When Prediction Is Not Prevention
Putting the focus on the population, not the individual.
Last month we noted that it took two celebrity suicides and a seemingly shocking CDC report demonstrating increasing suicide rates in the US to get professionals, the media, and the public to finally focus their attention on suicide. Suddenly, medical journals, newspapers, and social media are abuzz with the news that we have an “epidemic of suicide” on our hands. A smoothly-produced CNN program hosted by Anderson Cooper, whose brother died by suicide, may be one of the best places to see how the media is dealing with the news that suicide is a terrible problem in the US.
As usual, whenever we “discover” something that has always been a serious problem, there is a tendency to gloss over important facts in ways that can bedevil our desire to make the situation better. In this case, we are concerned that an emphasis in popular media on predicting which individuals are prone to suicide may be misplaced. As the public health community advocates, suicide prevention on a population level rather than prediction on an individual basis, should be our primary goal. If this focus is not adopted, we predict that health professionals will become even more reluctant to treat people who are at risk for suicide than they already are.
Everyone seems to be citing the statistic that suicide is the tenth leading cause of death in the US, hoping that this emphasizes its importance. But let’s look a little more carefully at that statistic. The top ten causes of death in 2014 were:
- Heart disease, 23.4 percent of all deaths
- Cancer, 22.5 percent
- Chronic lower respiratory disease (e.g. bronchitis, emphysema, asthma), 5.6 percent
- Accidents, 5.2 percent
- Stroke, 5.1 percent
- Alzheimer’s disease, 3.6 percent
- Diabetes, 2.9 percent
- Influenza and pneumonia, 2.1 percent
- Kidney disease, 1.8 percent
- Suicide, 1.6 percent
In 2014, 614,348 people died from the number one cause of death, heart disease, whereas 42,773 died by suicide. Now, we totally agree that every life is precious, and 42,773 is a big number. Our point here is not that we should consider deaths by suicide as negligible, but rather to recognize that on epidemiological grounds such deaths are uncommon, representing only 1.6 percent of the deaths in the US in 2014. Statisticians traditionally consider an event that occurs less than 5 percent of the time to be a rare event. By that criterion, suicide, though tragic, is a rare event.
This is important because everyone seems suddenly to be talking about suicide prevention. On a broad, population level, a public health approach to suicide prevention makes sense. But chasing the goal of predicting and preventing every individual case of possible suicide at the point of care is all the more difficult, given that suicide is a rare event in the first place.
Two assumptions are commonly made about suicide: one, that we can predict who is at risk and two that with that knowledge we can prevent individuals from attempting suicide with treatment. With respect to prediction, it is true that we know some factors that increase the risk that someone will attempt suicide. Given this information and the fact that people who attempt suicide are usually suffering from conditions like depression, schizophrenia, or substance use disorders, it is natural to jump to the assumption that mental health professionals should be able to predict who is at risk for suicide and intervene with effective treatment to prevent that from happening by instituting effective treatment.
In fact, the ability of any clinician to predict who is going to attempt suicide is notoriously poor. As psychiatrist Robert Simon wrote in 2002, “Psychiatrists cannot predict with certainty which patients will commit suicide. Suicide is a rare event. Attempts to predict suicide produce many false-positive and false-negative results.” In other words, most people we think might attempt suicide never do; only a small minority of people, even among those who suffer from mental illness, actually die by suicide.
With respect to individual-level prevention, almost all experts agree that mental health care in the US—and indeed around the world—is inadequate. Many patients suffering from psychiatric illness do not have access to evidence-based psychiatric care. And given the fact that studies show that 90 oercent of people who die by suicide have a mental disorder at the time of their deaths, it is understandable that we assume that if people get the treatment they need, deaths by suicide can be prevented. Sadly, however, even as treatment rates for depression increase in the developed world—more and more people are taking antidepressant medication than ever before—the suicide rate continues to increase. That doesn’t mean that antidepressants don’t work—they clearly do for people with moderate to severe levels of depression. Rather, it means that even for people who get mental health treatment, some will nonetheless kill themselves. Good mental health treatment can prevent many, but not all, suicides.
