A quantitative exploration of gastrointestinal bleeding in intensive care unit patients

Quantitative assessments of the severity of bleeding in patients with bleeds within the gastrointestinal tract (GIB) are generally limited to blood tests like the hematocrit. The varied and irregular nature of the data collected during such observations makes it difficult in retrospective data analysis to characterize the complete course of bleeding. We intend to quantify the rate of blood loss over the course of an ICU stay, facilitating more precise analysis of retrospective data, and to use this quantification to examine questions about the effects of GIB.

A population of 2,445 intensive care admissions across 2,266 patients with a diagnosis of GIB was studied. Using statistical techniques for smoothing data and accepted medical approaches for calculating blood loss, we are able to convert collections of individual laboratory readings that are difficult to understand into a simple, interpretable overview of the patient's bleeding status over time. To demonstrate this method, we compare patients' standard vital signs while bleeding heavily to times when they are not bleeding, finding a 3.0 ± 0.5% increase in heart rate, a 1.3 ± 0.4% decrease in systolic blood pressure and a 0.9 ± 0.5% decrease in diastolic blood pressure. After considering the effect of bleeding on standard vital signs, we demonstrate that patients with upper GIB have significantly elevated blood urea nitrogen levels while bleeding heavily, with a mean increase of 11.7 ± 7.2%, while patients with lower GIB do not, with a mean increase of 4.2 ± 6.6%.

This study introduces a novel method of processing retrospective laboratory data to characterize the course of bleeds within the gastrointestinal tract. This method is used to examine the direct effects of bleeding on a patient and can be deployed in future studies of bleeding using retrospective data.