Preliminary application of processed electroencephalogram monitoring to differentiate senile dementia from depression


Dr Norihito Oshima MD, Office for Mental Health Support, Division for Counseling and Support, University of Tokyo, Hongo 7-3-1, Bunkyo, Tokyo, Japan. Email:


Background:  It is difficult, but important, to distinguish between dementia and depression in old age because senile depression has atypical symptoms, including cognitive impairment and memory disorder. Now brain computed tomography, magnetic resonance imaging, single photon emission computed tomography, and positron emission tomography can be used to differentiate between these two conditions. However, these methods are expensive and not always available. In the present case series, we assessed the potential of monitoring the bispectral index to distinguish between dementia and depression.

Methods:  A processed electroencephalogram monitor (bispectral index (BSI) monitor) was used to assess brain activity during relaxed wakefulness in 12 participants (seven with Alzheimer's disease (AD), three with depression, and two healthy volunteers). Each recording lasted 5 min and four variables (i.e. BSI, 95% spectral edge frequency, electromyogram activity, and signal quality index) were monitored.

Results:  The BSI was significantly smaller in AD patients than in patients with depression (P < 0.05) and the 95% spectral edge frequency tended to be lower in AD patients than in patients with depression (P = 0.26). Slow waves were found in patients with AD and beta waves were predominant in patients with depression and healthy volunteers.

Conclusion:  In conclusion, the BSI and 95% spectral edge frequency were slightly smaller in dementia patients than in patients with depression. Paroxysmal slow waves may account for the low bispectral index. Thus, BSI monitoring may become a useful tool with which to distinguish AD from depression.


It is difficult to distinguish dementia from senile depression; both are associated with depressive mood, apathy, and memory disorder. However, it is important to distinguish between them because memory disorder and cognitive impairment in senile depression is treatable. A diagnosis of dementia is currently based on psychological interview, examination results and brain imaging, namely magnetic resonance imaging (MRI), single photon emission computed tomography (SPECT), and recently developed amyloid positron emission tomography (PET).1,2 Brain imaging techniques are very powerful in differentiating between Alzheimer's disease (AD) and other diseases; however, they are inconvenient as an initial step. For the diagnosis of depression, physiological examinations and brain imaging, namely near-infrared spectroscopy, have been used3,4 and convenient methods are always required for use in routine clinical practice. For example, the usefulness of the electroencephalogram (EEG) has been reported recently. The EEG is normal in most patients with depression or pseudodementia, but is moderately to severely abnormal in dementia patients.5 However, to conduct an EEG analysis requires specialized knowledge, skills, and facilities.

The bispectral index (BIS) is an EEG parameter that was approved by the US Food and Drug Administration (FDA) in 1996 to monitor anesthetic effects in the brain.6 The BIS has steadily gained clinical acceptance as a reliable measure of the effects of anesthesia and sedation on the brain during anesthetic management and intensive care.

Monitoring of the BIS is simple, non-invasive, and requires only a few minutes to gain results. A previous study showed that slow wave activity in AD increases with disease progression.7 We expected that BIS may be helpful in distinguishing senile dementia from depression, because the BIS could reflect delicate changes in EEG activity.

However, there are a few studies regarding the use of BIS in psychiatry. Abnormal BIS values have been measured after electroconvulsive therapy (ECT) or epileptic seizures.8,9 In the present study we monitored and compared the BIS of patients affected by major depression or AD and found that it could be used to distinguish dementia from depression.


Patients with AD or major depression (meeting ICD-10 criteria; who visited the Psychiatry Outpatient Department in National Hospital Organization Hanamaki Hospital for the first time were enrolled in the present study. Exclusion criteria included a history of psychiatric disorder, head trauma or drug or alcohol abuse, and the use of psychoactive medications, including night-time benzodiazepines. The study protocol was approved by the ethics committee of Hanamaki Hospital and written consent was obtained from all participants.

Twelve participants (seven with AD, three with major depression, and two healthy volunteers) were enrolled in the study. Assessments included both the Mini-Mental State Examination (MMSE) and Hamilton depression scale (HAM-D) for patients with major depression and healthy volunteers, and the MMSE for AD patients.

