AMONG ADOLESCENTS and young adults, Chlamydia trachomatis is the most common bacterial sexually transmitted infection (STI) in the United States.1 An estimated 3 million new infections occur each year. Chlamydial infection may cause pelvic inflammatory disease, ectopic pregnancy, and tubal infertility in women.2,3 Infection also increases susceptibility to and transmission of HIV in women and men.4 Despite prevention and treatment measures, chlamydial infection continues to cause substantial morbidity.
Sexual mixing between members of different groups facilitates the spread of STI.5–7 Partner mixing by age is common among both adolescents and adults.8,9 Adolescent girls with older male partners are more likely than girls with partners close in age to be younger at first sex10,11 and use condoms inconsistently,10,12 both being risk factors for STI. Additionally, several studies associate older male partners with diagnosis or self-report of STI in adolescents.13–15 This association, however, is not consistent when examining recurrent infection.16,17 Whether this potential relation between older partners and risky behavior continues beyond adolescence is unclear. Two studies reporting the effect of partner age difference on STI among adults found little association with older partners, but both studies had major limitations. One study diagnosed chlamydial infection by culture,5 a diagnostic method that is no longer the gold standard.18 The other study relied on self-report of STI test or treatment in the past year,9 which is again a suboptimal method of determining infection status. These findings suggest that older male partners may play a lesser role in risk of STI as adolescent girls mature into adulthood, although misclassification of infection status may have reduced the ability to detect differences by partner age.
The mechanism by which age difference affects the risk of STI may be 2-fold. First, age-discordant partnerships are purported to represent relationships with unequal distribution of power.10,19 A younger woman may have more difficulty confronting sexual pressures, negotiating condom use, or influencing her partner’s sexual behavior with an older man rather than someone her own age. Second, partnerships between members of different sexual networks or populations have been linked to higher rates of chlamydial infection.5,20 Age-discordant partnerships among adolescents may facilitate mixing between populations with low and high STI prevalence.5,21–23 In national surveillance data, age-stratified rates of reported chlamydial infection differ by gender, with the highest reported rates among females aged 15 to 19 and males aged 20 to 24.24
Wave III of the National Longitudinal Study of Adolescent Health (Add Health) provides the necessary demographic, personal history, and behavior data together with laboratory-diagnosed STI results to examine the association between partner age difference and prevalence of chlamydial infection among young adult women.
Materials and Methods
Study Design and Sample
Add Health is a 3-wave prospective cohort study that followed nearly 20,000 adolescents into adulthood from September 1994 to May 2002.25 For this study, we conducted a cross-sectional analysis of Wave III (April 2, 2001 to May 9, 2002), which targeted all Wave I participants. Our study population was restricted to female Wave III participants who answered “yes” when asked “Have you ever had vaginal intercourse?” The University of North Carolina institutional review board approved all study procedures.
Add Health’s 2-stage sampling has been described in detail elsewhere.25 In short, a systematic random sample of secondary schools was chosen with unequal probability of selection. The sample was stratified so that selected schools were representative of all US secondary schools with respect to region, urbanicity, and the percentage of black and white. The original study participants were drawn from students in grades 7 through 12 enrolled in the selected schools. Certain populations, including black students with a college- educated parent, Cubans, and Puerto Ricans, were oversampled to increase estimate precision for these groups. Poststratification sampling weights account for persons who refused to participate or could not be located. The Add Health Wave III cohort, when incorporating survey weights, provides a representative sample of young adults aged 18 to 26 years residing in the United States.
Interview and Specimen Collection
Wave III sought participation from all original Wave I respondents who could be contacted. The in-home interviewer recorded responses into a laptop computer for nonsensitive issues. The participant used computer-assisted self-interview to enter responses directly into the computer for sensitive issues such as sexual experiences and behaviors.
Respondents who consented to provide a urine specimen for Chlamydia trachomatis testing received $10. A more detailed report of Add Health STI testing is available elsewhere.26 Cooled urine specimens were shipped by overnight express to the University of North Carolina at Chapel Hill for diagnostic testing.
