IN SEXUAL RISK-REDUCTION RESEARCH, reliance on self-reported outcome measures has been a matter of continuing controversy. 1,2 Studies such as the National Longitudinal Study of Adolescent Health have used self-reported measures of sexually transmitted disease (STD). 3 Yet, self-reported measures of STD history have questionable validity. One implication of possible systematic STD underreporting is the confounding of findings from intervention trials that use self-reported STD incidence as one indicator of program efficacy.
Although a recent report contained position statements on the validity attributed to self-reported measures of condom use and sexual risk behaviors, it did not address issues related to the validity of self-reported STD history. 2 Studies have observed substantial underreporting of STDs in self-reports, as compared with the history of STDs described in corresponding medical records. 4–6 However, such studies have not been prospective, nor have they been conducted in the context of a randomized controlled prevention trial. Studies assessing the validity attributed to self-reported measures of STDs clearly are needed, particularly those used for high-risk populations such as African American female adolescents, in whom STD prevention is a priority.
The underreporting of STDs may result from any of three factors: (1) Patient was not tested nor given any diagnosis when having an STD; (2) patient was tested but unaware of the diagnosis because the results were not received or not understood; (3) patient was tested and aware of the diagnosis, but did not report it.
The objective of the current study was to test the third, and most stringent, of these algorithms by determining the accuracy of STD test results self-reported by African American female adolescents, of which they had been notified over a relatively brief period (≈28 days).
From December 1996 through April 1999, project recruiters screened 1130 female teens in adolescent medicine clinics, health department clinics, and school health classes to assess their eligibility for participation in an HIV/STD prevention trial. Adolescents were eligible to participate in the trial if they were African American females between the ages of 14 and 18 years at the time of enrollment, sexually active in the preceding 6 months, and willing to provide written informed consent. The recruitment sites were in neighborhoods characterized by high rates of unemployment, substance abuse, violence, and STDs. Of those screened, 609 were eligible to participate in the study. Of those not eligible to participate (n = 521), most (98%) were not sexually active.
The current study consisted of 522 (85.7%) eligible adolescents who completed baseline assessments. Most of the eligible teenagers who did not participate in the study were unavailable because of conflicts with their employment schedules. Of the 522 adolescents enrolled in the trial, 92% (n = 479) completed baseline and follow-up measures specific to the current analysis. The Institutional Review Board Committee on Human Research approved the study protocol. Before entering the study, on receiving a verbal explanation of the study protocol that included STD testing, participants read and signed an informed consent.
Data and Specimen Collection
Data collection was conducted at the Family Medicine Clinic. The adolescents were taken to private examination rooms and asked by trained African American female interviewers to provide two vaginal specimens for STD testing. These biologic specimens were analyzed using DNA amplification assays for Neisseria gonorrhoeae and Chlamydia trachomatis (Abbott LCx Probe System for N gonorrhoeae and C trachomatis assays) 7,8 and culture for Trichomonas vaginalis (In Pouch TV test). 9 The adolescents were reimbursed $20 for their participation.
An average of 28 days after baseline STD testing and immediately on completion of intervention workshops, the adolescents were asked to complete a brief self-administered survey. A portion of this survey asked whether, since baseline, a health-care provider had told the adolescent that she had any of a list of STDs, or whether she had received any STD treatment. The importance of responding honestly and the confidentiality of all responses were emphasized to the adolescents.
During specimen collection at baseline, interviewers explained the reason for the specimen collection and gave the participants a brochure describing the three STDs targeted by their tests, another brochure describing where free (program-paid) treatment could be obtained, and a card with phone numbers they could call for their test results. Also, the interviewers asked the adolescents where they could be reached by telephone the following Wednesday evening (4 days later) in the event their tests results were positive, so an appointment for treatment could be made. The adolescents then were randomly assigned either to the intervention condition, in which they received HIV/STD-related knowledge and skills training, or to a comparison condition, in which they received a structurally similar program emphasizing nutrition, fitness, and resume preparation. Both conditions met for intervention workshops on four consecutive Saturdays.
The adolescents with positive test results received a telephone call 4 days after the baseline assessment informing them of the results and a clinic appointment for free treatment the next day. Those who failed to keep their scheduled clinic appointment were contacted again for the scheduling of another appointment unless they reported having been treated elsewhere. The adolescents received a simply worded letter, hand-delivered when they returned to participate in the program workshops 1 week after testing, informing them of their STD test results (either positive or negative).
Treatment and Treatment Documentation
Treatment was given either within the project at two clinic sites, for which free transportation and care were given, or at a medical facility of the participant’s own choosing, for which they provided their own transportation and cost of care. Treatment was considered documented when confirmed by the staff of the medical facility at which the treatment was given.
Accuracy of reporting was determined by calculating the percentage of adolescents with positive test results and the percentage of those with negative test results, then reporting these at the 28-day follow-up interviews. Contingency table analyses were used to determine significant differences between these obtained values as well as possible differential underreporting of the three STDs, and to test for possible differences by assignment to condition. A t test was used to test for possible differences by age. Contingency table analysis also was used to determine significant differences in the accuracy of reported STD treatment between those accurately and those inaccurately reporting positive results for any of the STDs.
