Intra- and inter-observer reliability of ten major histological scoring systems used for the evaluation of in vivo cartilage repair.

PURPOSE

In the last two decades, many surgical techniques have been described for articular cartilage repair. Reliable histological scoring systems are fundamental tools to evaluate new procedures. Several histological scoring systems have been described, and these can be divided in elementary and comprehensive scores, according to the number of sub-items. The aim of this study was to test the inter- and intra-observer reliability of ten main scores used for the histological evaluation of in vivo cartilage repair. The authors tested the starting hypothesis that elementary scores would show superior intra- and inter-observer reliability compared with comprehensive scores.

METHODS

Fifty histological sections obtained from the trochlea of New Zealand Rabbit and stained with Safranin-O fast green were used. The histological sections were analysed by 4 observers: 2 experienced in cartilage histology and 2 inexperienced. Histological evaluations were performed at time 1 and time 2, separated by a 30-day interval. The following scores were used: Mankin, O'Driscoll, Pineda, Wakitani, Fortier, Selleres, ICRS, ICRSII, Oswestry (OsScore) and modified O'Driscoll. Intra- and inter-observer reliability were evaluated for each score. In addition, the pavement-ceiling effect and the Bland-Altman Coefficient of Repeatability were then evaluated for each sub-item of every score.

RESULTS

Intra-observer reliability was high for all observers in every score, even though the reliability was significantly lower for non-expert observers compared with expert counterparts. In terms of Coefficient of Repeatability, some scores performed better (O'Driscoll, Modified O'Driscoll and ICRSII) than others (Fortier, Seller). Inter-observer reliability was high for all observers in every score, but significantly lower for non-expert compared with expert observers.

CONCLUSIONS

In expert hands, all the scores showed high intra- and inter-observer reliability, independently of the complexity. Although every score has advantages and disadvantages, ICRSII, O'Driscoll and Modified O'Driscoll scores should be preferred for the evaluation of in vivo cartilage repair in animal models.