Does Adopting a Low-Fat Vegan Diet Reduce Greenhouse Gas Emissions?

Kahleova et al. (2025) asked whether adopting a low-fat vegan diet reduces greenhouse gas emissions and cumulative energy demand. They analyzed 16-week dietary records from overweight adults who were randomly assigned to either a vegan group or a control group. They linked each participant’s food intake to environmental impact databases to estimate emissions and energy use. They found that the vegan group reduced emissions by about 1313 g CO2-eq per person per day, while the control group reduced only 314 g. They also found that the vegan group reduced cumulative energy demand by about 8194 kJ per person per day, versus no real change in the control group.

Why This Article Was Selected for The Policy Scientist

This article addresses an important policy topic: the environmental burden of food systems at a time when climate impacts, land constraints, and dietary transitions are receiving heightened attention. The lead authors have published extensively on diet-related outcomes, and this study extends that line of work into environmental metrics. Its contribution is timely because many governments are evaluating dietary guidelines, procurement standards, and emissions targets that intersect with food choice. The data are credible, linking detailed food records to established environmental databases, though generalizability beyond U.S. diets may be limited. The randomized design is a methodological strength, offering stronger causal claims than observational studies.

Full Citation and Link to Article

Kahleova, H., Jayaraman, A., McKay, B., et al. (2025). Vegan diet, greenhouse gas emissions, and cumulative energy demand: A secondary analysis of a randomized clinical trial. JAMA Network Open, 8(11), e2543871. https://doi.org/10.1001/jamanetworkopen.2025.43871 

Central Research Question

The central research question in Kahleova et al. (2025) is whether adopting a low-fat vegan diet causally reduces greenhouse gas emissions (GHGE) and cumulative energy demand (CED) relative to a control diet over a 16-week period. The authors frame this as a secondary environmental analysis of a prior randomized clinical trial (RCT) focused on metabolic and physiological outcomes. Their primary hypothesis is that removing or sharply reducing animal-source foods would materially decrease both emissions and energy inputs associated with food consumption. They treat these environmental metrics as quantifiable downstream outputs of dietary choice, and they investigate magnitude, sources, and statistical significance of any differences between treatment and control groups.

Previous Literature

The article intersects with several strands of research concerning food systems, climate impacts, and nutritional epidemiology. One line of literature evaluates the environmental consequences of dietary patterns. Systematic reviews have repeatedly found that animal-source foods, particularly red meat, generate disproportionately high emissions relative to plant-based foods. Aleksandrowicz et al. (2016) synthesized results across multiple dietary interventions and estimated that shifting to vegetarian or vegan diets yields meaningful reductions in GHGE, land use, and water use. The EAT-Lancet Commission (Willett et al., 2019) advanced these findings by proposing global dietary targets intended to balance health and sustainability outcomes; that commission identified red meat as especially resource intensive and emissions heavy.

A second strand evaluates the health impacts of vegan or low-fat diets. Prior RCTs, including work by the same lead author, have examined body weight, insulin sensitivity, postprandial metabolism, and lipid content in muscle and liver tissue. These studies positioned vegan diets as metabolic interventions rather than environmental ones. The current article explicitly leverages that research base by reanalyzing dietary records for environmental outputs.

A third strand concerns life-cycle assessment (LCA) approaches linked to food consumption. LCA-based databases estimate GHGE and CED per food commodity or category. Prior studies using household or survey data have applied LCAs to infer the emissions footprint of dietary patterns, but these typically rely on observational data. This article differs by embedding environmental metrics within an experimental design. The alignment with existing literature is therefore partial: the environmental component draws on a relatively mature LCA literature, but the use of an RCT to generate causal claims about dietary change and environmental impact is less common, and thus constitutes a contribution.

Data

The data originate from a randomized clinical trial conducted between January 2017 and February 2019 in Washington, D.C. The analytical sample includes 223 overweight adults (mean age ≈ 55 years; approximately 87 percent women) who completed the 16-week intervention. Participants were randomized 1:1 into a vegan group (n=117) or control group (n=106). The vegan group was instructed to follow an ad libitum low-fat vegan diet consisting of fruits, vegetables, grains, and legumes, without caloric restriction. The control group was instructed not to modify their diet.

Environmental measurement relies on 3-day dietary records at baseline and week 16. Registered dietitians coded these logs using the Nutrition Data System for Research. Food items were then linked to two key data sources: (1) the USDA Food Commodity Intake Database (FCID), and (2) the University of Michigan database of Food Impacts on the Environment for Linking to Diets (dataFIELD). These databases provide GHGE estimates (in grams CO2-equivalent per person per day) and CED values (in kJ per person per day). Linkages were independently verified by three blinded reviewers, and a senior researcher reviewed accuracy.

