The prevalence of students who skipped school because they felt unsafe there increased among all students in aggregate (Table 1). There was an increase in the prevalence of feeling unsafe at school for both sexes and among most race/ethnicities, including American Indian/Alaska Native, Black, Hispanic/Latino and White students, but no trend among Asian and Native Hawaiian/Pacific Islander students (Table 1S). Many intersectional identities had an increase, including American Indian/Alaska Native females and males (Table 2), Black females and males, Hispanic/Latino females and males, and White females and males. Figure 1 facilitates comparisons between intersectional identities (where n>100) for feeling unsafe at school by displaying three key aspects of the data: first, the significance of linear trends; second, the prevalence estimates in 2015 and 2019 stratified by identity; and third, the ranking of prevalence estimates in 2015 from highest to lowest. In this figure, all identities, aside from Asian females and Asian males, had a trend increase. Hispanic males and Black males had the highest prevalence estimates in 2015 and 2019, respectively. White males had a significant increase, but their prevalence estimate in 2019 was lower than the prevalence estimate for the aggregate in 2015, highlighting the heterogeneity in the magnitude of student experiences.

The prevalence of students who were threatened or injured with a weapon at school increased in aggregate. An increase also occurred among female students and White students. Among intersectional identities, White males were the only group that had an increased trend. Notably, in 2015 and 2019 Black males had the highest prevalence and Asian females had the lowest, and both identities had extreme prevalence estimates that did not overlap with any other identity’s estimates (Figure 1S).

There was no change for getting into a physical fight at school in aggregate, for either sex, or for any race/ethnicity. Among intersectional identities, White males had an increase in this behavior. Black males had the highest prevalence in 2015 and 2019, and White females and Asian females had the lowest estimates (Figure 2S).

Experiencing dating violence decreased in aggregate with. There was no linear change among female or male students. Dating violence increased among American Indian/Alaska Native students and decreased among White students. Among intersectional identities, there was a decrease among White females decreased. The prevalence of experiencing dating violence was highest among White females in 2015 and Hispanic females in 2019. Hispanic females had a non-significant increase in dating violence, while most other identities had a non-significant decrease (Figure 3S).

There was no change in experiences of bullying in aggregate, when stratified by sex or race/ethnicity, or among any intersectional identity. However, White females had a notably higher prevalence of bullying compared to all other identities (Figure 4S).

There was an increase in the prevalence of persistent feelings of sadness or hopelessness in aggregate. There was increase among female and male students, and among Hispanic/Latino, and White students. Among intersectional identities, there was an increase among Black females, Hispanic/Latino males, Multiracial males, and White males. Notably, Hispanic females had the highest prevalence estimates and their prevalence in 2015 was greater than the prevalence estimates of all other identities in 2019. The 2019 prevalence estimates for Hispanic/Latino males, White males, Black males, and Asian males were lower than the aggregate prevalence estimate in 2015 (Figure 5S).

There was no change in the prevalence of students who attempted suicide in aggregate. There was no change among female or male students. There was a decrease in attempted suicide among Multiracial students and an increase among White students. Among intersectional identities, there was a decrease among Hispanic/Latino males and an increase among White males. The prevalence of attempted suicide was highest among Hispanic females in 2015 and Black females in 2019 and lowest among White males in 2015 and Hispanic males in 2019 (Figure 6S). Although most intersectional identities did not have significant trends, the prevalence estimates of attempted suicide were lower in 2019 than 2015 for Hispanic/Latino females, Asian females, and White females, while other identities had higher prevalence estimates in 2019 compared to 2015.

When grouped by sex, the prevalence of feeling unsafe at school and feeling persistently sad or hopeless increased from 2015 to 2019 among all female students; however, these trends did not hold true when stratified by race/ethnicity and the magnitude of these behaviors greatly varied by race/ethnicity (Figure 2; for males, see Figure 7S). When grouped by race/ethnicity, Black students had an increased linear trend in feeling unsafe at school and feeling persistently sad or hopeless; however, when stratified by sex, only the trend for skipping school was significant for Black females and Black males (Figure 3; for other race/ethnicities, see Figures 8S-10S). Black females and Black males had the same directional changes for all behaviors 2015–2019 (although most were non-significant), but which sex held the highest prevalence estimate varied by behavior.

