Systematic review and meta-analysis comparing heterogeneity in body mass responses between low-carbohydrate and low-fat diets

An important notion in personalized medicine is that there is clinically relevant treatment response heterogeneity. Low-carbohydrate (CHO) and low-fat diets are widely adopted to reduce body mass. To compare individual differences in responses between two dietary interventions, a formal statistical comparison of response variances between study arms in a randomised controlled trial (RCT) is crucial. We compared change variances in RCTs for the body mass responses to low-CHO dietary interventions vs change variances for the low-fat groups (typically considered as the comparator intervention). A literature search identified relevant RCTs (n=25, 3340 participants). We extracted the means and standard deviations (SD) of body mass change in low-CHO and low-fat study arms to calculate the variances of individual responses. These were meta-analysed in a random effects model and converted to the SD for individual responses (SDir). The pooled SDir for body mass response was 1.4 kg (95%CI: –1.1 to 2.3) with a wide 95% prediction interval of –6.3 to 10.4 kg. We conclude that evidence is insufficient to suggest the response heterogeneity to low-carbohydrate diets differs from that observed with low-fat diets.

Keywords: inter-individual response, randomized