Copyright 2005 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2005
Influencing adolescents’ sexual behaviors has the potential to influence trajectories of risk for sexually transmitted infections (STIs) among young adults.
To determine whether family, school, and individual factors associated with increased duration of virginity also protect against STIs in young adulthood.
Prospective cohort study. Wave I of the National Longitudinal Study of Adolescent Health occurred in 1995 when participants were in grades 7 through 12. Six years later, all wave I participants who could be located were invited to participate in wave III and provide a urine specimen for STI testing.
In-home interviews in the continental United States, Alaska, and Hawaii.
Population-based sample. Of 18 924 participants in the nationally representative weighted wave I sample, 14 322 (75.7%) were located and participated in wave III. Test results for Chlamydia trachomatis, Neisseria gonorrhoeae, and Trichomonas vaginalis were available for 11 594 (81.0%) of wave III participants.
Main Outcome Measure
Positive test result for C trachomatis, N gonorrhoeae, or T vaginalis.
Controlling for biological sex, age, race/ethnicity, family structure, and maternal education, adolescents who perceived that their parents more strongly disapproved of their having sex during adolescence were less likely to have STIs 6 years later (adjusted odds ratio, 0.89; 95% confidence interval, 0.81-0.99). Those with a higher grade point average during adolescence were also less likely to acquire STIs (adjusted odds ratio, 0.84; 95% confidence interval, 0.71-0.99). Stratified analyses confirmed these findings among female, but not male, adolescents. Feelings of connection to family or school, reported importance of religion, attending a parochial school, and pledges of virginity during adolescence did not predict STI status 6 years later.
Perceived parental disapproval of sexual intercourse and higher grades in school during adolescence have protective influences on the trajectory of risk for acquiring STIs, primarily among female adolescents. Most factors associated with increased duration of virginity in adolescence do not influence the trajectory of STI risk.
Sexually transmitted infections (STIs) cause considerable personal and societal burden in the United States.1 Strategies to reduce this burden must target adolescents and young adults.2 A recent population-based study3 found that 4.7% of young women and 3.7% of young men in the United States between the ages of 18 and 26 years are infected with Chlamydia trachomatis, and that the prevalence of infection is substantially higher among certain racial and ethnic minority groups.
Biological factors, sexual networks, partner characteristics, health care, and social context contribute to the risk of STIs.1,4- 8 Sexual behaviors are also key determinants of risk, and virginal adolescents and young adults are at low risk of acquiring STIs. Encouraging increased duration of virginity is one strategy to reduce STI prevalence because the immature cervix is biologically predisposed to becoming infected if exposed,9- 11 and those who delay initiation of sexual intercourse (sexual debut) until they are older often engage in less risky sexual behaviors.12- 15
Wave I of the National Longitudinal Study of Adolescent Health (Add Health) contributed to our understanding of the characteristics of families and schools associated with delaying sexual debut in a large nationally representative adolescent sample.16 By using a conceptual paradigm from the resiliency literature17 and focusing on factors that were potentially modifiable, cross-sectional analyses revealed that adolescents who felt connected to family, lived in families with good mental health, and perceived that their parents more strongly disapproved of them having sex or using contraception were more likely to delay sexual debut than comparison groups.16 School factors associated with delay in sexual debut included feelings of being connected to school and attending a parochial school. Individual factors associated with later sexual debut included adolescents’ feelings about importance of religion, making a pledge of virginity, perceived likelihood of early death, number of hours worked per week, perceived physical maturity compared with peers, and grade point average (GPA).
Wave III of Add Health provides an opportunity to determine whether factors that influence age of sexual debut also influence longitudinal trajectories of risk for STIs—using biological tests. In this study, we test the hypothesis that family, school, and individual factors associated with increased duration of virginity during adolescence at wave I16 longitudinally decrease risk of infection with C trachomatis, Neisseria gonorrhoeae, or Trichomonas vaginalis at wave III.
