# Risk factors associated with knee osteoarthritis severity and their correlation with Kellgren–Lawrence grade

## TL;DR

This study identifies C-reactive protein (CRP) as a strong indicator of knee osteoarthritis severity, suggesting it could be useful in clinical assessments.

## Contribution

The study demonstrates that CRP is strongly associated with KOA severity, even after adjusting for age, BMI, and pain scores.

## Key findings

- CRP showed the strongest correlation with Kellgren–Lawrence grade in both male and female patients.
- The combined model had good discriminatory accuracy for distinguishing mild from moderate-to-severe KOA.
- Female patients showed a numerically stronger association between CRP and KOA severity.

## Abstract

To identify the risk factors associated with knee osteoarthritis (KOA) radiographic severity in a defined cohort of patients with KOA and to examine their associations with Kellgren–Lawrence (K–L) grade.

In this retrospective study, 258 patients with confirmed KOA (100 males and 158 females) diagnosed between July and November 2024 at Shenzhen People’s Hospital were enrolled. Demographic and clinical data—including age, sex, body-mass index (BMI), K–L grade, C-reactive protein (CRP) level, and pain intensity assessed by the visual analogue scale (VAS)—were extracted. Sex-related differences in age, K–L grade, BMI, CRP, and VAS were analyzed using the Mann–Whitney U-test, Chi-square test, and independent t-test. Spearman rank correlation was employed to evaluate the associations between each variable and K–L grade within each sex. Multivariable logistic regression was performed to examine the associations of each factor with mild (K–L 1–2) versus moderate-to-severe (K–L 3–4) KOA. ROC curves were constructed to assess the discriminative accuracy for moderate-to-severe K–L grade.

(1) Significant gender differences were observed in height, weight, and K–L grade distribution among KOA patients (all P < 0.05). In contrast, age, BMI, CRP levels, and VAS pain scores showed no statistically significant gender variations (all P > 0.05). (2) CRP exhibited the strongest correlation with K–L grade in both sexes (r = 0.51 in males, r = 0.60 in females). (3) Multivariable logistic regression analysis, after adjusting for age, BMI, and VAS pain score, revealed that CRP showed the strongest association with K–L grade among the factors in both sexes. (4) ROC analysis revealed good discriminatory performance for the combined model in distinguishing mild from moderate-to-severe KOA, with AUC = 0.865 (95% CI: 0.793–0.938) in males and AUC = 0.880 (95% CI: 0.827–0.933) in females, indicating good discriminatory performance within this dataset. It should be noted that these results reflect the model’s performance in the current sample and internal validation would be needed to assess generalizability.

In this cross-sectional study, CRP showed a strong association with KOA radiographic severity after adjusting for age, BMI, and pain scores. A numerically stronger association was observed in female patients, though this difference requires further confirmation. These findings highlight the potential role of systemic inflammation in KOA and support the further investigation of CRP as a supplementary clinical assessment tool.