Subsequent Sexually Transmitted Infection in Urban Adolescents and Young Adults FREE

IN 1995, there were approximately 12 million new cases of sexually transmitted infections (STIs) in the United States; two thirds of these occurred in individuals younger than 25 years.¹ There are significant long-term consequences to untreated bacterial infection, including pelvic inflammatory disease, infertility, increased risk for ectopic pregnancy, complications of pregnancy, and chronic pelvic pain. Having one STI increases the risk of acquiring others, including human immunodeficiency virus.² Estimated direct and indirect costs of these infections reached $10 billion in 1995.³

Although many infections (chlamydia, gonorrhea, and trichomonas) are curable with appropriate antibiotic treatment, recurrent infection is common. As many as 40% of annual chlamydial and gonorrheal infections occur in people previously infected with those organisms^{4 - 7} ; little is known about recurrent infection with trichomonas.⁸ Most data on subsequent STI probably underestimate the disease burden, because they focus only on recurrence of the original infectious organism. Because these infections all have similar means of acquisition, and the presence of multiple simultaneous infections is not uncommon,⁷ it is important to broaden the scope of analysis for recurrent STIs.

Current recommendations include annual screening for chlamydia and gonorrhea among sexually active adolescents and young adults.^{9 - 11} Recent studies^{1 ,12} suggest that screening for STI on a more frequent basis (vs annual screening) may be indicated for certain of these populations. Little information is available to guide the clinician in screening other similar populations. The purpose of the present study is to determine rates and predictors of subsequent infection with Chlamydia trachomatis, Neisseria gonorrhoeae, or Trichomonas vaginalis among adolescents and young adults diagnosed as having an STI and those known to be uninfected contacts of STI.

Unmarried adolescents and young adults aged 14 to 21 years were recruited from a sexually transmitted disease (STD) clinic and 3 urban adolescent community clinics at the time of treatment for chlamydia, gonorrhea, or trichomonas infection or as a contact of a documented infection from March 1, 1996, through November 30, 1998. Contacts who presented for treatment concurrent with a subject who was enrolled because of a documented STI were eligible regardless of age. These clinics serve a predominantly urban, African American, and low to middle income population. Individuals were excluded from the study who did not plan to continue residence in the area for the next 3 months or who were pregnant. All eligible subjects were invited to participate. No information was collected from persons who declined to participate.

Diagnostic criteria for infection at enrollment in women included culture of endocervical secretions indicating the presence of C trachomatis or N gonorrhoeae, and microscopic examination of saline vaginal secretions revealing motile organisms consistent with T vaginalis. In men, endourethral swabs were cultured for chlamydia, and gram stain was performed to test for polymorphonuclear leukocytes and intracellular diplococci (if results were negative, a culture was performed). Nongonococcal urethritis was diagnosed in men when 10 or more polymorphonuclear leukocytes were observed per high-power field under light microscopy in the absence of gram-negative intracellular diplococci. Men were not screened for trichomonas. Women found to have trichomonas based on findings of the wet mount and men diagnosed as having gonorrhea or nongonococcal urethritis based on the results of the gram stain were enrolled at this visit. The remainder of subjects were enrolled at the time of antibiotic treatment based on the above diagnostic test results or when presenting for treatment as a sexual contact with one of these organisms. Contacts underwent a genital examination, and specimens were obtained before treatment. All subjects, including contacts, were treated with appropriate observed single-dose antibiotic therapy (1 g of azithromycin, 400 mg of cefixime, and/or 2 g of metronidazole) according to Centers for Disease Control and Prevention guidelines.⁹ Individuals with gonorrhea were also treated for chlamydia, regardless of the results of the chlamydia culture. As part of standard clinic policy, all subjects were informed that they required treatment either for an STI or because they were a contact to an unnamed individual who had been treated for an STI. Following treatment, subjects completed a questionnaire and structured interview about sexual behaviors, condom use, substance use, and sociodemographic characteristics. The study was approved by the Indiana University and Purdue University (Indianapolis) institutional review board; written informed consent was obtained from all subjects.