Add Health is a prospective cohort study following up almost 20 000 adolescents into adulthood.18 The original Add Health sample was identified from 1994 to 1995, using a stratified school-based sample of middle and high school students in the United States. Wave I interviews were conducted in 1995. Wave III occurred from 2001 to 2002, and included private interviews of original participants (N = 14 322) and testing for C trachomatis and N gonorrhoeae by ligase chain reaction and T vaginalis by polymerase chain reaction. The original sampling design, and all Add Health procedures, have been described in detail elsewhere.3,16,19- 21 For this study, we test whether the wave I variables associated with age of sexual debut16 predict STI status at wave III. The University of North Carolina at Chapel Hill Institutional Review Board approved all study procedures.
The outcome variable was a positive test result for C trachomatis, N gonorrhoeae, or T vaginalis on wave III urine specimens.
Demographic Variables. Age was categorized as 18 to 20, 21, 22, 23, or 24 years and older. All participants who reported that they were of Hispanic or Latino origin were coded as Latino. All other participants were coded as white, African American or black, Native American, or Asian American. Family structure during adolescence was measured by self-report at wave I, and coded as living in a 2-parent home (biological or nonbiological) or not. Highest level of maternal education was measured by the mother’s report if she was interviewed at wave I or by the adolescent’s report at wave I, and coded as less than high school, high school graduate, or college or beyond.
Wave I Family Variables. Previously described measures of family connectedness (Cronbach α = .88), perceived parental disapproval of adolescent sex (Cronbach α = .90), and perceived parental disapproval of contraception (Cronbach α = .91) were created from items scored on Likert-type scales.16,22 Family suicide attempts and/or completions were measured by the following question: “Have any family members tried to kill themselves or succeeded in killing themselves over the past 12 months?”
Wave I School Variables. A previously described measure for feelings of being connected to school was created using an 8-item scale (Cronbach α = .71).16,22 Measures for mean daily attendance for the participants’ school and school type (coded as parochial or not) were obtained from wave I school administrator questionnaires.
Wave I Individual Variables. The importance of religion was measured by the following questions: How important is religion to you? (There were 4 response categories, ranging from very to not at all.) How often do you pray? (There were 5 response categories, ranging from at least once a day to never.) What is your religion? Responses to the first 2 items were recoded from 0 to 1 so that lower values represented lower importance of religion, and the mean of the 2 items was calculated as an index of religious importance. Respondents who reported not having a religious affiliation and who did not answer these questions were coded as 0. Respondents who did report a religious affiliation had to answer at least 1 of the first 2 questions to receive a score. The index was then standardized to have a mean of 0 and an SD of 1.Respondents who reported feelings of attraction toward a person of the same sex or a romantic relationship with a person of the same sex were coded as having same-sex attractions or relationships. The perceived risk of an early death was measured by asking participants to indicate the chance that they will live until the age of 35 years. The number of hours engaged in paid work in a typical nonsummer week was coded as at least 20 per week or less. Self-report of physical appearance was used to measure whether participants believed they looked older or younger than most others their age. Grade point average was calculated from self-report of grades in the most recent grading period in English/language arts, mathematics, history/social studies, and science. Virginity pledge was measured by the following item: Have you taken a public or written pledge to remain a virgin until marriage?
Descriptive frequencies were derived from unweighted data. All other analyses used data weighted to reflect a nationally representative sample using survey estimation commands in computer software (Stata)23 that account for the stratification, clustering, and unequal probabilities of selection in the study design.20
We examined bivariate relationships between sociodemographic characteristics, wave I independent variables (family, school, and individual), and having a positive STI test result at wave III. We conducted multivariate analyses to determine the ability of family, school, and individual variables to predict STI status, while controlling for sex, age, race/ethnicity, family structure, and level of maternal education. Because many factors influencing sexual behaviors and risk of STIs may differ by sex, we assessed separate multivariate models for female and male adolescents.
In bivariate analyses, the range of each scaled variable was divided into thirds. In multivariate regression analyses, these variables and age, mean school attendance, and GPA were modeled as continuous variables after confirmation that the assumption of linearity was met. Multivariate models included only observations with no missing values for any included variable. For each multivariate model, correlations between independent variables were calculated to ensure the absence of strong colinearity. The 2 most strongly correlated variables were perceived parental disapproval of sex and contraception (ρ = 0.57 in the full model). All other correlations were below 0.40. Results did not change substantively when perceived parental disapproval of contraception was excluded from final models.
