Current Cardiovascular Risk Reports

, Volume 4, Issue 2, pp 86–88

Weight-Loss Diet Studies: Is the Most Rigorous Trial To Date Generalizable to Public Health?


  • Mindy Dopler Nelson
    • Stanford Prevention Research Center
  • Jennifer Otten
    • Stanford Prevention Research Center
  • Beibei Oelrich
    • Stanford Prevention Research Center
    • Stanford Prevention Research Center
Clinical Trial Report

DOI: 10.1007/s12170-010-0086-8

Cite this article as:
Nelson, M.D., Otten, J., Oelrich, B. et al. Curr Cardio Risk Rep (2010) 4: 86. doi:10.1007/s12170-010-0086-8

Sacks FM, Bray GA, Carey VJ, et al.: Comparison of weight-loss diets with different compositions of fat, protein, and carbohydrates. N Engl J Med 2009, 360:859–873.


•• Of major importance.


There is ongoing debate over the relative effectiveness of diets of varying macronutrient composition on weight loss. Previous trials have produced inconsistent results and have been limited by small sample size, short duration, under-representation of men, poor dietary adherence data, and large loss to follow-up.


The authors sought to compare the effect of four diets of varying macronutrient composition on change in body weight in overweight and obese adults after 2 years using 2 × 2 factorial comparisons of low-fat or high-fat X average protein or high-protein (resulting in four different levels of carbohydrate, ranging from lower to higher). Secondary outcomes included an extensive set of related metabolic and behavioral factors.


The investigators randomized 811 overweight participants from two diverse geographic locations (Boston, MA and Baton Rouge, LA) to one of four diets: low-fat, average-protein (20% fat, 15% protein, 65% carbohydrates); low-fat, high-protein (20% fat, 25% protein, 55% carbohydrates); high-fat, average-protein (40% fat, 15% protein, 45% carbohydrates); and high-fat, high-protein (40% fat, 25% protein, 35% carbohydrates). Participants were given group and individual instructions for 2 years with integrated behavioral counseling and a set goal for physical activity. The diets recommended similar foods and calorie-reduction goals (deficit of 750 kcal per day). All participants were given the same weekly goal of 90 min for physical activity. Urinary nitrogen and respiratory quotient were used as biomarkers of dietary adherence.


All four diets produced similar changes in body weight at 2 years. At 2 years, weight loss in the intention-to-treat analyses was as follows: 3.0 kg in those assigned to moderate protein and 3.6 kg for high protein; 3.3 kg for those in the low-fat and high-fat diet groups; and 2.9 kg for high carbohydrate and 3.4 kg for low carbohydrate. Among the 80% who completed the study, the average weight loss was 4.0 kg. Biomarkers and dietary records confirmed that participants altered their diets relative to baseline; however, the changes achieved fell considerably short of target macronutrient levels. Nonetheless, improvements in blood lipid parameters and fasting insulin levels were statistically significant and in the directions anticipated based on macronutrient content. Overall cardiovascular risk reduction was observed across all four diet groups, independent of the diet assignment. Self-reported diet acceptability, cravings, fullness, hunger, energy intake, and physical activity were similar across dietary assignments. Attendance at group sessions as opposed to individual sessions or diet assignment was the best predictor for weight loss success.


The authors concluded that reduced-calorie diets produce clinically meaningful and similar weight loss and maintenance after 2 years, regardless of macronutrient composition. Despite intensive long-term counseling and sophisticated web-based tracking tools, participants were unable to achieve their assigned macronutrient intake goals. These findings suggest that although individuals tend to revert back to their customary macronutrient intakes, they may be able to maintain behaviors, such as reduced-calorie intake or increased physical activity, which will produce and sustain weight loss.


To date, the study by Sacks et al. [1] represents the gold standard for evaluating the effects of weight-loss diets of varying macronutrient composition in a free-living population. Some of the field’s most accomplished investigators designed this trial to overcome many limitations of previous trials. Specifically, they extended the length of the study to 2 years; recruited a diverse population with respect to age, income, and geography; enrolled a large percentage of men (40%); retained 80% of the sample; extensively assessed dietary adherence; blinded study staff to diet assignment; and attempted to address questions about high fat versus low fat, high protein versus low protein, and high carbohydrate versus low carbohydrate in a single trial with sufficient statistical power. Careful measures were taken to increase the likelihood of adherence to dietary and physical activity recommendations by incorporating intensive counseling for the entire trial duration and novel tools designed to assist individuals in consuming their assigned diets.

However, in spite of the impressive overall rigor of the trial, the final outcome suffers from the same limitation that continues to plague virtually all free-living, long-term, weight loss–dietary intervention trials, and that limitation is poor adherence. Although the target difference in protein among diets was intended to be 15% versus 25% of energy, the average diet group difference achieved at 2 years was 20% versus 21%. The average diet group differences in fat and carbohydrate intake were greater than for protein, but at 2 years were still roughly one third of the target goals. This arguably fatal flaw in conduct draws into question whether the similarity in weight loss patterns among the four diet groups is due to a lack of effect of differences in dietary composition or simply due to a lack of dietary composition differences achieved.

In terms of study design and meticulous attention to detail, we feel the rigor in design and conduct of this study is unparalleled in the field. Despite this rigor, the difficulties with adherence suggest an important need to employ alternate approaches. An alternate design approach to maximize adherence is to employ feeding studies; however, high participant burden and cost make these prohibitive for large sample sizes and long durations, which then become the inherent limitations of those studies [2]. A novel potential strategy would be to choose a target diet composition of clinical interest to be maintained at the end of a study protocol, and then design the intervention diets to be initiated with more extreme differences in composition (ie, an induction phase), such that when the inevitable decrease in adherence occurs, what remains at the end of the protocol is still an important and worthwhile set of dietary comparisons. Although this approach has its own inherent challenges (eg, how to predict/gauge the long-term degree of deviation from adherence to the induction phase diet), this limitation appears no worse than those described for the more traditional dietary design approaches and might answer the question about whether following prescribed diets (ie, as intended) leads to differential outcomes, as well as information on the rate at which people regress.

The past 5 years have yielded a set of relatively long-term, large-sample, dietary weight loss trials in free-living populations that, in sum, suggest that a “low-fat” approach is not the most effective strategy for weight loss [37]. These same trials have also failed to demonstrate that any other macronutrient composition is strikingly superior, although a number of these trials have reported modest relative benefits for weight loss and/or related parameters for a “lower-carbohydrate” approach. The study by Sacks et al. [1], arguably the most rigorous trial to date, successfully tested whether being assigned to one of four diets differing in macronutrient composition would lead to clinically relevant differences in weight loss and related parameters, and it convincingly concluded that the answer is “no”.

Perhaps the greatest public health significance of these studies is the finding that the average person enrolled in these studies does not adhere meaningfully to weight loss diets that vary in macronutrient content over extended time periods. If researchers want to answer public health questions about the differential effects of varying dietary macronutrients in free-living populations, we first need to improve adherence. Whether this can be achieved around macronutrient goals remains to be seen.

No doubt the rich data sets generated by studies over the past 5 years will yield provocative secondary analyses, which will in turn lead to a period of new studies testing hypotheses in targeted subpopulations of overweight and obese individuals [4]. Of particular interest would be the elucidation of characteristics or behaviors that define adherent individuals, as well as the identification of subsets of these populations who might be differentially responsive to either “lower-carbohydrate,” “lower-fat,” “higher protein,” or other dietary patterns. We anticipate this next round of studies with guarded optimism.


No potential conflicts of interest relevant to this article were reported.

Copyright information

© Springer Science+Business Media, LLC 2010