The psychiatric acute and intensive care units serve a broad spectrum of acutely admitted patients. A limited number of patients presents with brief depressive periods intervened with rapidly alternating psychiatric symptoms more typical of other psychiatric diagnoses than depression. These patients fail to meet current diagnostic criteria for affective disorders and the optimal treatment is not established. For the purpose of the present study we call this condition Acute Unstable Depressive Syndrome (AUDS) and the specific criteria are defined in the methods section.

It is important to study subsyndromes of depression with EEG and other imaging techniques as possible differential diagnoses in such patients may be psychiatric disorders associated with epilepsy. Psychiatric disorders associated with epilepsy seem to be clinically distinct [1], and affective disorders and depression, often with atypical clinical presentations failing to meet current DSM-IV criteria, are the most frequent [2, 3]. Studies of epilepsy or other organic brain dysfunctions in psychiatric acute departments are sparse [4]. However, it is important to identify patients with atypical depressive syndromes because treatment with traditional antidepressants might induce cycle acceleration in some patients with affective disorders [5, 6]. Recognition of organic brain syndromes, like epilepsy, is important in acute wards in order to identify patients who may benefit from treatment with antiepileptic drugs [7].

Electroencephalography (EEG) is the best method to diagnose epileptic seizures and syndromes. EEG may also be useful in treatment selection [8, 9] and classification [10, 11] of mood disorders. Subgroups with different familiar characteristics, clinical features, and prognosis have also been identified by EEG methods within a psychiatric population [12]. Reasonably accurate separations of patients with primary depression from healthy individuals with quantitative EEG (QEEG) multivariable neurometric methods has been repeatedly demonstrated and replicated in large samples by one research group [13].

We aimed to use prospective and blinded analysis of EEG and QEEG [14, 15] to investigate whether the AUDS group had more epileptiform and slow (delta and theta) activity in the frontal, temporal and posterior brain regions than major depressive patients in the course of an acute admission. We also aimed to study if regional alpha activity, posterior alpha frequency and regional QEEG asymmetry differed between the two groups. By this study we hoped to clarifiy the pathophysiology and nosological status of these unstable depressive patients.

More AUDS than MDE patients had multiple abnormal features in their EEG. Other studies, suggest that individuals with EEG abnormalities within the affective disorder group may have another symptomatology and aetiology of the disorder. [12]. These patients may have an organic basis for the disease [22–25]. Our findings accordingly suggest that more AUDS than MDE patients may have an organic origin to their symptoms.

QEEG delta amplitude was significantly higher in AUDS than MDE patients. QEEG is a more sensitive method than visual EEG inspection [15]. Delta activity is a relatively unspecific finding. Slow waves between 0.5 and 4 Hz prevail naturally during sleep [26]. Pathological polymorphic delta activity can be elicited by lesions of the subcortical white matter, the thalamus and the mesencephalic reticular formation [27] most likely produced by the pyramidal neurons of the cortex [28]. In cats, localized delta activity appears in cortex overlying a circumscribed white matter lesion, but may also result from a localized thalamic lesion. Unilateral diffuse delta activity appears on the side of thalamic or hypothalamic lesions, whilst bilateral delta activity results from bilateral lesions of the midbrain tegmentum [29]. Presumably, diminished cholinergic influence from cortical projecting neurons of the basal forebrain induces the delta activity [26]. Localized lesions of the cerebral cortex do not produce delta activity on the other hand [29]. Generally, either a structural lesion, a metabolic abnormality, or lesional epileptiform activity should be considered [30].

Increased asymmetry with higher delta power on the right side was observed in the AUDS group. Hemispheric asymmetry with right sided dominance found in psychosis has been attributed to left side deficiency with the right side becoming dominant because of a lack of interhemispheric suppression [31]. Epilepsy should also be considered, as recurrent brief depressive episodes have been described both in postictal dysphoria and interictal dysphoric disorder in epilepsy [32]. A temporal epileptogenic lesion tend to be associated with polymorphic slow activity or runs of rhythmic delta or theta activity [33], but neither EEG-pattern was observed. Visual EEG did not provide evidence for significantly increased epileptiform activity. Lesional epilepsy is accordingly less likely in AUDS.

Decreased delta activity in MDE patients, as opposed to increased activity in the AUDS group, could have contributed to the observed group difference [34]. However, as more clinical EEG abnormalities were also seen in AUDS than in MDE patients, it seems likely that the group difference originates from excess delta activity in AUDS patients. Healthy controls must also be studied with a larger cohort of MDE and AUDS patients to settle this question. In addition it cannot be ruled out that a depression-related factor in the AUDS group explains or contributes to the observed EEG asymmetry [35], as elevated right side delta power has been reported in depressed patients [36]. However this is less likely, as depression mostly has been associated with increased alpha and beta power and asymmetry rather than increased slowing [37, 38]. The use of other EEG-analytical methods seems to indicate a more diverse broad-band dysfunction in depressives [39]. Asymmetry was not found among unselected depressives in a recent blinded study [34].

The co-existing psychiatric symptoms in the AUDS patients could naturally contributed to the results. Coexistent anxiety may be linked with right frontal EEG dysfunction in depression [40]. John et al [31] have found power asymmetry, with more power in the right side in all or almost all bands to be a common feature of psychotic patients independent of whether the psychosis were due to schizophrenia, depression or alcohol as well as medication status. Half of the AUDS population was co-diagnosed with psychosis; hence this state may have contributed to the elevated asymmetry numbers. Although the nosological status of AUDS patients presently is not clear, our results also suggest that presence of co-morbidity should be paid even more attention to in future EEG-studies in depression.

Other explanations for EEG-slowing should also be considered. First, substance use may affect EEG. Slow EEG activity after an alcohol-related seizure has been observed [41, 42], but in the present material, alcohol use was similar between groups. Moreover, EEG recorded 8–10 days after admittance, (free from any pre-hospital medication influence) confirmed our main results. Second, antiepileptic drugs, like phenytoin and carbamazepine are known to cause EEG slowing [43] or QEEG slowing [44]. We had a possible medication bias in our material caused by the use of antiepileptic drugs used as psychiatric stabilizing agents in the study group. Treatment was necessary for ethical reasons. However, the group without antiepileptic medication alone had significantly more delta activity than the controls even though our subgroup analyses were not powered to detect such differences. The results are therefore most likely not produced by the drug administration.

The limited number of patients renders our results vulnerable to type II errors. We emphasize that independent confirmation is needed. It might be argued that pre-hospital circumstances are hard to control, and it is uncertain if a high sensitivity and specificity of QEEG can be achieved in a clinical setting. However, this study aimed at investigating patients "as they are" in a realistic acute psychiatric setting because the diagnostic challenge arises here, not in a controlled environment. We have used a well-matched control group admitted under equal circumstances as these "real life patients". EEG is nevertheless useful for the diagnostic process in acute psychiatric settings [10] e.g. for evaluation of disorders like epilepsy, dementia and encephalopathy. With the development of large, commercially available QEEG databases, some QEEG techniques may grow towards clinical utility [13].

Our results suggest that a group of patients with rapidly fluctuating psychiatric symptoms including depression have more abnormal EEG, compared to patients with MDE. The findings suggest more pronounced organic brain dysfunction in AUDS- than in MDE patients. Although the hypothesized association between epilepsy and AUDS was not statistically confirmed in the present study, a subgroup of AUDS patients had epileptiform activity in their EEG. The nosological status of these patients is not yet clear and a more detailed study of clinical and neuroradiological features is underway.