Low-carb or low-fat? Long-term trials reveal modest weight differences, mixed blood-test results, and the importance of food quality and staying with a plan.
In a trial involving 609 adults over 12 months, the low-fat group lost an average of 5.3 kg and the low-carbohydrate group lost 6.0 kg. Read those figures alone and low-carb appears to win. Yet the difference was not statistically significant.[1] A decimal place can make a verdict look more definite than the evidence supports. Choosing dinner requires more than comparing the last digit.
The research does not deliver a sweeping victory for either approach. Some analyses find slightly greater average weight loss with carbohydrate restriction; some long-term trials find no clear difference between the diets.[1][12][18] To understand why, look at the comparison period, the actual foods people ate, and the support they received. A diet’s name is only the beginning of its description.
This feature compares carbohydrate and fat restriction as weight-loss approaches. The roles of protein and fiber are covered in our features on protein, appetite, and muscle and fiber and the gut. Here, the practical question is how to bring long-term research back to a household table in Tokyo: what does a study establish, what remains uncertain, and which changes could fit meals you actually share?
Low-carb and low-fat labels leave out important details
A low-carbohydrate diet in a research paper can overlap with what people casually call cutting carbs, but the term does not describe a single menu. Researchers distinguish diets by the amount of carbohydrate or its share of total energy. Similarly, low-fat does not mean removing all fat. Before comparing results, establish what each group was asked to reduce and by how much.[1][20] Without those details, apparently similar studies can be testing substantially different eating patterns.
One randomized trial recruited 148 men and women without cardiovascular disease or diabetes. Its low-carbohydrate diet specified less than 40 g of carbohydrate per day; its low-fat diet specified less than 30% of daily energy from fat. Both groups received dietary counseling at regular intervals.[20] Randomization means participants were assigned to groups by chance. That differs from comparing people who chose their preferred diet, where preferences and other differences can complicate interpretation.
The larger DIETFITS trial compared a healthy low-fat diet with a healthy low-carbohydrate diet. Participants worked toward the lowest fat or carbohydrate intake they could maintain over the long term, with an emphasis on food quality. Health educators provided 22 diet-specific small-group sessions during 12 months.[1] The intervention therefore included repeated help with changing behavior. Its results describe supported dietary changes, rather than a food rule handed out once and followed without further contact.
At 12 months, the average macronutrient proportions in DIETFITS were 48% carbohydrate, 29% fat, and 21% protein in the low-fat group. In the low-carbohydrate group, they were 30% carbohydrate, 45% fat, and 23% protein.[1] Macronutrients are the nutrients being compared here: carbohydrate, fat, and protein. These percentages describe what participants were eating on average at that point. They are not a requirement that every meal match the same distribution.
Notice what remained on both sides. The low-carb group still ate carbohydrates, and the low-fat group still ate fat.[1] Declaring a meal a failure because it contains rice, or because oil was used in cooking, does not fit the diets observed in this trial. At a supermarket prepared-food counter, the relevant choices include the portion of rice and the accompanying dishes. What replaces a reduced food is part of the diet too.
DIETFITS assigned 305 participants to low-fat and 304 to low-carbohydrate eating.[1] In the separate 148-person trial, 60 participants in the low-fat group and 59 in the low-carbohydrate group completed the intervention.[20] Group assignment and study completion answer different questions. The first describes how the comparison started; the second helps show how many people remained through the intervention. Both belong beside a report of kilograms lost.
For your own meal planning, the distinction between an initial prescription and a later eating pattern matters. An instruction to begin with strict restriction is not necessarily a description of what people maintained a year later. The DIETFITS approach explicitly sought a sustainable level.[1] If a headline leaves out that adjustment process, it can turn a supported, adaptable intervention into a rigid rule that the research did not test over the entire period.
初日の汗は、予約した人だけが持ち帰れる。
Large long-term trials do not show a clear weight winner
DIETFITS was a 12-month randomized trial of 609 adults aged 18–50 who had overweight or obesity and did not have diabetes. Their body mass index, or BMI, ranged from 28 to 40; average age was 40, and 57% were women. A total of 481 people completed the trial, representing 79% of those randomized.[1] The population matters: these results cannot simply be transferred to dietary decisions for someone receiving diabetes treatment.
Average weight change was −5.3 kg with low-fat eating and −6.0 kg with low-carbohydrate eating. The between-group difference was 0.7 kg, with a 95% confidence interval of −0.2 to 1.6 kg.[1] For DIETFITS, this interval describes the uncertainty in the estimated weight difference between the two assigned diets. Because this interval includes no difference, the researchers did not find a statistically significant difference in weight change between the groups. The observed averages and the uncertainty belong in the same sentence.
This result does not establish that everyone will lose more weight on low-carb. It also does not establish that both approaches were useless: average weight fell in both groups.[1] Those are separate judgments. A comparison asks whether one intervention outperformed another under the study conditions. It does not require the two groups to have achieved nothing whenever the difference between them is unclear.
