Association between serum calcium trajectories and 30-day mortality among critically ill stroke patients: a cohort study using the MIMIC-IV database

BACKGROUND AND PURPOSE: The prognostic value of serum calcium in stroke remains unclear, with prior studies relying on single measurements. We aimed to identify distinct longitudinal trajectories of serum calcium during the first 7 days after admission to intensive care unit (ICU) stay and investigate their association with 30-day mortality in stroke patients. METHODS: This retrospective cohort study analyzed adult stroke patients from the MIMIC-IV database. Group-based trajectory modeling (GBTM) was applied to serial serum calcium measurements over the first 7 days to identify characteristic patterns. The primary outcome was 30-day mortality, assessed using multivariable Cox proportional hazards models adjusted for confounders selected by a combination of clinical knowledge and algorithms (LASSO and Boruta). Sensitivity analyses included propensity score matching (PSM). RESULTS: Among 1,274 patients, GBTM identified three distinct trajectories: traj1 (> 9.5 mg/dL; 9.5%), traj2 (8.5–9.5 mg/dL; 58.3%), and traj3 (< 8.5 mg/dL; 32.2%). Patients in traj3 had higher illness severity scores and comorbidity burden. The 30-day mortality rate was highest in traj3 (48.5%) compared to traj2 (38.4%) and traj1 (38.0%). After multivariable adjustment, traj3 was independently associated with an increased risk of 30-day mortality compared to traj2 (hazard ratio 1.32, 95% confidence interval 1.09–1.61). Sensitivity analyses using PSM confirmed this association (HR 1.39, 1.11–1.74).​ Restricted cubic spline analysis revealed a U-shaped relationship between serum calcium levels and mortality risk. CONCLUSIONS: Longitudinal serum calcium trajectories are significantly associated with short-term mortality in critically ill stroke patients. A persistently low trajectory (< 8.5 mg/dL) identifies those at highest risk. Monitoring dynamic calcium patterns provides prognostic information beyond a single measurement. These findings support the role of serial​ monitoring of calcium homeostasis and generate the hypothesis​ that correcting sustained hypocalcemia might improve outcomes, a premise that warrants testing in future interventional studies.