Patellar height ratios. A comparison of four measurement methods.

Three blinded observers on three separate occasions calculated four commonly employed patellar height ratios on the knees of 15 patients who had three lateral radiographs each. The observers used the same measurement instrument, a hand-held goniometer ruler, to determine the relative reliability of each patellar height ratio. The measurements by the three observers were examined, and the error and reliability of the four methods of measurement were tested statistically. Among the four methods of measuring patellar height that we studied, the Blackburne-Peel method most consistently reproduced the patellar height index. Interobserver measurement error averaged 0.06 for all ratios. Values greater than 0.06 represented real patellar height changes. The ratios were not significantly affected by the change of knee flexion angle from 30 degrees to 50 degrees. Side-to-side patellar height measurement differences averaged 0.16, which suggests that the healthy contralateral limb is not reliable as a control. Mild arthritic changes decreased variability of measurements by an average of 24% because small osteophytes better defined the articular margins of the patella.