A Prospective Study of Postictal Psychoses with Emphasis on the Periictal Type


Address correspondence and reprint requests to Dr. T. Oshima at Department of Neuropsychiatry, Aichi Medical University, 21, Karimata, Yazako, Nagakute-cho, Aichi, 480–1195, Japan. E-mail: tomohirot@happytown.ne.jp


Summary: Purpose: To assess prospectively episodes of postictal psychosis.

Methods: We followed 108 consecutive patients with temporal lobe epilepsy, who were divided into three groups: those without psychotic episodes (n = 87, N group), those with interictal psychosis (n = 13, IIP group), and those with postictal psychosis (n = 8, PIP group). The first episode of postictal psychosis, which was defined as a psychotic episode that occurred within 1 week after the end or within 3 days before the beginning of seizure clusters, was assessed with the Brief Psychiatric Rating Scale (BPRS) and Social Dysfunction and Aggression Scale (SDAS) during the observation period.

Results: The duration of illness was significantly different between the N and PIP groups (p = 0.004) and between the N and IIP groups (p = 0.039). The average initial BPRS score (obtained 3.0 days after the end of the seizure cluster) was 19.7, and then decreased to 5.8 after 1 week, and finally normalized at 1.5 after 1 month. A statistically significant decrease in BPRS scores was found between the initial assessment and those obtained after 1 week (p = 0.011). Those who had psychotic episodes without a lucid interval tended to have episodes more often than monthly, and experienced additional seizure recurrence even during the psychotic episodes. Two patients exhibited a frank manic phase, and three patients showed excessively aggressive behavior, as determined by the SDAS.

Conclusions: Postictal psychosis should be subdivided into the nuclear type, with an established clinical picture as an indirect aftereffect of seizure activity, and the atypical periictal type, which is a direct manifestation of limbic discharge.

The concept of postictal psychoses, proposed by Logsdail and Toone (1988) became established in the 1990s, based on the results of several case–control studies (Devinsky et al., 1995; Kanner et al., 1996). Nearly all of the previous reports regarding postictal psychoses, including those of multiple cases, were retrospective, except for the study by Kanner et al. (1996), who investigated artificially induced episodes in patients waiting in seizure-monitoring units in preparation for epilepsy surgery. In addition, psychotic manifestations preceding seizure clusters were evident in a substantial number of patients who showed episodes of postictal psychosis, although no other previous studies have mentioned those. Such preictal psychiatric episodes in patients with postictal psychoses tend to be underestimated in institutions that lack participation by a psychiatrist in epilepsy treatment. In the present study, we assessed psychotic episodes, if any, that occurred before or after seizures in patients newly diagnosed with temporal lobe epilepsy. This study was performed to amplify the preliminary study presented by us previously (Oshima et al., 2005).


The subjects were patients referred to Dr. Kanemoto, one of the study authors, between January 2001 and December 2005, who were diagnosed as having temporal lobe epilepsy when they showed both temporal epileptiform discharge on EEG and complex partial seizures in principle, as a number of previous studies of postictal psychosis have pointed out a close relation between temporal lobe epilepsy and postictal psychosis (Kanemoto, 2002). Exceptionally, we also included patients who had typical limbic auras, such as déjà vu, according to international criteria defined in 1989, and complex partial seizures, even without temporal EEG foci. Overall, 108 patients were included in the present study.

