A recent JAMA: Psychiatry article makes some important points about the difficulties addressing suicide as a public health problem before sliding into the authors’ promotion of their personal agendas.
Christensen H, Cuijpers P, Reynolds CF. Changing the Direction of Suicide Prevention Research: A Necessity for True Population Impact. JAMA Psychiatry. 2016.
This issue of Mind the Brain:
- Reviews important barriers to effective approaches to reducing suicide, as cited in the editorial.
- Discusses editorials in general as a form of privileged access publishing by which non-peer-reviewed material makes its way into ostensibly peer reviewed journals.
- Identifies the self-promotional and personal agendas of the authors reflected in the editorial.
- Notes that the leading means of death by suicide in the United States is not even mentioned, much less addressed in this editorial. I’ll discuss the politics behind this and why its absence reduces this editorial to a venture in triviality, except that it is a call for the waste of millions of dollars.
Barriers to reducing mortality by suicide
Prevention of death by suicide becomes an important public health and clinical goal because of suicide’s contribution to overall mortality, the seeming senselessness of suicide, and its costs at a personal and social level. Yet as a relatively infrequent event, death by suicide resists prediction and effective preventive intervention.
Evidence concerning the formidable barriers to reducing death by suicide inevitably clashes with the strong emotional appeals and political agendas of those demanding suicide intervention programs.
Skeptics encounter stiff resistance and even vilification when they insist that clinical and social policy concerning suicide should be based on evidence.
A skeptic soon finds that trying to contest emotional and political appeals quickly becomes like trying to counter Ted Cruz or Donald Trump with evidence contradicting their proposals for dealing with terrorism or immigration. This is particularly likely after suicides by celebrities or a cluster of suicides by teenagers in a community. Who wants to pay attention to evidence when emotions are high and tears are flowing?
See my recent blog post, Preventing Suicide in All the Wrong Ways for some inconvenient truths about suicide and suicide prevention.
The JAMA: Psychiatry article’s identification of barriers
The JAMA: Psychiatry article identifies some key barriers to progress in reducing deaths due to suicide [bullet points added to direct quotes]:
- Suicide rates in most Western countries have not decreased in the last decade, a finding that compares unfavorably with the progress made in other areas, such as breast and skin cancers, human immunodeficiency virus, and automobile accidents, for which the rates have decreased by 40% to 80%.
- Preventing suicide is not easy. The base rate of suicide is low, making it hard to determine which individuals are at risk.
- Our current approach to the epidemiologic risk factors has failed because prediction studies have no clinical utility—even the highest odds ratio is not informative at the individual level.
- Decades of research on predicting suicides failed to identify any new predictors, despite the large numbers of studies.
- A previous suicide attempt is our best marker of a future attempt, but 60% of suicides are by persons who had made no previous attempts.
- Although recent studies in cognitive neuroscience have shed light on the cognitive “lesions” that underlie suicide risk, especially deficits in executive functioning, we have no biological markers of suicide risk, or indeed of any mental illness.
- People at risk of suicide do not seek help. Eighty percent of people at risk have been in contact with health services prior to their attempts, but they do not identify themselves, largely because they do not think that they need help.
- As clinicians, we know something about the long-term risk factors for suicide, but we are much less able to disambiguate short-term risk or high-risk factors from the background of long-term risk factors.
How do editorials come about? Not peer review!
Among the many privileges of being editor-in-chief or associate editors of journals is the opportunity to commission articles that do not undergo peer review. Editors and their friends are among the regular recipients of these gifts that largely escape scrutiny.
Editorials often provide a free opportunity for self-citation and promotion of agenda. Over the years, I’ve noticed that editorials are frequently used to increase the likelihood that particular research topics will become a priority for funding for the particular ideas will be given advantage in competition for funding.
Editorials of great opportunities for self citation. If an editorial in a prestigious journal cites articles published in less prestigious places, readers will often cite the article, without bothering to examine the original source. This is a way of providing false authority to poor quality or irrelevant evidence.
