The relationship between low-carbohydrate diet score, dietary macronutrient intake, and rheumatoid arthritis: results from NHANES 2011-2016.

BACKGROUND

This study sought to determine if dietary macronutrient consumption and the low-carbohydrate diet (LCD) score were linked to rheumatoid arthritis (RA).

METHODS

Participants ≥ 20 years were analyzed from the National Health and Nutrition Examination Survey (NHANES) 2011-2016. LCD score was calculated by summing the 11 quantiles values of the percentages of energy derived from carbohydrate, protein, and fat. Weighted logistic regression, eXtreme Gradient Boosting (XGBoost), and Light Gradient Boosting Machine (LightGBM) models were used to explore the relationship between LCD score, dietary macronutrient intake, and RA. Propensity score matching (PSM) were applied for sensitivity analysis.

RESULTS

Ultimately, 8118 participants (RA: 499, without RA: 7619) were analyzed. After fully adjusting for confounders, a negative association was found between the LCD score and the presence of RA [OR (95% CI), 0.97 (0.96, 0.99)]. A higher LCD score was also negatively associated with a lower likelihood of RA based on a categorical model. Among macronutrients, participants in the third and fourth quartiles had significantly increased odds of RA compared with the lowest carbohydrate intake. Regarding protein intake, individuals in the highest quartile of percentage of energy from protein had a 46% lower presence of RA compared with the lowest reference group. The relative importance of the LCD score on RA was determined based on XGBoost and LightGBM models. Moreover, the association between the LCD score, dietary macronutrient intake, and RA presence remained substantial after PSM.

CONCLUSIONS

LCD score was negatively associated with odds of RA in US adults. Moreover, a correlation was found between a lower likelihood of RA and high protein, and low carbohydrate consumption. Key Points • A significant negative association was found between LCD score and RA presence. • Machine learning models revealed the LCD score was a significant predictor of the presence of RA. • Low carbohydrate intake and high protein intake were correlated with a lower odds of RA.