'100% accurate' eye test for schizophrenia claim
“New ‘100% accurate’ test diagnoses schizophrenics simply by checking their gaze,” the Daily Mail reports. The newspaper goes on to say that the “tests are simple, cheap, and take only minutes to conduct” and (rather contrarily) “demonstrated 98 per cent accuracy” in distinguishing between those with and without schizophrenia.
This news is based on research into the ability of a series of eye movement tests to detect schizophrenia.
There is a lot of evidence highlighting the fact that many people with schizophrenia have abnormal eye movements. Until now, this fact has never been used to help diagnose schizophrenia. In this study, researchers recruited two groups of people:
- one group of people with a confirmed diagnosis of schizophrenia, that was judged to be well-controlled (their symptoms were responding to treatment)
- a control group without any serious mental health problems
Each group was then given the following visual tests:
- following a moving object with the eyes
- steadily holding a gaze
- viewing an image
They found that people who had significant difficulties with all of the above were far more likely to be from the schizophrenia group than the control group – the results of the testing allowed them to build a diagnostic model that they claimed was 98.3% accurate.
The researchers conclude that the tests may be a useful addition to current schizophrenia diagnostic practices that are based on the presence of symptoms. Though further research will be needed to validate the results and to see if abnormal eye movements are only limited to people with schizophrenia (ie the test can exclude all other conditions).
Where did the story come from?
The study was carried out by researchers from the University of Aberdeen, the University of Munich, and the National Institute of Mental Health in the US. The research was supported by the Royal Society of London, the Millar-Mackenzie Trust, the National Institute of Mental Health, the University of Aberdeen, the SGENE Consortium and the Scottish Chief Scientist Office.
The study was published in the peer-reviewed medical journal Biological Psychiatry.
While the Daily Mail coverage of the study was accurate overall, there were two main problems with the reporting.
Firstly, the use of the term ‘schizophrenic’ in the headline is unhelpful. As many mental health charities have argued, using such a term is essentially defining an individual by a disease. ‘People with schizophrenia’ would better reflect the experience of people who do have, often complex, mental health problems, but also have a life outside of those problems.
Secondly, an earlier online version of the article contained a picture of the actor Clare Danes, who is currently starring as CIA agent Carrie Mathison in US hit TV series Homeland, who the caption described as having schizophrenia. But as any fan of the show knows, Carrie actually has bipolar disorder. While on the surface this may be a trivial point, the mistaken image (now removed) does suggest a pattern of ignorance about mental health in some sections of the media.
What kind of research was this?
This was a case-control study that assessed the ability of eye movement tests to accurately predict whether or not a person has schizophrenia.
The researchers selected these tests because abnormal eye movements have long been reported to be a feature of psychotic illness, including schizophrenia.
The authors say that there has been little success in finding early warning signs of schizophrenia that could reliably predict the development of the disorder.
The aim of this research was to examine whether or not certain eye movement abnormalities could serve as stable markers of schizophrenia and accurately distinguish between cases and controls.
This study provides early evidence and helps the researchers construct models and identify the most useful parts of the test.
Case-control studies such as this are, in general, not ideal designs for evaluating the accuracy of diagnostic tests. A study where a test is performed before confirmation of a diagnosis in an unselected group of patients would be more reliable.
What did the research involve?
The researchers recruited 88 schizophrenia patients and 88 healthy controls. The two groups were matched on age, and all participants had normal vision. The researchers recorded the participants’ eye movements during a series of eye movement tests, which included:
- smooth pursuit, which involves smoothly tracking a moving object on a screen for 20 seconds
- fixation or gaze maintenance, which involves steadily holding one’s gaze on a single, unmoving object for five seconds while ignoring a distracting object to the side of the target
- free-viewing scanpaths, which trace how a person’s gaze moves around a picture of objects, faces, computer-generated images or everyday scenes that appeared on a screen for eight seconds
The researchers collected data on several features of each of these tasks and used this data to build a series of models intended to predict whether a person had schizophrenia or was a healthy control. They applied the model on a group of 26 people with schizophrenia and eight healthy controls that were retested nine months after the original tests, in order to check for any change in the model’s prediction over time.
A second group of 36 new cases and 52 new controls then completed the three eye movement tasks, and the models were used to predict whether or not each individual was a case or control. The researchers then built new models based on the data from all 298 tests and determined which model had the highest predictive ability.
What were the basic results?
The researchers found that performance on the smooth pursuit, fixation and free-viewing tasks were all abnormal in the schizophrenia group compared to the healthy control group.
When using the data from all 298 tests, the researchers found that the predictive accuracy ranged from 87.6% to 98.3% across the models. When looking at the model that resulted in approximately 98% accuracy, the researchers found that no people with schizophrenia had been misclassified as normal, while five control subjects had been misclassified as having schizophrenia.
In terms of individual tests, the researchers report that free-view scanning abnormalities were widespread among people with schizophrenia, and were the largest single discriminator between people with schizophrenia and healthy controls.
How did the researchers interpret the results?
The researchers say that their results suggest that eye movement tests have “considerable power to discriminate schizophrenia cases from control subjects” and that “they are cheap, easy to administer, and can be used in a hospital or clinic on all but the most severely disturbed”.
Conclusion
This case-control study suggests that a series of simple eye movement tests may be able to accurately predict whether or not a person has schizophrenia. The model will need to be tested on a broader range of people, especially those with early illness, before we can be sure that the high degree of accuracy seen in this study will hold in practice.
When using each of the models to predict schizophrenia status, the researchers report that some people with schizophrenia had eye movement abnormalities that would be considered borderline.
They say that each of the models performed differently depending on the group of participants included, and that it is not clear whether this variation in performance is due to the size of the groups on which the models were built, or the model structure itself.
One of the intriguing features of this test is that it can be carried out relatively quickly and without the extensive training currently required for schizophrenia diagnosis.
The authors of the study say that current symptom-based diagnostic practices involve “time-consuming neuropsychologic assessments carried out by expensive, highly qualified individuals”, whereas “eye-movement recordings can be performed by a technically competent assistant after a few hours of training”. Additionally, the eye movement data can be collected “in a few minutes and analysed in real time”.
There are, however, limitations to the current study. The authors note that cases and controls were drawn from different populations (people with schizophrenia from Scotland and Germany, and healthy controls from Scotland alone). While the two groups were similar clinically speaking, ideally one would recruit cases and controls from the same populations to reduce potential confounding.
The authors also note that they intentionally included a group of younger control subjects in the group of new participants. They say that this has the limitation of including control subjects that are still at an age where they are still at risk of developing schizophrenia.
While the model was able to accurately discriminate between schizophrenia cases and controls, the researchers point out that further research is necessary to see if the eye movement abnormalities correctly classify people with schizophrenia compared to people with other psychiatric disorders.
Finally, even if the claimed predictive accuracy of the test was correct, the test alone could never be used as a sole diagnosis for schizophrenia. However, this research does offer a potentially promising method – especially when used in combination of other well-established techniques – to improve diagnosis of schizophrenia.
