Imaging: do erosions heal?

It was shown previously that reparative changes in erosions can be seen in individual joints and individual patients. Whether repair may occur at a group level, and can be induced by treatment, is not known. This manuscript describes a means of visualising data obtained in a clinical trial by the use of probability plots to better understand the results. These probability plots give a good insight into the coherence of the data. They can also be used to make the interpretation of repair at a group level easier. Probability plots also explain the hazard of using binomial cut off points to compare treatment effects. The interference of true repair with measurement error is demonstrated. Repair at a group level is suggested if the mean progression score is statistically significantly different from zero, which can be visualised by a 95% confidence interval of this mean change score below zero. Application of this technique may give us better information on the effects of new drugs on the induction of repair.