Osmolarity trajectories and outcomes in patients with acute pancreatitis in the intensive care unit: group-based trajectory modeling

Acute pancreatitis (AP) is a common acute gastrointestinal disorder, with a subset of patients progressing to severe forms associated with high mortality. While osmotic pressure significantly influences its pathological processes, the relationship between trajectories of plasma osmolality changes and patient prognosis remains incompletely elucidated. Using the MIMIC-IV database, we analyzed 1,138 ICU-confirmed acute pancreatitis cases. Group-based trajectory models were applied to identify distinct trajectories of plasma osmolality changes. Cox regression models and Kaplan-Meier analyses were used to examine associations between these trajectories and all-cause mortality at 28, 90, and 365 days, with adjustment for potential confounding factors. Five distinct subgroups were identified and validated based on plasma osmolality trajectories: the “Moderately Stable Group” (39.5%), “Low Stable Group” (22.8%), “High Sudden Rise Group” (9.5%), “High Sudden Drop Group” (6.3%), and “High Gradual Drop Group” (21.8%). After adjustment, both the “High Sudden Rise Group” and “High Sudden Drop Group” demonstrated significantly increased mortality risks at 28, 90, and 365 days, with the latter exhibiting the highest risk. After adjusting for confounding factors, the “High Sudden Rise Group” was significantly associated with 28-day mortality (HR = 2.15, 95% CI: 1.10–4.21), and this association was even stronger for the “High Sudden Drop Group” (HR = 3.62, 95% CI: 1.82–7.19). Additionally, 24-hour plasma osmolality displayed a U-shaped association with 28-day mortality. Subgroup analyses confirmed that the “High Sudden Drop Group” consistently showed an elevated mortality risk across all subgroups (P < 0.05). This study identified distinct phenotypic subgroups of acute pancreatitis by analyzing plasma osmolality trajectories. By incorporating dynamic changes in plasma osmolality, we applied a group-based trajectory model to stratify mortality risk among patients with acute pancreatitis. These findings may facilitate the early identification of high-risk individuals, optimize clinical fluid management strategies, and ultimately improve patient outcomes.