We have read with great interest the article by Li et al. discussing the role and benefit of serum gastrin and epidermal growth factor (EGF) in predicting the early diagnosis of gastric ulcer and acute upper gastrointestinal bleeding (AUGIB).1 We believe that there are a few issues that need attention. As highlighted in the current literature, various scoring systems have been established, namely Rockall, AIMS65, Glasgow Blatchford and Progetto Nazionale Emorragia Digestiva score pertaining to AUGIB prediction.2 Among all, the Glasgow Blatchford score has the highest accuracy at predicting the need for an urgent endoscopy or outpatient management.2 The score consists of certain vital markers, in which, the important biochemical markers are serum haemoglobin and urea. We were wondering whether the addition of other serum markers can enhance the current well established predictive marker for AUGIB.
We believe that this method of “case-control” participant recruitment is subjected to selection bias and may overestimate the diagnostic accuracy.3,4 It would be more ideal for the authors to randomly select the subjects, or use a consecutive sampling method whereby all patients who were admitted for gastric ulcer in a specified period of time are recruited, irrespective of the outcomes. In Figure 1, it was also implied by the authors that the diagnostic accuracy of serum gastrin and EGF obtained from the ROC curves was able to predict gastric ulcer and AUGIB. We believe that this statement is inaccurate and may exaggerate the accuracy of both serum markers. Since both subject groups had gastric ulcer, the diagnostic accuracy values for serum gastrin and EGF in this study may only predict AUGIB in a population in which gastric ulcer is present, and not the general population. The more accurate title for Figure 1, in our opinion, would be “ROC curves of serum gastrin and EGF to predict AUGIB in patients with gastric ulcer,” rather than “ROC curves of serum gastrin and EGF to predict gastric ulcer and acute upper gastrointestinal bleeding.” This also applies to all other similar statements in the article.
CONFLICT OF INTEREST:
The authors declared no conflict of interest.
FH: Conceptualise the idea and drafted the letter.
MAAR: Provided input on statistical methods.
SSSAR: Involved in literature search.
All authors agreed on the final version of this letter.