In 121 people with cirrhosis, the biggest fat eaters died more often over five years. The headline figure is 'eight times the risk', and the group it is measured against had 10 deaths

Cirrhosis is the scarring left when the liver has been injured for years, by a hepatitis virus, alcohol, fat build-up or the body's own immune system. Diet advice for it has mostly been about getting enough: enough calories, enough protein, no long gaps without food, because so many patients lose muscle and weight. What kind of fat they should eat has hardly been studied. A team of nutrition researchers and liver specialists in Iran has now looked, in a paper published online in Experimental Physiology on 28 September 2026.

Between 2016 and 2018 they enrolled 166 outpatients at two teaching hospitals in Tehran, all diagnosed with cirrhosis in the previous six months. Dietitians sat down with each one and went through a 168-item questionnaire on what they had eaten over the past year. That was the only time diet was measured. The patients were then phoned once a year until April 2023. After setting aside people with implausible calorie counts, incomplete forms, an extreme body mass index or a cancer found in the first year, 121 were left (83 men and 38 women, average age 55). Fifty of them, 41%, died during follow-up.

The deaths were not spread evenly. Split into thirds by total fat, 11 people died in the lowest third and 23 in the highest. For saturated fat it was 10 against 26. For omega-6 fat, the kind in sunflower, corn and soybean oil, it was 6 against 29. After adjusting for age, sex, calories, body mass index, smoking, alcohol, the cause of the cirrhosis and two scores of how advanced it was, the paper puts the risk of dying in the top third at 8.2 times that of the bottom third for saturated fat, 6.7 times for total fat and 5.2 times for omega-6. Monounsaturated fat, the main fat in olive oil, showed no link either way, and neither did omega-3.

Risks that large are rare in diet research, and the authors say so. Every one of them comes with a very wide margin of error: for saturated fat the plausible range runs from 1.3 times to more than 20 times, which is what happens when the comparison group contains ten deaths. They did no sample size calculation, they say the models may be overfitted, and they describe the result as hypothesis-generating throughout. There is also something odd in who ate what. The people eating the least fat were eating far less of everything, about 1,600 calories a day against 2,500, were mostly women, and were sicker on paper: 38.5% had moderate or severe cirrhosis on the Child-Pugh scale against 18.5% of the top third. Yet they died less. Adjusting for calories and for how advanced the disease was is what pushed the total-fat figure from 2.3 up to 6.7, so the eye-catching numbers depend heavily on the statistical model, in a study with few deaths to feed it.

A few things in the tables are worth flagging, and this reading is ours, not the paper's. In this kind of analysis the estimate normally sits at the geometric middle of its range. For total fat the range given is 1.37 to 13.34, whose middle is about 4.3, not the 6.69 reported; saturated fat has the same mismatch (middle about 5.2, reported 8.2). The limitations section says protein intake was not adjusted for, while the appendix shows a model that did adjust for it, in which the total-fat figure falls to 3.4. And the omega-6 intakes look too small to be the whole picture: the top third for total fat averaged 4.4 g of omega-6 and 1.2 g of omega-3 a day, but 20.6 g of polyunsaturated fat in all, when those two families normally account for nearly all of it. None of this means the link is wrong. It is a fair reason to wait for the larger studies the authors ask for before anyone with cirrhosis cuts fat, and with it calories, on the strength of this one. The study had no funding and the authors report no conflicts of interest.