Dietary inflammation index and osteoarthritis in the elderly: is there a mediating role of physical activity?

Wang H, Liao R, Tang W, Su W, Zeng M, Yang J, Fan X, Xie J, Hu Y · The British Journal of Nutrition

The index designed to score a diet's inflammatory potential, applied to osteoarthritis. Adjusted for age, sex, BMI, education, race, income, marital status, hypertension, diabetes and serum calcium and phosphorus, OA risk rose across quartiles to OR 1.64 (95% CI 1.13-2.37) in the most pro-inflammatory (p trend 0.002). Adding physical activity to that model takes the same estimate to 1.46 (95% CI 0.92-1.92), p trend 0.102 — the association does not survive it. The authors read activity as a mediator rather than a confounder: it interacts with the index (p < 0.001) and carries 20.1% of the effect in a model allowing exposure-mediator interaction, and among the least active the top quartile runs at OR 2.12 (95% CI 1.46-3.26).

Population

1,249 adults aged 65 and over in NHANES 2011-2016

Intervention

Higher energy-adjusted Dietary Inflammatory Index — a score summarising a diet's inflammatory potential, ranging here from −5.32 to +4.26

Comparator

Lowest quartile of E-DII

Limitations

Cross-sectional NHANES data with self-reported osteoarthritis and diet from a food-frequency questionnaire, so a mediation analysis here cannot establish the order of events. The E-DII is a composite score built from nutrient-level inflammatory weights, a construct rather than an exposure anyone can eat. Whether activity is the pathway or the explanation is the whole question, and this design cannot separate them: a fifth of the effect is carried by activity in the mediation model while the direct estimate loses significance once activity is adjusted for. Parts of the methods and one results heading describe a mediation of depression risk by BMI rather than of osteoarthritis by activity, and several confidence intervals in the main table are not centred on their own point estimates.