Introduction: Acute myelogenous leukemia (AML) increases the chances of bleeding because of thrombocytopenia and dysfunction in platelet activity. The presence of pre-existing AML in patients with upper gastrointestinal bleeding (UGIB) complicates the hospital course and affects the incidence of multiple complications. Our retrospective analysis aims to study the statistical difference in the incidence of complications in AML and non-AML patients admitted to the hospital for UGIB.

Methods: Adult cases with a primary diagnosis of UGIB were extracted from the 2016 to 2021 National Inpatient Sample(NIS). The NIS is a set of annual hospital records released anonymously through HCUP, AHRQ, and partners. Cases with AML were also identified using the International Classification of Diseases, 10th Revision (ICD-10) codes. All other forms of cancer were excluded from our non-AML cohort, based on Elixhauser's codes. A propensity-score matched(PSM) sample was created at a caliper set at 0.01, and a 1:10 ratio for patients with and without AML. Logistic regression models estimated the differences in various complications between the two groups. STATA 18.0, SPSS 29.0, and R-studio were used for our analysis.

Results: Our pre-PSM sample contained 1806544 cases of UGIB amongst which 1360(0.1%) had AML. The AML group was older (median age 74.00(IQR: 66.00-81.00) vs. 69.00(IQR 57.00-80.00), p<0.01). Medicare was the main form of insurance in both groups (75.0% of AML and 62.7% among the non-AML group). After careful adjustments, 1360 AML cases were matched with 13440 non-AML patients. Differences in outcomes were noted as AML patients were less likely to undergo esophagogastroduodenoscopy (EGD) (aOR 0.514, 95 % CI 0.457- 0.578, p<0.001) and report events of acute myocardial infarction (AMI) (aOR 0.42, 95 % CI 0.267- 0.663, p<0.001) or cardiac arrest (aOR 0.398, 95 % CI 0.16- 0.99, p<0.047). However, AML was also linked with higher odds of sepsis (aOR 3.051, 95 % CI 2.151- 4.329, p<0.001, need for blood transfusion (aOR 2.081, 95% CI 1.856 - 2.333, p<0.001), and all-cause in-hospital mortality (aOR 4.648, 95% CI 3.575 - 6.044, p<0.001), as compared to the non-AML cohort. No differences in the use of mechanical ventilation (aOR 0.702, 95 % CI 0.479 - 1.03, p<0.07) and acute kidney injury (AKI) (aOR 1.046, 95% CI 0.913 - 1.2, p<0.516) were found. In general, AML patients experienced a longer stay (median stay 4.00 days, IQR 3.00-8.00 vs. 3.00, IQR 2.00-5.00, p<0.01), along with a higher median hospital charge ($47543 vs. $33336, p<0.01).

Conclusion: Our analysis confirms the poorer outcomes among UGIB patients with AML, with higher mortality, sepsis, and transfusion use. AML patients were also less likely to undergo UGIB and experienced fewer cases of AMI or cardiac arrest. It is vital to expand our results and identify the various causes of such disparities in the use of UGIB and the varying complications, as well as reduced cardiovascular complications in the AML group. Such additional studies will help improve admission protocols, whereby the gastroenterology teams can better coordinate care with the hemato-oncology teams for improved outcomes among AML cases.

Disclosures

No relevant conflicts of interest to declare.

This content is only available as a PDF.
Sign in via your Institution