Get access

Criterion-based Screening for Sexually Transmitted Infection: Sensitivity, Specificity, and Predictive Values of Commonly Used Questions


  • Patricia Aikins Murphy CNM, DrPH,

  • Janet Jacobson MD,

  • David K. Turok MD, MPH

Patricia Murphy, College of Nursing, University of Utah, 10 South 2000 East, Salt Lake City, UT 84112. E-mail:


Introduction: Practice protocols that mandate prescreening for sexually transmitted infection prior to insertion of an intrauterine device for contraception can pose obstacles for women seeking this highly effective method of birth control. Some practices screen for presumed risk factors for sexually transmitted infection in order to identify those who may be infected or those in whom laboratory testing should be obtained. The value of such criterion-based screening is unclear.

Methods: Data from a prospective observational trial of offering the copper intrauterine device for emergency contraception were used to assess the value of several screening questions in predicting the presence of sexually transmitted infection. Criteria evaluated were age under 25 years, history of a sexually transmitted infection, and having 2 or more sexual partners in the previous 3 months. The sensitivity, specificity, and likelihood ratio of both positive and negative test results and positive and negative predictive values were calculated for 3 separate questions as well as for combinations of these 3 questions.

Results: There were 197 women who received a copper intrauterine device for emergency contraception at the same time they were tested for sexually transmitted infection. In this sample, there were 8 cases of Chlamydia trachomatis identified and no cases of Neisseria gonorrhoeae. The sensitivity of individual and combined questions in identifying those who were infected ranged from 0% to 88%; specificity ranged from 37% to 97%. The positive predictive values for single or combined screening questions were in the range of 4% to 6%. Likelihood ratios for both positive and negative test results did not change posttest likelihood of disease in any appreciable way.

Discussion: Standard screening questions used to identify sexually transmitted infection risk could presume infection in large numbers of uninfected women and are of little value in identifying the women who are truly infected.