Audio computer-assisted self interviewing for sexually transmitted infection prediction

Thesis (Ph. D.)--University of Washington, 2003Background. Sexually transmitted infections (STI) cause considerable morbidity in the United States, but are under-assessed. We explored how to improve STI screening in STD clinics. Our aims were to: (1) identify efficient predictors of viral and non-viral STI, and (2) determine whether reporting of STI risk factors and characteristics varied by type of sexual history interview.Design. This cross-sectional study utilized an audio computer-assisted self-interview (ACASI) followed by a clinician-administered sexual history among 295 females and 314 males (n = 609) in a public STD clinic in Seattle, Washington.Methods. Participants were tested for herpes simplex virus type 2 (HSV-2), Neisseria gonorrhoeae, Chlamydia trachomatis, and among women, Trichomonas vaginalis. We assessed individual, sexual partnership, and biological risk factors to predict asymptomatic STIs using logistic regression with Akaike's information criterion (Aim 1). We assessed predictors of HSV-2 among individuals without genital lesions who had not been previously diagnosed with HSV-2; creating risk scores and examining their performance using receiver operating characteristic curves. We assessed predictors for non-viral STI among women without vaginal discharge or abdominal pain symptoms. We assessed data completeness, item reporting, and report concordance between the two interviews (Aim 2).Results. Asymptomatic HSV-2 seroprevalence in men (21.9%) was predicted by individual-level factors, and in women (27.3%) by both individual- and partnership-level factors. A low, medium, or high risk score was associated with HSV-2 prevalences of 8.5%, 21.3%, and 41.8% respectively in asymptomatic men and 10.0%, 43.3%, and 61.8% in asymptomatic women. There were fewer missing data in the ACASI (mean 5.4% vs. 25.5%, p < 0.001). Risk reporting and reporting concordance by interview largely corresponded to assumptions regarding social desirability bias, especially among women. Computerized sexual histories were acceptable to 89% of respondents.Conclusion. Risk assessment that includes individual and partnership factors can identify asymptomatic persons with a relatively high prevalence of STI, and may help efficiently target diagnostic resources. ACASI sexual histories were more complete than recorded clinician histories, and may reduce missed opportunities to counsel about high-risk behaviors. Combining ACASI and clinician-administered histories for risk assessment in STD care settings may improve data quality and clinical management