Of the 18 924 Add Health participants in the nationally representative weighted wave I sample, 14 322 (75.7%) were located and agreed to participate in wave III. Of these participants, 1130 (7.9%) refused to provide a urine specimen and 226 (1.6%) were unable to provide a specimen at the interview. Furthermore, 421 (2.9%) of the specimens could not be processed due to shipping or laboratory problems and 951 (6.6%) did not have results for all 3 STI tests.21 In sum, test results were available for C trachomatis, N gonorrhoeae, and T vaginalis for 11 594 (81.0%) of the wave III participants.
Including participants who did and did not undergo urine testing, just more than half (52.8%) of the study sample was female). Most participants (54.2%) were white, with substantial representation of African Americans (21.3%), Latinos (16.4%), Asian Americans (7.2%), and Native Americans (1.0%) (percentages do not total 100 because of rounding). Participants’ mean age was 22.0 years (SD, 1.8 years).
Overall, 6.2% (95% confidence interval, 5.3%-7.2%) of wave III respondents had a positive test result for C trachomatis, N gonorrhoeae, or T vaginalis.
The prevalence of a positive test result for C trachomatis, N gonorrhoeae, or T vaginalis was slightly higher among young adult females than males later (Table 1). The prevalence was highest among African Americans and Native Americans, intermediate among Latinos, and lowest among Asian/Pacific Islanders and whites. The prevalence was higher among young adults with mothers who did not graduate from high school and those with mothers who had only completed high school when compared with those with mothers who had some college education or a college degree. Young adults who lived in a 2-parent home during adolescence had a lower prevalence of positive test results than those who did not live in 2-parent homes. The prevalence of positive test results did not significantly vary by age.
In bivariate analyses, adolescents’ perceptions of their parents’ approval of sex predicted their likelihood of having STIs 6 years later (Table 2). Of the participants who perceived that their parents strongly disapproved of sex during adolescence, only 5.5% later tested positive compared with 8.0% and 8.9% of adolescents who perceived that their parents had moderate and low disapproval of sex, respectively. Findings related to perceived parental disapproval of contraception were similar.
In general, adolescents who attended schools characterized as having higher mean daily attendance were more likely to acquire STIs than comparison groups. Adolescents who attended parochial schools were less likely to acquire STIs than those who attended nonparochial schools. School performance during adolescence also predicted STI status 6 years later. Among adolescents with a GPA between 3.25 and 4.0, 4.2% later tested positive for STIs. In comparison, 7.2% of adolescents with a GPA between 2.25 and 3.0 and 7.8% of adolescents with a GPA between 1.0 and 2.0 acquired STIs. Adolescents who worked 20 hours per week or more were at lower risk of STIs 6 years later than those who did not, and adolescents who perceived looking younger than most peers were at higher risk of STIs than those who did not.
Adjusting for biological sex, age, race/ethnicity, family structure, and maternal education, adolescents who perceived that their parents more strongly disapproved of their having sex during adolescence were less likely to have C trachomatis, N gonorrhoeae, or T vaginalis 6 years later (adjusted odds ratio [95% confidence interval], 0.89 [0.81-0.99]) (Table 2). Adolescents with a higher GPA during adolescence were less likely to have these STIs 6 years later (adjusted odds ratio [95% confidence interval], 0.84 [0.71-0.99]). Other wave I family, school, and individual variables were not associated with wave III STI status in multivariate analyses.
In stratified multivariate analyses, female adolescents who perceived that their parents were more disapproving of their having sex during adolescence had a lower risk of STIs 6 years later (Table 3). Female adolescents with a higher GPA had a reduced likelihood of later STIs. Among male adolescents, family, school, and individual factors during adolescence did not predict STI status 6 years later.
