Higher Carbohydrate-to-Fiber Ratio Linked to Prevalent RA

Key Takeaways

• Among US adults surveyed in 2009–2018, the highest CFR quartile had 67% greater adjusted odds of prevalent self-reported RA than the lowest quartile.
• The CFR association was approximately linear across the main range, with uncertain curvature at sparse high ratios and no apparent clinical threshold.
• No sex, hypertension, diabetes, or carbohydrate-intake interaction survived correction for multiple testing.
• The exploratory share of the association attributed to SII was estimated at 5% and was not statistically significant.

Current diet can be difficult to interpret in adults who already have rheumatoid arthritis (RA). A carbohydrate-to-fiber ratio (CFR) expresses grams of carbohydrate consumed per gram of fiber; a higher value indicates more carbohydrate relative to fiber. Eating habits may have changed after diagnosis, complicating any connection between today's intake and existing disease. And so investigators compared current dietary composition with reported arthritis status.

Researchers analyzed 2009–2018 National Health and Nutrition Examination Survey (NHANES) data from 19,260 adults aged at least 20 years with valid two-day dietary weights in their cross-sectional study. CFR divided mean total carbohydrate grams by mean fiber grams across two 24-hour recalls. Population-derived quartiles placed the lowest group at ≤11.21 and the highest at >20.02. Respondents had prevalent RA if they reported a health professional's arthritis diagnosis and identified RA as its type; investigators did not clinically adjudicate diagnoses. Survey-weighted logistic regression compared those respondents with arthritis-free adults, excluding other arthritis types. Investigators pooled estimates after multiple imputation and used directed acyclic graph (DAG)-informed adjustment for demographic and socioeconomic characteristics, smoking, alcohol, activity, and energy intake. Expanded metabolic and medication adjustments were sensitivity analyses. Generalized additive models (GAMs), post hoc interactions, and systemic immune-inflammation index (SII) decomposition explored other aspects of the association.

In the primary DAG-informed model, the highest versus lowest CFR quartile had an odds ratio of 1.67 (95% CI 1.25–2.24; individual P < .001) for prevalent self-reported RA relative to arthritis-free adults. This comparison concerns the odds of an existing reported diagnosis, rather than the likelihood of developing RA later.

Survey-weighted prevalence of self-reported RA was 4%, and the distinct pooled global test across CFR quartiles yielded P = .002. The crude global test was not significant. Positive associations persisted in expanded metabolic and medication models and in selection-weighted and missing-data checks; those checks did not remove the possibility of bias. The adjusted GAM association was approximately linear across its main, outlier-excluded range, while curvature in the sparsely populated upper tail was uncertain. No sex, hypertension, diabetes, or total-carbohydrate-intake interaction survived Holm correction, including an initially nominal hypertension signal. Exploratory SII decomposition found a small, nonsignificant indirect association with uncertainty spanning no contribution; it did not establish causal mediation.

Concurrent measurement of diet and existing diagnosis leaves open whether RA preceded changes in eating habits. Two 24-hour recalls may not represent long-term intake, and CFR may reflect a broader, correlated dietary pattern. Residual confounding and selection bias remain possible. RA status lacked clinical adjudication, while the arthritis-free comparison excluded adults with other arthritis types. Quartile boundaries and the outlier fence were sample-derived, not clinical dietary thresholds. An association with prevalent disease does not establish incidence, the effects of changing diet, or an inflammatory mechanism.

According to the investigators, higher current CFR was associated with existing self-reported RA. They concluded that prospective research with repeated dietary assessment and confirmed incident RA would be needed to clarify the timing of the association.

Clinician Questions

Which patients are represented by the carbohydrate-to-fiber ratio and rheumatoid arthritis association in NHANES?

Survey weighting targets US civilian noninstitutionalized adults represented by NHANES, subject to its dietary-weight and arthritis-status eligibility rules. The carbohydrate-to-fiber ratio association with rheumatoid arthritis cannot automatically be extended to children, institutionalized patients, or non-US populations.

Were osteoarthritis and psoriatic arthritis also associated with dietary carbohydrate-to-fiber ratio?

In the primary adjusted model, global tests for osteoarthritis (P = .84) and psoriatic arthritis (P = .08) were not significant, and remained so after Holm correction. A nominally significant individual psoriatic arthritis comparison (OR 2.67) rested on only 66 cases and wasn’t supported by the global test.