Application of a Food-Based Nutritional Profiling System to Assess Diet Quality in Diet-Level Data: Evidence on Construct and Convergent Validity from the Hatoyama Cohort Study and Kusatsu Cohort Study

TL;DR

This study evaluates how well a nutritional profiling system works for assessing diet quality in older adults using questionnaire data from two Japanese cohorts.

Contribution

The study introduces the first nutritional profiling system for older adults and shows how modifying it improves alignment with established diet quality indices.

Findings

Removing nutrient caps in the MNPS-OA significantly improved correlations with HEI-2015 and NRF9.3.

The modified MNPS-OA (WEL-WC) showed strong convergent validity with existing diet quality indices.

The system is adaptable for large-scale epidemiological studies and public health applications.

Abstract

Nutritional Profiling Systems (NPSs) are designed to classify foods by nutritional quality, but most validations occur at the food level. Their applicability to diet-level data from dietary questionnaires—commonly used in epidemiology—remains unclear.

The study aims to evaluate the construct and convergent validity of the Meiji NPS for Older Adults (MNPS-OA), the first NPS developed for older populations, when applied to diet-level data.

Cross-sectional data from the Hatoyama Cohort Study and the Kusatsu Cohort Study involving 1102 Japanese adults aged ≥65 y were utilized. Dietary data were analyzed using a validated Brief Dietary History Questionnaire. Four MNPS-OA specifications were tested: 1) original, 2) without energy limits (WEL), 3) without nutrient caps (WC), and 4) without energy limit and nutrient caps (WEL-WC). Construct validity was assessed by intermodel correlations; convergent validity was evaluated against Healthy Eating Index-2015 (HEI-2015) and Nutrient-Rich Food Index 9.3 (NRF9.3).

Median MNPS-OA scores ranged from 124.5 (original) to 391.6 (WEL-WC). Correlations with HEI-2015 improved from r = 0.27 (original) to 0.58 (WEL-WC), and with NRF9.3 from r = 0.26 to 0.61. Removing nutrient caps substantially enhanced convergence, whereas removing energy limits had minimal effect.

MNPS-OA can be adapted for diet-level assessment with targeted modifications. Eliminating nutrient caps markedly improves alignment with established diet quality indices, supporting its potential use in large-scale epidemiological studies and public health applications.

Figures