Wenker et al reported a model for predicting Barrett's esophagus in a patient. This can be automated to screen a patient population. The authors are Baylor College of Medicine, Michael E DeBakey Veterans Affairs Medical Center, LTC Charles S Kettles Ann Arbor Veterans Affairs Medical Center, and University of Michigan.
BEST = Barrett's esophagus electronic screening tool
Parameters:
(1) sex
(2) age of the patient in years
(2) race or ethnicity
(3) smoking status
(4) body mass index (BMI) in kg per sq meter
(5) GERD
(6) family history of esophageal cancer
Parameter
Finding
Beta-Coefficient
sex of the patient
female
0
male
1.28
age of the patient
< 50 years
0
>= 50 years
0.52
race or ethnicity
Black
0
White non-Hispanic
1.6
White Hispanic
0.97
smoking status
never
0
former
0.05
current
0.22
BMI
< 25 kg per square meter
0
25 to 29.9
-0.01
>= 30
0.12
GERD
no
0
yes
0.42
family history of esophageal cancer
no
0
yes
0.65
value of X =
= SUM(all of the beta-coefficients) - 4.85
probability of Barrett's esophagus =
= 1 / (1 + EXP((-1) * X))
Performance:
• The area under the ROC curve was 0.68.
To read more or access our algorithms and calculators, please log in or register.