We confirm that STIs are common among young adults in the United States, and that prevalence varies substantially by nonmodifiable factors such as race and ethnicity. This study focuses on determining whether modifiable family, school, and individual factors associated with increased duration of virginity during adolescence also protect against having STIs 6 years later.
Adolescents who perceive that their parents more strongly disapprove of their having sexual intercourse as an adolescent are less likely to be infected with C trachomatis, N gonorrhoeae, or T vaginalis 6 years later. To our knowledge, this is the first report of a longitudinal link between perceptions of parental opinions about sex during adolescence and biologically measured STIs. This finding is consistent with previous research showing that the more disapproving adolescents perceive their mothers to be toward their engagement in sexual intercourse, the less likely adolescents are to become pregnant24 and to self-report STI diagnoses.25
From a practical perspective, if parents who disapprove of their adolescent children engaging in sexual intercourse convey this attitude effectively, then the long-term risk of STIs may be reduced. Unfortunately, effective conveyance of parental attitudes about sexual intercourse seems to be challenging. Adolescents often have inaccurate perceptions of their parents’ attitudes about their engaging in sexual intercourse, and the tendency is for adolescents to underestimate parental disapproval.24,26,27 Furthermore, the correspondence between parent and child reports of whether they have discussed sexual topics is often low, and many parents underestimate their adolescent children’s engagement in sexual activity.26,28,29 Taken together, this suggests a need to identify strategies to increase the congruence between adolescents’ perceptions of parental attitudes about sexual intercourse during adolescence and actual parental attitudes. Strategies will need to take into account the complex process of conveying parental attitudes to children, which likely involves nonverbal and verbal communication.
We found that adolescents with higher grades in school are less likely to have acquired STIs 6 years later than those with lower grades. This finding is consistent with previous research showing that adolescents with higher grades are less likely to participate in high-risk behaviors across many domains,16 and that adolescents with higher intelligence are less likely to engage in risk-associated sexual behaviors.30 We cannot exclude bidirectional relationships between academic achievement and long-term risk of STIs. Others31 have reported bidirectional influences between academic achievement and age of sexual debut, with variation in these relationships among adolescent boys and girls.
Our results show that the longitudinal risk of STIs is influenced by perceptions of parental disapproval of sex for adolescent girls, but not adolescent boys. The link between parental disapproval of sex and lower likelihood of STIs in late adolescence likely works through influences on sexual behaviors (eg, fewer partners, fewer incidents of sexual activity, or more consistent condom use) or some aspect of partner choice (eg, monogamous relationship, marriage, or “safer” partners). There is literature32 documenting sex-differentiated socialization by parents, with adolescent boys being more likely to be encouraged to be independent and adolescent girls more likely to be encouraged to be obedient. Borawski et al33 also found that adolescent girls report less negotiated unsupervised time than adolescent boys. The link between perceived parental disapproval of sex and lower longitudinal risk of STIs among adolescent girls may be, at least in part, the result of internalization of more conventional standards or a stronger desire among adolescent girls to behave in ways that will not result in parental disapproval.
We also found sex differences in the protective relationship between higher GPA and lower longitudinal STI risk. Previous research30 has shown that adolescents with higher scores on a standardized measure for vocabulary are less likely to engage in vaginal intercourse and other noncoital sexual behaviors than those with lower scores, and that this relationship is stronger among adolescent girls than adolescent boys. Our findings are generally consistent with this observation. However, GPA is influenced by many factors other than verbal ability, and it is possible that higher grades, especially for adolescent girls, may be another manifestation of internalization of conventional standards and a desire to avoid parental disapproval—which in the realm of sexual development may be linked to less risky sexual behaviors and partner choices. Further research will clearly be needed to better understand predictors of longitudinal risk of STIs for adolescent girls and boys, and how and why these predictors may differ by sex.
Overall, our negative findings are perhaps the most interesting. Family, school, and individual factors linked to delayed sexual debut among adolescents16 in large part did not predict STI status 6 years later. Perhaps the correlation between early sexual debut and risky sexual behaviors (and, therefore, risk of STIs)12- 15 weakens as adolescents approach adulthood. Furthermore, it may be unrealistic to expect broad protective models based on the adolescent resiliency literature to longitudinally predict complex health outcomes. Future research will be needed to better understand our negative findings.
