Derivation and validation of a clinical diagnostic model for chlamydial cervical infection in university women
B. A. Johnson, R. M. Poses, C. A. Fortner, F. A. Meier and H. P. Dalton
Department of Internal Medicine, Medical College of Virginia, Virginia Commonwealth University, Richmond.
We developed and prospectively tested a logistic regression model for
chlamydial cervical infection. Study subjects included 2271 women receiving
gynecologic care in our student health clinic. Clinical data were collected
in a standardized fashion. We identified cell culture--isolated Chlamydia
trachomatis from 133 (9%) of 1458 subjects in the derivation set and 73
(10%) of 729 subjects in the validation set. Model variables included a new
sexual partner within 2 months or more than one sexual partner within 6
months; cervical ectopy; cervical friability; at least 20 polymorphonuclear
leukocytes per high-power field in cervical secretions; white blood cells
in vaginal secretions; and use of an antibiotic active against C
trachomatis within a month. This model can distinguish women with low,
medium, and high risks of chlamydial infection (on derivation set: receiver
operating characteristic curve area, 0.710; SE, 0.026; on validation set:
area, 0.698; SE, 0.035) using simple clinical information obtained in the
office.
