Comparative strategies for using cluster analysis to assess dietary patterns

Area of Science:

• Nutritional epidemiology
• Dietary pattern analysis
• Gerontology

Background:

• Understanding dietary patterns is crucial for public health and disease prevention.
• Cluster analysis is a statistical method used to identify groups of individuals with similar dietary habits.
• Different analytical approaches may yield varying dietary patterns, impacting health outcome associations.

Purpose of the Study:

• To characterize dietary patterns in older adults using two distinct cluster analysis strategies.
• To compare the effectiveness of two cluster analysis methods in identifying health-associated dietary patterns.

Main Methods:

• Cross-sectional study involving 179 community-dwelling adults (aged 66-87) in rural Pennsylvania.
• Dietary intake assessed via five 24-hour recalls over 10 months, with foods classified into 24 subgroups.
• Cluster analysis performed using two different strategies to derive dietary patterns.

Main Results:

• Both cluster analysis methods identified dietary patterns with significant differences in diet quality (Healthy Eating Index scores).
• Higher Healthy Eating Index (HEI) scores correlated with increased micronutrient intake.
• The 'number of servings' method highlighted dietary patterns more strongly associated with favorable weight status and higher fruit/vegetable intake compared to the 'percent energy' method.

Conclusions:

• Different cluster analysis methodologies produce both similar and dissimilar dietary patterns.
• The 'number of servings' method appears more sensitive in identifying healthful dietary patterns linked to weight status and fruit/vegetable consumption in older adults.
• Methodological choices in dietary pattern analysis warrant careful consideration for public health recommendations.