Urine specimens were tested for C. trachomatis per manufacturer instructions using ligase chain reaction (LCR) amplification technology in the Abbott LCx Probe System (Abbott Laboratories, Abbott Park, IL). Specimens were tested even if they exceeded the recommended volume.27
The outcome variable was a positive test result for C. trachomatis. The primary exposure was the difference in years between each female study participant and her most recent male sex partner. A negative age difference meant the woman’s partner was younger; a positive age difference meant the woman’s partner was older. Additionally, as a secondary exposure, we used the age difference of the most age-discordant partner in the past year. The exposure measures were categorized to reflect substantively meaningful age differences with the idea that in this age range of young women, partners close in age may be associated with different risks than younger partners, slightly older partners, and much older partners. Potential modifying or confounding covariates were derived from the set of self-reported demographic, behavior, and health care factors available from the in-home interview. Covariates in the full model included age, race/ethnicity, marital status, employment status, ability to pay rent and utilities in the past year, highest attained education, antibiotic use in the past 30 days, age at first sex, number of sex partners in the past year, condom use in the past year, exchanging sex for money or drugs in the past year, and regretting sex in the past year.
Analysis was performed using Stata Version 7.0 (Stata Corporation, College Station, TX). Analyses accounted for Add Health’s complex survey design by using school as the primary sampling unit, region of the country as the stratification variable, and poststratification weights. In preliminary analyses, we examined the frequency distribution of the outcome, primary and secondary exposures, and other covariates. We calculated bivariate prevalence odds ratios (OR) and 95% confidence intervals (CI) to assess the relation between the covariates and chlamydial infection.
We used unconditional multiple logistic regression for survey data with a backward elimination strategy.28 Covariates that could alter the association between age difference and prevalence of chlamydial infection were included in the full model. We tested the null hypothesis that interaction terms between the exposure and each potential modifier were equal to zero, and looked for substantive changes between unstratified and stratified effect estimates. After examination and removal of interaction terms, we eliminated potentially confounding covariates one at a time from the model beginning with the variable with the largest P value. We evaluated confounding by comparing the crude, adjusted, and fully adjusted effect estimates. We retained the covariate if the change between crude and adjusted effect estimates was greater than 10%.29 Otherwise, the covariate was dropped. Backwards elimination stopped when all covariates that neither modified nor confounded the association between age difference and chlamydial infection were removed from the model. We examined the final model for collinearity and overly influential covariate patterns. Primary and secondary exposure measures were evaluated through separate but identical processes.
Mixing by age within partnerships can be assortative (like-with-like), disassortative (like-with-unlike), or random. The Q statistic, defined as (Σiwi − 1)/(N − 1), where wi is the matrix eigenvalue, uses an N × N mixing matrix to identify high or low within-group mixing.5,30 Q is zero if partner choice is random and approaches one in the assortative and −1/N in the dissassortative extremes. For our analyses, the matrix rows were the categorized age of the woman and the matrix columns were the categorized age of the male partner, calculated for the 2 partner age difference measures. This assessment could not incorporate survey weighting.
Of the 18,924 Add Health participants in the nationally representative Wave I sample, 1109 (5.9%) refused participation, 3493 (18.5%) could not be located or were unable to participate, and 14,322 (75.7%) were located and agreed to participate in Wave III. Of these, 6594 were females who reported ever having vaginal intercourse (87.2% of all Wave III females). C. trachomatis test results were available for 5854 (88.8%) of the sexually experienced females. Reasons for unavailable test results included inability or refusal to provide a urine specimen, processing errors due to shipping, or laboratory problems.
The overall prevalence of chlamydial infection among sexually experienced women was 5.1% (95% CI, 4.2%–6.0%; Table 1). The majority of women with test results were white (68.2%), with representation of black (17.1%), Latino (11.0%), Asian American (2.9%), and Native American (0.8%) women. The mean age was 21.8 years (SE, 0.1 year), the mean age at first sex was 16.3 years (SE, 0.1 year) and the mean number of sex partners in the past year was 1.5 partners (SE, 0.04 partners). Partner ages ranged from 15 to 60 years. The mean age difference of the most recent partner was 2.9 years (SE, 0.1 year) and mean age difference of the most age-discordant partner in the past year was 3.4 years (SE, 0.1 year).
The Q statistics for the age difference with the most recent partner (Q = 0.3) and the most age-discordant partner (Q = 0.2) indicated that mixing by age between women and their partners was random, with a tendency towards assortative (like-with-like) mixing.
