Choroidal thickness and granulocyte colony-stimulating factor in tears improve the prediction model for coronary artery disease

Table 3 Logistic regression and comparative study between area under curve (AUC) of the prediction models with the classic risk factors (CRF) including, or not, the selected ophthalmological variables

Variable	OR^a (CI 95%)	AUC^c (CI 95%)	AUC^d (CI 95%)
AL	1.13 (0.81–1.58)	0.83 (0.76–0.90)	0.83 (0.76–0.90)
CT	1.02 (1.01–1.03)^b	0.89 (0.84–0.94)^e	0.83 (0.77–0.90)
IL-8	0.99 (0.99–1.00)^b	0.86 (0.80–0.92)	0.84 (0.78–0.91)
G-CSF	0.97 (0.95–0.99)^b	0.91 (0.84–0.97)^e	0.82 (0.72–0.92)
MCP-1	0.96 (0.94–0.99)^b	0.87 (0.80–0.93)	0.84 (0.78–0.91)
MIP-1β	0.99 (0.91–1.08)	0.87 (0.80–0.93)	0.85 (0.79–0.92)
RANTES	1.02 (1.00–1.05)^b	0.87 (0.81–0.93)	0.84 (0.78–0.91)

Statistically significant results are shown in bold (p<0.05). OR Odds ratio; CI Confidence interval; AL Axial length; CT Choroidal thickness; IL Interleukin; G-CSF Granulocyte colony-stimulating factor; MCP Monocyte chemoattractant protein; MIP Macrophage inflammatory protein; RANTES Chemokine ligand 5
^aAdjusted by variable and CRF: sex, age, diabetes, high blood pressure, hypercholesterolemia, smoking habit, and obesity
^bp < 0.05 (likelihood ratio test)
^cAUC from model adjusted by variable and CRF
^dAUC from model adjusted only by CRF
^eStatistically significantly (p < 0.05) higher than AUC from model adjusted only by CRF

ISSN: 1475-2840