Case Study 1 — Two People, Same Diet, Opposite Years

A clinical case. Both patients are illustrative composites; the pattern is one I see constantly.

Setup

The most important number in DIETFITS was not the difference between groups. It was the spread within them — some participants lost thirty kilograms and some gained, on the same assigned diet, with the same counselling.

That number is abstract until you meet two of the people it describes.

Marguerite Oyelaran, 44, hospital administrator, two teenagers at home. Dominic Sarno, 51, self-employed builder, lives alone, eats out most lunchtimes.

Both started a whole-food Mediterranean-style pattern in the same month, with the same written plan, from the same clinic. Same targets: vegetables at two meals, legumes four times a week, olive oil as the cooking fat, fish twice a week, less refined snacking.

Twelve months later:

Same diet. Same instructions. Opposite outcomes.

If you believe the diet is the intervention, this result is inexplicable. If you believe §10.3, it is entirely predictable — and the interesting work is finding out what actually differed.

What differed

Not their biology, as far as anyone could measure. Similar age bracket, similar starting weight category, similar baseline diet quality, neither with a diagnosed metabolic condition beyond borderline lipids.

Here is what actually differed, in the order it mattered.

1. Who else was eating

Marguerite cooks for four. Her teenagers complained for about two weeks and then stopped, because the food was good and because they were fifteen and seventeen and had no alternative supply. Legumes went into things that already existed — chilli, pasta sauce, soup — rather than appearing as a separate dish anyone had to be persuaded about.

Critically: her household changed with her. She wasn't running a personal regime inside a kitchen that was doing something else.

Dominic eats alone at home and on site at lunch. His lunch is whatever is available near a job, and the job moves. He cooked in the evening for three weeks, found himself throwing away vegetables that had gone off because a job overran, and stopped.

The plan assumed a kitchen he was reliably in.

2. Where the failure point was

Marguerite's hard part was evenings — tired, teenagers, 6:30. Her adaptation was a rotation of four meals she could make without thinking, plus a standing Sunday batch of a legume-based thing that went into three lunches.

Dominic's hard part was lunch, five days a week, in a different location each time. He had no adaptation for that, and nobody had asked him where his hard part was.

The plan addressed the hard part of a life that wasn't his. It was a good plan, written by good people, for someone with a kitchen and a schedule.

3. What the first bad week did

Both had a bad week around week six — Marguerite's mother was ill; Dominic had a job overrun and ate takeaway five days running.

Marguerite treated it as a week. She'd been told to expect one, and her plan had a floor: if everything falls apart, the minimum is a tin of beans, a bag of frozen vegetables, and eggs. She used the floor for six days and resumed.

Dominic treated it as the end. He hadn't been given a floor, so a bad week read as evidence that the plan didn't fit — which, per §10.9, is a completely reasonable inference from the information he had.

4. Feedback timing

Marguerite weighs herself weekly and noticed a downward trend by week five, which sustained her through the bad week.

Dominic doesn't own scales and noticed nothing. Seven weeks of effort with no perceptible feedback, followed by a bad week, is a predictable stopping point — and the absence of feedback was a design oversight, not a character trait.

What we did differently the second time

Dominic came back fourteen months later. This time we designed around the constraint rather than around the diet.

Nothing about the macronutrient composition changed. It was still the same Mediterranean-ish pattern. What changed was that it now fitted a builder who eats lunch on site.

At nine months: −7 kg, triglycerides 184 → 122, and still going.

Analysis

1. The within-group spread is not noise. It is the whole signal, and it is made of things like whether your teenagers have an alternative food supply and whether your job overruns. Nutrition trials report it as variance; a clinician has to read it as information.

2. Ask where the hard part is, before you write a plan. Marguerite's was evenings; Dominic's was lunch. The same plan addressed one and ignored the other, and that single mismatch accounted for more of the outcome difference than anything in the food.

I now ask every patient one question before anything else: "Walk me through a bad day. Not a normal day — a bad one." The plan gets built for that day.

3. A floor beats a plan. Marguerite had one and used it; Dominic didn't and stopped. The intervention that survives is the one with a defined minimum, because the minimum is what actually runs during the weeks that determine whether you're still doing this in a year.

4. Household context is close to decisive. Marguerite's household changed with her. Dominic's context — a moving job site — could not be changed, so the plan had to. Individual dietary advice delivered into a household that isn't participating has a poor prognosis, and this is rarely assessed.

5. Dominic's first attempt was not a failure of adherence. It was a failure of design, and he carried it as a personal one for fourteen months. That's the cost §10.9 is about, and it's invisible in every trial.

Discussion Questions

1. DIETFITS reported enormous within-group variation and no between-group difference. If the variation is made of things like household composition and job schedule, what does that imply about what nutrition trials are actually measuring?

2. "Walk me through a bad day" is offered as the first question. What would you ask second? Design a five-question intake that predicts adherence better than any biological measure.

3. Dominic's second attempt kept the same macronutrient composition and changed only the logistics. Is that still "the same diet"? What does your answer imply about how diets should be named and compared?

4. Marguerite's household changed with her; Dominic's couldn't. How should a clinician handle a patient whose household is actively unsupportive — a spouse who cooks differently, or resents the change?

5. The floor — tinned fish, tinned beans, frozen veg, eggs — is arguably the most valuable element in Dominic's second plan. Why is it so rarely prescribed? What does it cost a clinician to say "here is your minimum, and it counts"?

Your Turn

Whatever you're currently doing or planning, answer these four before anything else:

1. Where is my hard part? Not "eating better" — the specific occasion. Weekday lunch? 9 p.m.? Weekends? Work travel? Being tired at 6:30? Be specific enough to name the time and place.

2. Who else is eating? Does your household change with you, or are you running a private regime inside a kitchen doing something else? If the latter, what's the plan for that?

3. What is my floor? Write the actual minimum — three or four items, requiring no cooking skill and no fresh ingredients — that you will fall back to when everything goes wrong. Write it down now, while things are fine. Nobody designs a floor during a bad week.

4. What feedback will I get, and when? If the answer is "none for three months," you have designed a plan that will stop at week seven. Pick something — weekly weight average, a waist measurement, a lab date, how a specific pair of trousers fits — and set the interval.

Most plans fail at question 1 or question 3. They are the two nobody writes down, and they are the two that determine whether you are Marguerite or Dominic.