A longer trial followed 307 participants, whose average age was 45.5 years, for 2 years. Both low-carbohydrate and low-fat diets were combined with a comprehensive program to change everyday behavior. Weight loss was approximately 11 kg at 1 year and 7 kg at 2 years. At every measurement point, there were no between-group differences in weight, body composition, or bone mineral density.[12] Body composition concerns what body weight consists of, including fat and nonfat components.
The time course deserves attention. The average loss at the end of the first year was not fully maintained at the end of the second.[12] A diet’s long-term result is therefore more than its early downward curve. When choosing a plan, consider how you might continue it, adjust it, and return to it when circumstances change. The study does not justify expecting a straight line of uninterrupted loss simply because the initial approach worked.
There were also important limits. Participants received intensive behavioral support; people with diabetes or abnormal blood lipids were excluded; and dropout was high by 2 years.[12] These conditions affect how closely the study resembles your own situation. A result obtained with substantial help does not tell us that the same average would follow from trying to reproduce only the nutrient targets without that help.
The low-carbohydrate group began with 20 g of carbohydrate per day for the first 3 months, then increased the daily allowance by 5 g each week. The low-fat group limited fat to no more than 30% of energy and also restricted total energy intake.[12] This was a comparison of interventions with different instructions, including a planned adjustment phase. It was not a test of holding every participant at the same starting restriction indefinitely.
Evidence favoring low-carb belongs in the picture too
DIETFITS cannot close the debate on its own. In the trial of 148 participants, the 12-month difference in weight change favored the low-carbohydrate group by −3.5 kg, with a 95% confidence interval of −5.6 to −1.4 kg. The low-carbohydrate group also had greater improvement in body fat.[20] A fair account must retain this favorable result alongside trials that found no significant weight difference. Selecting only the result that supports a preferred diet would narrow the evidence artificially.
A systematic review and meta-analysis combined 38 trials involving 6,499 adults. For this comparison, the researchers gathered randomized trials of the two dietary approaches and pooled their weight and blood-lipid results by follow-up period. At 6–12 months, the pooled average difference in weight change was −1.30 kg in favor of low-carbohydrate diets, with a 95% confidence interval of −2.02 to −0.57 kg.[18] This analysis found an average advantage, and its interval did not include no difference.
The size of that advantage is part of the finding. It means an average of 1.30 kg more loss than with low-fat diets within that follow-up period.[18] It is not the total amount lost on either diet, and it is not a promise about an individual. Keeping the comparison explicit prevents a small difference between two active approaches from being mistaken for a complete description of what either approach can achieve.
The review searched for trials published up to September 2019. Three reviewers independently extracted results, and the authors assessed study bias before combining the evidence. They also grouped results by follow-up duration and compared blood lipids as well as body weight.[18] This provides a way to consider weight and laboratory measurements over the same time window. It still combines a range of diets and participants rather than reproducing one uniform meal plan.
Another synthesis asked a different question. A network meta-analysis included 121 trials with 21,942 adults who had overweight or obesity. At 6 months, estimated weight loss relative to usual diet was 4.63 kg for low-carbohydrate diets and 4.37 kg for low-fat diets. The evidence was rated as moderately certain.[9] A network meta-analysis links comparisons across several interventions. Here, the figures refer to comparison with usual eating, rather than a direct difference between the two diets.
By 12 months, weight-loss effects across dietary approaches had weakened in that analysis.[9] This is relevant to expectations, but it does not turn the 6-month figures into a direct head-to-head trial result. Changing a usual diet and choosing between two changed diets are different questions. If you compare numbers without checking their reference point, you can unknowingly change the question halfway through the comparison.
An earlier meta-analysis of 5 trials and 447 participants compared low-carbohydrate diets without energy restriction against energy-restricted low-fat diets. The difference favored low-carbohydrate eating by −3.3 kg at 6 months. At 12 months it was −1.0 kg, with a 95% confidence interval of −3.5 to 1.5 kg.[11] The later interval included no difference. A substantial early advantage and a less clear later advantage can coexist in the same evidence base.
The pooled estimate from 38 trials and the DIETFITS estimate have different uncertainty ranges.[1][18] That does not make one automatically correct and the other incorrect. A pooled analysis covers different people and a broader range of dietary interventions than a single trial. Ask both whether a difference was detected and how large it was. Those questions help you decide what the evidence means without turning every paper into a declaration of victory.
Numbers: low-carb versus low-fat over time
609 adults, 12 months: −5.3 kg with low-fat and −6.0 kg with low-carbohydrate eating; no significant between-group weight difference.[1]
38 trials: the 6–12-month difference favored low-carbohydrate diets by −1.30 kg; 95% confidence interval −2.02 to −0.57 kg.[18]
148 adults, 12 months: a −3.5 kg difference favored low-carbohydrate eating; 95% confidence interval −5.6 to −1.4 kg.[20]
307 adults, 2 years: approximately 7 kg lost with behavioral support; no between-group difference in weight.[12]
121 trials: 6-month weight-loss effects versus usual diet were 4.63 kg and 4.37 kg; these are not direct head-to-head differences.[9]
初日の汗は、予約した人だけが持ち帰れる。
1
2