Urgent endoscopy within 6 hours of hospital presentation was an independent predictor of reduced mortality, but not rehemorrhaging, in high-risk patients with acute nonvariceal upper gastrointestinal bleeding (UGIB), according to Korean researchers.
In 961 consecutive high-risk emergency department (ED) patients (81% men, mean age 57) with Glasgow-Blatchford scores (GBS) of >7 (mean 12.1), the mortality rate was 1.6% with urgent endoscopy versus 3.8% with elective endoscopy, reported Yoon-Seon Lee, MD, PhD, of Ulsan College of Medicine in Seoul, and colleagues.
The two timing groups also differed in the number of transfused packed red blood cells (urgent 2.6-2.5 versus elective 2.3-2.1 packs), need for intervention (69.5% versus 53.5%), and embolization (2.8% versus 0.5%), they wrote in Clinical Gastroenterology and Hepatology.
The authors noted that endoscopy timing in high-risk patients is controversial. While the 2010 International Consensus Recommendations call for the procedure endoscopy within 24 hours of presentation, accumulating evidence has suggested that endoscopy within 12 hours may provide better outcomes. Data are sparse and conflicting on the benefit of much earlier endoscopy within 6 hours of hospital admission.
“Considering the poor outcomes of high-risk patients, it is critical to predict the clinical course of these patients in the ED, and provide timely and appropriate treatment,” they stated.
All patients underwent endoscopy at a tertiary-care center for acute nonvariceal UGIB during 2005-2014. “Urgent” referred to endoscopy performed earlier than 6 hours after presentation and “elective” to that done within 6-48 hours.
Overall, 571 (59.4%) underwent urgent endoscopic investigation, and these patients were somewhat older than their elective counterparts at a mean of approximately 58 versus 55, and were also unstable hemodynamically. Comorbidities were comparable in the two groups.
In the study’s two primary endpoints, the overall 28-day mortality rate was 2.5% (24 patients) and the rebleeding rate was 10.4% (100 patients).
Stratifying by urgent or elective endoscopy, the investigators found slight intergroup differences in rebleeding rate (11.4% versus 9.0%), ICU admission (4.4% versus 4.7%), vasopressor use (1.9% versus 0.8%), and length of stay (6.7 versus 6.4 days).
Surgery was performed in six patients with urgent endoscopy, but no statistically significant difference in surgery rate was evident between groups (1.1% versus 0%).
Among the diagnoses on endoscopy, gastric and duodenal ulcers were identified in 67.8% of patients, followed by Mallory-Weiss tear in 9.3%, Dieulafoy lesion in 4.6%, and vascular ectasia in 2.3%. Of the 100 patients who rebled within 28 days, 8% died.
Increased mortality was associated with malignancy (odds ratio 3.58, 95% CI 1.33-9.62) and cirrhosis (OR 4.67, 95% CI 1.85-11.76). Urgent endoscopy reduced the OR of death to 0.36 (95% CI 0.14-0.95).
In terms of malignancy and cirrhosis diagnosed at endoscopy, 26.8% of the cohort had cancer and 15% cirrhosis. GI cancer such as liver and stomach was by far the most common malignancy in patients (79.3%), followed distantly by lung cancer (1.5%).
Other independent predictors of mortality were failed primary endoscopic treatment (OR 15.03, 95% CI 4.63-48.82) and rebleeding (OR 2.77, 95% CI 1.03-7.45).
Rebleeding was associated with Forrest I ulcers (OR 7.67, 95% CI 2.71-21.69), Forrest II ulcers (OR 2.34, 95% CI 1.51-3.60), and coagulopathy (OR 2.34, 95% CI 1.51-3.60).
The authors stressed that patient selection for urgent endoscopy is an important issue, since many hospitals are unable to provide this procedure 24/7. And while many studies have defined high-risk patients as those having a GBS of >12, they said their findings support a GBS cut-off of >7 for urgent endoscopy.
‘Limited Number of Mortality Outcomes’
In an accompanying editorial, Ian Gralnek, MD, MHS, of the Rappaport Faculty of Medicine at Technion-Israel Institute of Technology in Haifa, said the results are not likely generalizable because the sample came from a single tertiary-care center with an unusual capacity to routinely offer round-the-clock urgent endoscopy .
Furthermore, the findings’ implications for mortality need careful interpretation because of the small number of deaths in the large cohort. “This limited number of mortality outcomes is problematic because this can lead the multivariable regression model to be overfit,'” Gralnek wrote. “Overfitting a regression model describes random error or ‘noise’ and not a true underlying association between events.” It occurs when a regression model is too complex, meaning that too many variables are put into the model relative to the number of outcomes.
Nevertheless, the results offer some suggestion of the optimal timing of early endoscopy. “These data may support the role of very early upper GI endoscopy in patients with acute upper GI hemorrhage and high-risk clinical features,” Gralnek wrote, defining “very early” endoscopy as performed once the patient is hemodynamically resuscitated and within 12 hours of presentation.
“We are awaiting high-quality data that support the performance of endoscopy sooner than this, ” he stated. Of the two studies evaluating immediate endoscopy in this patient population, neither showed improvement in mortality, rebleeding, surgery, blood transfusions, or for repeat endoscopy, Gralnek noted, although a 1999 randomized trial showed that endoscopy within 2 hours of ED admission reduced length of hospital stay by a day.
In line with current guidelines, Gralnek recommended that very early endoscopy within 12 hours of patient presentation may improve outcomes, “but probably only in select acute UGI hemorrhage patients with high-risk clinical features. The onus is therefore on gastroenterologists and endoscopists to be vigilant and recognize the patients with high-risk features, and when present, take the necessary steps to perform very early endoscopy.”
Study limitations included its retrospective observational nature, its single-center tertiary-care population with unusual access to urgent endoscopy, and the potential for information gaps in existing records. Furthermore, a large proportion of patients had malignancy or cirrhosis, making it difficult to isolate the effect of urgent endoscopy on outcomes.
Lee and co-authors, as well as Gralnek, disclosed no relevant relationships with industry.
Robert Jasmer, MD Associate Clinical Professor of Medicine, University of California, San Francisco and Dorothy Caputo, MA, BSN, RN, Nurse Planner