There was little difference in chlamydial prevalence between women with a most recent partner 2 to 8 years younger (“youngest”; OR, 1.9; 95% CI, 1.0, 3.7) and women with a most recent partner six or more years older (“oldest”; OR, 1.6; 95% CI, 0.9, 2.9) when compared with women with a most recent partner within 1 year age difference (the referent, “close in age”; Table 1). The similarity of the effect estimates across age difference categories persisted when examining the secondary exposure measure, age difference of the most age-discordant partner (youngest OR, 2.4; 95% CI, 1.2, 4.7; oldest OR, 2.1; 95% CI, 1.2, 3.4). Prevalence ratios were comparable to prevalence odds ratio for nearly all characteristics because of the overall low prevalence of chlamydial infection. For attributes where the prevalence of infection varied markedly across categories, such as race/ethnicity and exchanging sex for money or drugs, the prevalence ratio was attenuated slightly: black women had 5.7 times the odds of infection when compared with white women, whereas black women had 5.1 times the prevalence of white women.
In multiple logistic regression analyses, the odds of the prevalence of chlamydial infection among women with the youngest partners were approximately 2 times greater (most recent OR, 1.8; 95% CI, 0.9, 3.5; most discordant OR, 2.1; 95% CI, 1.1, 4.3) than among women with partners close in age, adjusting for number of partners in the past year (Table 2). Among women with older partners, the adjusted odds of infection differ little between partners 2 to 5 years older (most recent OR, 1.4; 95% CI, 0.9, 2.3; most discordant OR, 1.4; 95% CI, 0.8, 2.3) and partners 6 or more years older (most recent OR, 1.6; 95% CI, 0.9, 2.8; most discordant OR, 1.7; 95% CI, 1.0, 2.8) when compared with partners within 1-year age difference. The relation between partner age difference and chlamydial infection is consistent for the 2 exposure measures. These associations did not vary by women’s age or number of sex partners in the past year.
Stratified by Race
The relation between most discordant partner age difference and chlamydial infection did vary by women’s race/ethnicity (interaction P = 0.1). For white women, the greatest odds of infection are among those with the oldest partners (OR, 2.8; 95% CI, 1.2, 6.9) compared with partners close in age, adjusting for age, highest attained education, and number of partners in the past year (Table 2). Youngest partners have little effect on adjusted odds of infection for this group (OR, 1.2; 95% CI, 0.2, 6.4). The associations are reversed, however, among black women. For black women, the adjusted odds ratio is greatest for the youngest partners (OR, 3.2; 95% CI, 1.2, 9.0) and shows little effect for the oldest partners (OR, 0.7; 95% CI, 0.3, 1.4) when compared with partners close in age. The unstratified effect estimate showing the largest odds ratio for the youngest partners is clearly moderated by the experience of black women, who constitute only 17% of the study population.
Qualitatively similar results were seen for the stratified association between most recent partner age difference and chlamydial infection, although the variation across race/ethnicity was not statistically significant (interaction P = 0.3; Table 2). Only among Latino women is there a substantial difference between the 2 stratified exposure measures, with consistently stronger estimated effects for the most discordant, rather than most recent, partner age difference.
Among young adult women, older partners when compared with partners within 1 year age difference are moderately associated with the prevalence of chlamydial infection. A similarly increased association was observed for younger partners. These associations remain consistent when examining the age difference with the most recent partner and the most age-discordant partner in the past year. Overall, this finding in young adults contrasts with some previous studies of adolescent girls, for whom those with older partners were 2 to 4 times more likely to be have an STI.13–15
Among young adult white women, older partners are significantly associated with increased prevalence of chlamydial infection.