The resulting dataset contains detailed consumption patterns and matched environmental attributes, allowing decomposition by food category (e.g., meat, dairy, eggs, grains, legumes, vegetables, fruits, added fats). Importantly, the data rely on self-reported food records, a common approach in nutritional research but one that is vulnerable to underreporting or misclassification. Nonetheless, the environmental portion uses well-established LCA-derived databases, which yields consistency in relative rankings of foods. Generalizability is moderate: the sample is U.S.-based, overweight, predominantly female, and self-selecting as trial volunteers. However, the food items themselves reflect broadly available U.S. dietary commodities, allowing cautious extrapolation to similar contexts.

Methods

The study design is an RCT, which strengthens causal inference relative to cross-sectional or cohort designs. Randomization was conducted in a 1:1 allocation ratio. Participants in the vegan arm received dietary instruction and support; the control arm was asked not to change their diet. Environmental metrics were assessed at two time points: baseline and week 16.

Analytically, the authors used repeated measures analysis of variance (ANOVA) to compare within- and between-group changes. Statistical significance was set at p<0.05, and analyses were performed in SAS 9.4. Results were expressed as means with 95 percent confidence intervals. The authors also decomposed aggregate changes by food category to identify which components drove GHGE and CED reductions.

The choice of RCT design and repeated-measures ANOVA gives the study internal validity, but the statistical strategy remains relatively descriptive: it quantifies changes attributable to treatment without modeling heterogeneity or mechanisms beyond food category composition. For the research question at hand—estimating environmental impacts from experimentally induced dietary change—the approach is sufficient and superior to non-experimental methods used in prior literature. Other methods, such as structural causal models or instrumental variables, would not meaningfully enhance inference here given randomization.

Findings/Size Effects

The primary findings show substantial reductions in GHGE and CED in the vegan group relative to the control group over 16 weeks. For GHGE, the vegan group decreased emissions by approximately 1313 g CO2-equivalent per person per day (95% CI: −1486.6 to −1139.4; p<0.001). The control group decreased emissions by approximately 314 g CO2-equivalent per person per day (95% CI: −596.0 to −32.0; p<0.05). The between-group effect size is approximately −999 g CO2-equivalent per person per day (95% CI: −1329 to −669; p<0.001).

For CED, the vegan group decreased energy demand by roughly 8193.8 kJ per person per day (95% CI: −9369.3 to −7018.4; p<0.001), whereas the control group experienced no meaningful change. The between-group effect size is −6767 kJ per person per day (95% CI: −8718 to −4817; p<0.001).

Decomposition analyses show that reduced meat consumption is the largest driver of environmental impact reductions, followed by dairy and eggs. In the vegan group, meat-related GHGE dropped by roughly 800 g CO2-equivalent per person per day, dairy dropped by roughly 448 g, and eggs dropped by roughly 107 g. These category-specific results align with prior literature indicating that red meat and dairy products contribute disproportionately to GHGE. Plant-based foods (legumes, vegetables, fruits, and grains) increased or remained stable in consumption but contribute far less to emissions and energy use.

The magnitude of effects is environmentally meaningful because daily reductions of approximately 1 kg CO2-equivalent per person accumulate substantially over longer horizons. The intervention period is relatively short, yet the environmental signal is strong due to the immediate substitution away from emissions-intensive foods.

Conclusion

Kahleova et al. demonstrate that experimentally induced adoption of a low-fat vegan diet significantly reduces GHGE and CED relative to no dietary change. The study’s key contribution is its causal identification strategy: by embedding environmental measurement within an RCT, it provides stronger evidence than observational studies linking plant-based diets to lower environmental footprints. The findings corroborate prior LCA-based assessments but extend them by showing that real-world dietary shifts—rather than hypothetical menus—produce measurable environmental impacts over a modest time frame.

The trial’s internal validity is strong, aided by randomization and blinded dietary coding. The environmental data sources are widely used and methodologically coherent. Limitations concern external validity (overweight U.S. volunteers, short duration), self-reported dietary intake, and lack of exploration of long-run adherence. Despite these constraints, the study offers persuasive evidence that reducing animal-source foods can yield rapid and substantial environmental benefits. It thus aligns with and strengthens existing literature showing that food system transformations represent a meaningful lever in climate and energy policy.