TABLE 1.

Multivariable logistic regression coefficients of the full model.

Variable	β	SE	P-value
Sex	–0.051	0.231	0.824
Age	1.086	0.200	0.000
BMI	–0.240	0.162	0.139
Height	–0.361	0.211	0.086
Education	–0.241	0.146	0.099
Smoking	0.331	0.185	0.074
Drinking	–0.288	0.213	0.176
CAD	0.099	0.123	0.424
Hypertension	–0.068	0.166	0.683
Diabetes	–0.098	0.160	0.539
Depression	0.244	0.138	0.076
Stroke	0.201	0.128	0.116
APOE-ε4	0.352	0.139	0.011
Orange	0.055	0.151	0.715
Leather	–0.235	0.163	0.151
Cinnamon	–0.182	0.179	0.308
Peppermint	–0.371	0.118	0.002
Banana	–0.341	0.153	0.026
Lemon	0.202	0.168	0.229
Licorice	0.054	0.160	0.737
Coffee	–0.027	0.127	0.828
Cloves	–0.013	0.161	0.934
Pineapple	0.408	0.176	0.020
Rose	–0.244	0.158	0.124
Fish	0.159	0.148	0.280
MMSE	–0.719	0.138	0.000

SE, standard error; BMI, body mass index; CAD, coronary artery disease; OI, olfactory identification sum score; APOE, apolipoprotein; MMSE, Mini-Mental State Examination. Weight and olfactory identification sum score were excluded from the model because of collinearity. β corresponds to the change of women versus men for sex; per year for age and education; per unit for BMI, height, and MMSE; with the feature versus without the feature for smoking, drinking, and comorbidities; and positive versus negative for APOE-ε4 and individual odor tests.