Subjects were asked to return 1, 3, 5, and 7 months after treatment. At each visit, STI screening was performed. Chlamydia and gonorrhea were screened by urine sample, using polymerase chain reaction (Amplicor PCR test kit; Roche Diagnostic Corp, Indianapolis, Ind)^{13 - 14} ; women provided a self-obtained vaginal swab to culture for trichomonas, using modified Diamond medium.¹⁵ A preliminary study demonstrated that self-obtained vaginal swabs identified 25 of 25 vaginal infections with T vaginalis compared with provider-obtained vaginal swabs (D.P.O., oral communication, February 1996). The sensitivity and specificity of polymerase chain reaction for N gonorrhoeae in urine samples were 89.5% and 99.7%, respectively, for women, and 96.4% and 99.1%, respectively, for men. The sensitivity and specificity for C trachomatis were 87.2% and 97.5%, respectively, in women, and 94% and 98%, respectively, in men (B. Van Der Pol, BS, written communication, March 30, 2001).

Subjects completed a questionnaire regarding activity and behaviors in the preceding period. Subjects found to have an STI at return visits were treated with appropriate observed single-dose oral antibiotics. They were informed that their partners required similar treatment and that it was the policy of the county health department to contact and treat all sexual partners.

Summary statistics for continuous data are presented as mean ± SD, or as median (range) for skewed data. Categorical data are presented as frequencies and percentages. t Tests and Wilcoxon rank sum tests were used to compare means and medians between groups, respectively. Pearson product moment correlation was used to compare categorical data between groups. The median time to reinfection was estimated using the Kaplan-Meier product-limit method. Log-rank tests were used to compare the reinfection curves between those who were originally infected and those who were STI contacts at enrollment.

One of the objectives of this study was to determine risk factors for reinfection using baseline and interval sexual behavior data. Demographic variables included in the model were age at enrollment, ethnicity, sex, and initial infection status (infected vs contact). Data collected during the study captured interim sexual behaviors in the 2 months before each follow-up visit and included the number of partners, the number of events, the number of unprotected events, and whether the subject had acquired a new partner. For subjects who reported no sexual contacts during the interval 2 months, the numbers of partners, events, and unprotected events were set at zero. For analysis purposes, the numbers of partners, events, and unprotected events were each categorized into 3 groups: none, 1, or 2 or more.

Cox proportional hazards regression models were used to model the time to the first reinfection, using baseline predictors as the independent variables. We chose not to treat the time to reinfection as interval-censored data. Instead, we used a right-censoring approach and determined the total elapsed time from enrollment to subsequent infection or to the last known visit with negative test results. We defined the date of reinfection as the date of the visit on which the subject tested positive for an STI. If a subject never tested positive or was unavailable for follow-up, the subject was censored at his or her last known visit. This gives the most conservative time to subsequent infection by overestimating the actual time to reinfection. We explored using different time points as estimates for date of reinfection. We used the day after the last known visit with negative test results and the midpoint between the last known visit with negative test results and the visit with positive test results. Both of these definitions yielded the same results as the more conservative definition. The proportional hazards assumption was verified for each predictor variable using graphical methods. To assess the effect of sexual behavior during the follow-up, multiple logistic regression analysis was used to model the probability of infection at each subsequent visit as a function of baseline predictors and interim behavioral factors. Generalized estimating equations were used to account for the correlation between visits on the same subject. All available visits for each subject were included in the analysis. Commercially available statistical software (SAS version 8.0; SAS Institute, Cary, NC) was used to perform the analyses.

Four hundred forty-four subjects were enrolled. Seventy-five percent were women; 77% were African American. Men were older (mean ± SD, 18.3 ± 2.0 years) than women (17.1 ± 1.9 years) (P<.001). About half of the subjects attended school (49%) and were unemployed (47%). One hundred ten (25%) were enrolled as uninfected contacts of STI, 200 (45%) were infected with C trachomatis, 97 (22%) with N gonorrhoeae, 59 (13%) with T vaginalis, and 25 (6%) had nongonococcal urethritis. Simultaneous infection with more than one organism was observed: 29 subjects (7%) were coinfected with chlamydia and gonorrhea, 7 women (2%) with chlamydia and trichomonas, and 3 women (1%) with gonorrhea and trichomonas.

