Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Schizophrenia Research: Cognition

journal homepage: http://www.schizrescognition.com/

The evolution of illness phases in schizophrenia: A non-parametric item response

analysis of the Positive and Negative Syndrome Scale
Anzalee Khan a,b,⁎, Jean Pierre Lindenmayer a,c, Mark Opler b,c, Mary E. Kelley d, Leonard White e,
Michael Compton f, Zimeng Gao b, Philip D. Harvey g,h
a
Nathan S. Kline Institute for Psychiatric Research, Psychopharmacology Research, Orangeburg, NY, USA
b
ProPhase LLC, New York, NY, USA
c
New York University School of Medicine, New York, NY, USA
d
Emory University, Department of Biostatistics and Bioinformatics, Atlanta, GA, USA
e
Mount Sinai School of Medicine, New York, NY, USA
f
George Washington University, Washington, DC, USA
g
University of Miami School of Medicine, Miami, FL, USA
h
Miami VA Healthcare System, Miami, FL, USA

a r t i c l e i n f o a b s t r a c t

Article history: Background: The Positive and Negative Syndrome Scale (PANSS) assesses multiple domains of schizophrenia.
Received 5 July 2013 Evaluation of each of these domains was conducted to assess differences in the characteristics of psychopa-
Received in revised form 20 January 2014 thology and their relative predominance in sub-populations.
Accepted 24 January 2014 Method: Subjects (N = 1,832) with DSM-IV schizophrenia were represented in three sub-populations: First
Available online xxxx
Episodes, n = 305, Chronic Inpatients, n = 694, and Ambulatory Outpatients, n = 833. Nonparametric
Item Response Analysis (IRT) was performed with Option Characteristic Curves (OCC), Item Characteristic
Keywords:
Schizophrenia subgroup
Curves (ICC), slopes and item biserial correlation. Items were characterized as Very Good, Good, or Weak
Psychopathology based on specified operational criteria for item selection.
Item response analysis Results: First episode patients were represented by negative, disorganized hostility and anxiety. Some nega-
tive domain items (Poor Rapport, Passive/Apathetic Social Withdrawal) and most positive domain items
were scored as Weak. For chronic inpatients, all items of the anxiety domain and some items of the positive
domain (Suspiciousness/Persecution, Stereotyped Thinking, Somatic Concerns) were Weak; for all other do-
mains, items were Very Good or Good. For ambulatory outpatients, most items in the anxiety and hostility
domain were scored as Weak. The majority of PANSS items were either Very Good or Good at assessing
the overall illness severity: chronic inpatients (73.33%, 22 items), first episodes (60.00%, 18 items), and
only 46.67% (14 items) in the ambulatory group.
Conclusion: Findings confirm differences in symptom presentation and predominance of particular domains
in subpopulations of schizophrenia. Identifying symptom domains characteristic of subpopulations may be
more useful in assessing efficacy endpoints than total or subscale scores.
© 2014 Elsevier Inc. All rights reserved.

1. Introduction
Schizophrenia has marked heterogeneity in symptoms. The cur-
rent symptom domains contained in the Diagnostic and Statistical
Manual for Mental Disorders (DSM-V) are psychosis, negative symp-
toms, disorganization, abnormal motor behavior and social/occupa-
tional dysfunction (American Psychiatric Association, 2013).
Although not part of the formal diagnostic criteria for the illness,
mood symptoms including depression and anxiety are common
in many patients (Green et al., 2003; Moller, 2005), hostility and
⁎ Corresponding author at: Psychopharmacology Research Program, Manhattan
Psychiatric Center, 1 Wards Island Complex, 15th Floor, Wards Island, NY 10035.
E-mail address: akhan@nki.rfmh.org (A. Khan).
belligerence are present in some cases (Chen et al., 2001), and cogni-
tive impairments are present in nearly all patients (Green et al., 2004;
Keefe, 2008). Symptoms as domains or dimensions have been
examined in detail with factor analytic procedures (e.g., White et al.,
1997a, 1997b), including factor analyses of the current dataset
(Kelley et al., 2013), and the domains examined by the PANSS have
been widely validated.
Different subsets of patients, defined by their stage or course of
illness or their overall outcome, have different predominant symp-
toms (Bengston, 2006). For example, at an acute phase, patients
with schizophrenia routinely come to clinical attention because
of the emergence of psychosis, often accompanied by social with-
drawal. Since these experiences are new and unsettling, anxiety
and depression would be expected to be substantially present as
well. A clear case of a subgroup defined by long term outcome is

