Low-carb versus low-fat: what the weight-loss trials really found

For weight loss, low-carbohydrate and ketogenic diets do not beat low-fat diets by a meaningful margin once trials run long enough — the best head-to-head randomised evidence finds the two roughly tie [s1][s2]. Where low-carb eating does show a distinct signal is in blood lipids and, in some trials, glycemic control, which is a narrower and more defensible claim than the weight-loss promises the approach is usually sold on [s2].

The trial built to settle the weight question

The DIETFITS trial, published in JAMA in 2018, randomised 609 adults aged 18 to 50, without diabetes and with a body-mass index between 28 and 40, to a healthy low-fat or a healthy low-carbohydrate diet for 12 months [s1]. Both groups attended 22 small-group sessions and were coached to hit the lowest fat or carbohydrate intake they could sustain while emphasising food quality — vegetables and whole foods rather than processed "low-carb" or "low-fat" products [s1].

At 12 months the diets had genuinely diverged in composition: carbohydrates made up 48% of intake on the low-fat diet versus 30% on the low-carb diet, and fat 29% versus 45% [s1]. Weight change was −5.3 kg on the low-fat diet and −6.0 kg on the low-carb diet, a between-group difference of 0.7 kg with a 95% confidence interval from −0.2 to 1.6 kg [s1]. That interval crosses zero: the difference was not statistically significant.

DIETFITS was also designed to test a popular idea — that people can be matched to the "right" diet by biology. It looked at three genotype patterns and at insulin secretion measured after a glucose challenge [s1]. Neither predicted who did better: there was no significant interaction between diet type and genotype pattern (P = .20) or between diet and insulin secretion (P = .47) for weight loss [s1]. The premise behind much personalised-nutrition marketing failed its own test in the trial built to check it.

The longer trial, and where a real difference shows

The two-year DIRECT trial, published in 2008, adds the metabolic half of the picture. It randomised 322 moderately obese adults to a low-fat calorie-restricted diet, a Mediterranean calorie-restricted diet, or a low-carbohydrate diet with no calorie restriction [s2]. Adherence was unusually good for a diet trial — 95.4% at one year and 84.6% at two [s2].

Weight loss again clustered together: 2.9 kg on the low-fat diet, 4.4 kg on the Mediterranean diet and 4.7 kg on the low-carbohydrate diet over two years; among the 272 who completed the trial the figures were 3.3, 4.6 and 5.5 kg [s2]. The low-carb edge exists but is small.

The lipid finding is where the arms separated. The ratio of total to HDL cholesterol — a marker of cardiovascular risk — fell by 20% in the low-carbohydrate group versus 12% in the low-fat group [s2]. And among the 36 participants with diabetes, glycemic control improved more on the Mediterranean diet than on the low-fat diet [s2]. Different diets moved different markers; none was uniformly best.

What this means for the keto claims

Ketogenic diets are the strictest form of carbohydrate restriction, and much of what is marketed under that label rests on the general low-carb trials above rather than on long-term trials of ketosis specifically. DIRECT confirmed that its low-carb arm reached ketosis — that group had the highest proportion of participants with detectable urinary ketones [s2] — yet still lost weight comparable to the other diets over two years. Ketosis, in other words, was not a shortcut to greater weight loss in the trial.

The reasonable summary is that carbohydrate restriction is one effective option among several for weight loss, not a superior one, and that its clearest benefits are metabolic markers such as the cholesterol ratio rather than kilograms [s2]. What a given diet does to those markers, and whether it suits someone with diabetes or heart disease, is exactly the sort of question that belongs with a clinician — the American Diabetes Association updates its own dietary guidance as the evidence moves. For comparison, the Mediterranean pattern has the larger cardiovascular-outcome trial behind it, and intermittent fasting has been tested head-to-head against continuous calorie restriction.

None of this is a recommendation to adopt any particular diet. It describes what trials found, not what any individual should do.

Sources

• Effect of Low-Fat vs Low-Carbohydrate Diet on 12-Month Weight Loss in Overweight Adults and the Association With Genotype Pattern or Insulin Secretion — JAMA, 2018-02-20
• Weight Loss with a Low-Carbohydrate, Mediterranean, or Low-Fat Diet — New England Journal of Medicine, 2008-07-17