Psychotic episodes were defined according to ICD-10 criteria and documented in 21 patients, who were assessed directly or before entry into the study. Of those 21 patients, eight had psychotic episodes that occurred within 1 week after the end or within 3 days before the beginning of seizure clusters. In that period, six of the eight patients experienced psychotic episodes, the first of which was assessed by using the Brief Psychiatric Rating Scale (BPRS) at the first clinical examination, and then 1 week and 1 month after the end of the seizure cluster, in addition to evaluations with the Social Dysfunction and Aggression Scale (SDAS). The BPRS has been widely used and is considered to be one of the best rating scales (Morlan and Tan, 1998). Because nonpsychotic symptoms, such as affective disinhibition and extreme anxiety, are intrinsic components of postictal psychosis, we used the total score of BPRS, which included such nonpsychotic symptoms, as an index of the severity of postictal psychosis. The SDAS, consisting of nine items covering outward aggression and two items (SDAS-2) covering inward aggression, was developed as an attempt to construct an observer scale for aggression and is analogous to the Hamilton scale for depression. The interobserver reliability of the SDAS has been shown to be adequate (Wistedt, 1990).

We used SPSS version 11 for statistical analysis, with a level of p < 0.05 used to determine statistical significance.



The initial 108 patients were 51 men and 57 women, in whom mean epilepsy onset was 16.1 years (SD, 10.9). The mean duration of epilepsy was 15.2 years before the study (SD, 10.9).

Psychotic episodes

No psychotic episodes occurred in 87 of the patients before entry into the study or during the follow-up period (N group), whereas 21 patients experienced at least one psychotic episode during the course of illness. In the latter group, interictal psychotic episodes occurred in 13 patients (IIP group), and postictal or periictal episodes occurred in eight (PIP group). The mean duration of illness for the N group was 13.3 years (SD, 12.3), whereas it was 21.0 years (SD, 12.3) in the IIP group and 25.8 years (SD, 8.6) in the PIP group. A one-way factorial analysis of variance with a Bonferroni test revealed a statistically significant difference between the N and PIP groups (p = 0.004) and between the N and IIP groups (p = 0.039).

Clinical courses of postictal psychoses

The average initial BPRS score (determined a mean of 3.0 days after the end of the seizure cluster) was 19.7 (n = 6), which decreased to 5.8 (n = 5) after 1 week and finally normalized to 1.5 (n = 6) after 1 month (Fig. 1). A statistically significant decrease in BPRS scores was noted between the initial assessment and the assessment after 1 week, as shown by a paired t test (t= 5.608; p = 0.011). In five of the cases, sedatives were given from day 1 to 3 after the end of the seizure clusters. In one of those cases, diazepam (DZP; 10 mg) was given intramuscularly only once, the effect of which was doubtful. In the other four, a combined administration of flunitrazepam (FZP; 4 mg) and levomepromazine (25–105 mg) was needed to sedate the extremely agitated or dangerously aggressive patients and was continued for 3 to 14 days. Although psychiatric symptoms were ameliorated visibly the day after the administration of the drugs in two cases, two patients remained under custody for 1 week despite intensive medication, because of aggression toward their surroundings.

Figure 1.

Results of BPRS. The average of BPRS of 1 week after the end of seizures was lower than that of first time (mean, 3.0 days) after the end of seizures (p < 0.05).

Two patients exhibited a frank manic phase that corresponded to 6 and 9 points in total, respectively, when measured by the total sum of the BPRS scores (8, 17, and 18), and three patients showed excessively aggressive behavior, which totaled 14 points, 15 points, and 17 points, respectively, as measured by the SDAS. In addition, three cases had lucid intervals, whereas the others experienced psychotic episodes without a lucid interval. Additional seizures occurred in the midst of psychotic episodes in three patients without a lucid interval. Although more than one psychotic episode occurred monthly in patients without a lucid interval, the other three patients with lucid intervals experienced psychotic episodes yearly, twice, and only once throughout their entire life. Further, although an initial brief episode of manic state preceded the psychotic episodes in two patients with lucid intervals, no manic phase was observed in the patents without a lucid interval.

All of the patients with PIP experienced auras, intractable complex partial seizures, and secondarily generalized seizures. By using MRI, left medial temporal sclerosis was found in two patients, right medial temporal sclerosis was found in one patient, and no apparent lesion was found in the others. Electroencephalographic examinations showed a left-sided epileptiform discharge in two patients, whereas a right-sided epileptiform discharge was found in the other three patients. Only one patient had a psychiatric family history in the third-degree relatives. No developmental delay was found in any of the patients with PIP.