The relation between most age-discordant partner and chlamydial infection seems to vary by women’s race/ethnicity, although with stratification these results become imprecise and should be interpreted cautiously. If this is a true difference, the divergent effect on STI of older partners for white women and younger partners for black women may result from racial/ethnic differences in age-dependent sexual behaviors or rates of chlamydial infection.24,31
It is unclear how the diminished overall association between partner age and chlamydial infection develops from adolescence to young adulthood, but several explanations are possible. Physical and emotional maturity may prepare a young woman to better protect herself by exercising more power in sexual decision-making. Older partners for young adults may also represent a different population than older partners for adolescents, both in terms of age-specific prevalence and social networks. Adolescents likely need to reach beyond their social setting to meet older partners. These partners may be links to networks with higher risk behaviors such as multiple and overlapping partners, drug use, and transactional sex. Age is merely a marker for these risky behaviors. Alternatively, young adults may have more opportunities to meet older partners in their usual social network. Although the partners are older, they may be similar with respect to risk behaviors. Lastly, the diminished overall association may be artifactually low. Some of the women in this population with partners within 1 year age difference (the referent) are interacting with men aged 20 to 24 years, the group with the highest rate of chlamydial infection. A decrease in the odds ratio between adolescence and young adulthood may represent an increased risk among the referent group rather than a change in risk for those with older partners. The increased odds among white women with older partners is even more surprising, given the higher prevalence among the partners in the referent group.
Few studies have rigorously assessed the association between partner age difference and the prevalence of chlamydial infection in young adult, rather than adolescent, females. Our study improves over these earlier works by using nucleic acid amplification detection of chlamydial infection rather than culture5 or self-report of recent or lifetime STI test, treatment, or diagnosis.9 Nucleic acid amplification testing is approximately 90% sensitive and 99% specific.32 Chlamydial culture is a considerably less sensitive (60%–65%) diagnostic tool and is no longer the optimal diagnostic method.18 The high rate of false-negatives may have attenuated an association between partner age and infection. Results reliant on self-report of STI test are also subject to numerous biases. Moreover, evaluating chlamydial infection alone recognizes the distinct epidemiology and age distribution across the spectrum of STI.
Using the age difference with both the most recent and the most age-discordant partner in the past year mitigates a limitation common to earlier investigations of partner age difference and STI.33 Because untreated chlamydial infection can persist for an extended period of time,34 the most recent partner may not be the source of a prevalent infection. Our findings show comparable estimated effects when examining either measure. Additionally, we assessed the applicability of calculating the exposure measure as the absolute difference in ages rather than subtracting the woman’s age from that of her partner. The dose-response remained nonlinear for the absolute difference, with the maximum observed effect at a difference of 6 to 10 years (P = 0.05).
Implicit in a difference between the age of the most recent partner and the age of the most age-discordant partner is the existence of multiple partners. Although we controlled for the number of partners, other unmeasured confounding behaviors such as concurrency or shortened time between partners that can only occur in the presence of multiple partners might be important. Furthermore, type of partner may be of differential importance across the 2 exposure measures. Women with more than one partner in the past year may be more likely to have a main and a casual partner whereas women with a single partner may be more likely to have only a main partner. Type of partner may play a primary role in transmission of infection.35,36
Our study findings are limited by the quality of both the study sample and the diagnostic test. The quality of our study sample depends on how well the original school-based sample represented all students attending U.S. secondary schools, response to the Wave III follow-up survey, valid reporting of sexual experiences, and reason for missing diagnostic outcome. The original sample included only students on school registers, but bias in Add Health caused by school dropouts is thought to be small.37 Although 24% of Wave I participants could not be located for Wave III, the poststratification sample weight adjustment ensured that this missing data bias is also small.38 Computer-assisted self-interview likely improved both the frequency and validity of answers to sensitive questions about sexual experiences.39 Additionally, respondents who participated in STI testing were similar to those who did not participate.40 Prevalence estimates also are robust to differences between survey responders and nonresponders and to the performance of the diagnostic test for chlamydial infection.41 Unavailable information such as accurate measures of partner concurrency or standard definitions of partner type would have enriched these analyses. Reports of partner age also may be inaccurate. A more powerful study is needed to fully examine the potential modifying effect of a woman’s race/ethnicity on the association between partner age difference and chlamydial infection.
Understanding the role of partners in the social dynamics of chlamydial infection is central to developing successful public health interventions and treatment programs. Among young adults, older and younger partners convey essentially the same elevated risk of STI when compared with partners close in age. This association may vary by women’s race/ethnicity. Although partnerships are clearly important, it is doubtful that partner age difference is as substantial a contributor to the high rates of chlamydial infection among young women as it is to adolescent girls. Among adolescent girls, older partners are a strong risk factor for STI12,13 and likely facilitate maintenance of high infection rates in this group.42 This discrepancy in risk factor between adolescents and young adults may be exploited to better understand how adolescents in age-discordant partnerships may be helped to protect themselves from STI. Programs to assist with self-esteem, relationship communication, condom negotiation, and partner notification may benefit by understanding why age difference is so important in STI risk among adolescents, but not when these women are just a few years older.
