Nonlinear association between comorbidity burden and in-hospital mortality among adults hospitalized with acute kidney injury

Background Acute kidney injury (AKI) occurs across a heterogeneous population with substantial variation in comorbidity burden. Although multimorbidity is associated with adverse outcomes in AKI, whether mortality increases progressively with disease burden or varies according to the extent and composition of coexisting conditions remains unclear. We examined the association between selected comorbidity burden and in-hospital mortality among U.S. adults hospitalized with AKI. Methods We conducted a retrospective cross-sectional study using the 2022 National Inpatient Sample. Adult hospitalizations with AKI were identified using ICD-10-CM codes. Comorbidity burden was defined as the number of six conditions—diabetes, hypertension, heart failure, non-ESRD chronic kidney disease, liver disease, and malignancy—and evaluated as counts from 0 to 6. Survey-weighted logistic regression estimated associations with in-hospital mortality after adjustment for demographic, admission, and hospital characteristics. Functional form was evaluated using linear, spline, and categorical specifications. Population-standardized mortality probabilities were estimated, with additional analyses examining acute illness, disease composition, and hospital location/teaching status. Results The cohort included 1,066,687 hospitalizations, representing 5.33 million U.S. hospitalizations with AKI. The association between comorbidity burden and mortality was nonlinear. Relative to count 0, adjusted mortality odds were lower at counts of two (aOR, 0.83; 95% CI, 0.80–0.86), three (aOR, 0.77; 95% CI, 0.74–0.80), and four (aOR, 0.79; 95% CI, 0.76–0.82), but higher at counts of five (aOR, 1.27; 95% CI, 1.20–1.34) and six (aOR, 1.52; 95% CI, 1.27–1.82). Population-standardized mortality was 9.92% at count 0, 7.83% at count 3, 12.20% at count 5, and 14.25% at count 6. Adjustment for sepsis and mechanical ventilation attenuated differences at intermediate levels, whereas elevated mortality at counts five and six persisted across models. Disease composition influenced the high-burden association; elevated mortality at count five persisted after exclusion of malignancy but not liver disease. Hospital location/teaching status interacted with burden (P for interaction=0.028), although absolute mortality differences were modest. Conclusions Among U.S. adults hospitalized with AKI, selected comorbidity burden was nonlinearly associated with in-hospital mortality. The relationship reflected the number and composition of coexisting conditions and acute illness context, cautioning against interpreting comorbidity count as a uniform linear risk gradient or fixed clinical threshold.