Effects of low-carbohydrate and low-fat diets: a randomized trial
Lydia A. Bazzano, Tian Hu, Kristi Reynolds, Lu Yao, Calynn Bunol, Yanxi Liu, Chung-Shiuan Chen, Michael J. Klag, Paul K. Whelton, Jiang He - Annals of Internal Medicine, 2014
In this 12-month randomized trial, a low-carbohydrate diet (under 40 g/day) produced greater weight loss than a low-fat diet (a 3.5 kg difference) and better cardiovascular risk-factor changes: higher HDL, lower triglycerides, and a lower total/HDL cholesterol ratio. There was no significant difference in LDL cholesterol between the diets, and the trial measured risk factors only, with no clinical cardiovascular endpoints.
Key findings
- The low-carbohydrate diet produced greater weight loss than the low-fat diet: mean difference -3.5 kg (95% CI -5.6 to -1.4 kg; P = 0.002).
- Greater increase in HDL cholesterol on low-carbohydrate: mean difference +0.18 mmol/L (+7.0 mg/dL) (95% CI 0.08 to 0.28 mmol/L; P < 0.001).
- Greater reduction in triglycerides and in the total/HDL cholesterol ratio on low-carbohydrate.
- No significant difference in LDL cholesterol between the two diets; the trial measured cardiovascular risk factors only, with no clinical CVD endpoints.
Effect measures
- Other: Mean difference in weight change -3.5 kg (low-carb vs low-fat)95% CI -5.6 to -1.4 kg
- Other: Mean difference in HDL cholesterol change +0.18 mmol/L (+7.0 mg/dL)95% CI 0.08 to 0.28 mmol/L (3.0 to 11.0 mg/dL)
- Other: Mean difference in triglyceride change -0.16 mmol/L (-14.1 mg/dL)95% CI -0.31 to -0.01 mmol/L (-27.4 to -0.8 mg/dL)
- Other: Mean difference in total/HDL cholesterol ratio change -0.4495% CI -0.71 to -0.16
Why this evidence tier (Moderate)
- Risk of bias:
- A genuine randomized parallel-group trial with good retention, but unblinded by nature (diet) and reliant on self-reported intake.
- Precision:
- Small sample (148 participants); between-group differences are statistically significant but estimated with wide intervals.
- Directness:
- Measures surrogate risk factors (weight, lipids), not clinical cardiovascular events, so it does not directly speak to heart attacks or mortality.
- Consistency:
- Weight and triglyceride/HDL changes are consistent with other low-carbohydrate trials; the absence of an LDL difference here contrasts with the LDL rise seen in some other low-carbohydrate trials.
- Funding / COI:
- Funded by the National Institutes of Health; no industry or food-company sponsorship stated. No funding conflict identified.
Moderate certainty for the surrogate outcomes: low-carbohydrate produced more weight loss and better triglyceride/HDL profile, with no LDL difference, but the trial cannot speak to clinical events.
- Population:
- 148 men and women without clinical cardiovascular disease or diabetes at a large academic medical center (Tulane, New Orleans); racially diverse.
- Conflicts of interest:
- Funded by the NIH with no industry sponsorship stated. Author disclosure forms are on file with Annals of Internal Medicine; no specific competing interests were surfaced. None known.
- Funding:
- National Institutes of Health (NIH); no industry or food-company sponsorship stated.
Limitations
- Small single-center trial (148 participants) over 12 months.
- Measured cardiovascular risk factors only; no clinical cardiovascular endpoints.
- Diet was self-reported and the trial could not be blinded.
- The no-difference-in-LDL finding is not stated in the abstract; it is supported by the full-text results tables, where LDL is absent from the significant between-group differences.