The BIS is a processed EEG parameter represented by a numerical value that is derived from bispectral analysis, power spectral analysis, and time domain analysis data. The BIS appears to be correlated with the level of consciousness,10 with BIS values near 100 representing an ‘awake’ clinical state and 0 denoting a flat EEG.11 The BIS decreases during natural sleep, as well as during the administration of an anesthetic agent.12

In the present study, the BIS was recorded using an A-2000 EEG Monitor (Model A-2000, Bispectral Analysis ver. 3.12; Aspect Medical Systems, Natick, MA, USA). The EEG signal was acquired using disposable electrodes (BIS Sensor; Aspect Medical Systems) applied to the participant's forehead, as recommended by the manufacturer. The impedance of each electrode was <5 kΩ. The smoothing rate for BIS reports was set at 30 s. All participants were asked to lie down, be calm, and close their eyes. We confirmed the stability of the electrodes and started recording. The BIS was recorded during relaxed wakefulness for approximately 5 min and values from the last 3 min before termination of the recording were used for subsequent analysis. The variables recorded were BIS, 95% spectral edge frequency (SEF; which is the frequency at which 95% of the total power was at lower frequencies), electromyogram (EMG; power in decibels in the 70–110 Hz band corresponding to activity from the EMG) activity, and signal quality index (SQI; the percentage of good-quality EEG epochs entering the BIS system over the previous 1 min, scaled from 0 to 100 (optimal)).

The Mann–Whitney test was used for comparisons of BIS and 95% SEF. Statistical analyses were performed using spss 12.0 J (SPSS Japan, Tokyo, Japan).


The results are given in Table 1. The MMSE scores of all AD patients were under 21, but scores for patients with major depression and healthy volunteers were higher and within the normal range. The HAM-D score of patients with major depression was >18; in contrast, healthy volunteers showed low, normal scores. Because memory and cognition were impaired in AD patients, they could not be assessed by the HAM-D. However, none of the AD patients had any observable depressive mood, disinclination, or sadness.

Table 1.  Participant profile and results of bispectral index monitoring
  1. MMSE, mini-mental state examination; HAM-D, Hamilton depression scale; 95% SEF, 95% spectral edge frequency; BIS, bispectral index; EMG, electromyogram; SQI, signal quality index; AD, Alzheimer's disease; Dep, depression.


The EMG activity during recording of the BIS was <50. This means that the effect of the EMG on the BIS was small. However, the EMG of Case 6 was slightly high, so the results for this patient were excluded from analysis. The SQI of most patients was >70. This also confirms that the EEG signal we used for analysis was appropriate.

The BIS of AD patients ranged from 76.1 to 92.9 and was significantly less than the BIS values obtained for patients with major depression (P < 0.05). Values for 95% SEF tended to be lower in AD patients than in patients with major depression, but the difference did not reach statistical significance (P = 0.26). One of the patients with major depression (Case 9) had low BIS and 95% SEF alules.

Figure 1 shows EEG recordings from an AD patient (Fig. 1a), a patient with major depression (Fig. 1b), and a healthy volunteer (Fig. 1c). The 3–6 Hz slow waves are seen in the AD patient (Fig. 1a), but beta waves are dominant in both the patient with major depression and healthy volunteer (Fig. 1b,c). Figure 2 shows EEG recordings from Case 9.

Figure 1.

Screenshot of the bispectral index (BSI) monitor showing results obtained in (a) a patient with Alzheimer's disease AD patient (case 1), (b) a patient with senile depression (Case 8) and (c) a healthy volunteer (Case 11). The 3–6 Hz slow waves are seen in (a), but beta waves are dominant in the other two cases (b,c).

Figure 2.

Electroencephalogram of Case 9 (senile depression) showing 8–9 Hz alpha waves. Case 9 had a low bispectral index.