The group was at high risk for STI; overall, 62% reported a prior STI. Men were at higher risk than were women, reporting a younger age at their first episode of sexual intercourse (13.3 vs 14.2 years; P<.001) and more lifetime sexual partners (median, 12 [range, 1-100] vs 5 [range, 1-350]; P<.001) and partners in the 2 months before enrollment (median, 1.5 [range, 0-15] vs 1.0 [range, 0-12]; P<.001).

At enrollment, women and African Americans were more likely to be infected. Compared with uninfected contacts, adolescents and young adults with an STI were younger, more likely to be enrolled in school, reported fewer sexual partners in the 2 months before enrollment, and were more likely to have used a condom at their last sexual encounter. There were no significant differences between infected and contact groups in age at first intercourse or in number of lifetime partners (Table 1).

Compared with subjects enrolled from the STD clinic, those recruited from the adolescent clinics were more likely to be female, African American, in school, to have used a condom at last coitus, and to have reported fewer sexual partners in the 2 months before enrollment (data not shown).

Overall, 80% of subjects (n = 355) had at least one follow-up visit. Three hundred thirty-seven subjects (76%) returned at 1 month, 262 (59%) at 3 months, 225 (51%) at 5 months, and 222 (50%) at 7 months. Compared with those who were unavailable for follow-up, those who returned were more likely to be female, African American, enrolled in school, infected with an STI at enrollment, and to report more sexual partners in the 2 months before enrollment. There were no significant differences in mean age at first intercourse, number of lifetime partners, or reported condom use at last sexual encounter (data not shown).

Subsequent infections were common, irrespective of the enrollment status. Figure 1 and Figure 2 depict the Kaplan-Meier curves for the time to subsequent infection for women and men, respectively. Among the women, the median time to subsequent STI was significantly shorter for those who had an STI at enrollment (140 days) compared with the group who were enrolled as contacts of an STI (209 days) (P = .04). By 7 months, among women, an estimated 53% of the contacts and 73% of those with an STI at enrollment had subsequent STI. The same pattern was observed among men, although it was not statistically significant (P<.09). For men, the median time to infection was 161 days among those enrolled with an STI and 217 days among the contacts. The proportions of the subjects estimated to have an STI at each of the periods, derived using the Kaplan-Meier product-limit method, are shown in Table 2. The percentages of subjects infected are not prevalence-type estimates (ie, number of infected/number of subjects). Results of the proportional hazards regression analysis to estimate the time to subsequent infection as a function of baseline predictor variables are shown in Table 3. Women and those infected at enrollment had a shorter time to reinfection. The repeated-measures multiple regression model that included sexual behaviors in the 2 months before each follow-up visit (Table 4) demonstrated that being female and having at least one new sexual partner independently increased the likelihood of subsequent infection. Original infection status no longer was a significant predictor of subsequent STI (P = .05).

Subsequent infections with C trachomatis, N gonorrhoeae, or T vaginalis were common in this population of adolescents and young adults, despite adequate treatment with supervised antibiotics. Although the time to the acquisition of a subsequent infection was longer for men and for those enrolled as a contact of STI compared with those who had an STI at enrollment, 40% to 73% of subjects were estimated to have become infected within 7 months after treatment. These data add to the growing body of evidence demonstrating the substantial risk of subsequent infection within a few months of an initial infection. In addition, uninfected but at-risk adolescents and young adults also have high rates of subsequent infection. Previous research has demonstrated that subsequent infections with the same species are common among adolescent women. Preliminary work by Fortenberry and Evans¹⁶ indicated that early reinfection with C trachomatis was common among adolescent women. Blythe and colleagues⁶ found that 38% of adolescent women receiving reproductive health care had recurrent chlamydial infection following treatment; most were reinfected within 9 months of treatment. Orr et al¹⁷ reported that 17% to 26% of women were reinfected with chlamydia 6 months after treatment. More recently, Burstein and coworkers¹⁸ demonstrated a high prevalence of chlamydia infections among inner-city adolescent women, with a median time to subsequent infection of 6.3 months.

Studies examining subsequent infection with more than the original species are limited. Oh and colleagues¹⁹ followed up a group of adolescent women for 24 months with periodic examinations for chlamydia and gonorrhea (culture) and trichomoniasis (wet mount). Forty-seven percent of the women were infected with one of the organisms at enrollment (N gonorrhoeae, 11.6%; C trachomatis, 23.2%; and T vaginalis, 23.4%). Nearly 21% of the women had chlamydial and 17% had gonococcal cervicitis during the follow-up. Burstein et al¹² investigated subsequent infections with chlamydia and gonorrhea among adolescents in urban school-based clinics. The median times to subsequent infection with N gonorrhoeae and C trachomatis were 2.6 and 4.8 months, respectively, among the adolescent women.

