Teenagers often mention risky behaviours on social networking websites but a single warning email from a doctor may decrease those references, say U.S. doctors who used MySpace to conduct their research.

Members of social networking sites such as MySpace create personal web profiles that can include photos, text and audio. It's estimated that just under half of American adolescents use social networking sites.

But there are risks associated with the public display of information related to sex, substance use and violence, says Dr. Megan Moreno, formerly of the University of Washington and Seattle Children's Hospital Research Institute, and now of the University of Wisconsin-Madison.

It could attract unwanted attention from sexual predators or jeopardize future employment prospects as more and more potential employers seek online information about job candidates, says Moreno, lead author of a report in the January issue of Archives of Pediatrics & Adolescent Medicine.

She and her team examined the content of 500 publicly available profiles between July and September 2007. All of the profile owners reported being 18 years old and living in the United States.

Just over half the profiles made reference to risky behaviours, including about a quarter that mentioned sex, 41 per cent that mentioned substance use, 37 per cent that referred to alcohol use, and nearly 15 per cent that mentioned violence.

Moreno and her colleagues did not explore whether the mention of risky behaviour meant the teens were actually engaging in the behaviour. But the researchers suggest social networking sites may provide a new way to detect teens at risk.

In a second study, also published in Archives of Pediatrics & Adolescent Medicine, the team experimented with online intervention. They set up a MySpace profile with the user name "Dr. Meg," containing information about professional credentials and research interests.

They then identified 190 MySpace profiles of 18- to 20-year-olds that contained three or more references to sex or substance use, including at least one reference each to alcohol and tobacco use.

Half were randomly selected to receive a single email from the Dr. Meg profile, warning them that their profile contained risky information. The email also provided clinical resources, including a link to a website with information about sexually transmitted diseases.

Before the email was sent, 54.2 per cent of the profiles referred to sex and 85.3 per cent mentioned substance use.

Three months after the email was sent, references to sex decreased to zero on 13.7 per cent of profiles in the group that received the email. Such references decreased to zero in 5.3 per cent of profiles that did not receive the email.

References to substance use disappeared on 26 per cent of the intervention profiles versus 22 per cent of the controls. The difference is not considered statistically significant.