Public health experts have long identified race and ethnicity, family income and family structure--whether teens live with one parent or two--as important risk factors for health-risk behaviors during adolescence, from smoking and drinking to sexual intercourse and violent behavior, and suicide.

But according to the largest survey of adolescents ever conducted in the United States, these factors account for only a small fraction of individual differences in adolescents' health risk behaviors. More important, investigators say, are school performance, how young people spend their free time, friends' behavior and relationships with family.

"If we make predictions about kids based on the color of their skin, their parents' bank account or how many parents they live with, we're going to be right only a tiny percentage of the time," says Robert W. Blum, MD, PhD, a professor of pediatrics at the University of Minnesota and the study's lead investigator.

The new findings, the latest from the congressionally mandated National Longitudinal Study of Adolescent Health, familiarly known as Add Health, indicate that "conventional thinking on this issue may be wrong," remarks U.S. Surgeon General David Satcher, MD, PhD. The work "shows that if we really want to help American teens, we can't put them in neat little boxes based on whether they're black or white, rich or poor, or come from a one- or two-parent family."

Demographic effects

In the December issue of the American Journal of Public Health (AJPH) (Vol. 90, No. 12), Blum and colleagues reported the results of interviews with about 12,000 white, black and Hispanic students in seventh through 12th grades and their parents.

Consistent with earlier research, the researchers observed demographic effects on adolescents' reported behavior. For example:

  • 36 percent of Hispanic and 38 percent of black students in grades nine through 12 reported they carried or had used a weapon or had been involved in a weapon-related incident in the past year, compared with 22 percent of white youth.

  • 26 percent of seventh- and eighth-graders whose annual family income was less than $20,000 reported having had sexual intercourse, compared with 8 percent of young people whose families earned at least $41,000 per year.

  • 33 percent of seventh- and eighth-graders who lived with only one parent reported having smoked a cigarette in the past month, compared with only 16 percent of students in two-parent households.

Wanted: more structure

But demographic forces explained only a sliver of the differences that were observed in adolescents' health-risk behavior. In a monograph accompanying their December AJPH article, Blum's group reported that experiences that cut across demographics were strongly linked to whether or not teen-agers engage in unhealthy behavior. Examples of their preliminary findings include these:

  • Problems with school work were linked to cigarette smoking, alcohol use, involvement in weapon-related violence and, for white and Hispanic students, suicidal thoughts and attempts.

  • Frequently "just hanging out" with friends was associated with cigarette smoking, alcohol use and, for males and minority females, involvement in weapon-related violence.

  • Having close friends who drink or smoke was a risk factor for those behaviors. Friends' drinking was also related to teens' risk of being involved in weapon-related violence and, for females and Hispanic males, to suicide risk.

  • Having a friend who has attempted suicide was strongly associated with young people's own risk for suicide.

  • Positive family relationships were a protective factor against involvement in weapon-related violence, alcohol use among black youth, suicide among females and black males, and having sexual intercourse, among black females.

  • Being in a romantic relationship was related to having sexual intercourse. Among some subsets of teens, the perceived benefits and costs of having sex or getting pregnant and the perceived--but not actual--amount of knowledge about birth control also influenced sexual behavior.

These risk and protective factors explained teens' behavior considerably more than the demographic variables did. For example, individual, peer and family influences accounted for between 24 and 49 percent of individual differences in smoking, depending on race and gender. Similar patterns emerged for alcohol use and sexual behavior. Weapon-related violence and suicidal thoughts and attempts remain most difficult to explain.

Many of the factors his group identified tend to be amenable to change, Blum notes. For example, calling school failure "a 911 emergency" for young people, he urges communities to invest more in structured, supervised after-school and evening activities and educational mentorship for teens.

"Too many kids--rich and poor--are left to their own devices," Blum argues. "Kids need structure to grow and to be healthy."

The monograph, "Protecting teens: beyond race, income and family structure," appears on the Web at www.peds.umn.edu/peds-adol/di.html.