What the DIETFITS Trial Found: Low-Carb vs Low-Fat Over 12 Months, Read Plainly
What the Study Said, in One Sentence
In a 12-month randomized trial of 609 adults, a healthy low-carbohydrate diet and a healthy low-fat diet produced nearly identical weight loss — and neither a genetic pattern nor insulin secretion predicted who did better on which diet. That is the headline of the DIETFITS trial (Diet Intervention Examining the Factors Interacting with Treatment Success), led by Christopher D. Gardner of Stanford Prevention Research Center and published in JAMA (Gardner CD et al., JAMA. 2018;319(7):667-679). Evidence grade for our readers: a randomized controlled trial — the strongest single-study design we cover — funded by the National Heart, Lung, and Blood Institute and the Nutrition Science Initiative, registered as NCT01826591.
How the Study Was Run
Researchers enrolled 609 adults aged 18-50 with a BMI between 28 and 40 and without diabetes, then randomly assigned them to a healthy low-fat or a healthy low-carb diet for 12 months (NCT01826591). Each group received 22 small-group sessions with health educators over the year. Both diets started strict — around 20 g of their assigned macro per day for the first 8 weeks — then let participants add back 5-15 g per week until they reached the lowest level they believed they could sustain. Crucially, neither group was told to count calories.
The design detail that matters most: both diets were built on food quality. Both groups were instructed to maximize vegetables, minimize added sugars, refined flour and trans fats, and cook minimally processed whole foods at home. In interviews, Gardner put the low-carb version of this bluntly: eat avocado, nuts, seeds and salmon — not low-carb chips (Stanford University press release; TCTMD interview, 2018). About 79% of participants completed the trial — strong retention for a 12-month diet study.
What the Numbers Showed
At 12 months, the low-carb group lost a median of 6.0 kg and the low-fat group 5.3 kg — a difference of 0.7 kg that was not statistically significant (JAMA. 2018;319(7):667-679). Individual results ranged from a loss of 30 kg to a gain of 10 kg in both groups. The trial's second question was whether biology predicted success: roughly 30% of participants carried a genotype pattern associated with low-carb response and 40% with low-fat response, and baseline insulin secretion was measured for all. Neither factor changed the outcome — participants matched to their genotype did no better than unmatched ones, and insulin secretion did not favor either diet.
On other markers, the two diets moved in different directions: LDL cholesterol decreased in the low-fat group and increased in the low-carb group, while HDL rose and triglycerides fell more in the low-carb group (JAMA. 2018;319(7):667-679). Resting energy expenditure declined similarly in both. These are group averages from a trial in healthy adults seeking weight management — the paper's own limitations note that results may not generalize beyond a well-resourced, highly educated study population.
One detail of the food design deserves its own line, because most diet studies cannot claim it. Both arms were told to reach for the same quality floor — vegetables maximized, added sugars and refined flour minimized, home cooking emphasized — so the trial compared two versions of a good diet rather than a good diet against a straw man. In the published paper the authors stress that a soda can be low-fat and lard can be low-carb, and neither would qualify (Stanford press release, 2018). That design choice is why the results read as a statement about macro ceilings and not about junk food: the junk was excluded from both arms, so whatever separated the groups was carbohydrate itself, and at 12 months that separation measured 0.7 kg, within the noise.
What This Means for Low-Carb and Keto Eaters — Our Data Read
We are a data editorial team, so we read this trial as a pricing study. Its central finding is not "low-carb does not work" — the low-carb group lost weight, and individual losses in that arm reached 30 kg. The finding is that when both diets are built on quality whole foods, the carbohydrate ceiling itself did not decide the 12-month result. The variable the trial could not remove was adherence: the diet each person could actually sustain, at whatever carb level, was the diet that delivered.
For our audience the practical read is unglamorous. First, the trial's low-carb arm was built on the same foods our verdict pages rate highly — vegetables, avocado, nuts, seeds, fish — which is a worth-noting alignment between a top-tier trial's design and our verified food data. Second, LDL moved up in the low-carb arm: our own coverage of recent work on gene-dependent LDL response to low-carb diets (see our WashU liver-fat trial report for a separate recent example of diet-composition research) is a reminder that individual lipid response varies, and that anyone monitoring blood work should do exactly that. Third, no diet badge in our system is a promise of outcomes — our keto and low-carb badges describe carbohydrate content per 100 g, verified against USDA data, nothing more. Whether any eating pattern suits an individual is a question for that individual and a qualified professional.
One honest limitation of drawing any single-study lesson here: DIETFITS tested moderate healthy low-carb eating, not a ketogenic protocol — a distinction our keto vs low-carb guide draws in detail — and its participants were not asked to reach ketosis. Trials of very-low-carb ketogenic diets under controlled feeding — like the 2026 Cell Metabolism trial we covered — test a different question again. Read together, the pattern across both is consistent: food quality carried the results, and the carb level set the metabolic texture.
Source Box
Primary source: Gardner CD, Trepanowski JF, Del Gobbo LC, et al. "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: The DIETFITS Randomized Clinical Trial." JAMA. 2018;319(7):667-679. DOI: 10.1001/jama.2018.0245. Trial registration: NCT01826591 (ClinicalTrials.gov). Funding: National Heart, Lung, and Blood Institute; Nutrition Science Initiative. Evidence grade: randomized controlled trial, single site, parallel group, 12 months. Quotes from Christopher D. Gardner drawn from the Stanford University press release and a TCTMD interview (2018). Study summary also reviewed via the NIHR Evidence review (University of Southampton/Bazian). External links are provided for verification.
Did DIETFITS prove low-carb diets do not work?
No. The low-carb arm lost a median 6.0 kg over 12 months, with individual results ranging up to a 30 kg loss (JAMA. 2018;319(7):667-679). The trial showed that a healthy low-carb diet and a healthy low-fat diet produced statistically similar average weight loss when both emphasized whole foods — a statement about the difference between diets, not about whether either works.
Should I test my genes or insulin to pick a diet?
That is a medical decision outside our scope as a data team. What the trial data shows: neither the three-SNP genotype pattern nor baseline insulin secretion predicted which diet produced more weight loss (NCT01826591), which led the authors to conclude these factors should not be used to direct diet choice. For personal testing decisions, consult a qualified professional.
Was the DIETFITS low-carb diet ketogenic?
No. Participants cut digestible carbohydrates to about 20 g/day for the first 8 weeks, then added carbs back to the lowest sustainable level — the final diets were moderate low-carb, not maintained ketogenic protocols. Our own verdict pages price foods against both thresholds precisely because the two eating patterns are distinct.
Related reading
Filed under researchweight-loss