The Youth Risk Behavior Surveillance System (YRBSS), administered by the Centers for Disease Control and Prevention (CDC), monitors risky behaviors among children in the United States. State and local Youth Risk Behavior Survey (YRBS) data have been used to support policy development related to HIV and sexuality, physical fitness, and underage drinking in Florida, Texas, Mississippi, and Wyoming (1). YRBS data are reported nationally and for 42 states and 20 metropolitan areas.

Many of the leading causes of death for youth are behavioral. For example, in the United States, most deaths among youth and young adults result from motor-vehicle injuries, other unintentional injuries, homicide, and suicide (1). The leading causes of premature death for adults are also behavioral (tobacco use, lack of physical activity, and dietary behaviors); behavioral habits and preferences frequently are established in childhood. Other major health issues such as sexually transmitted diseases and immunization rates also have behavioral patterns frequently established in youth.

YRBS data have been used for policy analysis by comparing possible relationships between policy change and health-related behaviors. For example, secondary analysis of YRBS data was used to demonstrate that tobacco tax policies were associated with reduced tobacco use among youth (2). Analysis of surveillance data also has been used to assess the effect of mandatory seat belt laws on seat belt use, motor-vehicle fatalities, and crash-related injuries among youths (3); stricter enforcement of laws forbidding tobacco sales to minors (4); tobacco prices and taxes on youth tobacco use (5,6); and relationship of population density to youth alcohol use (7).

Behavioral surveillance has also been used to understand factors associated with unhealthy conditions and high-risk behaviors. For example, YRBS data for 1 condition or risk behavior is linked with data for other risk behaviors. Violence, substance abuse, sexual behaviors, obesity, suicidal behaviors, environmental exposure to tobacco, asthma, and insufficient sleep are some of the conditions or risk factors for which relationships have been studied using YRBS data (8-20).

We describe a way to analyze YRBS data at the subcounty level and explore the usefulness of this approach and the implications for community-level decision making.

YRBS comprises a core set of standardized questions that have been tested for validity and reliability and are administered nationally (21). In addition to the core questions, customized questions can be included in the survey. The Duval County Public Schools, Duval County Health Department, YRBS Survey Advisory Team, and Jacksonville community stakeholders revised the YRBS instrument to fit the needs of the local community. In addition to questions to address specific topics, questions were added to facilitate subcounty analysis.

CDC procedures were followed for selecting the sample and for administering the survey (21). All 27 middle schools and 19 high schools in Duval County were included in the 2009 YRBS survey; classes were randomly selected.

Because zip code areas are statistically unreliable for analysis of many types of data, the Duval County Health Department had previously created 6 multiple–zip code health zones to provide reliable and consistent data for subcounty analysis. This spatial analytic technique is consistent with those used to assess behavioral, morbidity, mortality, and health disparity surveillance data in Duval County (22). The health zones have different demographic characteristics; 1 area is more than 80% African American and 2 others are less than 20% African American. Health zones also have major differences in social determinants, morbidity, and mortality. For example, health zone 1 has higher poverty rates, lower high school graduation rates, higher infant mortality, and higher diabetes mortality. Public and private agencies use the health zones for health and disease surveillance in Duval County.

Confidence intervals are the primary tool that CDC uses for comparing differences in YRBS responses (1,21), although the t test is also used to increase precision for assessing significance for bivariate comparisons (23). The responses were weighted according to CDC procedures (21,24) to increase precision for comparing frequencies of Duval County to Florida and among the 6 health zones. The z test for proportions (25) was also used to increase precision of significance testing for the comparison of Duval County to Florida and to confirm statistical significance for bivariate differences among the health zones for which differences were identified with confidence intervals. Significance was set at P < .05. (Information on statistical significance and weighting can be obtained from the corresponding author.)

A total of 2,542 high school students from 19 public schools and 3,138 middle school students from 27 public schools participated in this survey. The proportionate responses ranged from 6.2% in health zone 6 to 26.8% in health zone 2 (Figure 1).

