Differences in Predictors of Mortality and Resource Utilization Among Adult Patients with Acute Lymphocytic Leukemia and Acute Myeloid Leukemia

Introduction: Both acute lymphocytic leukemia (ALL) and acute myeloid leukemia (AML) require hospitalization for intensive chemotherapy, stem cell transplantation, and disease or treatment-related complications. There is a dearth of evidence in prediction of inpatient resource utilization and hospital outcomes among patients with these conditions. The goal of this study is to identify predictors of average length of stay in the hospital (ALOS) and inpatient mortality in adult ALL and AML patients.

Methods: We performed a retrospective cohort analysis of the National Inpatient Sample 2014 Database (HCUP-NIS). Patients were included in the study if they had a principal diagnosis of ALL or AML and age 18 years or older. We used descriptive statistics to characterize the cohort in terms of personal demographic factors (age, race, sex, insurance type, community income level), hospital characteristics (size, geographical region, teaching status, and urban or rural location), and admission timing (weekend or weekday). We performed univariate and multivariate regression to analyze the association of these factors with mortality and ALOS. All analyses apply the HCUP-NIS weights.

Results:

The ALL cohort included 5,550 admissions. Most ALL patients were white (65%) males (60%), and approximately half were age 50 years or younger. The AML cohort included 18,930 admissions. Most AML patients were white (74%) males (54%), aged 60 years or older (59%). Nearly all (95%) of ALL patients had insurance coverage, either private (40%), Medicare (25.9%), Medicaid (25%), or another type (5%). In contrast, most AML patients had Medicare (46%), followed by private insurance (36%), Medicaid (11.0%), other insurance (3.8%) or no insurance (2.8%). Care for both cohorts occurred most often in large, urban, and teaching hospitals. While Charlson index was the only statistically significant predictor of mortality in the ALL cohort (AMD 1.34, 95% CI 1.11 to 1.63, p=0.002), age (OR 1.02; 95% CI 1.014 - 1.03), Charlson index (OR 1.24; 95% CI 1.16 - 1.34) and other type of insurance were associated with increased mortality for AML. ALOS was similar for both cohorts: ALL 18.5 days and AML 18.9 days. For ALL, multivariate analysis showed Charlson index (AMD 1.53, 95% CI 0.32 - 2.74, p=0.013), and hospital type (urban AMD 5.73; 95% CI 2.73 to 8.73, p<0.01 and teaching hospital AMD 6.86, 95% CI 4.36 to 9.36, p<0.01) to be independent predictors of ALOS. For AML, age (AMD -0.22, 95% CI -0.27 to -0.17), insurance type (Medicaid AMD 3.02, 95% CI 0.65 - 5.39, private insurance 4.10, 95% CI 2.60 - 5.60), Charlson index (AMD 0.63, 95% CI 0.25-1.01), and hospital type (urban AMD 6, 95% CI 2.65 to 8.72), and teaching AMD 7, 95% CI 6.06 to 9.09) predicted ALOS.

Discussion: Care for ALL and AML occurs primarily in large, urban teaching centers. Although ALOS is similar for the two, age and insurance carrier (Medicaid and private) are associated with increased ALOS only for AML. The increased LOS with age is possibly because of increased complexities associated with it. LOS is longer in AML patients with Medicaid and a few private insurances possibly because the placement to health care facilities tends to be significantly more difficult and time-consuming than for Medicare. Similarly, age and other type of insurance are associated with increased mortality only in AML. No regional or racial/ethnic differences were seen for either cohort for ALOS or mortality. Understanding factors influencing ALOS may help institutions in planning health care resource allocation. Health policy decisions that negatively impact the financial health of large, urban teaching centers may influence their ability to provide the complex and expensive care for this group of patients.

Table.

View largeDownload slide

Table.

View largeDownload slide

Close modal