Biomarkers predicting adverse outcomes in non-surgical intensive care patients have not been reported.

Data for 1,006 emergency department patients were prospectively analyzed. The serum heart-type fatty acid-binding protein (s-H-FABP) level was measured within 10 min of admission. The patients were assigned to intensive care (n = 835) or other departments (n = 171). The intensive care patients were divided into survivors (n = 745) and non-survivors (n = 90) according to the in-hospital mortality and assigned to four groups according to the quartiles of s-H-FABP (Q1, Q2, Q3 and Q4). The s-H-FABP levels were significantly higher in the intensive care patients (12.7 [6.1–38.8] ng/ml versus 5.3 [3.1–9.4] ng/ml) and in the non-survivors (44.9 [23.2–87.6] ng/ml versus 11.5 [5.6–32.6] ng/ml). A Kaplan–Meier curve showed a significantly higher survival rate in Q3 than in Q1 and Q2 and in Q4 than in the other groups. The multivariate Cox regression model identified Q3 (HR 4.646, 95 % CI 1.526–14.146) and Q4 (HR 9.483, 95 % CI 3.152–28.525) as independent predictors of 90-day mortality. The sensitivity and specificity of H-FABP for in-hospital mortality were 81.1 and 66.0 % (AUC 0.775) at 20.95 ng/ml. The in-hospitality rate was significantly higher in the high s-H-FABP patients than in the low s-H-FABP patients in each etiology group.

The s-H-FABP level is an effective biomarker for risk stratification in non-surgical intensive care patients.