Computed tomography to estimate cardiac preload and extravascular lung water. A retrospective analysis in critically ill patients

Table 4 Predictive capabilities of computed tomography-based estimation of global end-diastolic volume index

		CT-based estimation of GEDVI vs. TPTD-derived GEDVI
		GEDVI < 680 mL/m²	GEDVI = 680 - 800 mL/m²	GEDVI > 800 mL/m²
Radiologist 1	Sensitivity	0 (0 to 23)	44 (14 to 79)	86 (42 to 99)
	Specificity	100 (79 to 100)	52 (30 to 74)	57 (34 to 77)
	PPV	not estimable due to zero counts in numerator	29 (8 to 58)	38 (15 to 65)
	NPV	53 (34 to 72)	69 (41 to 89)	93 (66 to 99)
Radiologist 2	Sensitivity	0 (0 to 23)	44 (14 to 79)	57 (18 to 90)
	Specificity	100 (79 to 100)	62 (38 to 82)	39 (20 to 61)
	PPV	not estimable due to zero counts in numerator	33 (10 to 65)	22 (6 to 48)
	NPV	53 (34 to 72)	72 (47 to 90)	75 (43 to 95)

Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) given as percentages with 95% confidence intervals (95% CI) in parentheses for computed tomography (CT)-based estimations of global end-diastolic volume index (GEDVI) with regard to GEDVI derived from transpulmonary thermodilution (TPTD) are shown separately for radiologist 1 and radiologist 2.