Reproducibility and Validity of Plant-Based Dietary Indices in the Cancer Prevention Study-3 Diet Assessment Substudy.

Background: The Overall Plant-based Diet Index (PDI), the Healthful Plant-based Diet Index (hPDI), and the Unhealthful Plant-based Diet Index (uPDI) are relatively new tools for characterizing the quality of plant-based dietary patterns in epidemiologic studies. Reproducibility and validity of these indices have not been assessed across populations.

Objective: The objective of this study was to evaluate the reproducibility and validity of PDI, hPDI, and uPDI in a racially and ethnically diverse population.

Methods: Participants (n = 433 women, n = 244 men, n = 417 non-Hispanic White, n = 159 non-Hispanic Black, and n = 101 Hispanic) from the 2015-2016 Cancer Prevention Study 3 Dietary Assessment Substudy (CPS-3 DAS) completed the CPS-3 Food Frequency Questionnaire (FFQ) on 2 occasions (FFQ1 and FFQ2) and ≤6 24-h dietary recalls (24HRs) over 1 y. Plant-based diet scores were calculated using energy-adjusted food group intakes from each FFQ and mean 24HRs. Pearson correlations were used to evaluate reproducibility of each score between FFQs. Validity of the scores derived from FFQ2 was assessed relative to the mean of the 24HRs and blood and urine biomarkers.

Results: Pearson correlations for reproducibility between FFQs for men and women were 0.64 and 0.66 for PDI, 0.85 and 0.80 for hPDI, and 0.84 and 0.78 for uPDI. Pearson correlations for validity between FFQ2 and the mean 24HRs for PDI, hPDI, and uPDI for men were 0.54, 0.68, and 0.60, respectively, and for women were 0.49, 0.0.62, and 0.62, respectively. Biomarkers of most carotenoids, total protein, sodium, and potassium showed correlations ≥0.20 with certain plant-based diet scores. Reproducibility and validity were moderate to good in all racial/ethnic subgroups and were highest in non-Hispanic White adults.

Conclusions: There was good reproducibility and moderate to good validity of PDI within the CPS-3 FFQ. Future work should continue to validate these indices across populations with varied diets and dietary measurement tools.