Is timing everything in the management of acute upper GI bleeding?

Time trends of causes of upper gastrointestinal bleeding and endoscopic findings

BACKGROUND

Upper gastrointestinal bleeding (UGIB) is a frequent cause for emergency endoscopy and, in a proportion, requires the application of endotherapy. We aim to evaluate the proportion of variceal and nonvariceal upper gastrointestinal bleeding (NVUGIB), the endoscopic findings that were detected, as well as the temporal trends of endoscopic findings over a period of 13 years.

METHODS

This is a retrospective study of patients who underwent an esophagogastroduodenoscopy with an indication of UGIB or presented with hematemesis, melena, or both, as well as those who had hematochezia, from January 2004 to December 2016 (13 years).

RESULTS

A total of 2075 patients were included with a mean age of 56.8 years (range 18-113) and males constituted 67.9%, while 65.9% had at least one comorbidity. Symptoms on presentation included hematemesis (52.5%), melena (31.2%), both hematemesis & melena (15.1%), and hematochezia (1.2%). The majority of UGIB were from a NVUGIB source (80.5%) and a variceal source was found in 13.1%, while no endoscopic findings were found in 6.4% of cases. The most common endoscopic diagnosis was gastroduodenal erosions (23.8%), duodenal ulcers (23.5%), reflux esophagitis (16.0%), esophageal varices (12.1%), and gastric ulcers (10.8%). There was no change in the endoscopic findings over the time period of the study. A third of duodenal ulcers (33.3%) as well as 21.9% of gastric ulcers were actively bleeding at the time of endoscopy, while 3.3% of duodenal ulcers had an adherent clot.

CONCLUSIONS

NVUGIB composed the majority of cases presenting with UGIB and variceal bleeding was lower than that described in prior studies, but there were no clear trends in the proportion of causes of UGIB during the study duration.

Using a Multi-Task Recurrent Neural Network With Attention Mechanisms to Predict Hospital Mortality of Patients

Estimating hospital mortality of patients is important in assisting clinicians to make decisions and hospital providers to allocate resources. This paper proposed a multi-task recurrent neural network with attention mechanisms to predict patients' hospital mortality, using reconstruction of patients' physiological time series as an auxiliary task. Experiments were conducted on a large public electronic health record database, i.e., MIMIC-III. Fifteen physiological measurements during the first 24 h of critical care were used to predict death before hospital discharge. Compared with the conventional simplified acute physiology score (SAPS-II), the proposed multi-task learning model achieved better sensitivity (0.503 ± 0.020 versus 0.365 ± 0.021), when predictions were made based on the same 24-h observation period. The multi-task learning model is recommended to be updated daily with at least a 6-h observation period, in order for it to perform similarly or better than the SAPS-II. In the future, the need for intervention can be considered as another task to further optimize the performance of the multi-task learning model.

Advances in management of nonvariceal upper gastrointestinal bleeding

PURPOSE OF REVIEW

To review new advances in managing nonvariceal upper gastrointestinal hemorrhage.

RECENT FINDINGS

Implementation of various scoring systems in combination with video capsule endoscopy assists in stratifying and managing nonvariceal upper gastrointestinal bleeding. New techniques such as thermocoagulation and hemoclips are useful to treat bleeding.

SUMMARY

The advancement of methods and procedures in managing nonvariceal upper gastrointestinal bleeding has decreased mortality of patients presenting with this type of hemorrhage. In this chapter, we will be discussing various scores to stratify nonvariceal upper gastrointestinal bleeding and techniques to stop bleeding.