Computer-Assisted Surveys may Expand Understanding of STD Risks and Incidence
Telephone surveys administered through computer-assisted self-interviewing could yield a substantially better picture of Americans’ STD-related experiences, and of disease incidence, than telephone surveys conducted by human interviewers. In a study designed to assess differences in reporting between the two survey modes, respondents participating in computer-administered interviews were significantly more likely than those speaking with an interviewer to say that they had had a variety of STD symptoms and that a recent main partner had had an STD. This group also had elevated odds of reporting ever having had chlamydia and marginally elevated odds of reporting that they had had gonorrhea.
The survey was conducted in 1999–2000 among a sample of 18–45-year-old U.S. residents; more than 2,000 eligible men and women agreed to participate and were randomly assigned to take a computer-administered survey or have the same questionnaire administered by an interviewer. Survey items covered respondents’ communication with their partners about sexual behavior, STD-related knowledge and experiences, and demographic characteristics. Analysts used multivariate logistic regression to assess differences in responses by survey mode. They also calculated estimates of STD incidence nationwide on the basis of each sample’s responses.
Among respondents who had had a main partner in the past year, those taking the computerized survey were significantly more likely than those speaking with an interviewer to say that their partner had ever had an STD (odds ratio, 2.4). They also had higher odds of saying that they had never talked with their partner about STD prevention (1.3) and reported more frequent discussions with their partner about their sex life (1.4 on an ordered categorical measure).
Reports of several STD symptoms in the last year—painful urination, genital sores and genital discharge—were significantly more common in computerized surveys than in responses to an interviewer (odds ratios, 1.5–2.8). Nearly all participants in each survey group had heard of gonorrhea, but the odds of giving this response were reduced in the group taking the computerized interview (0.5). No differences were found in familiarity with chlamydia or pelvic inflammatory disease, but the sample taking the computer-administered survey had elevated odds of claiming that they had heard of a fictitious STD (1.5). The reported occurrence both of gonorrhea and of chlamydia in the past year appeared to be higher among respondents taking computerized surveys than among those speaking with an interviewer, but in the multivariate analysis, only the difference for chlamydia was significant (6.1).
Supplementary analyses revealed that differences by survey mode varied across population subgroups. Notably, respondents of all races and ages were more likely to report genital sores in computerized surveys than when speaking with an interviewer, but the differential was sharper for blacks than for respondents of other races (odds ratios, 9.2 and 1.9, respectively), and was greater for 18–25-year-olds than for older participants (13.5 vs. 1.9–2.2). The youngest respondents also showed a marginally greater difference in reporting a partner’s STD than older participants. Although differences do not reach conventional levels of significance, the data suggest that the effect of survey mode on reporting of genital discharge was greater among married or cohabiting respondents than among others, and that the effect on reporting of genital sores declined with increasing level of education.
Projected to the national level, the data collected by interviewers suggest an annual gonorrhea incidence of 0.1% and an annual chlamydia incidence of 0.3%; data from the computerized interviews yield rates of 0.7% and 0.8%, respectively. Similarly, the lifetime incidence of gonorrhea estimated from the survey data is 2.8% if responses to an interviewer are used and 4.9% if based on information provided in a computerized survey; for chlamydia, the incidence rates are 4.6% and 5.6%, respectively. The analysts emphasize that these estimates, which are not statistically reliable, are meant only to “provide…an appreciation of the likely understatement of populationwide STD burdens” that could result from use of interview-based surveys.
Some of the findings, the analysts acknowledge, were unexpected and are difficult to explain (e.g., the computerized survey’s eliciting reports of more frequent discussions with partners about a couple’s sex life, a desirable behavior in the context of STD prevention, and therefore one that respondents would likely not be reluctant to report). Yet, the researchers note that their findings overall “are consistent with a growing body of studies that find that [telephone computer-assisted self-interviewing] increases reporting of stigmatized and sensitive sexual behaviors.”
1. Villarroel MA et al., T-ACASI reduces bias in STD measurements: the National STD and Behavior Measurement Experiment, Sexually Transmitted Diseases, 2008, 35(5):499–506.