Depression is a common psychiatric disease in old age, with a prevalence of 22%.13 Senile patients affected by depression sometimes first visit the family doctor, who frequently diagnoses and treats their depression as senile dementia. However, senile depression is not easily diagnosed because of its atypical symptoms (e.g. hypochondria, autonomic symptoms).14 Thus, an easier and objective tool to distinguish dementia from senile depression would be desirable.

Senile depression and dementia are related. Many follow-up studies show that senile depression with cognitive impairment progresses to dementia.15,16 However, it is important to distinguish depression from dementia because almost all cognitive impairment in late-life depression is reversible.

In the present study, BIS and 95% SEF values were slightly smaller in AD patients than in patients with depression. In agreement with these results, a previous study showed that the mean BIS of a group of AD patients was 92.9 (i.e. lower than normal).17 One reason for the low baseline BIS in AD patients is the increase in slow wave activity concomitant with a decease in fast wave activity.18 In the present study, the BIS in patients with major depression (Cases 8 and 10) was almost normal, but the BIS in Case 9 was low. The EEG in Case 9 showed 8–9 Hz alpha waves, which suggests that a slow basic rhythm was one of the reasons for the low BIS in this patient. Generally, however, EEG abnormalities, and therefore abnormal BIS values, in depression are rare.5 The present study shows that a low BIS may be a useful marker for distinguishing AD from senile depression. Low 95% SEF is related to low BIS. Accordingly, AD patients with low BIS had low 95% SEF. The age of the participants in the present study ranged from 62 to 93 years, but studies have shown that age has no significant effect on BIS.17,19

In the present study, the SQI was over 70, indicating that the EEG signal we used for analysis was reliable, except in Case 12. The low SQI in Case 12 can be explained by high skin impedance due to sweating during recording. In addition, the EMG activity was below 50. An EMG ≥49 has a tendency to give rise to an artifactually high BIS.17 Thus, in the present study, the effect of EMG on BIS was not considerable, except in Case 6. In Case 6, EMG activity was high because the patient's severely impaired memory and cognition (MMSE score 4) prevented her compliance with our order to relax during recording.

The usefulness of the EEG has also been reported, with the EEG being normal in most patients with depression or pseudodementia, but moderately to severely abnormal in dementia patients.5 Delta power is decreased in depression and increased in posterior brain areas in AD.20 However, the conductance of these EEG analyses requires specialized knowledge, skills, and facilities. The BIS monitor is a portable, easier-to-use device that gives a quantitative value reflecting the level of consciousness and electrical brain activity. One study used BIS monitoring to evaluate mild cognitive impairment,21 but in the present study it was used it to distinguish between senile depression and dementia.

Regarding brain metabolism and BIS, it has been reported that in a study of volunteers receiving graded doses of anesthetics, PET revealed a significant, direct correlation between decreases in BIS and a reduction in whole-brain metabolic activity.22 This research suggests that reductions in cerebral metabolism caused by other factors will result in decreases in BIS. For example, physiological changes known to impact on cerebral metabolic activity, such as cardiac arrest and hypothermia, have been characterized by changes in BIS.23 A reduction in the brain metabolic rate in patients with advanced AD may cause low BIS values. Low brain metabolism has also been demonstrated in patients with depression, especially in the frontal lobe.24 However, the BIS in patients with depression is almost normal because the reduction in brain metabolic rate may be not so bad as to cause abnormal changes on the EEG.

Other methods used to differentiate senile depression from dementia include: (i) brain CT and MRI, which can detect frontal brain atrophy in some cases of depression and parietal and temporal atrophy in cases of AD;25 (ii) SPECT, which detects low frontal and temporal blood flow in patients with AD and pseudodementia and decreased frontal blood flow in patients with depression;26 and (iii) PET, which detects low blood flow and metabolism in the frontal lobe in patients with depression.24 However, the cost of these examinations is high and their availability is limited.

The results of the present study suggest that BIS monitoring has the potential to detect dementia and distinguish it easily from senile depression. However, there was only a small number of participants in the present study because we restricted the study to subjects without a history of pharmaceutical medication. Many depressive patients have already been given medication by their physicians. The use of BIS monitoring in such patients will require further clarification in a large-scale research study.