In aggregate, the most common experience was feeling persistently sad or hopeless, and the least common experiences were skipping school due to feeling unsafe and being threatened or injured with a weapon at school. In aggregate, one behavior decreased (experiencing dating violence) and three increased (feeling unsafe at school, threatened or injured with a weapon at school, and feeling persistently sad or hopeless) 2015–2019. No individual intersectional identity experienced all four significant trends.

We demonstrated that prevalence estimates and trends of the overall population or based on a single characteristic of sex or race/ethnicity may not offer the validity of an intersectional analytic approach, which recognizes that adolescents simultaneously hold multiple characteristics that constitute an identity (11, 12). Although skipping school due to feeling unsafe there increased in 2019 from 2015 in aggregate and for almost all identities, the heterogeneity of the magnitude of these experiences for each identity was muted. The 2019 prevalence was 8.5% among all students, but this behavior ranged from 5.2% among White males to 12.2% among Black males. Estimates based on the overall population or single demographic characteristics rather than intersectional identities do not describe the experiences of individuals with identities that have been oppressed by multiple and overlapping power structures (10). For example, three behaviors had a significant linear trend among all females, but each individual female identity experienced unique trends when stratified by race/ethnicity. The aggregate results indicated that one risk behavior decreased over time, but in fact, only White females decreased. It may not be revelatory that the best fitting line is one that averages all student experiences, but one best fitting line in aggregate, or stratified by sex or race/ethnicity does not tell a complete story of student experiences.

Public Health researchers and practitioners need to be thoughtful when assessing and interpreting data using an intersectional approach. The extant literature is rich with guidance on how to utilize intersectionality with fidelity but also flexibility for quantitative analysis (10, 11, 13, 14, 28–30). Because intersectionality brings a new perspective to analyzing surveillance data, it may include new modes of analysis that focus on summarizing thematic findings highlighting oppression rather than focusing on a singular risk behavior (29). For example, White males had the most changes out of any identity with increases in five risk behaviors. Although White males ranked among the lowest prevalence estimates for the behaviors where they increased, it is concerning that five out of seven risk behaviors have changed for the worse. Another perspective to consider is that we examined seven behaviors with three timepoints, yielding 21 total data points. Black males had the greatest number of the highest prevalence estimates for seven of these data points, spanning three risk behaviors. Hispanic females and Black females had five and four of the highest prevalence estimates, respectively, and each identity spanned three risk behaviors.

While surveillance data often does not allow us to identify the mechanism for health behavior, racism is an important factor to consider. Racism, in various forms, and as it intersects with sexism and other social determinants of health (31) may explain why Black students (females and males) and Hispanic females held the highest prevalence estimates across the most risk behaviors and did not experience any improvements in a five-year period. Yet, the intersectional framework illustrates that each identity has a unique experience. Hispanic males were unlike Black females and males and Hispanic females because they did not hold multiple highest prevalence estimates.

Surveillance data are used for action to intervene and promote health, but when looking at data for students overall, or piecemeal by demographic characteristics, we do not create evidence that reflects the experiences of students, but rather abstract categories. If we analyze surveillance data using an intersectional perspective, we can create policies and programs to meet all student needs because they recognize students for the whole of their identities.

School-based interventions that center students who experience oppression but are implemented among all students have universal benefit in reducing risk behaviors and promoting health. An intervention to create a safer and more supportive school environment for LGBTQ students found that the presence of a Gay/Straight alliance significantly reduced illicit drug use and reports of being threatened or injured with a weapon at school among lesbian, gay, bisexual, and heterosexual students (32). Presence of a Gay/Straight alliance conferred additional significant reductions in risk behavior exclusively among heterosexual students, including reports of feeling unsafe at school, forced sex, physical dating violence, making a suicide plan, attempted suicide, and suicide attempts that required medical treatment. A trauma-informed intervention to create a safer and more supportive school environment by reducing racial/ethnic, social class, and language-based disparities in student achievement found improvements in attendance and ability to learn, as well as reductions in disciplinary incidents and out-of-school suspensions among all students (33). Applying intersectionality to promote adolescent health and health equity dictates that we produce data that recognizes intersectional identities, which enables creating policies and programs to dismantle the systems of oppression that are perpetuated at the community, school, peer, and individual levels to improve health among all adolescents.

An overarching goal of public health surveillance is using data to spur action to prevent adverse health outcomes and promote well-being. Using intersectionality to display prevalence estimates and trends as a complement to traditional surveillance approaches is necessary to achieve the Healthy People 2030 objectives of eliminating health disparities and reducing adolescent experiences of violence, poor mental health, and suicidal risk behavior.