A prospective multicenter Canadian study found that the Oakland score was the most specific risk stratification tool for identifying patients with lower gastrointestinal bleeding who could be safely discharged from the emergency department, outperforming clinical judgment and other competing risk scores. The findings support using the score alongside physician assessment when making discharge decisions.
“This study matters because it provides physicians with a tool to help safely discharge patients from the emergency department,” first author Michael Sey, MD, MPH, FRCP(C), of the division of gastroenterology at Western University, London, Ontario, Canada, told GI & Hepatology News. “Although most patients with lower gastrointestinal bleeding do well, a small percentage can rebleed and develop serious adverse events. This tool, along with sound clinical judgement, helps the attending physician risk stratify and identify patients who can be safely managed as outpatients with appropriate follow up.”
For the study, published in Clinical Gastroenterology and Hepatology, Dr. Sey and colleagues included 344 adults who came to eight Canadian hospitals with signs of lower gastrointestinal bleeding, including bright red blood from the rectum or maroon-colored stool, between September 2019 and February 2024. The researchers followed patients for 28 days after presentation.
The average age of participants was 59 years. Women made up 48% of the group, and 42% were admitted to the hospital when they first sought care. Nearly 20% were taking antiplatelet medications, and 12% were using anticoagulants.
During the 28-day follow-up period, 16 patients were readmitted to the hospital, 41 required blood transfusions, and 24 underwent endoscopic procedures to stop bleeding. Two patients needed radiologic intervention, and 12 required surgery. Five patients died within 28 days of presentation; however, investigators reported that none of the deaths were directly caused by lower GI bleeding.
The researchers compared the Oakland score with several other tools used to assess bleeding risk, including the Strate, BLEED, Glasgow-Blatchford, and clinical Rockall scores. They also compared these tools with physicians’ judgments about whether patients could be safely discharged. The NOBLADS score was not included in the analysis because serum albumin levels were not available for most patients.
Using area under the receiver operating characteristic curve (AUROC) analysis, the Glasgow-Blatchford score demonstrated the highest overall discrimination at 0.81, followed by the Oakland score at 0.77 and clinical judgment at 0.76. The Strate score achieved an AUROC of 0.68, the clinical Rockall score 0.63, and the BLEED score 0.62. The Oakland score performed similarly to physicians’ clinical judgment, with no statistically significant difference between the two. However, the Oakland score was significantly more accurate than the BLEED and clinical Rockall scores.
Because clinicians rely on specific score cutoffs when making decisions, the researchers also assessed performance at validated thresholds. An Oakland score of 8 or less had the highest specificity (95%), followed by the Glasgow-Blatchford score (88%) and clinical judgment (81%). Clinical judgment had the highest sensitivity (70%), followed by the Glasgow-Blatchford score (50%) and BLEED score (48%). The Oakland score's lower sensitivity (23%) reflects its focus on avoiding unsafe discharges.
“We were surprised by how well the Glasgow-Blatchford Bleeding score performed,” Dr. Sey said. “Although it was developed almost three decades ago for upper gastrointestinal bleeding, it was still quite discriminative for the prediction of safe discharge in lower gastrointestinal bleeding.”
The researchers noted that specificity may be more important than sensitivity in this setting because incorrectly classifying high-risk patients as safe for discharge could lead to adverse outcomes. Clinical judgment incorrectly identified 16 unsafe patients as suitable for discharge, compared with four patients classified as safe by the Oakland score.
Dr. Sey added that the study’s overall findings “will provide reassurance to physicians, both GI and non-GI alike, that lower gastrointestinal bleeding patients who score 8 or less on the Oakland score can be safely discharged for outpatient care provided there are appropriate discharge instructions and follow-up plans.”
Several additional analyses supported the main findings. The results remained unchanged after accounting for missing data and after excluding digital rectal examination findings from the Oakland score. Researchers also found no meaningful differences in outcomes between patients enrolled during versus after the COVID-19 pandemic or between those enrolled on weekdays versus weekends.
“We conducted a large part of this trial during the early days of the COVID-19 pandemic,” Dr. Sey recalled. “Given we were dealing with gastrointestinal bleeding, we were often in the emergency department. As such, it was a very difficult prospective cohort study to conduct on a national scale due to the challenges of the pandemic [such as] shortages of PPE, lock downs, and restrictions on entering the ED.”
Among patients who underwent colonoscopy, hemorrhoidal bleeding was the most common diagnosis, accounting for 22% of cases, followed by diverticular bleeding at 20%. Colonoscopy was performed more frequently in patients who were ultimately classified as unsafe for discharge. An upper gastrointestinal source was identified in 5% of patients who underwent esophagogastroduodenoscopy.
The researchers noted several limitations, including missing digital rectal examination data, possible underreporting of minor rebleeding events treated outside the hospital, differences in transfusion practices, and variation in physicians’ discharge decisions.
The study was funded by an Innovation Grant from the Academic Health Sciences Centre in Ontario, Canada. One of the study authors, Harminder Singh, MD, MPH, reported advisory or consulting relationships with Pendopharm, Amgen Canada, AbbVie Canada, Innomar, Sandoz Canada, Takeda Canada, and Guardant Health unrelated to the study. All other authors reported no relevant conflicts of interest.