Energy balance

Using weight-dependent prediction equations (Goldberg or pTEE methods) to exclude implausible energy reporters in epidemiologic studies of diet and BMI introduces selection bias and reverses observed associations compared to using simple absolute caloric cutoffs.

When analyzing diet and weight data, avoid using complex formulas that exclude people based on their weight if your outcome is also weight-based (like BMI). This creates a statistical illusion. Simple cutoffs for extreme calorie reporting (e.g., <500 or >3500 kcal/day) are often sufficient and less prone to bias. For prospective changes in weight, the method of exclusion matters less.