Predicting Suicide: Return of a Scandal (Part 2)
10 bulan ago kesimpulan 0
What is happening in our world? Who is discovering what? What is going on now? These are questions that will be answered. Enjoy.
Predicting Suicide: Return of a Scandal (Part 2)
In the first post in this series, I looked at the work of Swedish psychiatrist Lars Thorell, who has developed a test which, he claims, is able to predict suicides in depressed patients. Thorell’s test is called electrodermal orientation reactivity (aka electrodermal hyporeactivity), and while Thorell’s work on the technique goes back to the 1980s, it has recently been commercialized by a company called Emotra AB, who named the product EDOR®.
Previously, I expressed scepticism over the published evidence purporting to show that electrodermal orientation reactivity can predict suicide. In this post, I’m going to examine Emotra and their claims about EDOR®.
Emotra is a Swedish company of which Thorell is head of research, as well as a board member and major shareholder. EDOR® is their only product and, according to them, it’s quite something. Here’s how they describe the test:
EDOR® An objective diagnostic tool for assessment of suicide risk. EDOR® identifies hyporeactive patients, related to imminent risk of suicide or violent suicide attempts.
Emotra claim that EDOR will provide benefits such as:
– Many lives can be saved if risk patients may be identified at an early stage
– If most suicide attempts can be prevented, high costs for hospital care and rehabilitation can be avoided
– The use of EDOR® makes it possible to support the caregivers in avoiding overtreatment (unnecessary preventive actions) of patients, that do not belong to the risk group
So where’s the evidence that it works? Emotra prominently say that “as many as 97 per cent of the studied patients who have committed suicide were hyporeactive. This represents an undisputable strong relationship between hyporeactivity and risk for suicide.” However, as I pointed out last time, it is very easy to create a test which predicts 97% of suicides: you simply predict suicide in 97% of everyone you meet. Such an absurd ‘test’ would have 97% sensitivity, but to be useful, a test needs not only sensitivity but also specificity.
On their ‘Scientific Studies‘ page, Emotra say that “Among depressed patients who were not hyporeactive, only 2 % took their life (98% specificity for suicide)3.”
But this just isn’t what specificity means. It’s wrong. Specificity is the proportion, of those people who didn’t commit suicide, who were not hyporeactive – not the other way around. The specificity value in the referenced paper was 33%, not 98%. Thorell has previously exhibited confusion on this point in a paper, which earned the professor a rebuke from an undergraduate student (amongst other people) but to see the same error repeated in a commercial context is rather extraordinary.
So I don’t think the published data on this test is very impressive… but it gets worse. In a September 2017 press release, Emotra reveal some results from EUDOR-A, their large ‘naturalistic’ trial of EDOR® in depressed patients, launched in 2014. The trial was announced in a 2017 paper, but the results remain unpublished.
The press release doesn’t provide exact data like sensitivity and specificity for EDOR-A, but it does state that there were very few suicides in either the hyporeactive or non-hyporeactive patients: “All in all, only three suicides occurred in the hyporeactive group. In the normally reactive group the number of suicides, as expected, was very low.” Reading between the lines, I suspect this to mean that the test didn’t predict suicides, because if it had, Emotra would surely have said so.
Emotra nonetheless suggest that the EUDOR-A results might mean that EDOR® does work – at preventing suicides! “The total ratio of documented suicides in EUDOR-A is a record low… this reduction can most probably be explained by the directed suicide prevention measures that the clinics by their own accounts implemented to protect hyporeactive patients.” A nice way of looking at it. Happily, this means that the EUDOR-A trial was all along a win/win proposition for Emotra: however the results had turned out, they could claim it as a success for their product.
Indeed, there’s reason to believe that EUDOR-A was intended to serve a marketing function. In early 2015, Emotra’s CEO described EUDOR-A as part of their strategy to “lay a solid foundation for a future launch” of EDOR®:
Many of Europe’s leading clinics are participating in [EUDOR-A]. If the study results turn out to be as compelling as we think they will, this fact should make these clinics want to assist us to establish the method in specialised psychiatric care as soon as possible. We expect that training efforts will mainly be handled by the ‘Key Opinion Leaders’ that have participated in our study…