Not only do authors of commissioned articles get to say what they wish without peer review, they can restrict what can be said in reply. Journals are less willing to publish letters to the editor about editorials rather than empirical papers. They often give the writers of the editorial veto power over what criticism is published. Journals always give the writers of the editorial last word in any exchange.
So, editorials and commentaries can be free sweet plums if you know how to use them strategically.
Helen Christensen, PhD Black Dog Institute, University of New South Wales, Randwick, New South Wales, Australia.
Pim Cuijpers, PhD Department of Clinical, Neuro, and Developmental Psychology, Vrije Universiteit Amsterdam, the Netherlands
Charles F. Reynolds III, MD Department of Psychiatry and Neurology, Western Psychiatric Institute and Clinic, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania.
The authors’ agendas
Helen Christianson is the Chief Scientist and Director of the Black Dog Institute, which is described at its website:
Our unique approach incorporates clinical services with our cutting-edge research, our health professional training and community education programs. We combine expertise in clinical management with innovative research to develop new, and more effective, strategies for people living with mental illness. We also place emphasis on teaching people to recognise the symptom of poor mental health in themselves and others, as well as providing them with the right psychological tools to hold the black dog at bay.
A key passage in the JAMA: Psychiatry editorial references her work.
Modeling studies have shown that if all evidence-based suicide prevention strategies were integrated into 1 multifaceted systems approach, about 20% to 25% of all suicides might be prevented.
Here is the figure from the editorial:
The paper that is cited would be better characterized as an advocacy piece, rather than a balanced systematic review.
Most fundamentally, Christiansen makes the mistake of summing attributable risk factors to obtain a grand total of what would be accomplished if all of a set of risk factors were addressed.
The problem is that attributable risk factors are dubious estimates derived from correlational analyses which assume that the entire correlation coefficient represents a modifiable risk. Such estimates ignore confounding. If one adds together attributable risk factors calculated in this manner, one gets a grossly inflated view of how much a phenomenon can be controlled. The attributable risk factors are themselves correlated and they share common confounds. That’s why it is bad science to combine them.
Christiansen identifies the top three modifiable risk for suicide as (1) training general practitioners in detection and treatment of suicidal risk, and notably depression; (2) training of gatekeepers such as school personnel, police, (and in some contexts, clergy) who might have contact with persons on the verge of dying by suicide; and (3) psychosocial treatments, namely psychotherapy.
Training of general practitioners and gatekeepers has not been shown to be an effective way of reducing rates of suicide. #Evidenceplease. I’ve been an external scientific advisor to over a decade of programs in Europe which emphasized these strategies. We will soon be publishing the last of our disappointing results.
Think of it: in order to be effective in averting death by suicide, training of police requires that police be on the scene in circumstances where they could use that training to prevent someone from dying by suicide, say, by jumping from a bridge or self-inflicted gun wounds. The likelihood is low that it would be a police officer with sufficient training being in the right place at the right time, with sufficient time and control of the situation to prevent a death. A police officer who had received training would unlikely encounter only a few, if any situations in an entire career.
The problem of death by suicide being an infrequent event that is poorly predicted again rears its ugly head.
Christiansen also makes a dubious assumption that more readily availability of psychotherapy will substantially reduce the risk of suicide. The problem is that persons who die by suicide are often in contact with professionals, but they either break the contact shortly before death or never disclose their intentions.
Christiansen provides a sizable estimate for the reduction in risk for suicide by means restriction.
]. Yet, I suspect that she underestimates the influence of this potentially modifiable factor.
She focuses on restricting access to prescription medications used in suicides by overdose. I don’t know if death-by-overdose data holds for even Australia, but the relevant means needing restriction in the United States is access to firearms. I will say more about that later.
So, Christiansen makes use of the editorial to sell her pet ideas and her institute markets training.
Pim Cuijpers doesn’t cite himself and doesn’t need to. He is rapidly accumulating a phenomenal record of publications and citations. But he is an advocate for large-scale programs incorporating technology, and notably the Internet to reduce suicide. His interests are reflected in passages like
Large-scale trials are also needed. Even if we did all of these things, large-scale research programs with millions of people are required, and technology by itself will not be enough. Although new large trials show that the effects of community programs can be effective,1,6 studies need to be bigger, combining all evidence-based medical and community strategies, using technology effectively to reduce costs of identification and treatment.
