Rates of psychosis in epilepsy may not be as high as previously reported, says new systematic review

For many years, psychiatry has highlighted that people with epilepsy appear to have an elevated risk for psychosis. However, studies exploring this relationship (of which there are many) seem to disagree on just what the prevalence of psychosis is in this group. For example, Gudmundsson (1966) interviewed every patient with epilepsy in Iceland and concluded a prevalence rate of psychosis of 7.2%, whereas Swinkels et al (2005) report a rate of just 0.48%.

The connection between these two conditions might be an important one in helping us to understand how psychosis develops within the brain’s structure and impacts upon it in the long term. Several theories have been proposed to explain the apparent connections between psychosis and epilepsy, but as yet, there has not been much consensus on how much of a risk factor epilepsy is for developing psychosis.

What we need is a systematic review! Cue Clancy et al (2014), who have undertaken the first systematic review of the prevalence of psychosis in epilepsy.

The reviewers aimed to estimate the prevalence of psychosis for different types of epilepsy in a variety of population groups

Methods

The aims of the systematic review were to estimate the prevalence of:

Psychosis in epilepsy
Psychosis in Temporal Lobe Epilepsy (TLE) in particular (because it is one of the most common forms of epilepsy)
Postictal and interictal psychosis
- Postictal psychosis is when psychotic symptoms develop at the time of a seizure or during the week following it and last for 24 hours up to 3 months
- Interictal psychosis is a psychotic syndrome that would otherwise fulfil the criteria for schizophrenia, except that there is epilepsy also present
- Psychosis in specific subgroups:
  - Patients who also have learning disability
  - Children and adolescents
  - Patients who also have a genetic vulnerability to psychosis

They also had an aim to:

Calculate an odds ratio to determine the exact level of risk epilepsy confers for developing psychosis

Clancy and colleagues searched four electronic databases for any studies published in English giving prevalence rates or data which would allow prevalence rates to be calculated. It’s unclear, but as far as the elves can tell from Clancy et al’s published paper, they did not seek out unpublished studies or use the references of the papers they found from their initial searches to uncover further papers. 58 papers were used in the final analyses.

Data was then extracted and analysed using a random-effects meta-analysis, weighted by study sample size, to give effect sizes for their first four aims. Odds ratios were also calculated.

Results

Clancy et al found that the prevalence rates of:

Psychosis in epilepsy was 5.6% (95% Confidence Intervals (CI) = 4.8 to 6.4)
Psychosis in TLE epilepsy specifically was 7.0% (95% CI = 4.9 to 9.1)
Interictal psychosis was 5.2% (95% CI = 3.2 to 7.2)
Postictal psychosis was 2.0% (95% CI = 1.2 to 2.8)

When it came to specific subgroups, very few papers were available, but they did find:

In the subgroup of patients with a comorbid learning disability, psychosis had a 7.4% prevalence (95% CI = 2.6 to 12.2; four studies)
In the child and adolescent subgroup, the prevalence of psychosis was 5.4% (95% CI = 0.6 to 10.2; three studies)
In epilepsy patients with a family history of psychosis, the prevalence was 5.4% (95% CI = 1.0 to 9.8; three studies).

Again, the number of available studies was low when analysing the risk of developing psychosis as an epilepsy patient. Using four studies which reported psychosis prevalence in both an epilepsy group and a control group, Clancy et al determined that the odds ratio for risk of psychosis among epilepsy patients was 7.83.

The number of available studies was low when analysing the risk for developing psychosis, and exploring subgroups.

Discussion

This review has produced some interesting results compared to the existing research in this field. Whilst the 7% prevalence rate for psychosis in TLE epilepsy fits with previous studies, the 5.6% pooled rate for psychosis in epilepsy is slightly lower than previous reports.

Lower still is the 2% prevalence rate of postictal psychosis in epilepsy. Clancy et al highlight that this form of psychosis has previously been described as the most common form of psychosis in epilepsy patients – at a rate of 2%, which now seems unlikely. However, this form of psychosis may also be one of the most difficult to gain an accurate picture of due to its transient presentation making diagnosis; challenging due to the symptoms being explained as a different condition, or not being presented to a healthcare professional at all.

Inaccuracies may have crept into the available data, not just because of the difficulty in diagnosing postictal psychosis. Clancy et al note that some of the older studies included in the review may have used less stringent diagnostic criteria for “schizophrenia” and “psychosis” than more recent papers, possibly increasing previous estimates of prevalence rates. This may explain why Clancy et al’s figures match or fall lower than previous studies, but does highlight a weakness of the review. Nine of the 58 included studies were published before 1990, and 22 were from before the year 2000; which means DSM-II and ICD-8 criteria were used in some studies.

In fact, there was a lot of variation between studies in some aspects. Some of this variation is a strength (such as the fact that not all studies came from Western sources), whilst other variation is less useful (such as the range of time periods assessed by different papers).

More importantly, however, is the fact that where there were similarities between studies, these threaten the representability of the findings. Most studies were:

cross sectional in design (77.6%), and based on samples from tertiary referral centres (71%) and these factors may limit the interpretation of the results with respect to the general population.

This means that it is unclear how accurate the 5.6% or 7% prevalence rates are to the wider population of epilepsy patients, but Clancy et al suggest that the rates found here may be over-estimates.

The usual problem of too few studies also reared its ugly head in this review. In attempting to analyse the prevalence of psychosis in different subgroups of epilepsy patients, Clancy et al ran into difficulties finding enough appropriate studies, particularly when exploring the influence of genetic vulnerability. This subgroup may be of particular interest given the proposal by some that epilepsy and psychosis could share a common aetiology. Unsurprisingly, then, Clancy et al conclude their review by calling for more family-based studies to rectify this, and suggest that future studies tighten up the diagnostic criteria of psychotic syndromes found in epilepsy.

So what have we made of this review? Well, it seems that the risk of developing psychosis is increased 8-fold if you have epilepsy compared to if you do not. However, this result and the figure of 5.6% prevalence of psychosis in epilepsy is to be taken with a fair pinch of salt because of diagnostic variability within studies, sample unrepresentativeness and plain ole lack of available data. It looks like Clancy et al have taken the first step but there might yet be a long way to go!