A search for truth in dietary epidemiology

Area of Science:

• Nutritional Epidemiology
• Biostatistics

Background:

• Epidemiologic studies on diet provide valuable insights but often yield inconsistent results.
• This inconsistency is largely attributed to confounding factors and measurement errors in dietary assessments.

Purpose of the Study:

• To identify the primary reasons for controversy and inconsistency in dietary epidemiologic studies.
• To explore advanced analytical methods for more accurate and reliable dietary risk assessment.

Main Methods:

• Critically evaluate traditional analytical approaches in dietary research, highlighting their reliance on an often-unmet assumption of error-free dietary assessment.
• Introduce and discuss regression calibration as a novel statistical technique to address measurement error in dietary data.
• Examine variants of regression calibration, particularly those employing biological surrogates, for improved accuracy.

Main Results:

• Traditional analyses are significantly distorted when dietary assessment errors are present, potentially reducing risk estimates by half or more.
• Regression calibration, especially with biological correlates, offers a more realistic approach to correcting for measurement errors.
• The presence of correlations between nutrients and foods complicates the identification of specific dietary risk factors.

Conclusions:

• Addressing confounding and measurement error is crucial for resolving inconsistencies in dietary epidemiology.
• Regression calibration represents a promising advancement for producing less biased and more consistent results in dietary studies.
• Future research should focus on refining and applying advanced methods like regression calibration to enhance the validity of dietary risk assessments.