Help-seeking may well be assisted by using social media. Online social networks such as Facebook can be used to provide peer support and to change community attitudes in the ways already used by marketing industries. We can use the networks of “influencers” to modify attitudes and behavior in specific high-risk groups, such as the military, where suicide rates are high, or “captive audiences” in schools.
Disseminating effective programs is no longer difficult using online mental health programs. Although some early suicide apps and websites have been tested, better online interventions are needed that can respond to temporal fluctuations in suicide risk. The power of short-term prediction tools should be combined with the timely delivery of unobtrusive online or app personalized programs. However, if these development are not supported by government or industry and implemented at a population level, they will remain missed opportunities.
Pim Cuijpers is based the Netherlands and writing at a time when enthusiasm of the European Research Council is waning in funding large-scale suicide prevention programs, especially expensive ones requiring millions of participants. Such studies have been going on for over a decade and the yield is not impressive.
The projects on which I consulted adopted the reasonable assumption that because suicide is a rare event, a population of 500,000 would not be sufficient to detect a statistically significant reduction in suicide rates of less than 30%. Consider all the extraneous events that can impinge on comparisons between intervention and control sites in the time period in which the intervention could conceivably be influential. this is too low an estimate of the sample that would be needed.
The larger the sample, the greater the likelihood of extraneous influences, the greater the likelihood that the intervention wouldn’t prove effective at key moments when it was needed to avert a death by suicide, and the greater the cost. See more about this here.
Pim Cuijpers has been quite influential in developing in evaluating web-based and app-based interventions. But after initial enthusiasm, the field is learning that such resources are not effective if left unattended without users being provided with a sense that they are in some sort of a human relationship within which they are consistent use of this technology is being monitored and appreciated, as seen in appropriate feedback. Pim Cuijpers has contributed the valuable concept of supportive accountability. I have borrowed it to explain what is missing when primary care physicians simply give depressed patients a password to an Internet program and leave it at that, expecting they will get any benefit.
Evaluations of such technology have been limited to whether they reduce depressive symptoms. There is as much a leap from evidence of such reductions, when they occur, claims about preventing suicide, as there is from leaping from evidence that psychotherapy reduces the depressive symptoms to a case that psychotherapy prevents suicide.
Enlisting users of Facebook to monitor and report expressions of suicidality is not evidence based, It is evaluated by some as a disaster and a consumer group is circulating a petition demanding that such practices stop. A critical incident was
Although Charles Reynolds does not reference his paper in the text of the editorial, but nonetheless cites it.
I have critiqued the study elsewhere. It was funded in a special review only because of political pressure from Senator Harry Reid. The senator’s father had died by suicide shortly after a visit to a primary care physician. Harry Reid required that Congress fund a study showing that improving the detection and treatment of suicidality in the elderly by primary care physicians would reduce suicide.
I was called by an NIMH program officer when I failed to submit a letter of intent concerning applying for that initiative. I told her it was a boondoggle because no one could show a reduction in suicides by targeting physician behavior. She didn’t disagree, but said a project would have to funded. She ended up a co-author on the PROSPECT paper. You don’t often see program officers getting authorship on papers from projects they fund.
The resulting PROSPECT study involved 20 primary care practices in three regions of the Northeastern United States. In the course of the intervention study, one patient in the intervention group died by suicide and two patients, one in each of the intervention and control group, made serious attempts. A multimillion dollar study confronted the low incidence of suicide, even among the elderly. Furthermore, the substantial baseline differences among the practices dwarfed any differences in suicidal ideation in the intervention versus control group. And has of discussed elsewhere [ ], suicidal ideation is a surrogate end point that can be changed by factors that do not alter risk for suicide. No one advocating more money for these kind of studies would want to get into the details of this one.
