic diets — and even for him, the diet required eating a full tub of sour cream per day to hit 4,500 calories and became socially difficult with his children.

The CGM data he shares from a representative day illustrates the approach: peak glucose of 117 mg/dL about an hour after dinner (smoked salmon, Greek salad, lamb with lemon potatoes), returning to 98 mg/dL by bedtime, peak nighttime glucose of 95 mg/dL. He notes the nighttime peaks are almost always cortisol-driven, not food-driven — a mechanistic insight he validates by watching the timing pattern on CGM. The simplicity of the new framework — 'just try not to eat junk' — contrasts sharply with the rigid macro tracking of therapeutic ketosis.

On his representative day, Attia's peak glucose of 117 mg/dL came about an hour after dinner. By bedtime it had dropped to 98 mg/dL. But his peak nighttime glucose was 95 mg/dL — appearing during the night, not traceable to food. He specifically tracks 'peak nighttime glucose' as a separate metric from peak post-meal glucose, treating them as different signals about different physiological processes.

Attia mentions that Quest Nutrition had partnered with the Charlie Foundation (supporting children with epilepsy on ketogenic diets) to produce pre-made 2:1, 3:1, and 4:1 ketogenic meals — and that he tried them. He says: 'I could mainline those things if someone were like feeding me those all day.' This underscores that the primary barrier to sustained therapeutic ketosis is practical food preparation, not metabolic tolerance.