This was one of the first attempts to use YRBS to conduct subcounty analysis. We observed few differences among the subcounty health zones of Duval County, despite the differing demographic characteristics of these zones. Duval County's youth had a higher prevalence of risky behaviors than the rest of Florida's youth. Both similarities and differences at the subcounty level have policy implications; in some cases, districtwide policies are needed, and in other cases, more specific policies and resource allocation are needed for specific parts of the county. When a risk is significant throughout the county, identifying, monitoring, and communicating its severity is the first step in creating the community awareness that is necessary to mobilize community resources to address it.

However, some risk behaviors were more prevalent in some parts of the county. For example, the percentage of middle school students who reported fighting and carrying a weapon was highest in health zone 1, the urban core. Middle school students in the urban core also reported less physical activity, more sedentary behaviors, and poorer nutrition behaviors. Disparities between middle school students living in the urban core and those in the other health zones may be linked to other health disparities in health zone 1, such as high rates of diabetes deaths, hospitalization, and emergency department use (25). The lack of opportunity for physical education in conjunction with the lack of physical activity in the urban core has implications for state and local education policies, which have tended to reduce physical activity. On the other hand, students in other zones were more likely to engage in tobacco or alcohol use. Identifying and monitoring where these problems are most prevalent can allow policy makers to direct resources appropriately. For example, increased physical education and opportunities for physical activity may be a higher priority for parts of the community where disparities exist. In particular, the review and revision of policies that decrease opportunities for physical activity as punishment or increase time for remedial education should be considered.

Our analysis of YRBS data in the middle school population showed the importance of middle school behavioral surveillance. Florida does not collect data on sexual activity for middle school students, and our county data showed that middle school students were engaging in high-risk sexual behaviors. Another surprising result from YRBS was the high percentage of Duval County middle school students who used tobacco. It is unclear whether this finding is an anomaly or represents a reversal of the accomplishment in reducing youth tobacco use a decade earlier. This high rate of tobacco use needs to be monitored at the middle school level to determine whether the prevalence of tobacco use increases as these middle school students advance to high school; such a trend may support reinstituting tobacco control policies that were weakened early in the decade.

Middle school YRBS is conducted much less frequently across the country than high school YRBS. Tobacco use also illustrates how middle school YRBS can be a valuable tool to assess the effect of policy. Changes in tobacco policies and practices related to youth tobacco use appeared to reduce youth tobacco use a decade ago. Although many of those policies and the funding for those practices have been discontinued, youth tobacco use has remained consistent. Davis et al (26) found that cuts in the Florida Tobacco Control Program budget resulted in significant declines in exposure to some interventions, particularly tobacco countermarketing. Monitoring the effect of these reductions in exposure to tobacco countermarketing efforts may be necessary to determine whether there is a residual effect of the tobacco control program, and the residual effect is wearing off.

Physical activity is another risk behavior that may be influenced by policies. In Florida, students whose performance on state assessments in reading or math is low are required to enroll in a remedial course, which may waive their physical education requirement. This policy may have a greater effect on racial/ethnic minority students in Duval County, who tend to lag in academic progress and have much lower graduation rates (27). The lower rates of regular physical activity and high rates of television viewing and video gaming in the high-minority urban core (health zone 1) indicate that these students are at highest risk of the detrimental effects of low physical activity, and these are the students most likely to be deprived of organized physical activity during the school day. This finding has policy implications related to school support for physical activity, given the high rates of death, hospitalization, and emergency department use for health conditions linked to sedentary behaviors in the same area (health zone 1) of the city (22). Where behavioral risk is documented to be more persistent, resources and policies such as school policies that affect physical activity, can be adjusted to address those problems. Analysis at the subcounty level may be useful.

Previously identified limitations of YRBS apply to this study; all YRBS data are self-reported, students who were absent during survey administration were unaccounted for, and the extent of underreporting or overreporting of behaviors cannot be determined, although YRBSS data are consistently of acceptable quality (21). Limitations specific to this study include the smaller sample sizes for 2 of the health zones, which reduced the reliability of some results of the subcounty analysis. These lower sample sizes also decreased the number of health zone comparisons that could have been significant.

This study documented the feasibility of customizing the YRBS to enable analysis of behavioral risk factors at the subcounty level. Such an analysis can inform policy decisions and resource allocation on the basis of which high-risk behaviors are common to subcounty health zones and which are concentrated in specific zones. Many of these behaviors can be addressed through school- or community-based interventions.