Exploring Biases of the Healthy Eating Index and Alternative Healthy Eating Index When Scoring Low-Carbohydrate and Low-Fat Diets

Background: The Healthy Eating Index 2010 (HEI-2010) and Alternative Healthy Eating Index 2010 (AHEI-2010) are commonly used to measure dietary quality in research settings. Neither index is designed specifically to compare diet quality between low-carbohydrate (LC) and low-fat (LF) diets. It is unknown whether biases exist in making these comparisons. Objective: The aim was to determine whether HEI-2010 and AHEI-2010 contain biases when scoring LC and LF diets. Design: Secondary analyses of the Diet Intervention Examining the Factors Interacting With Treatment Success (DIETFITS) weight loss trial were conducted. The trial was conducted in the San Francisco Bay Area of California between January 2013 and May 2016. Three approaches were used to investigate whether biases existed for HEI-2010 and AHEI-2010 when scoring LC and LF diets. Participants/setting: DIETFITS participants were assigned to follow healthy LC or healthy LF diets for 12 months (n = 609). Main outcomes measures: Mean diet quality index scores for each diet were measured. Statistical analysis: Approach 1 examined both diet quality indices' scoring criteria. Approach 2 compared scores garnered by exemplary quality LC and LF menus created by registered dietitian nutritionists. Approach 3 used 2-sided t tests to compare the HEI-2010 and AHEI-2010 scores calculated from 24-hour dietary recalls of DIETFITS trial participants (n = 608). Results: Scoring criteria for both HEI-2010 (100 possible points) and AHEI-2010 (110 possible points) were estimated to favor an LF diet by 10 points. Mean scores for exemplary quality LF menus were higher than for LC menus using both HEI-2010 (91.8 vs 76.8) and AHEI-2010 (71.7 vs 64.4, adjusted to 100 possible points). DIETFITS participants assigned to a healthy LF diet scored significantly higher on HEI and AHEI than those assigned to a healthy LC diet at 3, 6, and 12 months (all, P < .001). Mean baseline scores were lower than mean scores at all follow-up time points regardless of diet assignment or diet quality index used. Conclusions: Commonly used diet quality indices, HEI-2010 and AHEI-2010, showed biases toward LF vs LC diets. However, both indices detected expected changes in diet quality within each diet, with HEI-2010 yielding greater variation in scores. Findings support the use of these indices in measuring diet quality differences within, but not between, LC and LF diets.