Abstract
Our study aims to investigate if the ratios proposed by Insall-Salvati and Caton-Deschamps follow the theory of normal distribution levels in a healthy population. 434 skeletal mature, healthy knees were obtained from a CT-scan-based modelling system (SOMA). Patellae height ratios were measured using the Insall-Salvati ratio and the Caton-Deschamps index. The patella height ratios of the sample population were plotted on a quantile plot and diagnosed as patella alta, baja or normal using the original ratio definitions. The study population patella height values at the 95th, 98th and 5th centiles were identified and compared to those described as alta or baja by Insall-Salvati and Caton-Deschamps. This meant that if the patient had patella alta, this would be defined as a ratio of ≥ 1.2 and hypothesised that this would align at the 98th centile of the study population, whilst if diagnosed as patella baja (≤ 0.74), this should align at the 5th centile of the population. The inter-rater reliability of both ratios was calculated using kappa statistics. Two authors made all calculations and compared them for consistency using the intraclass correlation coefficient. For the Insall-Salvati ratio, the study population's value at the 98th centile was equivalent to what Insall-Salvati describes as alta, a ratio of ≥ 1.2. In the study population, patients are overdiagnosed if using the Insall-Salvati ratio as patella baja ≤ 0.74, as the value at the 5th centile in the study population was 0.59. Using the Caton-Deschamps index, the 95th centile was 1.3, higher than the patella alta ratio determined by Caton as a ratio ≥ 1.2. Using Caton-Deschamps, patients are being underdiagnosed if utilising a patella baja ratio of ≤ 0.6. In our population, the 5th centile was 0.75. The two ratios had a kappa coefficient of 0.01, which indicates poor inter-rater reliability. Depending on the ratio used, caton-Deschamps and Insall-Salvati ratios diagnose patients as alta or baja differently. The current patella height ratios have originated from very small, select population samples. The future of patella height discussion should begin with data representing large populations. Then, we can discuss the upper and lower limits of abnormality.