and LSM by TE) would not be applicable, whereas the other proposed model, platelet-to-spleen ratio (PSR; which includes platelet count and the bipolar diameter of the spleen, first proposed by Dr. Giannini and colleagues) would be always applicable, given that spleen diameter can be measured in 100% of the cases. Indeed, one of the aims of the ANTICIPATE study was to provide different noninvasive tools to predict varices needing treatment, so that in case of failure or unavailability of one tool, an alternative tool could be used. This was done having in mind that the availability of LSM by TE is very limited in some areas of the world. The main conclusion of our study was that "LSPS, a combination of LSM by TE-platelet count, or PSR could be used to identify a relevant subset of patients with compensated cirrhosis and Child-Pugh A with a very low probability of VNT (<5%) in which endoscopy could be avoided." Therefore, we think we sufficiently addressed this concern in our report.The second issue raised by Drs. Giannini and Savarino was that the results of our study should be corrected in an "intention-to-test" analysis, taking into account the technical failures. This is especially important when using traditional measurements of diagnostic performance, such as sensitivity, specificity, and positive and negative predictive values. The approach that we followed in our study was, however, different. We used risk prediction modeling to assign a probability of the target condition for each value of the noninvasive predictor. The model (or its nomogram) can be used if the predictors are available and cannot be used if they are not. The performance statistics of the model only apply to the group of patients in which the model can be applied. Our approach therefore overcomes the question of how to factor technical failures on diagnostic accuracy parameters. The only value reported in our study that could be impacted by the rate of technical failures would be the percentage of endoscopies that would be avoided by a certain noninvasive test. Because our study was a blend of prospective and retrospective data, we cannot provide an exact applicability figure for our models. In a theoretical worst-case scenario of 20% unreliable results/technical failures of LSM by TE, the number of endoscopies saved would decrease from 30% to 24%.