Reply to Drs. Giri and Sundaram

We would like to thank Drs. Giri and Sundaram for their interest in our study and for their comments.

In this study, we reviewed data for 1679 patients with colonic diverticular hemorrhage (CDH) who were treated with endoscopic band ligation (EBL) or clipping at 49 hospitals across Japan (CODE BLUE-J study), and the association between endoscopic treatment and clinical outcomes was analyzed using logistic regression models [1]. The treatment strategy was chosen at the discretion of each hospital and endoscopist.

The reason why we used logistic regression analysis and not propensity score matching was in consideration of the number of confounders relative to the number of events. When there are at least eight events per confounder, logistic regression analysis is more precise and less biased than propensity score matching, and the empirical coverage probability and empirical power are adequate [2]. The total number of events (early rebleeding) in the two groups was 340, and the number of confounders was 30, which works out to 11 events per confounder. Therefore, the magnitude of bias would be smaller with logistic regression analysis than with propensity score matching, and we considered that logistic regression analysis would be more appropriate than propensity score matching.

Furthermore, we do have propensity score matching data for the patients in this study. We compared 29 patient characteristic factors in the two groups and found significant differences in 11 factors, including systolic blood pressure, hemoglobin level, and the use of antiplatelet agents and corticosteroids. Propensity score matching analysis of 525 pairs was used to compare outcomes, but the main conclusions of the study remained unchanged.

We agree that further randomized studies are needed to evaluate the appropriate endoscopic treatments for CDH, but it is also true that such trials are challenging in the emergency setting. We hope that our study will be of great help in selecting an endoscopic treatment method for CDH.

Publikationsverlauf

Eingereicht: 18. Dezember 2021

Angenommen: 27. Dezember 2021

Artikel online veröffentlicht:
24. Februar 2022

• References

• 1 Kobayashi K, Nagata N, Furumoto Y. et al. Effectiveness and adverse events of endoscopic clipping versus band ligation for colonic diverticular hemorrhage: a large-scale multicenter cohort study. Endoscopy 2021;
• 2 Cepeda MS, Boston R, Farrar JT. et al. Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders. Am J Epidemiol 2003; 158: 280-287