Since 1991, the national YRBS has been conducted biennially as part of the Youth Risk Behavior Surveillance System developed by CDC. These surveys collect cross-sectional data from independently selected nationally representative samples of public and private school students in grades 9 to 12. Each national YRBS uses a similar 3-stage probability sampling methodology.¹³ Weighting factors are applied to each student’s record to adjust for varying probabilities of selection at each stage of sampling, student nonresponse, and the oversampling of black and Hispanic students. Questionnaires contain approximately 98 items and are administered in the classroom during a regular class period by trained data collectors. Responses are recorded directly on computer-scannable questionnaire booklets or answer sheets. Student participation in the survey is anonymous and voluntary, and local procedures are used to obtain parental permission. The national YRBS has been reviewed and approved by an institutional review board at the CDC. This study analyzes national YRBS data collected from 2003 to 2015. Overall response rates ranged from 60% in 2015 to 71% in 2009 and 2011.

The analytic sample was restricted to currently sexually active students (i.e., students who had sexual intercourse during the past 3 months; n = 33,940), assessed with the question: “During the past 3 months, with how many people did you have sexual intercourse?” Response options were as follows: I have never had sexual intercourse; I have had sexual intercourse, but not during the past 3 months; 1 person; 2 people; 3 people; 4 people; 5 people; and 6 or more people. The outcome measure, condom use during last sexual intercourse, was assessed with the question: “The last time you had sexual intercourse, did you or your partner use a condom?” Response options were as follows: “I have never had sexual intercourse,” “yes,” and “no.” We created a dichotomous variable (yes vs. no) that excluded nonsexually experienced adolescents.

Four sexual risk behaviors were examined as potential moderators of declines in self-reported condom use. All study variables were dichotomized to be consistent with the standard reporting of YRBS data.⁹ Alcohol or drug use before last sexual intercourse was assessed with the question: “Did you drink alcohol or use drugs before you had sexual intercourse the last time?” A dichotomous (yes vs. no) variable was created. Number of lifetime sex partners was assessed with the question: “During your life, with how many people have you had sexual intercourse?” Response options were as follows: I have never had sexual intercourse, 1 person, 2 people, 3 people, 4 people, 5 people, and 6 or more people. Number of recent sex partners was assessed with the same response options; students were asked about number of sex partners during the prior 3 months. We created one variable for 4 or more sex partners during their life (≥4 people vs. <4 people) and another variable for 2 or more recent sex partners (≥2 people vs. 1 people). First sexual intercourse before the age of 13 years was assessed with the question: “How old were you when you had sexual intercourse for the first time?” Response options were as follows: I have never had sexual intercourse, 11 years or younger, 12 years old, 13 years old, 14 years old, 15 years old, 16 years old, and 17 years or older. Age of first sexual intercourse was dichotomized to create an indicator for early sexual initiation (first sexual intercourse before age 13 years vs. ≥13 years).

Demographics included race/ethnicity, sex, and grade. Race/ethnicity was a nominal variable with 4 categories: black non-Hispanic, white non-Hispanic, Hispanic, or other. Sex was assessed with the question: “What is your sex?” with the options “female” or “male.” Grade of enrollment was assessed with the question: “In what grade are you?” with the options “9th grade,” “10th grade,”“11th grade,”“12th grade,” or “ungraded or other grade.”

Because the YRBS continuously used the same recruitment and sampling procedures, we were able to combine multiple year’s data into one analytic data set. Data from 7 surveys, conducted from 2003 to 2015, were analyzed using SUDAAN release 11.0.0 (Research Triangle Institute, Research Triangle Park, NC). All analyses were stratified by sex because of large differences in the prevalence of condom use among female and male students. First, using logistic regression with self-reported condom use at last sex as the dependent variable, we tested racial/ethnic and grade differences by examining interactions between linear contrast for survey year and race/ethnicity or grade. We used a similar approach for testing for variation in linear trends in self-reported condom use by self-reported drinking alcohol or using drugs before last sexual intercourse, having 4 or more sex partners during their life, having 2 or more recent sex partners, and having first sexual intercourse before age 13 years, controlling for race/ethnicity and grade. We controlled for race/ethnicity and grade because of the co-occurrence of racial/ethnic and grade differences between self-reported condom use and other sexual risk behaviors. For significant interactions, we stratified analyses by sexual risk behavior. Prevalence estimates with 95% confidence intervals were calculated using Taylor series linearization. Logistic regression t tests for linear trend β’s and Wald F statistics for interaction terms were considered statistically significant if P < 0.05. Cases with missing data were removed from the analysis (i.e., listwise deletion).

