Both the FDA and the CDC have recently taken steps to address an epidemic of opioid overdose and addiction, which is now killing some 29,000 Americans each year. But these regulatory efforts will fail unless we acknowledge that the problem is actually driven by illicit—not medical—drug use.

You’ve probably read that 80 percent of heroin users started with prescription medications—and you may have seen billboards that compare giving pain medication to children to giving them heroin. You have probably also heard and seen media stories of people with addiction who blame their problem on medical use.

But the simple reality is this: According to the large, annually repeated and representative National Survey on Drug Use and Health, 75 percent of all opioid misuse starts with people using medication that wasn’t prescribed for them—obtained from a friend, family member or dealer. 

And 90 percent of all addictions—no matter what the drug—start in the adolescent and young adult years. Typically, young people who misuse prescription opioids are heavy users of alcohol and other drugs. This type of drug use, not medical treatment with opioids, is by far the greatest risk factor for opioid addiction, according to a study by Richard Miech of the University of Michigan and his colleagues. For this research, the authors analyzed data from the nationally representative Monitoring the Future survey, which includes thousands of students.

While medical use of opioids among students who were strongly opposed to alcohol and other drugs did raise later risk for misuse, the overall risk for this group remained small and their actual misuse occurred less than five times a year. In other words, it wasn’t actually addiction. Given that these teens had generally rejected experimenting with drugs, an increased risk of misuse associated with medical care makes sense since they’d otherwise have no source of exposure.  

But for the majority of students, who weren’t morally opposed to recreational chemicals, medical use made no difference. Here, heavy recreational drug use was what mattered, and that was probably a sign that this group was was at highest risk of addiction in the first place.

In general, new addictions are uncommon among people who take opioids for pain in general. A Cochrane review of opioid prescribing for chronic pain found that less than one percent of those who were well-screened for drug problems developed new addictions during pain care; a less rigorous, but more recent review put the rate of addiction among people taking opioids for chronic pain at 8-12 percent.

Moreover, a study of nearly 136,000 opioid overdose victims treated in the emergency room in 2010, which was published in JAMA Internal Medicine in 2014 found that just 13 percent had a chronic pain condition.

All of this this means that steps to limit prescribing opioids for chronic pain run a great risk of harming pain patients without doing much to stop addiction. The vast majority of people who are prescribed opioids use them responsibly—recent research on roughly one million insurance claims for opioid prescriptions showed that just less than five percent of patients misused the drugs by getting prescriptions for them from multiple doctors. 

If we want to reduce opioid addiction, we have to target the real risk factors for it: child trauma, mental illness and unemployment. Two thirds of people with opioid addictions have had at least one severely traumatic childhood experience, and the greater your exposure to different types of trauma, the higher the risk becomes. We need to help abused, neglected and otherwise traumatized children before they turn to drugs for self-medicatation when they hit their teens.

Further, at least half of people with opioid addictions also have a mental illness or personality disorder. The precursors to these problems are often evident in childhood, too. For example, children who are extremely impulsive are at high risk—but on the opposite end of the scale, so, too are children who are highly cautious and anxious. To reach these kids, we don’t need to label them, but we do need to provide tools that are tailored to their specific issues to prevent them from using drugs to manage those issues.

The final major risk factor for addiction is economic insecurity and poverty, particularly unemployment and the hopelessness, social marginalization and lack of structure that often accompany it. For example, heroin addiction rates among people who make less than $20,000 a year are 3.4 times higher than in people who make over $50,000. To those who study the effects of inequality on health, it is no coincidence that the collapse of the white middle class has been accompanied by a rise in all types of addictions, but especially addiction to opioids.

Many people would prefer it if we could solve addiction problems by busting dealers and cracking down on doctors. The reality, however, is that as long as there is distress and despair, some people are going to seek chemical ways to feel better. Only when we can steer them towards healthier—or at least, less harmful—ways of self-medication, and only when we reach children before they develop this type of desperation, will we be able to reduce addiction and the problems that come with it.