Researchers from the NIH’s Metabolic Research Program led by Kevin Hall examine continuous glucose monitoring in thirty adults without diabetes and report weak-to-moderate correlations (r≈0.45) and low reliability (ICC<0.3) in duplicate-meal postprandial glucose responses. Using linear correlations, ICC, and Bland–Altman analyses, they demonstrate that CGM lacks sufficient consistency to serve as a standalone proxy for personalized dietary advice and metabolic health optimization.

Key points

• Weak-to-moderate linear correlation (r≈0.45) between duplicate-meal 2-hour postprandial iAUCs recorded by Abbott Freestyle Libre Pro and Dexcom G4 Platinum CGMs
• Low intra-subject reliability with intra-class correlation coefficients (ICC: Abbott 0.28, Dexcom 0.17), indicating high within-individual glycemic variability
• Bland–Altman analysis reveals wide limits of agreement (±30 mg/dL) around near-zero bias, undermining CGM consistency for personalized dietary feedback

Why it matters: These findings challenge the reliability of CGM-based biohacking for precision nutrition, underscoring the need for more robust metabolic monitoring methods.