Fig. 1

Graphic illustration of the PISA method. Original inverse Kaplan-Meier graphs were first captured from the PDF of the article as high-definition images (.png) and then converted into data using the UN-SCAN-IT Graph Digitalizer software. Digitalized data were visualized and misplaced points (< 0.1%) manually shifted or erased and replaced. Time (x-axis) spacing was then forced to 0.25 months, and when missing the incidence (y-axis), data was automatically interpolated with the linear method using the two closest points. The integral function of the inverse KM curves (Time-Lost) was obtained by applying piecewise integration using the trapezoid rule with equal x segments. Then, the Time-Gain curve was obtained from the difference between the integrals of placebo and active treatment, plotted as a function of time. Finally, we performed curve fitting on data up to the time when 50% of the total population was still being followed up