http://dx.doi.org/10.1016/j.scog.2014.01.002
2215-0013/© 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
2 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx
difficult to discharge patients, who have been found to have ele-
vated levels of hostility, aggression, and positive psychotic symp-
toms (Bartels et al., 1991; White et al., 1997b). High levels of
disorganization and communication disorders are also associated
with this particular poor outcome subgroup (Davidson et al.,
1995). Conversely, successful adaption to living in the community
may be contingent on lower levels of disorganization, aggressive-
ness, and flagrant psychosis and patients with sustained communi-
ty tenure are also likely not to manifest substantial aggressive and
hostile behavior, which would bring them into contact with the
legal system or induce readmissions.
The long term course of schizophrenia suggests longitudinal
changes in patients with different outcomes developing (e.g., Chronic
inpatients vs. stable outpatients) and finding similarities within a dis-
order may help in the expansion of models to explain stages or the
course of the disorder. While the Positive and Negative Syndrome
Scale (PANSS) assesses multiple dimensions of schizophrenia, evalua-
tion of each of its individual domains has not been systematically
targeted at differences in the characteristics of psychopathology in
subpopulations. There is considerable interest in identifying the
course and treatment needs across different stages of illness. One
symptom domain that seems likely to be present across all of the dif-
ferent subgroups of patients is negative symptoms (see prevalence
review in Buchanan, 2007). These symptoms are temporally stable
in follow-up studies (Putnam et al., 1996) and are associated with im-
pairments in functional outcomes in both community dwelling and
institutionalized patients (see Chemerinski et al., 2006; and Harvey
et al., 2006 for a review). They are found to be present in many pa-
tients when other symptoms are in relative remission (particularly
in cases of the deficit syndrome). Negative symptoms are also present
at the time of the first episode (Lindenmayer et al., 1986; Milev et al.,
2005) and have been reported to be moderate or greater in severity in
a substantial proportion of community dwelling patients (Kurtz,
2005). Studies of older patients have reported even higher levels of
negative symptoms (Harris, 1991; Roseman et al., 2008), although
the longitudinal detection of change over the lifespan is challenging.
In a cross-sectional study comparing symptom severity in chronic pa-
tients across 8 decades (Davidson et al., 1995), negative symptoms
were more severe in older patients and manifested a greater age-
associated difference than positive symptoms. That said, studies of
older patients discharged from long-stay psychiatric care found great-
er improvements in negative symptoms than cognitive deficits post-
discharge (Leff and Trieman, 2000), implicating environmental fac-
tors to an extent.
In this paper, we present the results of an analysis of assessment
data from a large collated sample of people with schizophrenia, in-
cluding data from studies of first episode patients, community dwell-
ing patients, and long-stay patients from two New York State
Psychiatric facilities. These patients were all examined with a clinical
psychiatric rating instrument, the Positive and Negative Syndrome
Scale (PANSS) and for this paper we examined several features of
their clinical symptoms. Our hypotheses were that various domains
Table 1
Clinical description of data used in the investigation.
Chronic Patient Sample
Series Kay & Sevy, Caton et al., Bell et al.,
1990 1994, 1995 1994
Setting Inpatient Urban Veterans Hospital
Community Rehabilitation
n 239 400 150
Age 33.1 (10.2) 38.8 (10.6) 40.2 (8.6)
Mean (sd)
% Male 77 50 95
Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
of symptoms would be differentially prominent in different sub-
groups, as described above. Prominence was defined in terms of
symptom severity. Using Item Response Theory (IRT) models, we ex-
amined the extent to which an individual item contributed to the
overall severity scores for each domain and the extent to which
items were consistently sensitive to differing levels of severity for
each separate subgroup.
We hypothesized primarily that negative symptoms would be
found to be consistently validly measured and similarly prominent
in all three subgroups. We also hypothesized that institutionalized
patients would have more severe and validly measured symptoms
of hostility and disorganization, and psychosis, compared to the
other two groups. First episode patients were hypothesized to have
greater severity and measurement validity for anxiety/depression
and psychosis. Community dwelling patients were hypothesized to
be less impaired in other symptom domains (such as Anxiety and Dis-
organized domains), with resulting alterations in the patterns of do-
main structure and IRT findings.
2. Methods
2.1. Data source
This study uses data from 5 different observational studies (see
Table 1) aimed at cognition, functioning, and the course of illness
in people with schizophrenia. Subjects (N = 1,832) were all diag-
nosed with DSM-IV schizophrenia or schizoaffective disorder, and
were combined into three groups: First Episodes, n = 305, Chronic
Inpatients, n = 694, and Ambulatory Outpatients, n = 833. Studies
were carried out in accordance with the latest version of the Declara-
tion of Helsinki. Study procedures were reviewed by appropriate
ethics committees and informed consent (with specific exceptions
as seen below) was obtained after the procedures were fully
explained.
The First Episode group was defined as: consenting 18–45-year-
old patients who met DSM-IV criteria for schizophrenia, or
schizoaffective disorder for no more than 1 year prior to the assess-
ment and during which period they had no more than two psychiatric
hospitalizations for psychosis; and who did not have another axis I di-
agnosis, including substance dependence or abuse (Compton et al.,
2009; Compton et al., 2011). Diagnoses were determined using the
Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I;
First et al., 1998). Patients with first episode were not determined
by age but rather the course of symptom presentation and to ensure
that all courses during first episode (illness onset, episode onset,
end of episode, relapse of episode) were covered. The Chronic Inpa-
tient group was defined as: 18–85-year-old patients who met the
DSM-IV criteria for schizophrenia or schizoaffective disorder and
staying in a chronic psychiatric ward for N 6 months. This group was
examined with a waiver of signed informed consent because all pa-
tients in the hospital received the assessment and information was
First Episode
Sample
Davidson et al., Bowie et al., Harvey et al., Total
1995 2008 2010
Geriatric Outpatient Outpatient
inpatient
305 238 195 1527 305
75.7 (7.0) 56.6 (9.7) 44.0 (5.2) 48.9 (15.1) 23.6 (4.9)
44 73 69 64 73
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 3

