Reply to DS Ludwig et al.

Dear Editor:

Our recent secondary analysis [1] of cross-over diet studies explored differences in ad libitum energy intake and body composition between diet order groups. This publication provided a more complete picture of our original experiments, neither of which indicated significant within-participant order effects [2,3]. The letter by Ludwig et al. [4] objects to the concept of a within-participant diet order effect (defined as the interaction between diet and diet order group), but it has a straightforward interpretation because it simply evaluates the extent to which diet order assignment affected the estimate of the within-participant diet difference. In the case of our low-carbohydrate (LC) compared with low-fat (LF) diet study, every participant consumed more energy on the LC diet by ∼700 kcal/d on average (P < 0.0001), and the mean diet difference was nominally ∼140 kcal/d greater in those who received the LF diet first than those who received the LC diet first. Although this within-participant order effect was not statistically significant (P = 0.3), we fully acknowledged that our studies were not powered to detect diet order or carryover effects.

In our LC compared with LF diet study, we documented substantial differences in energy balance between diet order groups and explored several potential biological mechanisms for the observed carryover effects. However, the letter by Ludwig et al. [4] argues that data from the second diet period should be disregarded as invalid because of the significant carryover effects. However, previously, Ludwig et al. [4] purported to show that the carryover effects during the second diet period should be regarded as supporting their carbohydrate-insulin model of obesity [5]. Although several aspects of the reanalysis by Ludwig et al. [4] were shown to be incorrect and failed to support the carbohydrate-insulin model [6], we agree with their previous sentiment that data from the second diet period should not be ignored as Ludwig et al. [4] now apparently suggest. Rather, we believe that these data provide interesting information likely related to the mechanisms by which these diets affect energy balance. Indeed, to further investigate such diet order effects we recently launched a cross-over study employing controlled feeding of LC and LF diets over longer durations (clinicaltrials.gov ID NCT06325722).

Funding

This work was supported by the Intramural Research Program of the NIH, National Institute of Diabetes & Digestive & Kidney Diseases under award number 1ZIADK013037.

Conflict of interest

The authors report no conflicts of interest.