Diabetic ketoacidosis and hyperosmolar hyperglycemic state as high-acuity phenotypes in U.S. cancer hospitalizations: National Inpatient Sample, 2018–2022.

Abstract

e23197 Background: Diabetic ketoacidosis (DKA) and hyperosmolar hyperglycemic state (HHS) are uncommon but high-acuity events that may occur during cancer hospitalizations. National estimates describing their association with inpatient physiologic severity, critical illness, mortality, and resource utilization in oncology hospitalizations remain limited. Methods: A serial cross-sectional, hospitalization-level analysis of the 2018 to 2022 Healthcare Cost and Utilization Project National Inpatient Sample (NIS) was performed using discharge-level survey weighting with hospital clustering and stratification to generate nationally representative estimates. Adult hospitalizations with a principal diagnosis of malignancy were identified using ICD-10-CM codes C00 to C97 and D45 to D47. Hyperglycemic crises were defined using any-diagnosis ICD-10-CM code families for diabetic ketoacidosis (DKA; E10.1, E11.1, E13.1, E08.1, E09.1) and hyperosmolar hyperglycemic state (HHS; E10.0, E11.0, E13.0, E08.0, E09.0). Outcomes included All Patient Refined Diagnosis Related Group (APR-DRG) severity of illness subclass, with extreme severity defined as APR-DRG level 4, in-hospital mortality, length of stay, hospitalization cost, and ICU proxies including mechanical ventilation and shock. Survey-weighted multivariable models adjusted for demographics, payer, socioeconomic status, admission characteristics, cancer type, and hospital characteristics. Results: Among 961,848 unweighted cancer hospitalizations representing 4,809,239 hospitalizations nationally, DKA occurred in 0.11%, HHS in 0.02%, and any hyperglycemic crisis in 0.12%. Compared to hospitalizations without crisis, hyperglycemic crisis admissions had higher mean APR-DRG severity (3.52 vs 2.56), longer length of stay (10.80 vs 6.63 days), and higher cost ($35,491 vs $22,792). Hyperglycemic crises were also associated with higher rates of extreme severity (10.41% vs 3.45%), in-hospital mortality (14.15% vs 4.37%), mechanical ventilation (9.99% vs 2.58%), and shock (7.24% vs 1.38%). After adjustment, hyperglycemic crisis remained independently associated with higher inpatient severity (β 0.72, 95% CI 0.64 to 0.80), extreme severity (OR 4.79, 95% CI 3.62 to 6.35), in-hospital mortality (OR 2.80, 95% CI 2.35 to 3.33), mechanical ventilation (OR 3.50, 95% CI 2.89 to 4.25), and shock (OR 4.65, 95% CI 3.71 to 5.83). Conclusions: Although uncommon, hyperglycemic crises create a high-acuity subset of cancer hospitalizations with profound increases in extreme physiologic severity, critical illness, mortality, length of stay, and cost. These findings support hyperglycemic crises as high-priority inpatient risk phenotypes with implications for early recognition, escalation planning, and ICU resource utilization in hospitalized oncology populations.