More recently, Fortenberry et al⁷ described subsequent infections among adolescent women treated for C trachomatis, N gonorrhoeae, or T vaginalis. More than 40% of the women were subsequently infected with one of these organisms within 12 months following treatment; reinfection with the same species and different species organisms was common.

Among adult men treated in STD clinics, recurrent infection with gonorrhea is common.^{4 - 5} "Repeaters" are more likely to be younger, poor, to report "casual" sexual contacts, fail to use condoms, and be African American. Little information is available about the risk for subsequent STI among adolescent men. Cohen and coworkers²⁰ reported declining prevalence rates of chlamydia among adolescent boys screened and treated in public school health settings, suggesting that screening and treatment may be effective in reducing incident disease; the prevalence of gonorrhea did not change over the several years of study. Recurrence rates were not available. In a multicenter STD prevention study, Kamb et al²¹ reported that 12.8% of adolescent men developed an incident STD within 12 months. These limited data suggest that subsequent STI is probably common among high-risk adolescent men and that incident STI is unlikely to be limited to the original infecting species.

Investigators have been unable to identify consistent predictors of initial or subsequent infection among adolescents when multivariate analytic techniques are used.^{6 - 7 ,12 ,18 - 19 ,22} In our study, being female, being infected at enrollment, and reporting an unprotected coital event were associated with a shorter time to subsequent STI. However, multivariate analysis that included sexual behaviors that took place before each visit demonstrated that only being female and having any new interval sexual partner predicted diagnosis of an STI at that visit. The confidence intervals were large for several of the risk variables, and interactions could not be tested because of the sample size. Age was not a predictor of subsequent STI in our study.

The inability to identify clinically important predictors of subsequent infection has led several investigators to recommend screening sexually active adolescents for STI on a more frequent basis vs annually. Expanding screening for subsequent infections beyond the original species would also be important, because subsequent infections are not limited to the original species and are not predicted by characteristics available at the time of treatment.^{7 ,19} Asymptomatic infection is common, and the presence of one STI may increase the risk for another.^{23 - 26} Recurrent infection with chlamydia increases the risk for hospitalization for pelvic inflammatory disease and ectopic pregnancy.²⁷

Considerable research indicates that chlamydia screening and control programs decrease the prevalence of chlamydia in these populations and would be cost-effective for sexually active women, primarily by decreasing complications of short- and long-term sequelae.^{28 - 29} One report,³⁰ completed before the availability of nucleic acid amplification tests, suggests that screening adolescent men would also be cost-effective. We could find no information about the effectiveness of screening adolescents for other STIs, nor do there appear to be consistent useful predictors of reinfection.

There are limitations to our study. The diagnostic tests used at follow-up visits were more sensitive than those used at enrollment, resulting in the potential for misclassification bias at enrollment and an inflated rate of incident infection at the 1-month visit. We would expect this error to be demonstrated by observing an increased rate of incident infections at 1 month among subjects who were not treated at enrollment for a specific organism, because the less sensitive test did not identify an infection when truly present (false-negative test results). This would inflate the rates of incident infection at 1 month, because the more sensitive tests were used at all visits subsequent to enrollment. To identify any misclassification bias, we reexamined the data, looking for differential rates of incident STI at 1 month among subjects who were classified as free of infection at enrollment. For example, subjects who did not have evidence of chlamydia at enrollment or who were not contacts to chlamydia or gonorrhea (and therefore did not receive azithromycin) were no more likely to have chlamydia at 1 month than were those who were treated with azithromycin at enrollment. The results were the same for N gonorrhoeae and T vaginalis, suggesting that the use of more sensitive tests was not a serious problem.

