Previous quantitative EEG (QEEG) studies of spectral power derived from Fast Fourier Transformation¹ of the recorded background EEG signal in schizophrenia patients have revealed abnormalities in all the EEG frequency bands.²

Several studies have demonstrated a decrease in total alpha, low alpha, and high alpha power³—⁵ as well as reductions in alpha frequency⁶ in chronic schizophrenia. However, studies with neuroleptic-naive schizophrenia patients have shown higher alpha power in schizophrenia patients relative to comparison subjects.⁷,⁸ An increase in alpha1 power and a decrease in alpha2 activity in the posterior regions has been reported in untreated/drug-free schizophrenia patients realtive to healthy comparison subjects.⁷,⁹,¹⁰ Increased delta power has been shown in unmedicated/drug-free schizophrenia patients relative to healthy comparison subjects.⁶,⁹,¹⁰—¹² Even though some studies have shown that there is a frontal predominance of the delta increase in unmedicated schizophrenia patients during rest,¹³,¹⁴ others have reported that the delta increase is widespread and does not show a frontal accentuation.⁶,⁹,¹⁰,¹² Omori et al.⁷ did not find theta power abnormalities in never-treated schizophrenia patients relative to comparison subjects. Most previous studies have shown an increase in beta power in schizophrenia patients relative to comparison subjects,⁶,⁹,¹⁰,¹²—¹⁴ even though there have been reports of decreased beta power as well.¹⁵

The possible reasons for inconsistencies in spectral power findings in schizophrenia include confounding effects of medications, heterogeneity in symptom profiles of the study samples, chronicity, and gender differences of the patients studied. Differences in recording techniques and data analyses may also interfere with the comparability of results across studies. In a majority of studies, the EEG was recorded in medicated patients with varying degrees of chronicity of illness following a drug-free period. However, there are reported differences in EEG spectral power between drug-naive and medicated patients⁵,⁷,¹⁰,¹⁶ as well as between chronic and acute schizophrenia patients.⁴,⁸,⁹,¹⁷ Gender differences have also been reported in spectral power.¹⁸,¹⁹

Categorical subtyping and dimensional approaches have been proposed in the search for a satisfactory alternate conceptualization of schizophrenia psychopathology that could also reflect the pathophysiological heterogeneity of the disease, given the substantial symptom heterogeneity in schizophrenia.²⁰ Crow,²¹ Andreasen and Olsen,²² and Liddle²³ have been the pioneers of the above approaches. The most consistently replicated among the dimensional approaches is the three-dimensional model proposed by Liddle²³ comprising reality distortion (positive), psychomotor poverty (negative) and disorganization dimensions, whereas the classical type I (positive)-Type II (negative) dichotomy put forth by Crow²¹ and Andreasen and Olsen²² still holds the maximum heuristic value for subdividing schizophrenia into homogeneous symptom subgroups. EEG spectral power shows different profiles in the various subtypes of schizophrenia.⁸,²⁴,²⁵

In an attempt to overcome the above-mentioned limitations of previous studies of resting EEG in schizophrenia patients, we studied EEG spectral power in neuroleptic-naive, recent-onset schizophrenia (NRS) patients, and healthy comparison subjects in the resting eyes closed condition. We further subdivided the larger schizophrenia sample comprised of heterogeneous symptom profiles, into more homogeneous subgroups with predominance of positive or negative symptoms for spectral power comparisons, while controlling for gender effects. In our sample of NRS patients, we hypothesized that higher alpha1, lower alpha2 and higher delta power would be noted relative to comparison subjects. In view of the equivocal findings from previous research with respect to theta power and beta power in the resting eyes closed condition in neuroleptic-naive schizophrenia patients, we assumed the null hypothesis in those cases. Additionally, we hypothesized that, in keeping with putatively distinct neurobiological mechanisms underlying positive and negative syndromes,⁸,²⁴,²⁵ the spectral power would show characteristic profiles in the positive symptom and negative symptom subgroups.

Twenty-eight consecutively recruited patients fulfilling the DSM-IV criteria for schizophrenia or schizophreniform disorder²⁶ were selected for the study from among those who attended the outpatient department the National Institute of Mental Health and Neurosciences (NIMHANS), in Bangalore, India. The diagnosis of schizophrenia was arrived at using criteria from the DSM-IV based on the consensus of a research psychiatrist who conducted a semistructured interview and a trained research assistant who used the Mini International Neuropsychiatric Interview (MINI) Plus.²⁷ The inclusion criteria were: age between 17 and 45 years; age of onset of illness more than or equal to 17 years; duration from onset less than or equal to 3 years; right-handedness (as assessed by modified Annett’s inventory²⁸) and an Mini-Mental State Examination (MMSE) score of ≥21. The exclusion criteria were: other DSM-IV axis I disorders (as ascertained by administration of MINI-Plus); previous exposure to psychotropic drugs; significant suicidal or homicidal risk or other disruptive behavior which warranted immediate interventions; history of ECT within the previous 6 months; history of institutionalization, defined as continuous hospital stay for more than 6 months and prominent catatonic features.