then obtained from medical records. The Ambulatory group was de-
fined as: consenting 18–80-year-old outpatients who met the DSM-
IV criteria for schizophrenia, or schizoaffective disorder, who were
living in the community in a residence less restrictive than a nursing
home, and receiving maintenance drug treatment, at least with an an-
tipsychotic agent as the main drug treatment.

2.2. Instrument

The PANSS (Kay et al., 1987) is a 30-item rating instrument evaluat-

ing the presence/absence and severity of Positive, Negative and General
Psychopathology of Schizophrenia. All 30 items are rated on a 7-point
scale (1 = absent; 7 = extreme). The PANSS is administered by a cli-
nician trained in psychiatric interview techniques, with experience
Fig. 1. Example of an ‘ideal’ option characteristic curve (OCC). Note: Symptom severity
working with populations with Schizophrenia (e.g. psychiatrists, men-
refers to the total PANSS score of the domain being measured.
tal health professionals) and takes approximately 45 minutes to com-
plete. PANSS interviews were conducted by trained interviewers who
had at least 1 year experience using the PANSS. The PANSS-derived
PANSS-derived domains (positive, negative, disorganized thought,
Marder domain score was used. The study by Marder et al. (1997) fac-
hostility/excitement, anxiety/depression) and to examine the ability
tor analyzed the PANSS scores and produced five dimensions: negative
of each domain items to differ with levels of severity. A nonparamet-
symptoms, positive symptoms, disorganized thought, uncontrolled
ric kernel smoothing method and software (TestGraf), developed by
hostility/excitement, and anxiety/depression. All assessments were
Ramsay (2000) to estimate Option Characteristic Curves (OCC) was
performed with raters who were trained prior to rating, with the train-
used. The smoothing parameter was selected conditional on the bal-
ing described in the individual publications. See Table 1 for a descrip-
ance desired between bias and the variance of estimation, two com-
tion of the samples of patients in the study.
ponents of the mean square error of the estimator in Item
Characteristic Curves (ICC) estimation (Härdle, 1990; Ramsay,
2.3. Data analysis
2000). These methods have been used previously in studies on the
performance of PANSS items (Khan et al., 2011; Santor et al., 2000).
2.3.1. Assessment of unidimensionality of the PANSS-derived domains
TestGraf software (Ramsay, 2000) was used to fit the model. TestGraf
One important assumption of non-parametric IRT is that the con-
produces OCCs, ICC, an Item Information Function (IIF), and a Test In-
struct being measured (i.e., the domains of psychopathology) is uni-
formation Function (TIF) to assess the measurement precision of each
dimensional, meaning that the covariance among the items can be
item in each domain across the range of severity.
explained by a single underlying dimension. The percentage of the
Option Characteristic Curves (OCC) are graphical representations
total variance explained by the first component is regarded as an
of the probability of rating the different scores on the PANSS for a
index of unidimensionality. A Principle Components Analysis (PCA)
given item across the range of domain severity (or the latent variable,
without any rotation was conducted to assess unidimensionality as
theta H). Therefore, the probability of choosing a particular response is
follows: (1) a PCA was conducted on each of the five domains of
plotted against the range of symptom severity. If the probability of
the PANSS for each group (First Episode, Chronic Inpatients, Ambu-
rating an item changes as a function of symptom severity, the option