Although the concept of postictal psychosis was introduced in the 19th century (Jackson, 1875), it did not become established as a clinical entity until the most recent decade, as a series of investigations in the 1990s dedicated to this topic revealed clinical features of the nuclear group of postictal psychosis, as follows. A lucid interval exists between the end of the seizure cluster and the appearance of psychosis. The time course of the episode is acute or subacute, and the latent period between the onset of epilepsy and the onset of psychosis exceeds 10 years (Kanemoto, 1995), which is longer than that of interictal psychosis. The correlation between postictal psychosis and temporal lobe epilepsy is even closer than that with interictal psychosis, and it is also closely related to mood disorders (Logsdail and Toone, 1988; Devinsky et al., 1995; Kanner et al., 1996).

The time course of the psychotic episodes was acute in all of the present cases. Further, emotional aspects were noticeable, and the latent period between epilepsy onset and psychosis onset proved to be longer than that in patients with IIP, which agrees well with the hallmarks of postictal psychosis listed earlier. Our semiquantitative follow-up revealed that, although a statistically significant decrease in psychotic activity was noted, as measured by BPRS scores as early as 1 week after the end of the seizure cluster, the scores did not return to their former state until 1 month after the cluster. This suggests that a trailing note of psychotic symptoms might linger even after apparent recovery, which may explain why some patients who have shown repeated PIP episodes become increasingly more eccentric, even during the interictal period, over time (Adachi et al., 2002).

The exact demarcation of the nuclear group of postictal psychosis highlights the presence of another type of psychotic episode, which appears to be closely related to seizure clusters, as in the nuclear group, but shows different clinical features. In the setting of artificially induced seizures during presurgical evaluations, neither psychotic symptoms before a seizure nor seizures during postictal psychotic episodes have been adequately documented. However, the present results, based on a prospective study protocol, demonstrated that both can occur in patients with postictal psychosis. Indeed, in the nuclear group of postictal psychosis, with patients who have a lucid interval and are characterized by an initial emotional outbreak, frank psychiatric manifestations never precede seizure clusters, and seizures never occur while psychotic symptoms prevail. This is in sharp contrast to the observations obtained in the present study. A possible explanation for this discrepancy could be twofold: artificially induced versus naturalistic ways of sampling and methodologic neglect of psychiatric manifestations before seizures as a result of excessive emphasis on the postictal nature of postictal psychoses.

So et al. (1990) confirmed increased spiking activities in the medial temporal region of patients with depth-EEG recording during episodes of PIP; however, they were skeptical about the causal relation between their findings and the occurrence of frank psychosis, because increased spiking is commonly observed as a postictal phenomenon in patients without psychosis. In a similar study using depth EEG, Mathern et al. (1995) failed to confirm increased postictal spike activity. In contrast, studies reported by Wieser et al. (1985) and Takeda et al. (2001) as well as our previous report (Kanemoto, 1997) demonstrated ictal epileptic activity occurring concurrent with postictal or periictal mental aberrations.

Our suggestion is that PIP should be subdivided (Kanemoto, 2002). Episodes belonging to the nuclear group of PIP are not likely to be attributed to direct seizure activity but may be caused by some rebound alterations in the neurotransmitter pathways in response to seizure activity, those occurring as preictal or with accompanied seizures even during ongoing psychotic states. This hypothesis might particularly apply to instances of prominent peculiar feelings, such as depersonalization, mental diplopia, or déjà vu, might be caused by circumscribed limbic status epilepticus. Although this suggestion is speculative, because the present sample size was too small for additional analysis with subdivided groups, we believe that it warrants further examination.


Acknowledgment:  This study was partially supported by the ILAE subcommittee chaired by Dr. Bettina Schmidt.