Eating versus skipping breakfast has no discernible effect on obesity-related anthropometric outcomes: a systematic review and meta-analysis

Background: Eating or skipping breakfast for weight interests scientific and lay communities. Our objective was to systematically review and meta-analyze causal effects of eating versus skipping breakfast on obesity-related anthropometric outcomes in humans. Methods: Six databases were searched for obesity- and breakfast-related terms (final search: 02 JAN 2020). Studies needed to isolate eating versus skipping breakfast in randomized controlled trials. Mean differences were synthesized using inverse variance random effects meta-analysis for each outcome. Positive estimates indicate higher outcomes in breakfast conditions (e.g., weight gain). Leave-one-out sensitivity analysis, secondary baseline habit-by-breakfast assignment analysis, and study duration cumulative analysis were performed. Risk of bias was assessed using Cochrane risk of bias tool. Results: Ten articles (12 comparisons; 6d-12wk) were included. Conditions included recommendations to eat versus skip breakfast, or provision of some or all meals. 95% confidence intervals of all main analyses included the null value of no difference for each outcome: body weight (0.17 kg [-0.40,0.73], k=12, n=487, I (2)=74.5), BMI (0.07 kg/m (2) [-0.10,0.23, k=8, n=396, I (2)=54.1), body fat percentage (-0.27% [-1.01,0.47], k=6, n=179, I (2)=52.4), fat mass (0.24 kg [-0.21,0.69], k=6, n=205, I (2)=0.0), lean mass (0.18 kg [-0.08,0.44], k=6, n=205, I (2)=6.7), waist circumference (0.18 cm [-1.77,2.13], k=4, n=102, I (2)=78.7), waist:hip ratio (0.00 [-0.01,0.01], k=4, n=102, I (2)=8.0), sagittal abdominal diameter (0.19 cm [-2.35,2.73], k=2, n=56, I (2)=0.0), and fat mass index (0.00 kg/m (2 )[-0.22,0.23], k=2, n=56, I (2)=0.0). Subgroup analysis showed only one statistically significant result. The interaction effect for BMI (–0.36[-0.65,-0.07]) indicates assignment to conditions consistent with baseline habits had lower BMI. Leave-one-out analysis did not indicate substantial influence of any one study. Conclusions: There was no discernible effect of eating or skipping breakfast on obesity-related anthropometric measures when pooling studies with substantial design heterogeneity and sometimes statistical heterogeneity. Registration: PROSPERO CRD42016033290.