latory); (2) the variance explained by the first component produced
is useful; that is, it discriminates differences in domain specific symp-
by the PCA was examined; (3) if the variance explained by the first
tom severity. For OCC, individuals were ranked according to a maxi-
component was N20.00%, unidimensionality was assumed (see
mum likelihood estimation of their expected total scores on the
Reckase (1979) for methods of assessing unidimensionality using
derived symptom domain. Fig. 1 presents a graphical description of
PCA). Suitability of the data for factor analysis was tested by
the OCC of an ‘ideal’ item.
Bartlett's Test of Sphericity (Bartlett, 1954), which should be signif-
To validate that items of the symptom domain are either Very Good
icant, and the Kaiser-Meyer-Olkin (KMO) measure of sampling ade-
or Good at assessing the overall severity, Item Characteristic Curves
quacy, which should be N0.50 (Kaiser, 1970, 1974). It is important
(ICC) were examined. ICCs are graphical illustrations of the expected
to have unidimensionality to ensure that all items in the domain
total domain score on an item as a function of overall symptom sever-
are adding to the overall construct being measured. Multi-
ity. Fig. 2 displays an example of an item with an ideal ICC. Both OCCs
dimensionality would indicate that items within the domain are
and ICCs were using a Gaussian kernel smoothing technique of the
measuring more than one construct. However, it should be noted,
a set of items may have multiple eigenvalues greater than 1 and
still be sufficiently unidimensional for analysis with IRT (Orlando
and Thissen, 2000).

2.3.2. Non-parametric item response analysis

Nonparametric IRT models (Petersen, 2004; Sijtsma & Molenaar,
2000) provide a broad-spectrum and flexible data analytic framework
for investigating a set of polytomously scored items and for determin-
ing ordinal scales for measurement that include items that have
changeable locations and sufficient discrimination power (Sijtsma et
al., 2008). A nonparametric approach to modeling responses for the
PANSS items would allow for no a priori expectation about the form
of rating distributions, and items with non-monotonic item response Fig. 2. Example of an ‘ideal’ item characteristic curve (ICC). Note: Expected score refers
functions can be identified. The IRT examines the probability of to the total PANSS score on the domain being measured. Item Score refers to the 7 PANSS
choosing each response (PANSS item score) in relation to severity of option scores.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
4 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Table 2
Distribution of baseline score for PANSS-derived domains.

Marder Domain n Mean Std. Error Std. Deviation Variance Range

Anxiety First episode 305 10.8375 .31372 3.97 15.75 4,21

Chronic inpatient 694 8.2320 .14288 3.76 14.17 4,19
Ambulatory chronic outpatients 831 8.4481 .12880 3.69 13.59 4,22
Disorganized First episode 305 18.5750 .45094 5.70 32.54 7,38
Chronic inpatient 694 21.2017 .29482 7.77 60.32 7,46
Ambulatory chronic outpatients 831 13.5665 .17686 5.06 25.62 7,32
Hostility First episode 305 9.8063 .34958 4.42 19.55 4,22
Chronic inpatient 694 8.8012 .15308 4.03 16.26 4,27
Ambulatory chronic outpatients 831 6.1368 .10120 2.90 8.39 4,24
Negative First episode 305 21.3125 .53854 6.81 46.41 7,42
Chronic inpatient 694 20.8040 .31043 8.18 66.88 7,46
Ambulatory chronic outpatients 831 14.7192 .22224 6.36 40.45 7,37
Positive First episode 305 28.7063 .45855 5.80 33.64 14,47
Chronic inpatient 694 23.2579 .27161 7.16 51.20 8,43
Ambulatory chronic outpatients 831 18.2759 .25294 7.24 52.40 8,42