Twenty-five randomly selected healthy participants, between the ages of 17 and 45 years old served as the comparison group. These participants, predominantly constituted by hospital staff and bystanders of hospitalized patients, were interviewed by an experienced psychiatrist and ascertained to have no present or past neuropsychiatric disorders; in addition, there was no history of psychotic disorders in their first degree relatives.

Both groups of participants were ascertained to have no unstable medical or neurological conditions, history of significant head injury or substance dependence (excluding tobacco) meeting DSM-IV criteria. Written informed consent was obtained from all the subjects after complete description of the study to the subjects and their significant relatives. The study was approved by the ethical review board of NIMHANS.

Both neuroleptic-naive, recent-onset schizophrenia (NRS) patients and healthy comparison subjects were asked to avoid alcoholic beverages for at least 3 days prior to the EEG recording as well as to avoid use of nicotine for at least 3 hours prior to the recording. In the case of patients, all of whom were hospitalized and accompanied by a close relative (as is the case with all patients in our setting), abstinence from the above substances was ensured under supervision by the accompanying relative as well as the nurse-in-charge. However, for both patients and comparison subjects, we did not carry out a urine drug screen prior to the EEG recording. Further, no specific instructions regarding caffeine consumption prior to EEG recording was given. The EEGs were recorded between 3 p.m. and 6 p.m.

The participants were seated in a comfortable reclining chair with a head rest in a dimly lit, sound-attenuated, and electrically shielded room. They were given a half hour to acclimatize with the conditions. The EEG Electro-cap was fixed and electro-conducting gel was applied for all the electrodes. Three minutes of EEG was acquired while the subjects were seated with their eyes closed in an alert, quiet, and relaxed state. The participants were instructed to minimize blinking and roving movements of their eyes.

The EEG recording was performed with a 32-channel bioamplifier system (band pass filter: 0.5—30 Hz) of the Nihon-Kohden type using the EEG-SYS software.²⁹ The amplifiers were calibrated with a 50 μV sin wave in the 4—8 Hz frequency band, generated by the Nihon-Kohden signal generator. The EEG was recorded from 30 cephalic locations (FP1, FP2, AF3, AF4, F5, F3, F1, Fz, F2, F4, F6, T1, FCz, T2, T3, C5, C3, C1, Cz, C2, C4, C6, T4, T5, P3, Pz, P4, T6, O1, O2) positioned according to the 10/20 International System (Figure 1). Eye movements were monitored using electrodes placed 1/2 inch lateral to the outer canthi of both eyes and connected to the horizontal oculogram channel, as well as with electrodes positioned 1/2 inch above and 1/2 inch below both eyes and connected to the vertical oculogram channel. The electronically linked averaged ear-lobe signal ([A1 + A2]/2) was used as the reference signal. The impedance of all electrodes was kept below 5 kilo ohms. A ground electrode was placed 3 cm anterior to Fz. The signals were digitized at a rate of 256 Hz and converted into ASCII format for further processing and analysis.

