Framingham State Food Study

Principal Investigators: David Ludwig and Cara Ebbeling (Boston Children’s Hospital; Harvard Medical School)

Ebbeling CB, Feldman HA, Klein GL, Wong JMW, Bielak L, Steltz SK, Luoto PK, Wolfe RR, Wong WW, Ludwig DS.  Effects of a low carbohydrate diet on energy expenditure during weight loss maintenance: randomized trial.  BMJ. 2018; 363: k4583.  Read >>

Losing weight is the first challenge in treating obesity. The second challenge is maintaining that loss. Only sustained weight reduction can prevent the negative health consequences associated with obesity. However, maintaining significant weight loss indefinitely is notoriously difficult, making the development of strategies that facilitate maintenance of a reduced body weight critical for the effective long-term treatment of obesity.

An earlier study1 led by Drs. David Ludwig and Cara Ebbeling of 24 subjects suggested that eating a low-carbohydrate, high-fat diet over a 30 day period increased energy expenditure after weight loss more than did eating a high-carbohydrate, low-fat diet. This finding challenged the conventional thinking that obesity is an energy balance disorder in which the amount of calories consumed, independent of the source of those calories (i.e., carbohydrates and fat), is the critical factor in fat accumulation.

NuSI funded the Framingham State Food Study, or (FS)2, a larger, longer, and more robust trial, to more definitively address the question of whether the amount or type of calories eaten is the determining factor for preventing fat accumulation after weight reduction.

(FS)2 was a “feeding study,” meaning that the researchers supplied participants with all of their food throughout the study in order to control the amount and composition of food eaten and monitor the amount consumed. In the first phase of the trial2, the total food intake (carbohydrates, fats, and protein) of these overweight and obese participants was restricted to produce, on average, a 12% loss in body weight. At the start of the second phase of the study, the 164 participants who achieved this weight loss goal were randomly assigned to either a low-, moderate-, or high-carbohydrate diet containing (by calories) 20% protein and, respectively, either 20% carbohydrate and 60% fat, 40% carbohydrate and 40% fat, or 60% carbohydrate and 20% fat. For the next 20 weeks, the number of calories of each diet provided to each participant was adjusted to maintain their reduced body weight.

Total daily energy expenditure, the primary outcome for the trial, was measured before and after participants were randomly assigned to the different diets. If only the amount of calories consumed was important for weight maintenance, then energy expenditure should be the same for all three groups. This was not the case. Instead, energy expenditure increased to a greater extent in participants eating the low- or moderate-carbohydrate diets compared with those consuming the high-carbohydrate diet. Those eating the low-carbohydrate diet showed the largest increase.

Limiting calorie intake and increasing expenditure through physical activity is the standard prescription for both losing weight and preventing weight gain. However, weight loss due to restriction of calorie intake typically results in a decrease in total bodily energy expenditure. This drop in expenditure restrains continuing weight loss and may contribute to weight regain unless calorie intake is even further restricted. Maintaining energy expenditure during weight loss, or increasing it after weight reduction, as observed in the (FS)2 trial in association with the lower-carbohydrate, higher-fat diets, would therefore be expected to facilitate further weight loss and prevent weight regain.

The results also suggest that the type or mix of calories (i.e., carbohydrate or fat) is a critical factor for maintaining a healthy body weight. These findings, while not supportive of an energy balance model of obesity, are consistent with an alternative hypothesis that emphasizes the different effects of carbohydrates and fats on fat accumulation (known as the carbohydrate-insulin model or the hormonal/regulatory model of obesity).

1Ebbeling CB, Swain JF, Feldman HA, Wong WW, Hachey DL, Garcia-Lago E, Ludwig DS.  Effects of dietary composition on energy expenditure during weight-loss maintenance.  JAMA. 2012; 307: 2627-34.  Read >>

2Ebbeling CB, Klein GL, Luoto PK, Wong JMW, Bielak L, Eddy RG, Steltz SK, Devlin C, Sandman M, Hron B, Shimy K, Heymsfield SB, Wolfe RR, Wong WW, Feldman HA, Ludwig DS. A randomized study of dietary composition during weight-loss maintenance: Rationale, study design, intervention, and assessment.  Contemp Clin Trials. 2018; 65: 76-86.  Read >>


Secondary analyses and ancillary studies

Hall KD, Guo J, Speakman JR.  Do low-carbohydrate diets increase energy expenditure?  Int J Obesity. 2019; 43: 2350–2354. Read >>

This comment questions the main finding of the Framingham study that participants eating the low-carbohydrate diet showed increased energy expenditure as measured using doubly labeled water (DLW).  The issues raised included (i) a change in the statistical analysis plan regarding baseline measures, (ii) inclusion of participants whose energy expenditure exceeded energy intake resulting in unaccounted energy and suggesting non-adherence to the diet and (iii) the validity of respiratory quotient (RQ) values used to calculate expenditures using DLW.

Ludwig DS, Lakin PR, Wong WW, Ebbeling CB. Scientific discourse in the era of open science: a response to Hall et al. regarding the Carbohydrate-Insulin Model.  Int J Obesity. 2019; 43: 2355–2360.  Read >>

In rebuttal to Hall et al.’s comment, Ludwig et al. note that the change in the statistical plan for the Framingham trial was registered before the data were unblinded and were more appropriate to the study design than that described in the original plan. The issues of unaccounted energy and participants’ nonadherence to their diets are addressed in additional analyses, including a sensitivity analysis that also addresses the concern about the use of appropriate RQ values to calculate expenditure using DLW.

Ludwig DS, Greco KF, Ma C, Ebbeling CB.  Testing the carbohydrate-insulin model of obesity in a 5-month feeding study: the perils of post-hoc participant exclusions.  Eur J Clin Nutr. 2020; 74: 1109–1112.  Read >>

This paper provides a further statistical evaluation of the Hall et al. 2019 secondary analysis that questioned the main result of the Framingham study.  The evaluation finds that the post hoc exclusion of data on which the secondary analysis was based biased the outcome to minimize the effect of the low-carbohydrate diet on energy expenditure.  Statistically controlling for this bias produced results consistent with the original Framingham study.

Ebbeling CB, Bielak L, Lakin PR, Klein GL, Wong JMW, Luoto PK, Wong WW, Ludwig DS.  Energy Requirement Is Higher During Weight-Loss Maintenance in Adults Consuming a Low- Compared with High-Carbohydrate Diet.  J Nutr. 2020; 150: 2009-2015.  Read >>

During the weight stabilization phase of the Framingham study, each participant in the different diet groups was fed an amount of calories sufficient to maintain their reduced body weight. Energy requirements determined by such “caloric titration” were significantly higher in the low-carbohydrate group.  In keeping with previous work, this increase in energy requirements corresponded closely with estimates of energy expenditure using DLW in the original Framingham study paper and provides additional, independent support that a low-carbohydrate diet increases energy expenditure.

Shimy KJ, Feldman HA, Klein GL, Bielak L, Ebbeling CB, Ludwig DS.  Effects of Dietary Carbohydrate Content on Circulating Metabolic Fuel Availability in the Postprandial State.  J Endocr Soc. 2020; 4:bvaa062.  Read >>

In this ancillary study to the Framingham trial, a sub-set of participants who had stabilized their weight loss were admitted to a metabolic ward and fed breakfast, lunch and dinner according to their study diets.  Analysis of blood samples taken before and after the meals showed that the circulating availability of metabolic fuels 3-5 hours after a meal was lowest in those eating the high-carbohydrate diet.  This and other results are consistent with the hypothesis that high-carbohydrate diets promote storage of metabolic fuels in fat.