Profiling the socioeconomic characteristics, dietary intake, and health status of Korean older adults for nutrition plan customization: a comparison of principal component, factor, and cluster analyses

TL;DR

This study compares three statistical methods to profile the socioeconomic, dietary, and health characteristics of older adults in Korea to help customize nutrition plans.

Contribution

The study compares PCA, FA, and CA for profiling older adults, revealing consistent key factors across methods.

Findings

PCA identified 4 components explaining 71.6% of the variance.

CA grouped participants into 4 clusters with significant differences in diet, health, and income.

All methods highlighted dietary intake, social support, and health status as key factors.

Abstract

This study was conducted to establish profiles of socioeconomic characteristics, dietary intake, and health status among Korean older adults by employing 3 multivariate analysis techniques.

Data were obtained from 1,352 adults aged 65 years and older who participated in the 2019 Korea National Health and Nutrition Examination Survey. Principal component analysis (PCA), factor analysis (FA), and cluster analysis (CA) were utilized for profiling, with data preprocessing undertaken to facilitate these approaches.

PCA, FA, and CA yielded similar results, reflecting the high common variance among the variables. PCA identified 4 components, accounting for 71.6% of the accumulated variance. FA revealed 5 factors, displaying a Kaiser-Meyer-Olkin value of 0.51 and explaining 74.3% of the total variance. Finally, CA grouped the participants into 4 clusters (R2=0.465). Both PCA and FA identified dietary intake (energy, protein, carbohydrate, etc.), social support from family (incorporating family structure, number of family numbers, and engagement in social eating), and health status (encompassing oral, physical, and subjective health) as key factors. CA classified Korean older adults into 4 distinct typologies, with significant differences observed in dietary intake, health status, and household income (p<0.01).

The study utilized PCA, FA, and CA to analyze profiling domains and derive characteristics of older adults in Korea, followed by a comparison of the results. The variables defining the clusters in CA were consistent with those identified by PCA and FA.

Figures