So, the editorial acknowledges the difficulties studying and preventing suicide as a public health issue. It suggests that an unprecedented large study costing millions of dollars would be necessary if progress is to be made. There are formidable barriers to implementing an intervention in a large population of the complexity of the editorial suggests is necessary. Just look at the problems that PROSPECT encountered.
Who will set the direction of suicide prevention research?
The editorial opens with a citation of a blog by the then Director of NIMH
Insel T. Director’s Blog: Targeting suicide. National Institutes of Health website. Posted April 2, 2015.
The blog calls for a large increase in funding for the research concerning suicide and its prevention. The definition of the problem is shaped by politics more than evidence. But at least the blog post is more candid than the editorial in making a passing reference to the leading means of suicide in the United States, firearms.
51 percent of suicide deaths in the U.S. were by firearms. Research has already demonstrated that reducing access to lethal means (including gun locks and barriers on bridges) can reduce death rates.
Great, but surely death by firearms deserves more mentioned than a passing reference to locks on guns if the Director of NIMH is serious about asking Congress for a massive increase in funding for suicide research. Or is he being smart in avoiding the issue and even brave in the passing reference that he makes to firearms?
Firearms deserve not only mention, but thoughtful analysis. But in the United States, it is politically dangerous and could threaten future funding. So we talk about other things.
Banning research on the role of firearms in suicide
For a source that is much more honest, evidence-based, and well argued than this JAMA: Psychiatry editorial, I recommend A Psychiatrist Debunks the Biggest Myths Surrounding Gun Suicides.
In 1996, Congress imposed a ban on research concerning the effects of gun ownership on public health, including suicide.
In the spring of 1996, the National Rifle Association and its allies set their sights on the Centers for Disease Control and Prevention for funding increasingly assertive studies on firearms ownership and the effects on public health. The gun rights advocates claimed the research veered toward advocacy and covered such logical ground as to be effectively useless.
At first, the House tried to close down the CDC’s entire, $46 million National Center for Injury Prevention. When that failed, [Congressman Jay Dickey to whom the Dickey amendment is named] Dickey stepped in with an alternative: strip $2.6 million that the agency had spent on gun studies that year. The money would eventually be re-appropriated for studies unrelated to guns. But the far more damaging inclusion was language that stated, “None of the funds made available for injury prevention and control at the Centers for Disease Control and Prevention may be used to advocate or promote gun control.”
Dickey proclaimed victory — an end, he said at the time, to the CDC’s attempts “to raise emotional sympathy” around gun violence. But the agency spent the subsequent years petrified of doing any research on gun violence, making the costs of the amendment clear even to Dickey himself.
He said the law was over-interpreted. Now, he looks at simple advances in highway safety — safety barriers, for example — and wonders what could have been done for guns.
The Dickey amendment does not specifically ban NIMH from investigating the role of firearms in suicide, but I think Tom Insel and all NIMH directors before and after him get the message.
Recently an effort to repeal the Dickey amendment failed:
Just hours before the mass shooting in San Bernardino on Wednesday, physicians gathered on Capitol Hill to demand an end to the Dickey Amendment restricting federal funding for gun violence research. Members of Doctors for America, the American College of Preventative Medicine, the American Academy of Pediatrics and others presented a petition against the research ban signed by more than 2,000 doctors.
“Gun violence is probably the only thing in this country that kills so many people, injures so many people, that we are not actually doing sufficient research on,” Dr. Alice Chen, the executive director of Doctors for America, told The Huffington Post.
Well over half a million people have died by firearms since 1996, when the ban on gun violence research was enacted, according to a HuffPost calculation of data through 2013 from Centers for Disease Control and Prevention. According to its sponsors, the Dickey Amendment was supposed to tamp down funding for what the National Rifle Association and other critics claimed was anti-gun advocacy research by the CDC’s National Center for Injury Prevention. In effect, it stopped federal gun violence research almost entirely.
So, why didn’t the Associate Editor of the JAMA: Psychiatry, Charles Reynolds exercise his editorial prerogative and support this effort to repeal the Dickey amendment, rather than lining up with his co-authors in a call for more wasteful research that avoids this important issue?