To investigate the predictive value of calcium for in-hospital death in ICU patients with acute stroke.

A total of 10 097 hospitalized acute stroke patients in 208 hospitals in the United States from January 1, 2014 to December 31, 2015 were enrolled in this retrospective study. We applied the logistic regression to analyze the correlation between serum calcium levels and in-hospital death in patients with acute stroke. Then the generalized additive models(GAM)and smooth curve fitting were used to analyze the nonlinear relationship between serum calcium level and in-hospital mortality risk in acute stroke patients.

Among the 10 097 stroke patients included with acute stroke, the mean ± standard deviation was 8.87±0.67 mg/dL, and 1 430 (13.90%) patients died in hospital. Results based on the regression analysis showed that serum calcium level was an independent influencing factor of in-hospital mortality(OR: 0.81, P<0.001). GAM analysis showed a nonlinear relationship between serum calcium levels and in-hospital death upon admission to acute stroke. The saturation threshold effect and nonlinear relationship plot indicated that the inflection point of serum calcium was 9.50 mg/dL, of which the corresponding in-hospital mortality was lowest. Both high and low serum calcium levels increased the risk of in-hospital death in patients with acute stroke.

Serum calcium could be used as a biomarker to predict the risk of in-hospital death in patients with acute stroke.