There were several limitations of this study. First, we did not investigate the process by which family, school, and individual factors influence risk of later STIs, including the potential role of sexual behaviors. Second, the original Add Health sample was school based, so dropouts were excluded. However, Add Health has followed up all originally enrolled participants, including dropouts, and previous research suggests that potential bias because of missing out-of-school youth in Add Health is small.34 Third, 24.3% of the original wave I sample could not be located at wave III, and results of all 3 STI tests were not available for 19.0% of wave III respondents. However, careful assessment has shown that risk of wave III nonresponse bias seems minimal,35 and comparisons of wave III participants who did and did not provide urine specimens for STI testing have shown few differences.36 Fourth, the performances of STI tests on urine specimens are not perfect,21 which influences the accuracy of our main outcome variable. We are reassured by previous analyses quantifying the effect of nonresponse and test performance on prevalence estimates that show the risk of this potential bias seems small.3 Fifth, if wave III positive STI test results represent infections that have persisted since wave I, we would overestimate the influence of our predictive model. However, it is unlikely that these STIs would persist for 6 years because of the propensity of infections to resolve spontaneously37 or be effectively treated with antibiotics,38 which may be given for other reasons. Of more concern, our STI outcomes most likely reflect recent sexual behaviors; no serologic studies were available to test for antibody responses to STIs, which would more closely approximate cumulative risk of infection. Sixth, the reliability and validity of most independent variables have not been evaluated and measures included few items (eg, perceived parental disapproval of sex). Seventh, our independent measures may not have captured important dimensions of family and school context during adolescence that may influence later risk of STIs (eg, familial patterns of health care use), limiting the predictive ability of our model.
In summary, our results show that adolescents’ academic success and perceptions of parental disapproval of sexual intercourse diminish the risk of acquiring STIs in young adulthood, particularly for females. Further research will be needed to understand how these factors diminish STI risk, and to identify other strategies to reduce STI prevalence among adolescents and young adults. Parents should be encouraged to convey clear messages that they do not approve of their children having sexual intercourse during adolescence. Finally, prevention and intervention programs will need to account for the complex phenomena that contribute to health problems faced by adolescents and young adults, including STIs and the human immunodeficiency virus.
Correspondence: Carol A. Ford, MD, Adolescent Medicine Program, Campus Box 7220, The University of North Carolina, Chapel Hill, NC 27599-7220 (firstname.lastname@example.org).
Accepted for Publication: February 14, 2005.
Funding/Support: This study was supported in part by grant UO131496 from the National Institute of Allergy and Infectious Diseases, Bethesda, Md (STD Cooperative Research Center at The University of North Carolina at Chapel Hill); grant HD38210 from the National Institutes of Health, Bethesda; and The Robert Wood Johnson Foundation Generalist Physician Faculty Scholars Program. Mr Pence is a Howard Hughes Medical Institute Predoctoral Fellow.
Additional Information: This research uses data from Add Health, a program project designed by J. Richard Udry, PhD, Peter S. Bearman, PhD, and Kathleen Mullan Harris, PhD, and funded by grant P01-HD31921 from the National Institute of Child Health & Human Development, Bethesda, with cooperative funding from 17 other agencies. Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W Franklin St, Chapel Hill, NC 27516-2524 (http://www.cpc.unc.edu/addhealth).
Acknowledgment: We thank Carolyn Tucker Halpern, PhD, for her contributions to our interpretation of sex differences; and Ronald R. Rindfuss, PhD, and Barbara Entwisle, PhD, for their assistance in the original Add Health design.
Ford CA, Pence BW, Miller WC, Resnick MD, Bearinger LH, Pettingell S, Cohen M. Predicting Adolescents’ Longitudinal Risk for Sexually Transmitted InfectionResults From the National Longitudinal Study of Adolescent Health. Arch Pediatr Adolesc Med. 2005;159(7):657-664. doi:10.1001/archpedi.159.7.657