Background: Acute Myeloid Leukemia (AML) is the most common acute leukemia in adults and represents a heterogeneous group of clonal hematopoietic stem cell disorders with varying prognosis based on cytogenetic and host factors. Success in treatment of AML is thought to have only improved modestly in recent decades. We aimed to evaluate trends in hospital cost, length of stay, in-hospital mortality, and complication rates in adult patients admitted with active AML. We also sought to elucidate differences in these outcomes in teaching versus non-teaching institutions.

Methods: Using ICD-9 codes for acute myeloid and acute monocytic leukemias, all adult admissions with a primary diagnosis of active AML between 1999-2013 were identified from the Nationwide Inpatient Sample (NIS). Admission information including length of stay (LOS), total charges, and mortality were extracted. Total cost was adjusted for inflation using data from the U.S. Bureau of Labor Statistics. ICD-9 codes were selected to reflect the most common etiologies of in-hospital complications such as clostridium difficile infection (CDI), bacteremia, sepsis, pneumonia, venous thromboembolism (VTE), neutropenic fever, candidiasis, urinary tract infection (UTI), and acute respiratory failure. Rates of these complications were determined over the 15-year interval and compared in subsets of teaching and non-teaching hospitals. Rates of bone marrow transplant were also queried.

Results: We identified 51,684 admissions (weighted N = 247,747) with a primary diagnosis of AML from 1999-2013. Most of the admissions were at teaching hospitals (N = 32,982; weighted N = 158,952). Overall in-hospital mortality was determined to be 19.54%. LOS (days) was found to be longer in teaching (21.04 ±0.10) than in non-teaching (12.25 ±0.11) hospitals (p = .0001). Total charges were also greater in teaching ($157,709 ±1,089) versus non-teaching ($79,167 ±965) hospitals (p = .0001). Of note, after correcting for age, multivariate analysis yielded higher mortality in teaching than in non-teaching hospitals (OR = 1.11, CI: 1.04-1.19). Rates of CDI, bacteremia, neutropenic fever, sepsis, acute respiratory failure, and VTE were higher in teaching hospitals (p < .0001). On the other hand, rates of UTI were lower in teaching (7.31%) than in non-teaching (8.31%) hospitals (p=.0026). Rates of pneumonia and candidiasis did not have a statistically significant difference when comparing the two settings. Bone marrow transplant was more frequently performed at teaching (1.36%) than in non-teaching hospitals (0.56%) (p=.0001). Over the 15-year interval, in-hospital mortality has declined by greater than one third for all AML admissions (p <. 0001). Rates of nearly all of the complications, excluding candidiasis, but including CDI, neutropenic fever, bacteremia, UTI, pneumonia, VTE, and acute respiratory failure have increased during this interval, however. Total charges increased during this time period from $66,678 (±1,567) in 1999 to$197,439 (±4,532) in 2013 (p = .0001), which was greater than the expected inflationary increase to \$93,235 over the same time period.

Conclusions: Most admissions for AML occurred at teaching institutions. This may be due to increased resource requirements to care for this patient population. In-hospital mortality appears to have improved markedly from 1999-2013 for all admissions for AML, which may be a testament to well-established chemotherapy guidelines, use of less toxic chemotherapy regimens in the elderly, and standardized preventative practices such as the use of high-efficiency particulate air filtration and prophylactic antibiotics. On the other hand, rates of nearly all measured complications have increased during this interval. Given the opposite trend in mortality, we believe this may be in part due to improved surveillance and reporting. Rates of mortality as well several complication rates appear to be higher in teaching than in non-teaching institutions, which may be due to increased medical complexity and more aggressive therapy offered at teaching institutions. Further research is required to determine what additional factors and practice differences are contributing to these discrepancies. Total charges were higher at teaching institutions, which may be due to increased LOS, complication rates, medical complexity and resource consumption.

Disclosures

No relevant conflicts of interest to declare.

## Author notes

*

Asterisk with author names denotes non-ASH members.