Classification schemes across orthopaedics provide a common language for the discussion, prognostication, and study of bone and joint pathology. The current study represents an impressive large-scale effort to evaluate the reliability of a number of classifications of one disease condition, make specific recommendations to improve that reliability through the choice of appropriate radiographs, and correlate the findings with arthroscopic reality. The study holds both narrow and broad lessons for practitioners of knee reconstruction and for orthopaedic practice in general.

The authors have identified and evaluated no fewer than six classification schemes with a varying degree of reliance on the measurement of joint space narrowing and other descriptive interpretations. The report solidly supports three contentions.

First, the best combination of good interobserver reliability and medium correlation with arthroscopic findings is found with the International Knee Documentation Committee (IKDC) scale, the classification scheme among the six that most directly relies on a numerical measurement of the width of the joint space.

Second, the reliability and accuracy of imaging are substantially improved with the use of a radiographic technique in which the beam is tangential to the most common locations of chondral damage. For the knee, this view is the Rosenberg radiograph, a posteroanterior radiograph made with the patient’s knee flexed to 45° to tangentially capture more posterior portions of the femoral condyles than can be seen with use of the standing anteroposterior radiograph.

Third, the availability of bilateral images for a normal comparison view significantly improves interobserver reliability when Rosenberg radiographs are used.

The wise selection of the revision anterior cruciate ligament patient population ensures both the availability of uninvolved contralateral knees for radiographic comparison and a wide range of relatively early-stage osteoarthritis cases. Also noteworthy is consensus-building among the three raters on the more subjective judgments involved in applying the scales, a critical step in all such interobserver reliability studies that in itself has been shown to improve results¹.

The paper, then, provides a clear vision of the current state of the art in a mutually agreed upon assessment of one very common orthopaedic pathology. However, in its subtext, the paper also illustrates the relative failings of this state.

Rating scales that are designed to interpret physical phenomena are ultimately subject to the same limitations as those measuring psychometric outcomes. Reliability does not necessarily imply sensitivity or accuracy. Even when expertly applied, the scales here fail to provide more than a medium correlation with the arthroscopic findings. This outcome begs the question: To what degree is an arthroscopic evaluation an accurate reflection of the global degree of arthritis? It is true that the arthroscopic scoring system with use of the Outerbridge classification has been demonstrated to be reliable in several studies, including the study by Marx et al. cited in the current paper². However, arthroscopy, like radiography, still represents only an image of clinical reality. Few have explored this issue. Using an open arthrotomy as the gold standard, Cameron et al. found an accuracy of arthroscopically determined Outerbridge classification of only 68% in a small study of cadaveric knees³. More problematic, and a likely prime contributor to the limited correlation seen in this and other papers, is the difficulty of marrying the maximum degree of chondral pathology in each compartment with the total extent of disease. To use a mathematical analogy, the current approach uses a local maximum as a substitute for the integrated area under the curve. A more nuanced and thorough arthroscopic evaluation may be required.