Apply


Overcoming the “DSM Fallacy”: Is It Time to Start Crowdsourcing Mental Health Treatments?

LinkedInFacebookTwitterGoogle+PrintFriendlyShare

Most of us are familiar with the Global Burden of Disease studies emanating from a research collaboration that since 1990 has tracked overall metrics on causes of worldwide morbidity and mortality. In the United States, the “Healthy People” collaborative analyzes trends and sets aspirational goals on a decade-by-decade basis. Anyone who has followed these data, even superficially, will recognize a few trends: Many of the top causes of mortality have strong psychosocial and behavioral components (e.g., suicide or homicide by firearm, cirrhosis, opiate overdose, diabetes, COPD). Links to psychosocial antecedents are even more pronounced when considering causes of disability: Major depression has for more than 25 years been rated as the overall 2nd cause of disability, and other mental disorders such as bipolar disorder, schizophrenia, alcohol and drug dependence, anxiety and associated mental disorders comprise one third of the top 25 causes of disability. But what is surprising about the most recent version of these lists published in the Journal of the American Medical Association (U.S. Burden of Disease Collaborators) is how persistent many of these conditions are. Even for those causes of morbidity for which we have well-established treatments, 25 years of awareness and intervention has not resulted in any reduction in the public health burden they impose. For example, rankings for major depression, anxiety, and dysthymia are essentially unchanged. In both 1990 and 2016, major depression ranked 2nd, anxiety was ranked 5th and 7th, and dysthymia was 22nd and 23rd, respectively. In terms of mortality, key indices have actually eroded, with non-gun related suicide jumping from 16th to 12th and alcohol related deaths rising from 27th to 19th.

Why the dismal numbers? The usual culprits share most of the blame: Inadequate funding of preventive services, inattention to social determinants of health, and lack of insurance/access to healthcare services. As a nation, we have not yet learned the well-established lesson of other developed economies—the more you spend on health-related social programs, the less you spend on mental health treatment. As a clinical profession, we remain resolutely supply-sided in our outlook: We expect our patients to come to us, rather than meet them at the point of care where they seek most services—the primary care environment. No matter how rigorously tested your intervention, it is useless if those who need it can’t get it, and the vast majority of patients with mental disorders never seek specialty care. As academicians and researchers, we over-value the importance of data based on between-group analyses of efficacy. In the real world, neither patients nor psychologists behave like patients or psychologists in randomized trials.

On one level, we understand this. On others, our research and practice paradigms are increasingly antiquated and remain stuck on what I have elsewhere termed the “DSM Fallacy.”  Recalling that the DSM, when first published in the early 1950s, was a slender tome of approximately 50 pages and has grown to a sprawling compendium just shy of 1,000 pages, the “DSM Fallacy” goes like this: Because we can now describe mental disorders with increasing degrees of specificity, our treatments must therefore be increasingly specific.

Like all good fallacies, this one is patently untrue. As I pointed out in my March column where I discussed the APA treatment guidelines for PTSD, any guideline that recommends a range of interventions from cognitive processing therapy to brief narrative therapy is, by definition, nonspecific. The DSM Fallacy, however, demands specificity, so it is no surprise that our best psychological researchers attempt to present us with solutions that are precise and targeted at specific problems. Sometimes we succeed, if only modestly. CBT for insomnia (CBT-i) is if only slightly more effective than plain old CBT in dealing with sleep problems. Sometimes we don’t succeed. CBT for psychosis (CBT-p) does not seem to be more effective for managing psychotic thinking than unmodified CBT (it’s hard to be terribly precise when we really don’t have a grasp on the active component of CBT).

It is the nature of scientific inquiry to seek greater specificity in understanding natural phenomena, and our clinical research colleagues strive valiantly to do so. But the scientific method can lead us astray in that it demands a focus on problems, not patients. And this, in a world where nonspecificity reigns, may not be the best approach.

What would happen if we turned our method on its head? Instead of demanding answers from the research community, why don’t we demand them from patients themselves? In an era of big data collection, “crowdsourcing” mental health treatments is increasingly feasible.

Crowdsourcing can be quite effective in solving a variety of problems. Wikipedia utilizes crowdsourcing to produce knowledge that is generally of the same accuracy as knowledge derived from academic expertise. Crowdsourcing mental health solutions also presents several advantages. First, it is empirical, in that patients are not likely to recommend an intervention that they have not found to be helpful. Second, knowledge gained from crowdsourcing is as amenable to rigorous analysis as is knowledge derived from hypothesis testing if we use the right tools.

A variation of crowdsourcing called cascade screening is used in developing precision gene therapies for inherited diseases. Cascade screening collects data on relatives of patients with certain heritable disorders, this information can be used to identify and assist others at potential risk (Roberts, M., et al.; 2018). Naslund et al. (2015) identified a small number of randomized trials that were based on internet responses of patients with serious mental illness. While most such studies solicited epidemiological survey data, Naslund and colleagues argued that is now feasible to use crowdsourcing to test interventions for serious mental illness. Internet-based crowdsourcing would have the advantage of reaching patients with SMI who are typically difficult to identify and engage in study protocols. Amassing results from the large number of participants that might be willing to participate in an online intervention would increase confidence in the accuracy of findings and would guide development of interventions. Importantly, as Naslund et al. emphasized, anonymous, internet-delivered interventions might provide a way of extending effective interventions to patients for whom stigma, suspicion of providers, or lack of access to care has previously limited involvement in the mental health treatment system.

Crowdsourcing is no panacea. It introduces an abundance of methodological and ethical problems. Informed consent is a particular challenge, as complete informed consent requires patient knowledge of the ways that collected data might be used. This is relatively unpredictable when data are collected from very large numbers of anonymous individuals, particularly since crowdsourcing would not make a priori assumptions about the efficacy of any intervention. Finally and perhaps most saliently, anonymity in the world of the internet is not guaranteed, as many have discovered to their great chagrin. Unscrupulous actors might seize the most personal data (DNA profiles, diagnoses, and family information) and turn these against individuals or groups of patients, potentially imperiling insurance coverage for individuals or ethnic groups with genetic predisposition to a life-threatening condition. Numerous legal and regulatory protections will be needed to govern protocols for crowdsourced interventions and the conduct of extremely powerful companies that shape much of our online behavior. But the potential benefits are equally compelling.

At present, we understand that the quality of the psychologist-patient relationship is a key determinant of positive outcomes.  In a crowdsourced intervention, we presume no doctor-patient relationship. Extracting interpersonal variables might better position us to discover the true active components of treatment and guide us to more effective interventions, whether delivered in the context of the therapeutic relationship or the anonymity of the internet. And the combination of more effective outreach and engagement, along with better understanding of the active components of treatment, might provide at least a partial solution to finally reducing the contribution of psychological illness to the global burden of disease.

Copyright © National Register of Health Service Psychologists, All rights reserved.

References

Roberts, M., et al. (2018). Delivery of cascade screening for hereditary conditions…Health Affairs, 37, 800–808.

Naslund, et al. (2015). Crowdsourcing for conducting randomized clinical trials of internet delivered interventions in people with serious mental illness: A systematic review; Contemporary Clinical Trials, 44, 77–88.

1200 New York Ave NW, Ste 800

Washington DC 20005

p: 202.783.7663

f: 202.347.0550

Endorsed by the National Register