Managing Outliers in Adolescent Food Frequency Questionnaire Data

Area of Science:

• Nutrition Science
• Biostatistics
• Adolescent Health

Background:

• Accurate dietary assessment is crucial for understanding adolescent health.
• Food frequency questionnaires (FFQs) are common dietary assessment tools.
• Outlier data can significantly influence FFQ analysis results.

Purpose of the Study:

• To evaluate the impact of five distinct outlier removal decision rules on adolescent FFQ data.
• To assess how these rules affect the analysis of FFQ data and its relationship with weight-related variables.

Main Methods:

• Secondary analysis of data from a weight loss intervention trial involving African American adolescents.
• Utilized baseline and 3-month data, including self-reported FFQs, mediators of weight, caregiver-reported executive functioning, and measured weight status.
• Employed descriptive statistics and correlational analyses to examine variable patterns and relationships.

Main Results:

• Applying outlier decision rules reduced sample size and data range compared to no outlier removal.
• Baseline FFQ-mediator relationships were attenuated, while follow-up relationships showed varied effects (increase, decrease, reversal).
• Outlier rule application strengthened the relationship between estimated energy intake change and weight change under specific fixed-range rules.

Conclusions:

• The choice of outlier decision rules can substantially alter findings from adolescent FFQ data.
• Sensitivity analyses are recommended to report the effects of different outlier handling methods.
• Careful consideration and transparent reporting of outlier treatment are essential for robust dietary assessment research.