Mylène Jansen

56 Chapter 3 far the only cohort of KJD patients followed for such a long period of time. While the small number of patients could have resulted in larger confidence intervals, the intervals of parameters significantly associated with survival do not cross 1. As the data does not allow the usage of multivariable modeling, Cox regression analyses had to be performed separately for every parameter. This may have increased the rate of type I errors as a result of multiple testing. A best-worst case Kaplan-Meier and Cox regression analysis was performed, where the 3 patients lost to follow-up were either defined as failures at the moment of drop-out or as survivors for at least 9 years. These results did not change the outcome of this study significantly (see supplementary data). In conclusion, joint distraction for knee OA patients considered for TKA shows long-lasting clinical and structural improvement and an overall survival of the native knee 9 years after treatment in half of the treated patients and over 2/3 in males and those with the better initial structural response after KJD.