A decision rule to identify adolescent females with cervical infections.

OBJECTIVE

To develop a clinical decision rule to direct empiric treatment of adolescent females with Neisseria gonorrhoeae (GC) or Chlamydia trachomatis (CT) (or both) cervical infections in a pediatric emergency department.

METHODS

This was a cross-sectional study of adolescent females with symptoms necessitating sexually transmitted infection (STI) testing. The outcome was defined as cervical specimens positive for either GC on culture or CT on nucleic amplification assay test. Clinical variables included demographic, historical, physical, and laboratory findings. Bivariate associations were assessed using chi-square for categorical data and Student's t test for continuous variables. Variables significant at p < 0.5 were eligible for logistic regression (LR). Recursive partitioning (RP) analysis was used to create a clinical decision rule.

RESULTS

Of the 250 subjects, 83 (33.2%) were positive for GC/CT. The adjusted odds ratios (aOR) and 95% confidence intervals (CI) of the LR model were African American race (aOR = 3.2, CI 1.3-7.9), new partner within 3 months (aOR = 1.9, CI 1.0-3.5), cervical discharge (aOR = 2.0, CI 1.1-3.7), absence of yeast forms (aOR = 3.3, CI 1.3-10), and >10 white blood cells (WBCs) (aOR = 2.5, CI 1.3-4.6) on vaginal gram stain. Variables comprising the RP analysis included partner penile discharge, >10 WBCs on vaginal gram stain, African American race, absence of yeast forms on vaginal gram stain, and no hormonal birth control use. This algorithm was 75% sensitive and 71% specific, with a negative predictive value of 85%.

CONCLUSIONS

The LR model confirmed associations seen in other populations. Although STI testing is imperative, the RP model can be used to direct empiric treatment among high-risk adolescent females.