expected total score on an item based on the distribution of overall for the biserial correlation is +1 to − 1. A correlation is “medium”
PANSS-derived scores. Items' OCCs and ICCs were then examined, and at 0.30–0.49 and “large” at (0.50–1.00). Ratings for Criterion 6: A
items with Weak discrimination were identified. biserial correlation of ≥ 0.50 (i.e., large).
Similar to Khan and colleagues (2011) and adding an examination
2.3.3. IRT based item selection of the biserial correlation, items were scored as Very Good if all of the
To assess measurement precision dependent on the latent trait six criteria were fulfilled. Items were scored as Good if at least ≥ 4 of
(i.e., H, PANSS-derived domains), the Item Information Function (IIF) the 6 criteria were met. Items were scored as Weak, if they ful-
was plotted, indicating the range over the total severity of symptoms filled ≤ 3 of the 6 criteria. Items judged as Weak were considered as
(H) and to assess the amount of information about the specific domain contributing least to the domain symptomatology of the specific
(e.g. positive domain) that is provided by each item. The sum of the schizophrenia subgroup examined. All items were reviewed and
IIFs is provided as the Test Information Function (TIF), represented scored by two psychometricians, independently, and discrepancies
as I(H), indicating the amount of information in the scale about the in ratings were reviewed in a consensus meeting.
specific symptom at various severity levels. Items were characterized
as Very Good, Good, or Weak based on the following operational
criteria (adapted from Khan et al., 2011) and examination of the 3. Results
Item biserial correlation.
Criterion 1: The extent to which OCCs increase rapidly with changes Data analysis included PANSS item scores from 1,840 patients (2
in overall symptom severity. Basis of rating: Examination of the OCCs. patients were removed from the Ambulatory group due to missing
Ratings for Criterion 1: The probability (y-axis of OCC curve) of PANSS item data). Scores for each PANSS-derived Marder domain
selecting an option increases with increasing levels of severity, e.g., score is provided in Table 2.
the probability of option 2 being selected doubles from 0.5 to 0.25
when severity increases from a score of 12–18 (based on Fig. 1). 3.1. Unidimensionality of domains
Criterion 2. The region in which each option is more likely to be
selected is ordered, left to right, in accordance with their option Principle Components Analysis, with no rotation, was performed
scores on the OCC graphs. Basis for rating: Examination of the OCCs. for each schizophrenia group. For all three groups, all five PANSS-
Ratings for Criterion 2: The severity regions and corresponding se- derived domains had N 20.00% variance explained in the first
verity scores, e.g., the region in which option 2 is most likely to be se-
lected, falls between the regions in which option 1 and option 3 are Table 3
Eigenvalues of PANSS-derived domains (without rotation).
most likely to be selected.
Criterion 3. Options for an item span the full continuum of severity First episode KMO Bartlett's # % variance Unidimensionality
from the lowest score to the highest score. Basis of rating: Examina- test of Components of first assumed
sphericity loading component
tion of the OCCs.
Ratings for Criterion 3: For a particular item, all seven options span Anxiety 0.514 p b 0.001 2 35.749% Yes
the entire range of severity (e.g., from 5 to 45 in Fig. 1). Disorganized 0.621 p b 0.001 3 32.538% Yes
Hostility 0.500 p b 0.001 2 32.401% Yes
Criterion 4. There is a range of severity in which items are Negative 0.600 p b 0.001 3 26.975% Yes
expected to be scored. This is represented by the number of scores Positive 0.660 p b 0.001 2 28.343% Yes
(1–7) for which the item was more likely to be scored than all other Chronic
options. Basis of rating: Examination of the ICCs. inpatients
Anxiety 0.646 p b 0.001 1 48.941% Yes
Ratings for Criterion 4: Items for which ≥ 5 of the 7 choices are
Disorganized 0.673 p b 0.001 2 33.284% Yes
considered acceptable. Hostility 0.530 p b 0.001 2 41.231% Yes
Criterion 5. The steeper a slope of the ICC, the more discriminant Negative 0.642 p b 0.001 3 32.257% Yes
the item is. Slopes were computed in TestGraf for ICC graphs of each Positive 0.721 p b 0.001 2 34.643% Yes
item from the median option score (i.e., four (Moderate) on the Ambulatory
Anxiety 0.638 p b 0.001 1 44.365% Yes
PANSS). Basis of rating: Slope of the ICCs. Ratings for Criterion 5: Disorganized 0.663 p b 0.001 2 30.197% Yes
Slope is expected to be ≥0.40. Hostility 0.547 p b 0.001 2 36.823% Yes
Criterion 6. Biserial correlations were examined. Biserial correla- Negative 0.671 p b 0.001 2 34.141% Yes
tions are classical test theory estimates of item discriminating. In Positive 0.709 p b 0.001 3 33.886% Yes
TestGraf, the biserial correlation is the correlation between an item KMO: Kaiser-Meyer-Olkin measure of sampling adequacy. Bartlett's Test of Sphericity
and overall symptom psychopathology. The possible range of values significant (p b .001).