Of the 522 adolescents providing specimens at baseline, 43 did not return to complete the follow-up survey, thus providing a sample of 479 adolescents for the current analyses. Adolescents who did not complete the follow-up survey were as likely as those who completed the survey to receive positive test results for one or more of the assessed STDs (P = 0.81). Therefore, loss to attrition did not differ by STD diagnosis. Of the 479 adolescents in the sample, the average age was 16 ± 1.2 years. The prevalence of STDs at baseline was high, with 28% testing positive for at least one disease. By category, 4.8% received positive results for N gonorrhoeae, 17.1% for C trachomatis, 12.3% for T vaginalis, 4.5% for all three STDs, and 13.4% for two of the three STDs.
Approximately 75% of the adolescents with a positive test result were successfully notified by telephone 4 days after specimen collection. The remaining 25% of STD-positive adolescents were notified successfully within the following 7 days. Of the 134 adolescents with a positive test result for at least one STD, treatment was documented for 101 before follow-up assessments.
Table 1 displays the percentage of adolescents accurately reporting both positive and negative test results. Nearly all of the adolescents with negative test results accurately reported these results. Conversely, less than 75% of the adolescents who tested positive accurately reported these results. Underreporting was significantly different by type of STD (P = 0.001).
In contrast to adolescents with negative STD test results, those with positive results were significantly more likely to report their STD status inaccurately (P < 0.0001). Age (t value with 477 df = 0.70;P = 0.48) and assignment to condition (χ 2 with 1 df = 0.66;P = 0.42) were not significantly associated with underreporting of positive STD status.
As compared with the adolescents who inaccurately reported their STD status, those who accurately reported their positive STD status were significantly more likely also to report receipt of treatment accurately (P < 0.001). Of the 99 adolescents who accurately reported having any STD, 84% accurately reported receiving documented treatment, 7% reported no treatment when no treatment was documented, and 9% reported treatment that was not documented. Of the 35 adolescents who inaccurately reported their positive STD status, 54% did not report treatment when treatment was documented, and 46% reported no treatment when no treatment was documented.
Among African American female adolescents participating in an HIV/STD prevention intervention, there was substantial underreporting of STD incidence. The adolescents who did not report their recent STD diagnosis also failed to report treatment. Therefore, survey questions asking African American female adolescents about recent STD diagnosis or treatment may not produce accurate data. Consequently, self-reports of STD diagnosis or treatment may not be valid, possibly introducing systematic error into the evaluation of behavior research studies that include self-reported STDs as an outcome.
The current findings also suggest that the use of self-reported STD incidence as an outcome in HIV/STD prevention interventions may well inflate the possibility that the null hypothesis will be falsely accepted (type 2 error). For example, an efficacious intervention using self-reported measures of STDs would yield a lower STD incidence in the intervention group than in the control group. However, if STDs are underreported, between-group differences in self-reported STD incidences would be reduced and possibly not attain statistical significance, thereby obscuring the identification of efficacious risk-reduction interventions. Thus, reliance on self-reported STD incidence may underestimate intervention effects.
Likewise, the reliance on self-reported STD incidence may pose problems for longitudinal studies. In such studies, the use of self-reported STD incidence as an outcome variable results in a systematically biased underestimation of STD incidence. This can jeopardize the validity of epidemiologic, mathematical, and psychosocial models designed to predict the incidence of STDs.
The findings also suggest that adolescents may underreport their STD history in a clinical setting. If clinic protocols call for more intensive counseling and education efforts for adolescents reinfected with either the same or another STD, this underreporting may result in missed opportunities for prevention. Therefore, the sexual and treatment histories of adolescents should be checked against any available medical records.
The reasons for underreporting of STDs in a confidential setting, with its emphasis on honesty, are unclear. Other researchers have proposed several explanations for inaccurate reporting 10 of sexual behaviors and STD status: failure to remember or recall bias, 11,12 lack of understanding about an STD, 4,5,13,14 and intentional denial because of concern about confidentiality 5,15 or social desirability. 4,5,12,15,16 Several steps were taken to minimize the possible influence of these dynamics. Failure to remember was minimized by the short period covered (i.e., 28 days) and dual notification of STD status by letter and telephone. Repeated explanations of the testing and notifications of results were designed to reduce any lack of understanding about STDs. Because participants receiving STD education through the intervention were no more accurate than the comparison participants in reporting STDs, it seems unlikely that lack of understanding was a significant factor. Repeated assurances of confidentiality, with coded identification of data, and involvement of the same staff in all steps of the testing, notification, and collection of self-report status were attempts to reduce concerns about confidentiality and social desirability.
