Can people with schizophrenia or bipolar disorder keep to a ketogenic diet? In a Queensland trial, 26 of 40 assigned to it finished 12 weeks, and those who did spent 70% of it in ketosis. Whether their symptoms improved is still unreported

Ketogenic diets for serious mental illness have gathered plenty of enthusiasm and a shortage of controlled data. Before anyone can test whether the diet helps symptoms, there's a more basic question: can people living with these conditions, often on several medicines, keep to something this strict in their own kitchens? The North Queensland Dietary Intervention Trial in Townsville, Australia, set out to find out.

Adults with schizophrenia, schizoaffective disorder or bipolar disorder were randomised to a ketogenic diet or to the Australian Guide to Healthy Eating. Both groups had weekly contact with a dietitian and practical help with shopping and cooking, but they prepared their own food; no ready meals were delivered. The trial enrolled 71 people. The headline number in the abstract is 52 completers, split evenly at 26 and 26, and that even split hides something.

Allocation wasn't even: 40 people were assigned to the ketogenic diet and 31 to the guideline diet. So 65% of the keto group completed the 12 weeks, against 84% of the comparison group. Among those assigned to keto, three never started it, two were removed for not adhering or attending, and nine withdrew. Five of those nine couldn't stay in ketosis. Others had reasons that any real-world trial can recognise: a needle phobia that made daily blood tests impossible, a move out of the area, domestic violence unrelated to the participant's illness. One left on day three when a flood destroyed their kitchen.

Those who stayed did remarkably well on the measures the trial tracked. On average they spent 70% of the 12 weeks with blood ketones of at least 0.5 mmol/L, and 19 of the 26 were in ketosis on more than 80% of measurements. They attended 92% of dietitian appointments, against 85% in the comparison group. Counting everyone who began the diet and had the chance to show adherence, 30 of 36 (83%) managed at least partial adherence for as long as they took part, which is a fairer number than the completer figure. BMI fell by 6.5% on keto and 2.7% on the guideline diet, though the authors treat that as a check on adherence, not as a result in itself.

Gut symptoms were common in both groups and mostly settled by week six. Among completers, headache was reported by 35% on keto and 15% on the comparison diet, and nausea by 31% against 19%; hunger went the other way, 15% against 38.5%. None of those differences was statistically significant, in a trial this small, and no serious adverse events were recorded in either group. Nearly everyone completing the trial, 94%, was on at least one psychiatric medicine at the start.

Here is what this paper doesn't say. It reports nothing on psychosis, mood, cognition or metabolic health; those results are promised separately, and the authors state that no conclusions on efficacy can be drawn. The people who sign up for a trial like this are likely to be more motivated than most, and they had support few clinics could offer. Funding came from philanthropy, including the Baszucki Brain Research Fund, and one author has consulted for the Baszucki Group among others. So the answer to the narrow question looks like a qualified yes, for some people, with a lot of help. The bigger question is still open.