Prognosis of brief psychotic episodes

This paper (Fusar-Poli et al, 2016) is one of a series of meta-analyses by this group looking at the prognostic significance of differing types of “At Risk Mental States (of Psychosis)” or ARMS for short.

This particular paper looks at the prognostic significance of some subtypes of ARMS compared to different types of brief psychotic illnesses compared to First Episode Schizophrenia.

There is an interesting diagram looking at the different historical terms used (from 1857 onwards) to describe psychotic illnesses that were not felt to be episodes of schizophrenia and were thought to have a better outcome. These historical diagnostic constructs may overlap with the subtypes of ARMS discussed below.

There is an excellent table that summarises the characteristics of the different diagnostic constructs included in the study. I will summarise the abbreviations used for these different diagnostic constructs.

BLIPS: Brief Limited Intermittent Psychotic Symptoms (an ARMS subtype)
BIPS: Brief Intermittent Psychotic Symptoms (an ARMS subtype)
ATPD: Acute and Transient Psychotic Disorder (an ICD-10 brief psychotic illness)
BPD: Brief Psychotic Disorder (a DSM brief psychotic illness)
FES: First Episode Schizophrenia (ICD/DSM schizophrenia: remission defined as having stopped antipsychotics)

The primary hypothesis of the paper was that the risk of recurrence of psychosis was ranked in order ATPD > BPD > BIPS > BLIPS.

The secondary hypothesis of the paper was that the risk of recurrence of psychosis in FES was greater than ATPD/BPD/BIPS/BLIPS.

We don’t know the prognostic significance of competing constructs for brief psychotic episodes.

Methods

A thorough search and selection strategy was carried out to identify relevant papers on recurrence of psychosis for the different diagnostic constructs. Some studies were naturalistic follow-ups after discontinuing antipsychotics, whilst other studies used more experimental designs.

There was also use of different criteria to define remission not just between the different diagnostic constructs but also within the same diagnostic constructs. The effects of this on results were examined.

As the FES subjects were more likely to be on medication, the effects of exposure to antipsychotics on the results was also examined.

A random effects model was used to reflect potential heterogeneity of results and publication bias was looked for (no evidence of this was found). I loved the term “the Freeman-Tukey double arcsine transformation”, which sounds like an episode title of “The Big Bang Theory”.

Results

The primary hypothesis was not proven
- The risk of recurrence of psychosis was remarkably similar for ATPD/BPD/BIPS/BLIPS at 6, 12, 24 and 36 months.
The secondary hypothesis was accepted
- FES had a significantly higher risk of recurrence at 24 and 36 months than ATPD/BPD/BIPS/BLIPS.

Recurrence (reported as mean [95% CI]) was significantly higher in the FES group (0.78 [0.58 to 0.93] at 24 months and 0.84 [0.70 to 0.94] at 36 months; P < .02 and P < .001, respectively) compared with the other 4 groups (0.39 [0.32 to 0.47] at 24 months and 0.51 [0.41 to 0.61] at 36 months).

To turn the recurrence figures into a % (multiply by 100) so for FES group at 24 months recurrence percentage was a mean of 78% with a 95% confidence interval of 58% to 93%.

Further analysis showed that female gender (p=0.002) and “exposure to antipsychotics” (p=0.03), i.e. taking antipsychotics, significantly reduced the risk of recurrence of psychosis. This later information is not surprising as we have long known that female gender is associated with better outcomes for psychotic illnesses and similarly have known that antipsychotics reduce recurrence of psychotic illnesses.

There was no difference in the risk of developing an affective psychosis between any of the groups. This risk was much lower than the risk of recurrence of psychosis in general.

Supplemental data showed that methodological concerns about different types of outcomes used in the papers did not affect the results of the meta-analysis and neither did different study designs such as naturalistic follow-ups compared to experimental studies.

There is consistent meta-analytical evidence for better long-term prognosis of brief psychotic episodes compared with remitted first-episode schizophrenia.

Strengths and limitations

Reasonably comprehensive search strategy and selection criteria
Use of random effects
Combining different definitions of recurrence of psychosis into a combined single outcome; even though this was tested to see if affected the outcomes
Similarly, combination of different study designs such as naturalistic follow-ups and experimental designs

Conclusions

“Diagnosis” comes from the Greek word meaning ”to tell apart”. This study suggests there is little value in separating out BIPS or BLIPS from the “brief psychotic illness” diagnostic constructs of BPD or ATPD from a prognostic point of view.

Treatment is another issue as it is not recommended to offer antipsychotics to BIPS/BLIPS but to try CBT instead (NICE Clinical Guideline 78, 2014). In contrast, antipsychotics will often be used to treat ATPD or BPD. The study makes the intriguing suggestion that ATPD/BPD/BLIPS/BIPS could all be amalgamated into one diagnostic category. This may imply that BIPS/BLIPS should be treated with antipsychotics but this would require research to confirm that the benefits outweigh the risks.

Similarly, this study suggests it is worth differentiating schizophrenia from ATPD/BPD/BIPS/BLIPS due to the higher recurrence rate of psychosis in FES. FES differs from the other diagnostic constructs by the potential presence of negative symptoms in the diagnostic criteria. In FES duration of symptoms (according to ICD-10) overlaps with BPD and ATPD.

An alternative to making a diagnosis of FES from ATPD/BPD/BLIPS/BIPS would be to use a dimensional system, but not just including multiple symptom dimensions but also a dimension of duration of symptoms. This type of system would require a great deal of time in the consultation so is unlikely to suit doctors, who usually have brief amounts of time to see their patients. Doctors are best suited to using a diagnostic system with adding on brief symptom dimensions if they have time.