Baseline correction of the digital EEG signal was achieved using an autoregressive filter with coefficient 0.97. The 50 Hz and other high frequency noises were removed using a bidirectional 150-order Finite Impulse Response (FIR) digital filter with a bandwidth of 0.5—30 Hz (time constant: 0.3 second). Epochs contaminated by artifacts due to eye movements, eye blinking, drowsiness, muscle tension, sweat, or technical errors were excluded offline by careful visual inspection of the record by a researcher who was blind to the group status of the subjects. Prior to this, interrater reliability of artifact rejection was established. An average artifact-free EEG length of 75.43±26.05 seconds and 72.88±35.59 seconds was obtained in the comparison group and schizophrenia group, respectively. It has been shown by previous research that reliable EEG power values in all frequency bands could be obtained in both schizophrenia patients and healthy comparison subjects from a minimum of 32 seconds of artifact-free data.³⁰ For each participant, the artifact-free EEG signals were then subjected to Fast Fourier Transformation¹ by averaging over each 2-second epoch using a Hanning window to obtain the spectral power in the following nonoverlapping frequency bands: delta: 0.5 Hz—4.0 Hz; theta: 4.5 Hz—8.0 Hz; alpha1: 8.5 Hz—10 Hz; alpha2: 10.5 Hz—12.5 Hz; beta1: 13.0 Hz—17.5 Hz; beta2: 18.0 Hz—21.0 Hz; beta3: 21.5 Hz—30.0 Hz. These band widths were defined in accordance to the factor solution of Herrmann et al.,³¹ generated from resting eyes closed EEG. Both absolute and relative power (absolute power/total power) in all the bandwidths were computed, following which the weighted relative power for each band³² at each electrode position was computed by multiplying (i.e., weighting) the relative power value of a given band and site by the number of artifact-free 2-second epochs of EEG available within each 180-second eyes closed EEG record of every subject. The relative power was used for statistical analysis considering that NRS patients had significantly higher total power compared to healthy comparison subjects (Table 1). EEG signal processing and computation of power values were carried out using MATLAB, version 5.3.1.³³

The symptoms in schizophrenia patients were assessed using the Scale for the Assessment of Negative Symptoms (SANS)³⁴ and the Scale for the Assessment of Positive Symptoms (SAPS).³⁵ The psychopathology ratings for all the patients were carried out by one psychiatrist who had sufficient training and experience in administering the scales. The schizophrenia patient sample was subgrouped based on their SANS and SAPS scores into "positive" and "negative" according to the guidelines suggested by Andreasen et al.³⁶ Briefly, those subgrouped as positive schizophrenia had at least one of the following present to a marked degree (rating of ≥4 on SAPS): (1) hallucinations; (2) delusions; (3) formal thought disorder; or (4) bizarre behavior. Additionally, none of the following SANS items were present to a marked degree (rating of ≥4): (a) alogia; (b) affective flattening; (c) avolition-apathy; (d) anhedonia-asociality; or (e) attentional impairment. On the other hand, in those subgrouped as "negative" schizophrenia, at least two of the SANS items mentioned as (a)-(e) above were present to a marked degree, while none of the SAPS items mentioned as (1)-(4) above were present to a marked degree. The mixed subgroup (n=9) which included those patients who did not meet criteria for either positive or negative schizophrenia or who met criteria for both were excluded from subgroup analysis. Further, the psychopathological dimension scores were calculated from the total scores of SANS and SAPS subitems as follows: psychosis = delusions + hallucinations; negative = affective blunting + alogia + avolition + anhedonia; disorganization = formal thought disorder + bizarre behavior + inappropriate affect.³⁷ A more detailed description of the pattern of psychopathological dimensions observed in neuroleptic-naive, recent-onset schizophrenia patients from India is given elsewhere.³⁸

Normal distribution of the weighted relative spectral power values was assessed using Shapiro-Wilks test of normality. Since most of the power values were non-normally distributed, nonparametric statistical tests (Mann-Whitney U test and Kruskal-Wallis test) were used initially to make group and subgroup (Bonferroni corrected) comparisons of delta, theta, alpha1, alpha2, beta, and total power at individual electrode locations for purposes of graphical representation of power values in the individual bands.

Log10x-transformed weighted relative power in the total, delta, theta, alpha1, alpha2, and beta bands were analyzed using a repeated measures analysis of covariance (ANCOVA) model containing seven regions (left frontal, right frontal, central, left temporal, right temporal, left centro-parieto-occipital, and right centro-parieto-occipital), each comprised of four channels, with group and gender as between-subject factors, electrode locations as within-subjects factors, and age as covariate. This method of data analysis has been previously employed for analysis of individual frequencies of the resting EEG.³⁹,⁴⁰ Similarly, group comparisons were carried out between the symptom subgroups as well.

Following the above, a two-way analysis of covariance (ANCOVA) with group and gender as between-subject factors and age as covariate was used as a post-hoc assessment to analyze topographic group differences for log-transformed power at the seven scalp regions mentioned above. Secondary analyses also included assessing differences in relative weighted power in the seven scalp regions between the positive symptom subgroup, the negative symptom subgroup, and healthy comparison subjects using a similar procedure that included the Bonferroni correction for multiple comparisons. The statistical analyses were carried out using SPSS, Version 11.5.0.

The sociodemographic and clinical characteristics of the subjects are summarized in Table 2. The patient group and the comparison group did not differ significantly with respect to age and male:female ratio. However the NRS patients and healthy comparison (HC) subjects differed significantly (NRS