"Sufferers of psychotic illnesses 'may have treatable immune disorder'," The Independent reports.
Researchers from Oxford University found around 9% of people presenting with psychotic symptoms also had signs of immune dysfunction.
They found these people had antibodies in their blood linked to a condition called antibody-mediated encephalitis.
In this condition, antibodies made by the immune system mistakenly attack the surface receptors of brain cells, causing hallucinations, paranoia and delusions – a group of symptoms collectively known as psychosis.
Psychosis caused by antibody-mediated encephalitis can sometimes be treated successfully with drugs that suppress the immune system.
Researchers took blood from 228 people diagnosed with a first episode of psychosis and 105 people without any mental health condition.
They found 9% of people with psychosis had antibodies to a brain cell receptor, compared with 4% of people without psychosis. But this difference was so small it could have been down to chance.
The results of this study are not clear-cut. The presence of certain antibodies in the blood doesn't mean that people's psychosis was definitely caused by encephalitis, which also triggers symptoms like seizures and movement disorders.
Even so, the researchers advise that people presenting with symptoms of psychosis should be given a test for antibodies as part of their overall diagnosis.
The study was carried out by researchers from the University of Oxford, King's College London and the University of Cambridge.
It was funded by the Medical Research Council and published in the peer-reviewed journal, The Lancet Psychiatry.
Two of the researchers and the University of Oxford hold patents for tests to identify neuronal antibodies, which could be seen as a conflict of interest as they have a financial incentive to encourage the use of these tests.
The study was widely covered by the UK media, but the stories were very selective in their use of statistics.
BBC News, ITV News and the Mail Online all reported that researchers had found relevant antibodies in 1 in 11 (9%) patients.
However, none of them reported the important fact that researchers had also found these antibodies in 4% of people without psychosis, and that the difference between the two groups was too small to be statistically significant.
This case control study compared levels of antibodies with brain cell receptors found in the blood of people without mental illness and people with a first episode of psychosis.
Case control studies can find patterns linking factors, but cannot tell us whether one factor (such as antibodies) caused the other (such as psychosis).
In this case, we don't know whether the antibodies were present before the symptoms started, for example.
Researchers recruited people aged 14 to 35 who were treated at one of 35 English mental health service sites for a first episode of psychosis, and took blood samples.
They also used blood samples from 105 people without a mental illness, who were similar in age, sex and ethnic background.
They screened the blood samples for antibodies to brain cell receptors and compared the results between groups.
The control group came from another study, so they weren't specifically matched to the patients in this group, although they were the same average age, from a broadly similar ethnic background, and there were similar proportions of men and women.
The researchers say they used a method of blood antibody screening different from that used by other researchers, although it's not clear whether this made a difference to the results.
They adjusted their figures to take account of people's use of tobacco, alcohol and illegal drugs, as this differed between the two groups.
They also measured the symptom scores of the people who'd been treated for psychosis to see whether people with and without antibodies to brain cell receptors could be identified by their symptoms.
The researchers found:
The researchers say that, "Some patients with first-episode psychosis had antibodies against NMDAR that might be relevant to their illness."
Because the symptoms were similar whether someone had relevant antibodies or not, "the only way to detect those with potentially pathogenic antibodies is to screen all patients with first-episode psychosis" when they are first seen by doctors.
The implication of the headlines reporting on this study is that many people diagnosed with schizophrenia or another psychiatric illness might have been misdiagnosed, and that they need treatment for an immune disease instead.
If true, that would be a massive concern. But the results of this study don't really bear out these fears.
There's no evidence that most of the antibodies tested are more common in people with psychosis than in people without mental illness.
Only one antibody, NMDAR, was significantly more common in people with psychosis than in the control group. Only 3% of people with psychosis had this antibody, and none in the control group.
However, the control group of 105 was pretty small for this type of research, so it's hard to know if the results would hold true for a bigger group.
We'd need to test many more people to be sure that nobody without a mental health condition had antibodies against NMDAR.
Because the study only looked at people's antibodies at one point in time, we don't know whether they might have caused the symptoms of psychosis or not.
If the antibodies only appeared after the symptoms started, they might be an effect of the disease, not a cause of it.
Psychosis is not the only symptom of antibody-mediated encephalitis. Although it may be the first symptom, people also have neurological symptoms such as seizures and movement disorders.
It's likely that these would be noticed by doctors treating people for psychotic diseases over a long period of time.
The researchers call for universal blood testing of people with a first episode of psychosis. But it's not clear that these results support this.