Intentional denial appears to be the most likely explanation for the underreporting found in this study. However, with almost half of the adolescents not reporting their STD also not seeking treatment, it seems plausible that this segment of the sample may be in unintentional denial regarding their disease state. That is, instead of consciously denying their positive STD status, they did not believe they had the infection. Yet, the differential underreporting of the three STDs seems to support intentional denial, especially with gonorrhea, the most stigmatizing of the STDs, found to be the most underreported. Gonorrhea generally is considered very serious and something only “bad girls” would get.
One strategy for reducing the systematic error created by underreporting of STDs is to complement self-report with biologic assessments. 2 The advent of self-administered methods for obtaining biologic specimens maximizes the utility of including biologic outcomes in diverse clinical and community settings. In addition, these methods yield highly sensitive and specific test results for C trachomatis, N gonorrhoeae, and T vaginalis, with high rates of participant acceptance. 17
Given the minimal overreporting of STD diagnoses, this study is limited by the possibility that adolescents may have been diagnosed with an STD outside this program. Additionally, the sample was limited to economically disadvantaged African American female adolescents. Therefore, the findings may not generalize to males, other age or racial/ethnic groups, or adolescents from different socioeconomic strata.
Further research should explore other potential influences on the accuracy of self-reported STD history such as method, promptness, and setting of STD treatment. Because this study was limited to a specific population of females, further study is needed to determine the extent of underreported STD incidence and treatment among other populations (e.g., males, adults, and other ethnic groups).
The substantial underreporting of STD incidence in this study suggests that reliance on self-reported STD history may introduce misclassification bias, potentially leading to false conclusions in survey research and research evaluating the efficacy of STD/HIV prevention interventions.
1. Fishbein M, Pequegnat W. Evaluating AIDS prevention interventions using behavioral and biological outcome measures. Sex Transm Dis 2000; 27: 101–110.
2. Pequegnat W, Fishbein M, Celentano D, et al. NIMH/APPC Workgroup on Behavioral and Biological Outcomes in HIV/STD Prevention Studies: A position statement. Sex Transm Dis 2000; 27: 127–132.
3. Crosby RA, Leichliter JS, Brackbill RM. Longitudinal prediction of STDs among adolescents: results from a national survey. Am J Prev Med 2000; 18: 312–317.
4. Clark LR, Brasseux C, Richmond D, Getson P, D’Angelo LJ. Are adolescents accurate in self-report of frequencies of sexually transmitted diseases and pregnancies? J Adolesc Health 1997; 21: 91–96.
5. Hornberger LL, Rosenthal SL, Biro FM, Stanberry LR. Sexual histories of adolescent girls: comparison between interview and chart. J Adolesc Health 1995; 16: 235–239.
6. Hoffman S, Morrissey A, Walts A, et al. Validating Intervention Effects, Behavior Change, and Self-Reports of Condom Use: Using the Medical Record: A Cautionary Tale. American Public Health Association. Poster Presentation ed, Washington DC, 1998.
7. Hook EW III, Ching SF, Stephens J, Hardy KF, Smith KR, Lee HH. Diagnosis of Neisseria gonorrhoeae
infections in women by using the ligase chain reaction on patient-obtained vaginal swabs. J Clin Microbiol 1997; 35: 2129–2132.
8. Hook EW III, Smith K, Mullen C, et al. Diagnosis of genitourinary Chlamydia trachomatis
infections by using the ligase chain reaction on patient-obtained vaginal swabs. J Clin Microbiol 1997; 35: 2133–2135.
9. Biomed Diagnostics I. A Selective Culture System for the Diagnosis of Human Trichomoniasis. Document no. 100–001. Revision B ed. October, 1995.
10. Gordon SM, Mosure DJ, Lewis J, Brown S, McNagy SE, Schmid GP. Prevalence of self-medication with antibiotics among patients attending a clinic for treatment of sexually transmitted diseases. Clin Infect Dis 1993; 17: 162–165.
11. Catania JA, Gibson DR, Chitwood DD, Coates TJ. Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull 1990; 108: 339–362.
12. Zenilman JM, Weisman CS, Rompalo AM, et al. Condom use to prevent incident STDs: the validity of self-reported condom use [see comments]. Sex Transm Dis 1995; 22: 15–21.
13. McHugh MT, Palusci VJ. Assessing prior history of sexually transmitted disease [letter, see comments]. JAMA 1992; 267: 1610–1611.
14. Fleisher JM, Minkoff HL, Senie RT, Endias RE. Assessing prior history of sexually transmitted disease [letter, see comments]. JAMA 1991; 266: 1646.
15. Hingson R, Strunin L. Validity, reliability, and generalizability in studies of AIDS knowledge, attitudes, and behavioral risks based on subject self-report [see comments]. Am J Prev Med 1993; 9: 62–64.
16. Turner CF, Miller HG. Zenilman’s anomaly reconsidered: fallible reports, ceteris paribus, and other hypotheses [see comments]. Sex Transm Dis 1997; 24: 522–527.
17. Smith K, Harrington KF, Wingood G, Oh MK, Hook EW III, DiClemente R. Self-obtained vaginal swabs for treatable STD diagnosis in adolescent women. Arch Pediatr Adolesc Med (in press).