Many habits—both healthy and unhealthy—are established during adolescence. For this reason, strong data regarding youth behaviors can be of immense value to health policymakers, as well as schools and parents. The Youth Risk Behavior Surveillance System (YRBSS) seeks to gather precisely this type of data. Yet the survey, conducted by the Centers for Disease Control (CDC), has methodological deficiencies that currently limit the value of its findings.

On August 21, the CDC released its report on the 2019 YRBSS, which surveyed a nationally representative sample of almost 14,000 students in grades 9 through 12. While the study is ambitious in scope, its design leaves something to be desired. Below, I elaborate on issues that should be resolved in future versions of the survey. Namely: (1) metrics should be consistent and comparable across behaviors; (2) questions should be improved to enhance the specificity of data; and (3) behaviors should be treated as entangled, rather than independent.

One obvious flaw of the present survey is the use of long and inconsistent recall periods. For decades we have known that the length of recall used in questionnaires has a powerful impact on the validity of results. Currently, YRBSS uses a range of periods to define whether a student has engaged in a behavior. Sexual activity, for example, is measured over a 3-month period; current marijuana, alcohol, and tobacco use is defined as use happening at least once over a 30-day period; and dietary intake and physical activity rely upon 7-day recall periods. In future analyses, all behaviors should be reported over a 7-day period—both because shorter periods are generally better, and because this will simplify comparisons across behaviors.

To make studies like YRBSS as useful as possible, we also need clear data regarding the frequency with which behaviors are performed; and, on questions of substance use, it is vital to additionally collect data on the quantity or dosage of consumption. In the case of tobacco, for example, there is a big difference between someone who says they smoke one cigarette per month versus someone who smokes 15 cigarettes every day. Technically, however, both have smoked at least one cigarette in the past 30 days.

The health effects of both good and bad behaviors are a function of frequency with which they are practiced.  As such, dose and frequency data can be incredibly telling; and current YRBSS data on alcohol consumption speaks to the importance of clarifying these details. The survey shows that while current alcohol use (defined as use on at least one of the past 30 days) is 54.8%, that number drops drastically when respondents are asked about more frequent use: 36.6% reported drinking on 3-9 out of 30 days and only 3.5% reported drinking 20 out of the past 30 days. These extra data points suggest that most young drinkers are extremely infrequent users. Epidemiological studies comparing their long-term risks to those of more regular users should be provided. This type of clarity is also needed for marijuana users, where the gradient between frequent and infrequent use is less dramatic.

Future iterations of the YRBSS should also distinguish between sporadic and regular use of electronic vapors, cigarettes, cigars, and smokeless tobacco. Specifically, it should include data on how many tobacco products are used per day and week. Further, the present survey lacks vital acknowledgement of the large difference in epidemiological risks that exist between tobacco harm reduction products (e.g., electronic vapors, some smokeless tobacco products) and toxic tobacco products (e.g., combustible cigarettes, cigars). Failure to differentiate these products may confuse readers (and policy makers) hoping to develop strategies for reducing long term health risks.

In addition to improving the measurement of individual behaviors, future versions of the YRBSS must go further to identify links between these habits. The current data shows that many unhealthy behaviors are common in high school students. These include: marijuana, alcohol and tobacco use; physical inactivity; and low consumption of healthy foods. We need to move away from addressing each behavior as if it occurs independent from the others. This siloed approach has for too long been the hallmark of chronic disease risk factor interventions, despite our knowledge that unhealthy behaviors cluster and health behaviors can be motivated by a few “triggers.”

We need additional analyses to understand the ways in which unhealthy and healthy behaviors cluster in students, and among specific race and gender groups. The next iteration of the YRBSS should thus seek to answer questions that address these issues, such as: What percentage of the marijuana users are also alcohol and cigar users? Are students with the highest reports of physical activity less likely to use opioids, alcohol, and tobacco? Do these behaviors cluster differently in boys and girls?

The goal of public health policy is to strengthen healthy behaviors and ensure that “risky” behavior is addressed well before it manifests as premature disease and death. The first step toward improving public health is to ascertain a clear picture of current practices. The present YRBSS takes strides in this direction but misses the mark. To obtain an accurate understanding of youth behavior, we must ask the right questions. Only then can we hope to develop science-based policies that effectively reduce risky behaviors and the consequences they yield.