Download Full Text (259 KB)


Internal Medicine

Training Level

Resident PGY 1


Henry Ford Hospital


BACKGROUND: Patients with left ventricular assisted devices (LVADs) are at considerable risk of gastrointestinal bleeding (GIB) of about 23%. A significant portion of GIB occur in the stomach, duodenum or small intestine as compared with lower intestinal tract. A traditional work up for such patients differs between institutions but generally includes an esophagogastroduodenoscopy (EGD) and colonoscopy +/- RBC tagged scan. If the cause of GIB is not found, a capsule endoscopy or push enteroscopy (PE) is then pursued to evaluate for small intestinal bleeding, an area not accessible by other devices/ procedures including EGD and colonoscopy. The traditional approach requires considerable time and effort leading to a significant length of hospital stay. It also exposes the patient to multiple procedures with additive potential adverse effects and cost. AIM: Our goal is to compare the traditional work up/ management of GIB with an innovative approach of performing PE as the first diagnostic/therapeutic procedure to assess if the latter increases the diagnostic yield of GIB site detection with fewer procedures per hospital admission, shortens the length of hospital stay, and decreases all-cause mortality. METHODS: This is a retrospective study was performed in Henry Ford Hospital in Detroit, MI. ICD-9 and ICD-10 diagnosis codes were used to generate a list of LVAD patients who were admitted with an overt GIB or worsening anemia in the period from 1/1/2013 to 12/25/2018. Our primary outcomes were the rate of detection of GIB lesion/site and all-cause mortality. Secondary outcomes were the number of packed red blood cell (pRBC) units transfused during the hospitalization and the length of hospitalization. Chi-square, Fisher exact, paired-T tests and Pearson correlation were used for statistical analysis. The study protocol was approved by the hospital’s IRB. RESULTS: A total of 227 patients were reviewed. 89 patients were included with a mean age 61.36 years-old. The majority of patients (75.28%) were > 55 years-old and 70.78% of patients were males. All patients were on anticoagulation and 53 patients were on antiplatelets as well. The patient’s prestation were as follows: 38 patients presented with melena, 11 with hematochezia, 7 with hematemesis or coffee ground emesis and 33 patients with worsening anemia without overt GIB. A total of 71 patients underwent the traditional approach at the first index endoscopy, whereas 18 patients started with PE +/- colonoscopy. The source of GIB was detected at the first index endoscopy in 51 patients (36 traditional approach and 15 in PE approach). Arteriovenous malformation was the most common lesion detected (29 patients) and the two most common sites of bleeding were gastroduodenal followed by the small bowel. Doing PE at the first index endoscopy was associated with a higher rate of GI site detection, OR 4.861 (95% CI (1.293-18.271), P = 0.012), this was true, especially when patients presented with worsening anemia without overt bleeding, OR 11.2 (95% CI (1.202-104.33), P = 0.015). There was no statistically significant difference between both approaches in terms of all-cause mortality (P = 0.163). Patients in the PE group did have a shorter hospital stay (x̅ (SD) = 10.78 (13.97) days compared to 18.8 (25.58) days for the traditional approach) with P value = 0.034. No statistically significant difference in the number of pRBC units (P = 0.121). Finally, INR value on presentation was not associated with a higher risk of all-cause mortality P = 0.905 and didn’t correlate with a statistical significance with number of pRBC units and length of stay (P = 0.839 and 0.644 respectively). CONCLUSION: PE is a safe procedure. It increases the GIB site detection and shortens the length of hospital stay when considered on the initial evaluation of LVAD patients presenting with GIB in general and worsening anemia in specific.

Presentation Date


Comparison of clinical outcomes in traditional gastrointestinal hemorrhage work up versus direct utilization by push enteroscopy in patients with a left ventricular assist device