Empirical Approach to HEI scores in postmenopausal women

From the Nutrition and Physical Assessment Study (NPAAS) Group, Public Health Sciences Division

In recent years, researchers have highlighted the relationship between eating patterns and various nutritional benefits. The majority of the prominent health patterns including updated versions of the Health Eating Index (HEI) are similar to US Dietary Guidelines and expert reports in scoring systems. Recent literature reports an association between these indices and incidence and mortality rates among various chronic disease outcomes. In a past study, the NPAAS Group from the Public Health Sciences Division developed biomarker measures of HEI scores using Women’s Health Initiative (WHI) data to analyze chronic disease outcomes. Dr. Prentice elaborated, “Our work over the past 18 years in the Nutrition and Physical Activity Assessment Study aims to introduce strengthened measures of dietary intakes into dietary assessment and related chronic disease association studies, which have heretofore relied mainly on uncorrected self-reported diet.” In the current study, they applied an empirical scoring approach based on individual HEI components plus total energy consumption. The HEI components were each defined as ratios of intakes to total energy.

Their new study proposed a new Empirical-Scores Healthy Eating Index (E-HEI), combining dietary variables more efficiently and allowing for the influence of dietary composition on total energy consumption to determine health implications. Dr. Prentice explained the objective of the study, “The present study is a step in this direction that allows an adjustment for measurement error in the self-reported diets of 67,000 WHI participants, for the intakes of 15 dietary variables thought to be among the most important for determining chronic disease risk. These variables derive from those recommended for either increase or decrease in the Healthy Eating Index-2010, along with total calorie consumption.” In the current study, the NPAAS Group reported the association between E-HEI quintiles and all-cause mortality in WHI cohorts, using a split-sample cross validation procedure, and the resulting associations were compared with those using HEI-2010 variables and scores in the same split-sample cross-validation analyses. The study is published in the Journal of Nutrition, Nutritional Epidemiology.

The WHI Nutrition Biomarker Study (NBS) and NPAAS include doubly labeled water (DLW) assessments of total energy intake during a 2-wk period for each participant. Regression equations were used to calculate DLW-calibrated total energy intakes during baseline for NBS and NPAAS. The goal of this calculation was to correct systemic biases in FFQ total energy assessment in both studies. Multivariate calibration equations, regression models, and measurement models were used for dietary intake calibrations in NBS and NPAAs. Cox regression models were utilized to analyze all-cause and disease category-specific mortality rates in the participant data.

Dr. Prentice summarized the results: “Following measurement error adjustment, we found favorable associations with lower risk of death for higher vegetable and fruit intake, for whole and refined grains, and for unsaturated fat consumption, and unfavorable associations with saturated fat and total calorie intake, and especially with salt consumption. Results were similar for risks of cardiovascular death and cancer death specifically. We used these findings, which are novel to the dietary patterns area for refined grains and total calories, to define a new Empirical-scores Healthy Eating Index, which turned out to relate considerably more closely to risk of death than did the previously proposed Healthy Eating Index -2010.”

The authorship group elaborated on the importance of these analysis: “The new feature of these analyses was the simultaneous allowance for measurement error in the 15 dietary variables considered, using replicate dietary assessments in our WHI nutrition biomarker studies along with an excellent biological measure (i.e. biomarker) of total calorie intake” This paper implies a stronger role for diet in determining chronic disease mortality risk than had previously been appreciated, since earlier disease association studies made little or no allowance for measurement error in the self-reported dietary data utilized. The NPAAS Group’s next study brings a biological assessment of the Healthy Eating Index-2010 overall into chronic disease risk associations. Dr. Prentice explains, “We are also at the early stages of bringing biological assessments using high-dimensional small molecule measures in blood and urine into the dietary assessment for the same 15 dietary variables, toward a further improved Empirical-scores Healthy Eating Index.”


Graphical Representation of Dietary pattern indices used in WHI publications and other cohorts
Dietary pattern indices used in WHI publications and other cohorts Image from Dr. Prentice

This research was supported by the National Heart, Lung, and Blood Institute, National Institute of Health, National Cancer Institute, and the US Department of Health and Human Services.

Fred Hutch/UW Cancer Consortium members Dr. Ross L. Prentice, Dr. Ying Huang, Dr. Lesley Tinker, Dr. Garnet Anderson, Dr. Johanna Lampe, Dr. Shirley Beresford and Dr. Marian Neuhouser contributed to this work.

Prentice RL, Aragaki AK, Van Horn L, Thomson CA, Tinker LF, Manson JE, Mossavar-Rahmani Y, Huang Y, Zheng C, Beresford SA, Wallace R. Mortality Associated with Healthy Eating Index Components and an Empirical-scores Healthy Eating Index in a Cohort of Postmenopausal Women. The Journal of Nutrition. 2022 Mar 15.