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 5

Table 4
Item selection for First Episode patients derived from the Item Response Analysis of the PANSS.

Criteria met (Yes/No) Overall rating

Based on OCCs Based on ICCs Biserial r

1 2 3 4 5 6

Anxiety
G2. Anxiety Yes Yes Yes Yes No 0.795 (Yes) Good
G3. Guilt Feelings Yes Yes Yes No Yes 0.651 (Yes) Good
G4. Tension No No Yes No Yes 0.645 (Yes) Weak
G5. Depression Yes Yes Yes Yes Yes 0.769 (Yes) Very good
Disorganized
P2. Conceptual Yes Yes Yes Yes Yes 0.660 (Yes) Very good
Disorganization
N5. Difficulty in Abstract Yes Yes Yes Yes Yes 0.606 (Yes) Very good
Thinking
G5. Mannerisms No No Yes No No 0.618 (Yes) Weak
and Posturing
G10. Disorientation Yes No Yes No No 0.479 (No) Weak
G11. Poor Attention Yes Yes Yes Yes Yes 0.733 (Yes) Very good
G13. Disturbance of Volition Yes Yes Yes Yes Yes 0.649 (Yes) Very good
G15. Preoccupation Yes Yes Yes Yes Yes 0.716 (Yes) Very good
Hostility
P4. Excitement Yes Yes Yes No Yes 0.698 (Yes) Good
P7. Hostility Yes Yes Yes Yes Yes 0.820 (Yes) Very good
G8. Uncooperativeness Yes No Yes No No 0.774 (Yes) Weak
G14. Poor Impulse Control Yes Yes Yes Yes Yes 0.799 (Yes) Very good
Negative
N1. Blunted Affect Yes Yes Yes Yes Yes 0.660 (Yes) Very good
N2. Emotional Withdrawal Yes Yes Yes Yes Yes 0.606 (Yes) Very good
N3. Poor Rapport No No Yes Yes No 0.618 (Yes) Weak
N4. Passive/Apathetic No No Yes Yes No 0.479 (No) Weak
Social Withdrawal
N6. Lack of Spontaneity Yes Yes Yes Yes Yes 0.733 (Yes) Very good
and Flow of Conversation
G7. Motor Retardation Yes Yes Yes Yes Yes 0.649 (Yes) Very good
G16. Active Social Yes Yes Yes Yes Yes 0.716 (Yes) Very good
Avoidance
Positive
P1. Delusions No No Yes No Yes 0.433 (No) Weak
P3. Hallucinatory Yes Yes Yes No Yes 0.501 (Yes) Good
Behavior
P5. Grandiosity No No Yes No No 0.471 (No) Weak
P6. Suspiciousness/ No No Yes No No 0.560 (Yes) Weak
Persecution
N7. Stereotyped Thinking No No Yes No No 0.532 (Yes) Weak
G1. Somatic Concerns No No Yes No No 0.125 (No) Weak
G9. Unusual Thought No Yes Yes No Yes 0.596 (Yes) Good
Content
G12. Lack of Judgment No No Yes No No 0.317 (No) Weak
and Insight

Biserial r = Biserial correlation.