The hazard presented by the way these two assumptions are generally framed is that they place the burden of predicting and preventing suicides squarely on the shoulders of individual psychiatrists, psychologists, and other mental health professionals as they treat individual patients. An April 2018 Scientific American editorial points out that a substantial fraction of people who go on to die by suicide see mental health professionals or primary care doctors shortly before their deaths. “Yet,” the Scientific American editors note, “there are no national standards requiring these workers to know how to identify patients at serious risk of suicide or what techniques help them survive. If there were, perhaps some of those deaths could have been avoided.” The Scientific American editors go on to call for mandatory training in suicide risk assessment and prevention for all mental health care professionals and primary care physicians.
Training and evidence-based standards are all good things, and it is hard to argue against them when it comes to suicide prevention. But the editorial implies that it is the lack of such expertise that is responsible for the rising rates of suicide. There is absolutely no evidence to support the contention that this is the case. And advancing it as a cause of suicide has potentially devastating unintended consequences.
Imagine a cardiologist who tells a patient with severe heart disease that she won’t accept him into her practice because he has a high risk of dying of a heart attack. Or an oncologist saying, “the type of cancer you have is potentially fatal and I don’t take on patients who might die.” These seem absurd, of course. We know that many patients who cardiologists and oncologists take care of will die because the illnesses they treat are serious ones that kill people. As long as doctors in those specialties do the best job they can in treating their patients, we do not fault them when some succumb.
But we do not treat mental illness and suicide in the same way. Instead of acknowledging that depression, schizophrenia, and substance abuse are potentially fatal illnesses, we hold psychiatrists and psychologists accountable for suicide deaths by expecting them to make accurate predictions. This leads many to refuse to treat patients who express suicidalthoughts or plans. As psychiatrist H. Steven Moffic recently wrote, “For psychiatrists, the suicide of one of our patients is probably the most distressing event in our career.” According to journalist Sulome Anderson, “Many psychiatrists refuse to treat chronically suicidal patients, not only because of the stigma that surrounds it even in their profession, but because suicide is the number-one cause of lawsuits brought against mental-health treatment providers. Even though it is far harder to predict suicide than heart attacks, we seem to think that psychiatrists ought to be able to do it and that their failure to do so is what puts people at risk.
This problem extends to the research that is done to find better treatments for mental illness. Many clinical trials testing new antidepressant medication specifically exclude any patients who have suicidal thoughts or are at risk for suicide. The drug companies sponsoring these studies do not want the liability risk in case someone dies by suicide during a trial of an experimental medicine. This means that we have virtually no data from rigorous clinical trials about whether and which medications might actually decrease suicide risk. Only two psychiatric medications, lithium and clozapine, have ever been shown to reduce suicide risk and neither is specifically an antidepressant. Fortunately, this situation may change as the FDA recently proposed new guidelines that would permit including suicidal patients in antidepressant clinical trials. But whether drug companies will do so even with new guidelines is uncertain.
There are some promising research leads that may improve prediction, but they are not ready for clinical use and may never be. What we need to do instead is to reassure mental health care professionals that because it is impossible for them to know who will die by suicide, we want them to treat patients with psychiatric illness to the best of their ability regardless of the risk for suicide, knowing that some of these patients may tragically and unavoidably die. Depression, like heart disease and cancer, is a potentially fatal disease and not all deaths can be either predicted or prevented.
Rather than putting the burden on clinicians to sort out who might attempt suicide, experts in the field of suicide prevention, like those at the CDC, increasingly emphasize a public health approach. While preventing suicide on an individual basis is nearly impossible, there are a number of things we can to reduce the suicide rate on a population basis. These include putting up barriers to prevent jumping from bridges and tall buildings, restricting access to lethal means like firearms and opioids, and creating mental health educational programs in our schools.
The rate of suicide in the US is rising at a shockingly high rate, but suicide remains a rare event. We must be very careful not to scare away psychiatrists and psychologists from treating people with serious mental illness because they fear a patient will die. Rather, we must reassure them that prediction on an individual basis is not an evidence-based method of suicide prevention. Ensuring that suffering people have access to high-quality mental health care regardless of their suicide risk is much more likely to help.