Background Respiratory failure is a significant determinant of early mortality in cases of acute pancreatitis; however, the relative contributions of various pathophysiological processes remain ambiguous. Comparing the associative and discriminative values of inflammatory, coagulation, hypoxia-perfusion, and electrolyte-metabolic indices for respiratory failure holds substantial clinical implications. Methods We conducted a retrospective cohort study involving adult patients hospitalized with acute pancreatitis from 2019 to 2023. Four composite indices, which reflect systemic inflammation, coagulation activation, hypoxia-perfusion impairment, and electrolyte-metabolic disturbance, were constructed from routine laboratory measurements obtained within 24 hours of admission. The primary outcome was defined as respiratory failure occurring within 3 to 7 days of hospital admission, or respiratory−related death. Associations, dose-response relationships, and discriminative performance were evaluated using multivariable logistic regression, restricted cubic splines, and receiver operating characteristic analyses. Results Among 438 eligible patients, 54 (12.33%) developed respiratory failure or died within 3 to 7 days of admission. All four indices were associated with the primary outcome in univariable analyses. After mutual adjustment, electrolyte-metabolic disturbance emerged as the only dimension independently associated with the primary outcome, with an adjusted odds ratio of 1.32 per 1-SD increase (95% CI 1.16-1.51). This index demonstrated a near-linear dose-response relationship and outperformed the inflammatory, coagulation, and hypoxia-perfusion indices in terms of discrimination (AUC = 0.699). A combined four-dimension model achieved the highest overall discrimination (AUC = 0.744) but provided limited incremental value over electrolyte-metabolic disturbance alone. Conclusions Early electrolyte-metabolic disturbance exhibits the strongest and most consistent association with respiratory failure in acute pancreatitis during the first week of hospitalization, underscoring its potential utility for early risk stratification using routinely available laboratory data.
Lv et al. (2026) studied this question.