We studied a population of adolescents and young adults at high risk for STI because of personal sexual behaviors and residence in neighborhoods with high STD prevalence. Although we enrolled subjects from 2 types of clinics, there were no differences in the median time to subsequent infection based on the type of clinic from which adolescents and young adults were enrolled (data not shown). Although 80% of the subjects returned for at least one subsequent visit, men, those not attending school, adolescents and young adults reporting fewer sexual partners in the 2 months before enrollment, and uninfected contacts were more likely to be unavailable for follow-up. The 2 groups did not differ at enrollment in behaviors that have been associated with STI. The effects of attrition on our findings remain unknown.

We believe that the increasing body of evidence supports more frequent STI screening for high-risk, asymptomatic, sexually active adolescents and young adults. Newer nucleic acid amplification tests using urine or self-obtained vaginal swabs are more acceptable to this population than are those tests that require urethral or endocervical samples.^{31 - 33} A single sample may be used to detect chlamydia and gonorrhea; nucleic acid amplification tests for trichomonas are available for research purposes and could be made available for public use. The number of men in our study is small, limiting the power to detect statistical differences in the time to second infections. However, the findings are similar to those observed among women, suggesting that the failure to detect differences is related to inadequate statistical power. We believe our data are consistent with and expand on those of Burstein^{12 ,18} and Fortenberry⁷ and their colleagues. Adolescents and young adults are at substantial risk for subsequent STI if they present with a documented infection or are a contact of an STI.

Several unanswered questions remain that will require additional prospective research. What is the optimal frequency of testing? It appears that every 6 months would be sufficient, based on median times to reinfection of about 6 months in several studies. However, this would leave a substantial proportion of infected individuals at risk for transmitting infection for several months. How would testing best be accomplished? Given the availability of urine-based and vaginal swab–based tests, it would be possible to screen adolescents in nontraditional settings¹² and independent of contact with physicians.^{12 ,33 - 37} Home-based testing has been demonstrated to be feasible and acceptable in several smaller European studies.^{32 ,38 - 39} Although this is attractive in that it might overcome the stigma associated with delay in seeking STI-related care,⁴⁰ one loses the opportunity for education if there is no contact with a health provider. It is unclear how one would enlist the adolescent or young adult as an ally in programs for more frequent screening, particularly if divorced from an office or clinic visit.

Which organisms should screening detect? Given our and others' experience, we believe that gonorrhea, chlamydia, and trichomonas are obvious candidates based on prevalence, sequelae, and the availability of curative antibiotic therapy. Who should be screened? Our and other investigators' research has focused on adolescents and young adults at high risk for STI. Little is known about other populations of sexually active youth. Limited national data suggest that about 6% of older adolescent and young adult men have asymptomatic infection with C trachomatis.⁴¹ Research using computer models suggests that it should be cost-effective to screen at this level of prevalent infection. How will such screening be financed, especially if one is to maintain confidentiality in these populations? Cooperation of adolescents, parents, payers, and private and public providers of health care will be required.⁴²

Last, what is the desired outcome of more frequent screening? One hopes to reduce transmission and complications (secondary prevention). When applied to large populations, would this reduce complications, as demonstrated with chlamydia? Would it decrease the incidence of other STIs, including human immunodeficiency virus?⁴³ Additional research is required in each of these areas if we are to move forward in reducing STIs among adolescents and young adults.

Sexually active adolescents are at high risk for STIs. Current recommendations include annual STI screening in this population. Previous research has largely focused on organism-specific screening following an incident infection, ignoring the common mode of acquisition of an STI.

This study demonstrates that incident STIs are common among high-risk adolescents, independent of original organism and infection status. We believe that our data support a recommendation to screen high-risk youth every 6 months for C trachomatis, N gonorrhoeae, and T vaginalis.

Accepted for publication April 10, 2001.

This work was supported in part by grants U19AI31494 and U19AI43924 from the National Institute of Allergy and Infectious Diseases, Bethesda, Md.

Presented in part at the biannual meeting of the International Society for Sexually Transmitted Disease Research, Denver, Colo, July 13, 1999.

We thank Cathy Roberts, MA, and Patricia Brooks for their invaluable assistance in collecting the data for this study.

Corresponding author and reprints: Donald P. Orr, MD, Section of Adolescent Medicine, Department of Pediatrics, Indiana University School of Medicine, Riley Outpatient Garage, Room 070, 575 N West Dr, Indianapolis, IN 46202 (e-mail: dporr@iupui.edu).