Identifying novel acute pancreatitis sub-phenotypes using total serum calcium trajectories

Abstract Background Acute pancreatitis (AP) has heterogeneous clinical features, and identifying clinically relevant sub-phenotypes is useful. We aimed to identify novel sub-phenotypes in hospitalized AP patients using longitudinal total serum calcium (TSC) trajectories. Methods AP patients had at least two TSC measurements during the first 24 h of hospitalization in the US-based critical care database (Medical Information Mart for Intensive Care-III (MIMIC-III) and MIMIC-IV were included. Group-based trajectory modeling was used to identify calcium trajectory phenotypes, and patient characteristics and treatment outcomes were compared between the phenotypes. Results A total of 4518 admissions were included in the analysis. Four TSC trajectory groups were identified: “Very low TSC, slow resolvers” ( n = 65; 1.4% of the cohort); “Moderately low TSC” ( n = 559; 12.4%); “Stable normal-calcium” ( n = 3875; 85.8%); and “Fluctuating high TSC” ( n = 19; 0.4%). The “Very low TSC, slow resolvers” had the lowest initial, maximum, minimum, and mean TSC, and highest SOFA score, creatinine and glucose level. In contrast, the “Stable normal-calcium” had the fewest ICU admission, antibiotic use, intubation and renal replace treatment. In adjusted analysis, significantly higher in-hospital mortality was noted among “Very low TSC, slow resolvers” (odds ratio OR, 7.2; 95% CI, 3.7 to 14.0), “moderately low TSC” (OR, 5.0; 95% CI, 3.8 to 6.7), and “Fluctuating high TSC” (OR, 5.6; 95% CI, 1.5 to 20.6) compared with the “Stable normal-calcium” group. Conclusions We identified four novel sub-phenotypes of patients with AP, with significant variability in clinical outcomes. Not only the absolute TSC levels but also their trajectories were significantly associated with in-hospital mortality.