Young adults, unlike adolescents, are studied infrequently even though they continue to remain at high risk of chlamydial infection. This investigation and other recent work suggest that risk factors such as partner age difference and age at first sex,43 which contribute to the spread of STI among adolescents, play distinctly lesser roles in the epidemiology of infection among young adults. Additional research focused on young adults is needed to understand the high STI rates in this age group and to identify how these rates may be reduced.
1. Weinstock H, Berman S, Cates W Jr. Sexually transmitted diseases among American youth: Incidence and prevalence estimates, 2000. Perspect Sex Reprod Health 2004; 36:6–10.
2. Westrom L, Eschenbach D. Pelvic inflammatory disease. In: Holmes KK, ed. Sexually Transmitted Diseases, 3rd ed. New York: McGraw-Hill Health Professions Division, 1999:783–809.
3. Holmes KK. Sexually Transmitted Diseases, 3rd ed. New York: McGraw-Hill Health Professions Division, 1999.
4. Fleming DT, Wasserheit JN. From epidemiological synergy to public health policy and practice: The contribution of other sexually transmitted diseases to sexual transmission of HIV infection. Sex Transm Infect 1999; 75:3–17.
5. Aral SO, Hughes JP, Stoner B, et al. Sexual mixing patterns in the spread of gonococcal and chlamydial infections. Am J Public Health 1999; 89:825–833.
6. Rothenberg RB, Potterat JJ. Temporal and social aspects of gonorrhea transmission: The force of infectivity. Sex Transm Dis 1988; 15:88–92.
7. Potterat JJ, Rothenberg RB, Muth SQ. Network structural dynamics and infectious disease propagation. Int J STD AIDS 1999; 10:182–185.
8. Ford K, Sohn W, Lepkowski J. American adolescents: Sexual mixing patterns, bridge partners, and concurrency. Sex Transm Dis 2002; 29:13–19.
9. Kraut-Becher JR, Aral SO. Patterns of age mixing and sexually transmitted infections. Int J STD AIDS 2006; 17:378–383.
10. Miller KS, Clark LF, Moore JS. Sexual initiation with older male partners and subsequent HIV risk behavior among female adolescents. Fam Plann Perspect 1997; 29:212–214.
11. Rickert VI, Wiemann CM, Berenson AB. Health risk behaviors among pregnant adolescents with older partners. Arch Pediatr Adolesc Med 1997; 151:276–280.
12. DiClemente RJ, Wingood GM, Crosby RA, et al. Sexual risk behaviors associated with having older sex partners: A study of black adolescent females. Sex Transm Dis 2002; 29:20–24.
13. Begley E, Crosby RA, DiClemente RJ, et al. Older partners and STD prevalence among pregnant African American teens. Sex Transm Dis 2003; 30:211–213.
14. Boyer CB, Shafer MA, Teitle E, et al. Sexually transmitted diseases in a health maintenance organization teen clinic: Associations of race, partner's age, and marijuana use. Arch Pediatr Adolesc Med 1999; 153:838–844.
15. Ford K, Lepkowski JM. Characteristics of sexual partners and STD infection among American adolescents. Int J STD AIDS 2004; 15:260–265.
16. Kissinger P, Clayton JL, O'Brien ME, et al. Older partners not associated with recurrence among female teenagers infected with Chlamydia trachomatis.
Sex Transm Dis 2002; 29:144–149.
17. Magnus M, Schillinger JA, Fortenberry JD, et al. Partner age not associated with recurrent Chlamydia trachomatis
infection, condom use, or partner treatment and referral among adolescent women. J Adolesc Health 2006; 39:396–403.
18. Centers for Disease Control and Prevention. Screening tests to detect Chlamydia trachomatis
and Neisseria gonorrhoeae
infections—2002. MMWR Recomm Rep 2002; 51:1–38.