Table 1 contains sample characteristics and prevalence of risk behaviors. The percentage of currently sexually active students ranges from a high of 35.0% in 2007 to 30.1% in 2015. The percentage of black non-Hispanic students ranges from a high of 19.9% in 2007 to a low of 13.9% in 2015. The percentage of Hispanic students ranges from a high of 22.2% in 2015 to a low of 15.7 in 2005.

Overall, significant declines were observed in condom use during last sexual intercourse for sexually active female and male students between 2003 and 2015 (Table 2). Condom use during last sexual intercourse declined from 57.4% in 2003 to 52.0% in 2015 among female students (P = 0.01). Condom use during last sexual intercourse declined from 68.8% in 2003 to 61.5% in 2015 among male students (P = 0.001). However, variability in trends in condom use during last sexual intercourse was observed by race/ethnicity and by grade for both female and male students (Table 2). For female students, significant declines in self-reported condom use occurred only among black students (63.6% in 2003 to 46.7% in 2015; P < 0.01) and students in 9th (66.1% in 2003 to 56.7% in 2015; P < 0.01) and 10th grades (66.4% in 2003 to 54.0% in 2015; P = 0.02). For male students, significant declines in self-reported condom use occurred among white students (69.0% in 2003 to 58.1% in 2015; P < 0.01) and students in 9th (71.2% in 2003 to 63.3% in 2015; P = 0.03) and 12th grades (67.0% in 2003 to 57.4% in 2015; P = 0.01).

Among female students, interaction analyses suggested that trends in self-reported condom use varied significantly depending on other sexual risk behaviors, including drank alcohol or used drugs before last sexual intercourse (Wald F = 6.4, P = 0.01), had 4 or more sex partners during their life (Wald F = 8.2, P < 0.01), had 2 or more sex partners during the past 3 months (Wald F = 4.7, P = 0.03), or had their first sexual intercourse before age 13 years (Wald F = 4.6, P = 0.03; data not shown). Among male students, trends in condom use during last sexual intercourse did not significantly vary by involvement in other sexual risk behaviors.

Table 2 presents condom use trends for female students, stratified by other sexual risk behaviors. Across 3 risk behaviors (drank alcohol or used drugs before last sexual intercourse, 4 or more sex partners during their life, and 2 or more sex partners during the past 3 months), condom use decreased significantly only among female students who engaged in the risk behavior. Condom use decreased significantly for female students regardless of early sexual initiation, but the decreases were more pronounced among those who initiated sexual activity before the age 13 years compared with those who did not initiate sex before age 13 years. These findings are presented graphically in Figure 1.

This study not only extends what is known about recent declines in self-reported condom use among adolescents but also points to the need for future research to explain underlying drivers, variation by sexual risk behavior among female students, and differential effects by sex. Longitudinal data that explore developmental trajectories in condom use could be particularly useful. Qualitative studies to better understand motivations, barriers, and facilitators to using condoms among male and female students from racially diverse groups could also provide insight into declines in self-reported condom use. Such research could help identify strategies for strengthening sexual risk reduction programs, ensuring maximum reach and effectiveness for all adolescents, including those at highest risk.

These results demonstrate that public health and clinical efforts to increase condom use among young people are warranted. Many sexual risk reduction interventions have been found to increase condom use, and these can be implemented in a variety of settings including schools, community-based organizations, and clinics.⁶ Evidence also exists that provider counseling is associated with increased condom use, which suggests that an additional avenue for increasing condom use may be to improve adolescents’ access to quality sexual health and reproductive health services.²⁶ Increasing access to care also provides an opportunity for providers to promote STD prevention strategies for sexually active adolescents in addition to condom use, such as STD/HIV testing and human papillomavirus vaccination.⁷ Given that adolescent receipt of preventive counseling during adolescent well-visits is low, provider training to ensure adherence to existing guidelines may be important.^27,28 Ultimately, a combination of individual- and systems-level approaches will be needed to increase condom use among sexually active adolescents.