Inter- and intra-observer reliability of radiological grading systems for knee osteoarthritis

Radiographs are the first-line investigation for knee osteoarthritis (OA). Antero-posterior (AP) and Rosenberg views have demonstrated varying sensitivities in detecting severe OA. There is a multitude of radiological grading systems described. The aim of this study was to assess four commonly used grading systems in terms of inter- and intra-observer reliability and correlate the results with patient-reported symptom severity.

This study included 100 patients who had undergone both AP and Rosenberg X-ray views of the knee and evaluated using four radiological grading systems. These included Kellgren & Lawrence (K&L), International Knee Documentation Committee (IKDC), Ahlbäck and Fairbank. Evaluations were performed by two observers, independently, at two separate time points, 8 weeks apart. The data was then statistically analysed and inter- and intra-observer reliability calculated.

A significant difference was found between AP and Rosenberg views for all four grading systems (p < 0.001). Both inter- and intra-observer reliability were highest for IKDC (0.79 and 0.86 respectively) and lowest for Fairbank (0.20 and 0.27 respectively). No statistically significant correlation was found between any grading system and the Oxford knee score (OKS).

IKDC was found to be the most reliable grading system and Fairbank was the least. The Rosenberg view was overall more sensitive at detecting severe OA than the AP view. Radiological findings did not correlate with patient-reported symptoms (OKS).