19. Abma J, Driscoll A, Moore K. Young women's degree of control over first intercourse: An exploratory analysis. Fam Plann Perspect 1998; 30:12–18.
20. Aral SO, Foxman B. Spatial mixing and bridging: Risk factors for what? Sex Transm Dis 2003; 30:750–751.
21. Wylie JL, Jolly A. Patterns of chlamydia and gonorrhea infection in sexual networks in Manitoba, Canada. Sex Transm Dis 2001; 28:14–24.
22. Kerani RP, Golden MR, Whittington WL, et al. Spatial bridges for the importation of gonorrhea and chlamydial infection. Sex Transm Dis 2003; 30:742–749.
23. Bunnell RE, Dahlberg L, Rolfs R, et al. High prevalence and incidence of sexually transmitted diseases in urban adolescent females despite moderate risk behaviors. J Infect Dis 1999; 180:1624–1631.
24. Centers for Disease Control and Prevention. Sexually Transmitted Disease Surveillance, 2004. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2005.
25. Bearman PS, Jones J, Udry JR. The National Longitudinal Study of Adolescent Health: Study Design. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill, 2004. Available at: http://www.cpc.unc.edu/projects/addhealth/design
27. Van Der Pol B, Ferrero DV, Buck-Barrington L, et al. Multicenter evaluation of the BDProbeTec ET System for detection of Chlamydia trachomatis
and Neisseria gonorrhoeae
in urine specimens, female endocervical swabs, and male urethral swabs. J Clin Microbiol 2001; 39:1008–1016.
28. Maldonado G, Greenland S. Simulation study of confounder-selection strategies. Am J Epidemiol 1993; 138:923–936.
29. Kleinbaum DG, Klein M. Logistic Regression: A Self-learning Text, 2nd ed. New York: Springer, 2002.
30. Gupta S, Anderson RM, May RM. Networks of sexual contacts: Implications for the pattern of spread of HIV. AIDS 1989; 3:807–817.
31. Auerswald CL, Muth SQ, Brown B, et al. Does partner selection contribute to sex differences in sexually transmitted infection rates among African American adolescents in San Francisco? Sex Transm Dis 2006; 33:480–484.
32. Lee HH, Chernesky MA, Schachter J, et al. Diagnosis of Chlamydia trachomatis
genitourinary infection in women by ligase chain reaction assay of urine. Lancet 1995; 345:213–216.
33. Kissinger P. Do older partners place adolescent girls at higher risk for STDs? Sex Transm Dis 2003; 30:214–215.
34. Golden MR, Schillinger JA, Markowitz L, et al. Duration of untreated genital infections with Chlamydia trachomatis
: A review of the literature. Sex Transm Dis 2000; 27:329–337.
35. Lee JK, Jennings JM, Ellen JM. Discordant sexual partnering: A study of high-risk adolescents in San Francisco. Sex Transm Dis 2003; 30:234–240.
36. Tschann JM, Adler NE, Millstein SG, et al. Relative power between sexual partners and condom use among adolescents. J Adolesc Health 2002; 31:17–25.
37. Udry JR, Chantala K. Missing school dropouts in surveys does not bias risk estimates. Soc Sci Res 2003; 32:294–311.
39. Kurth AE, Martin DP, Golden MR, et al. A comparison between audio computer-assisted self-interviews and clinician interviews for obtaining the sexual history. Sex Transm Dis 2004; 31:719–726.
40. Ford CA, Jaccard J, Millstein SG, et al. To pee or not to pee? Who refuses to provide a urine specimen for Chlamydia trachomatis
and Neisseria gonorrhoeae
testing? J Adolesc Health 2004; 34:144–145.
41. Miller WC, Ford CA, Morris M, et al. Prevalence of chlamydial and gonococcal infections among young adults in the United States. JAMA 2004; 291:2229–2236.
42. Gorbach PM, Drumright LN, Holmes KK. Discord, discordance, and concurrency: Comparing individual and partnership-level analyses of new partnerships of young adults at risk of sexually transmitted infections. Sex Transm Dis 2005; 32:7–12.
43. Kaestle CE, Halpern CT, Miller WC, et al. Young age at first sexual intercourse and sexually transmitted infections in adolescents and young adults. Am J Epidemiol 2005; 161:774–780.