S.R. Ward and C.M. Powers reply:
We would like to thank Benedict A. Rogers, MA, MSc, MRCS, for his insightful questions about our recent manuscript. Each point brought forth in the original letter is addressed below.
First, two papers by Møller et al.1,2 were referenced in our study. As Mr. Rogers correctly points out, one study (reference 6 in our manuscript) demonstrates the relationship between patellofemoral malalignment and histological signs of articular cartilage damage in a rabbit model1. The second paper (reference 23 in our manuscript) demonstrates the relationship between patellar height and patellofemoral incongruence in human knees2. Reference 6 was incorrectly cited in the introduction of our paper (line 18). We apologize for this oversight; however, a thorough discussion of our findings relative to Dr. Møller's human data (correctly cited) is included in the Discussion section.
Second, one investigator measured the Insall-Salvati index in all cases. However, this investigator was blinded to the alignment and contact area measurements. The Insall-Salvati index has demonstrated interobserver reliability in our hands and has been shown to be comparable with measurements made on lateral radiographs of the knee3.
Third, the Insall-Salvati index4 was originally measured on lateral radiographs made with the knee flexed to 20° to 30° as Mr. Rogers correctly notes. The purpose of flexing the knee was to remove slack from the extensor mechanism and to allow the patella to engage with the trochlea. In our experiment, the leg was loaded in 0° of knee extension with 25% of the subject's body weight, which removes slack from the extensor mechanism. Regarding engagement with the femoral trochlea, we measured the height of the patella with the knee in extension and at 20° of flexion in each subject and found these measurements to be nearly identical (intraclass correlation coefficient = 0.92).
Fourth, the Insall-Salvati index4 does lack sensitivity to patellar morphology as Mr. Rogers correctly notes. Interestingly, our original hypotheses about the lack of correlation between the height of the patella and the amount of patellofemoral malalignment included suboptimal measurements of patellar height. For this reason, we originally measured patellar height with use of a variety of published indices, including those noted in the report by Mr. Rogers and colleagues5. However, none of them had associative values with malalignment and contact area that were as strong as the Insall-Salvati index. This was in direct contradiction to our original hypothesis. In response, we went back to our original data and determined that large Insall-Salvati indices were always driven by long patellar tendon lengths and not by patellar geometry6. Although we acknowledge that this has been reported in the literature, it was not apparent in our data. Therefore, the most simple measurement (the Insall-Salvati index) explained the largest amount of the variance in alignment and contact area. Perhaps this is why the measurement has survived in practice for more than thirty-five years.
These letters originally appeared, in slightly different form, on . They are still available on the web site in conjunction with the article to which they refer.