LAUSR

We appreciate the reading and commenting on our article by Sabour Siamak regarding the accuracy of oral 67Gallium citrate scintigraphy in the assessment of inflammatory activity of Crohn’s Disease. In his letter, Siamak performs contributions about accuracy estimate by likelihood ratio (LR) and assess of agreement [1]. The methodological analysis of our work involved additional complexity. The major challenge, in this study, was to consider multiple observations (intestinal segments) per individual, which implies that the observations within a patient are grouped when the level of interest involves these segments. Observations within patients (clusters) are often positively correlated, which can lead to skewed estimates of sensitivity and specificity when the nature of the data within each cluster is ignored. Hence, an analysis that ignores this correlation produces an estimated standard error equivocally, when considering all observations as independent ones [2]. Thus, data analysis that includes multiple observations per patient requires some form of adjustment to explain the possible correlation between the observations. On purpose to correct this underestimated or overestimated variance problem (bias), a ratio estimator for the grouped binary data variance was calculated [3, 4]. In Crohn’s disease activity, the result was typically counted as true-positive if the standard exam (histology) and the scintigraphy show the presence of significant activity inflammation anywhere in the same part of the segment. The result of a patient’s test is typically counted as false-positive if the standard reference examination (histology) did not demonstrate the significant presence of activity inflammation and scintigraphy showed at least one significant inflammation in any segment part. LRs combine sensitivity and specificity to quantify how useful a new diagnostic test is to change (increase or decrease) the likelihood of having a disease compared to the prevalence of that disease (pretest probability) in the population studied. Despite their many advantages, however, LRs are rarely used, primarily because interpreting them requires a calculator to convert back and forth between the probability of disease (a term familiar to all clinicians) and odds of disease (a term mysterious to most people other than statisticians and epidemiologists) [5]. The assumption of the evaluation of our article was to identify disease activity in patients previously identified with Crohn’s disease, so the disease was known, and prevalence was not. Thus, we believe for now, that the validity of the unexplored test is more important to report than likelihood ratio. Obviously, any kind of accuracy of a diagnostic test depends on the prevalence but multiple measures of diagnostic accuracy exist to describe the performance of a medical test and their calculation from the collected data. In our opinion, the authors should report the methods used for calculating the measures that they considered appropriate for their study objectives. In STARD recommendations, the term diagnostic accuracy is using when at least one measure is described (sensitivity, specificity or area under curve) [6]. All the calculations performed support the results and conclusions of this unprecedented modality of activity evaluation in inflammatory bowel disease. This reply refers to the comment available online at https ://doi. org/10.1007/s1214 9-020-01470 -x.