Authors
Zonghong Wu, Zhiyang Du, Dong Liu, Huihui Li, Runan Pan
Published in
Medicine. Volume 105. Issue 47. Pages e51267. Nov 20, 2026.
Abstract
Osteoarthritis (OA) is clinically heterogeneous. Most epidemiologic studies classify OA as present or absent based solely on self-report or radiographic criteria (binary diagnosis), which overlooks the wide variation in symptom burden and functional impact among individuals with OA. We aimed to define a high-burden OA phenotype in a nationally representative sample of US adults and to evaluate its clinical correlates and prognostic significance. This cross-sectional and longitudinal study used nationally representative National Health and Nutrition Examination Survey 2011-2018 data from adults aged ≥20 years, with mortality follow-up through the 2019 public-use Linked Mortality File. OA was defined by self-reported physician-diagnosed arthritis with self-classification as osteoarthritis/degenerative arthritis. High-burden OA was identified using work limitation, activity limitation, walking difficulty, need for special equipment, and difficulty with routine physical tasks. Conventional regression and survival analyses incorporated National Health and Nutrition Examination Survey sampling weights, strata, and primary sampling units. Least absolute shrinkage and selection operator (LASSO), random forest, and extreme gradient boosting models were trained with sampling weights; held-out performance was additionally evaluated using survey-weighted metrics. Among 22,436 eligible adults, 14,772 were included in the complete-case analysis: 12,918 without OA, 662 with low-burden OA, and 1192 with high-burden OA. Most weighted baseline characteristics differed across the 3 groups (all P ≤ .005), whereas lifetime alcohol use did not (P = .568). Age- and sex-specific prevalence differed by age-group and sex (both P < .001), without evidence of an age-by-sex interaction (P = .395), and prevalence increased across body mass index categories (P for trend < 0.001). In weighted held-out evaluation, least absolute shrinkage and selection operator had the highest area under the receiver-operating-characteristic curve (0.848), followed by random forest and XGBoost (both 0.839). Among 1853 participants in the OA survival cohort, 177 deaths occurred during a median observed follow-up of 4.58 years. High-burden OA was associated with all-cause mortality in the fully adjusted survey-weighted Cox model (hazard ratio 1.70, 95% confidence interval 1.08-2.68; P = .022). High-burden OA identifies a clinically vulnerable subgroup with greater functional limitation, worse metabolic and psychological health, and poorer survival. Burden-based stratification may better capture clinically meaningful heterogeneity than OA diagnosis alone.
PMID:
43063114
Bibliographic data and abstract were imported from PubMed on 21 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 3
- Comments 0