Therapeutics
Delivery Technology Paves Way for RNAi Therapies

Neuroscience
Gene Clue to Brain Asymmetry Revealed on Right Side

Social Medicine
Gun Violence May Be Viewed as Contagious

Neurology
Fetal-cell Transplants Reverse Parkinson’s in Two Patients

Clinical Research
Discord Found in Clinical-trial Contracts

Health Care Policy
National Mental Health Survey Shows Mixed Results on Progress

Genetics
Disease Mutation Tracked Down, Ending ‘Curse’ for Colombian Families

New Books
The Spring Bookshelf

Education
HMS Teaching Awards Presented for 2005

Accolades
Students Laud Gardner as Champion of Humanism in Medicine

Medical Ethics
Debate at HMS Frames Ethics of Online Organ Donation

Brain Chemical Serotonin Linked to Left–Right Patterning of Embryo

Rising Leaders in Minority Health Research Turn Data into New Directions

New Chairs Honor Federman and Egan Family

Honors and Advances

In Memoriam

Let Consumers Drive Progress in Health Care Quality

Front Page

Gun Violence May Be Viewed as Contagious

In a study designed to isolate the root causes of violent behavior, HMS researchers found that young teens who witnessed gun violence were more than twice as likely as nonwitnesses to commit violent crime themselves in the following two years. The research appears in the May 27 issue of Science.

Photo by Tai Viinikka

Based on his findings that personal contact with violence doubled the chances of committing violence, Felton “Tony” Earls said the best model for violence appears to be that of a socially infectious disease.

“Based on this study’s results showing the importance of personal contact with violence, the best model for violence may be that of a socially infectious disease,” said Felton “Tony” Earls, HMS professor of social medicine at Harvard University, HSPH professor of human behavior and development, and principal investigator of the study and of the Project on Human Development in Chicago Neighborhoods. “Preventing one violent crime may prevent a downstream cascade of ‘infections.’ And the lessons learned in Chicago should be broadly applicable. Generalizing this to any large city should be valid.”

The investigation, a five-year project that included interviews of more than 1,500 children and teenagers from 78 Chicago neighborhoods, used statistical advances and extremely detailed information about the study subjects to go beyond correlations and associations and estimate causation. “We have a broad range of factors and a long course of study, so we can tease out the causal mechanisms,” said first author Jeffrey Bingenheimer, currently a doctoral candidate at the University of Michigan who will be joining HSPH in September as a Robert Wood Johnson Health and Society Scholar.

Uncovering Causes
Previous work has shown that a large network of factors pushes or pulls young people away from or into violent crime. Researchers suspected that exposure to violence in the community played a role, but many argued that a common factor, perhaps in family structure or personality, might be the common cause of both exposure to violence and later acts of violence. Demonstrating cause and effect with a controlled experiment, deliberately exposing some children to mayhem, would be ethically impossible. But by grouping together and comparing teens with similar likelihood of exposure, some of whom were and some of whom were not actually witnesses to violence, the researchers were able to isolate the independent contribution made by seeing gun violence. And it turned out to be large, swamping previously identified influences like poverty, drug use, or being raised by a single parent.

In the 1980s, this technique, known as propensity stratification, was developed in part by Donald Rubin, now the John L. Loeb professor of statistics at Harvard. This tool has recently enabled clinicians, economists, and social scientists to design studies that isolate causal factors, despite the impracticality or ethical problems inherent in performing traditional blinded, controlled experiments in these areas. Within each propensity group, some subjects are exposed and fall into the treatment group; by chance others are not, and form the control. Experimenters must carefully measure and compensate for other factors that influence this lottery.

The researchers studied the subject teens at three points in their adolescence. Initially they and their caregivers were intensively interviewed and data was collected about their families, personalities, neighborhoods, school performance, and many other factors; this allowed the researchers to group the teens by their propensity to witness gun violence. Two years later, the subjects were interviewed to see which of them had actually seen someone being shot or shot at. Finally, almost three years later, they were interviewed again to determine who had participated in gang violence or other violent acts.

Social Contagion
The team found that witnessing violence more than doubled the risk that teens would commit a violent offense. Nine percent of those not exposed went on to commit serious violent acts, while about 26 percent of the exposed group acted violently, suggesting a raw odds ratio of about 3.55. Even adjusting for race, socioeconomic status, delinquency, and other known factors gave an adjusted odds ratio just below 2.5, so the exposure effect was still strong. The researchers then looked at their statistics to check whether an unrecognized factor could be confounding the results. Under the critical eye of senior statistician and co-author Robert Brennan, an HMS research associate in social medicine, they found that a powerful force would have to be at work to change the findings substantially, and this hypothetical influence would have to be uncorrelated with the factors they did examine. “And honestly, it’s very difficult to think what we might have left out,” Earls said, pointing to the 153 variables that were included in the study.

To address violence effectively, the challenge for social medicine researchers is to define its fundamental nature—is it a product of families, akin to a hereditary disorder? Or is it like an environmental contaminant, lurking in some communities and leaving others unscathed? Based on this study’s results, showing the importance of personal contact with violence, Earls believes the best model may be that of a socially contagious disease.

This study was part of the Project on Human Development in Chicago Neighborhoods, a major interdisciplinary study aimed at deepening society’s understanding of the causes and pathways of juvenile delinquency, adult crime, substance abuse, and violence. Earls says that there are many more lessons yet to be learned, waiting in the mountains of data gathered in Chicago.

—Tai Viinikka

top