component and Bartlett’s Test of Sphericity was significant for each
component in each group (all p values b .001). See Table 3 for the
variance accounted for in each of the first principal component for
each domain in each of the three groups.
3.2. PANSS-derived domain performance
3.2.1. First episodes
3.2.1.1. Option Characteristics Curves. OCCs are used to rate criterion 1–3.
Figs. 3, 4, and 5 shows OCCs for items for First Episode patients on Anxiety
and Hostility domains, Negative and Disorganized domains, and the Pos-
itive domain, respectively. In general, the OCCs for option 1 (absent) were
clearly less likely to be rated than were other options at higher severity
scores. Option 7 (extreme) was also used infrequently; the range of dis-
crimination was above the 95th percentile for all items. Items not meeting
Criteria 1 are presented in Table 4.
All items met Criterion 2, except, Anxiety domain (Tension), Hostility
domain (Uncooperativeness), Negative domain (Poor Rapport, Passive/
Apathetic Social Withdrawal), Disorganized domain (Disorientation,
Mannerisms and Posturing), Positive domain (Delusions, Stereotyped
Thinking, Somatic Concerns, Grandiosity, Suspiciousness and Persecution,
Lack of Judgment and Insight), which did not receive a score of “Yes.”
3.2.1.2. Item characteristic curves. ICCs were used to rate Criterion 4.
Figs. 3, 4, and 5 shows ICCs for items for First Episode patients on Anxiety
and Hostility domains, Negative and Disorganized domains, and the Pos-
itive domain, respectively. Table 4 presents items for which Criterion 4
was not met.
For Criterion 5 and 6, the slopes and biserial correlations of all
items are presented in Table 4.
3.2.2. Chronic inpatients
3.2.2.1. Option characteristics curves. Figs. 6, 7, and 8 shows OCCs for
items for Chronic inpatients on Anxiety and Hostility domains, Nega-
tive and Disorganized domains, and the Positive domain, respectively.
Similar to First Episode patients, the OCC for option 1 (absent) were
clearly less likely to be rated than were other options at higher se-
verity scores. Option 7 (extreme) was also used infrequently; the

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
6 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Table 5
Item selection for Chronic inpatients derived from the Item Response Analysis of the PANSS.

Criteria met (Yes/No) Overall rating

Based on OCCs Based on ICCs Biserial r

1 2 3 4 5 6

Anxiety
G2. Anxiety Yes Yes Yes No No 0.852 (Yes) Good
G3. Guilt Feelings No No Yes No No 0.706 (Yes) Weak
G4. Tension No No Yes No No 0.714 (Yes) Weak
G5. Depression No No Yes No No 0.761 (Yes) Weak
Disorganized
P2. Conceptual Yes Yes Yes Yes Yes 0.642 (Yes) Very good
Disorganization
N5. Difficulty in Yes Yes Yes Yes Yes 0.689 (Yes) Very good
Abstract Thinking
G5. Mannerisms and Yes Yes Yes Yes Yes 0.645 (Yes) Very good
Posturing
G10. Disorientation Yes Yes Yes Yes Yes 0.788 (Yes) Very good
G11. Poor Attention Yes Yes Yes Yes Yes 0.781 (Yes) Very good
G13. Disturbance of Volition Yes Yes Yes Yes Yes 0.649 (Yes) Very good
G15. Preoccupation Yes Yes Yes Yes Yes 0.640 (Yes) Very good
Hostility
P4. Excitement Yes Yes Yes No No 0.691 (Yes) Good
P7. Hostility Yes Yes Yes Yes Yes 0.813 (Yes) Very good
G8. Uncooperativeness Yes Yes Yes Yes Yes 0.740 (Yes) Very good
G14. Poor Impulse Control Yes Yes Yes No No 0.812 (Yes) Good
Negative
N1. Blunted Affect Yes Yes Yes Yes Yes 0.724 (Yes) Very good
N2. Emotional Yes Yes Yes Yes Yes 0.864 (Yes) Very good
Withdrawal
N3. Poor Rapport Yes Yes Yes Yes Yes 0.772 (Yes) Very good
N4. Passive/Apathetic Yes Yes Yes Yes Yes 0.859 (Yes) Very good
Social Withdrawal
N6. Lack of Spontaneity Yes Yes Yes Yes Yes 0.853 (Yes) Very good
and Flow of Conversation
G7. Motor Retardation Yes Yes Yes Yes No 0.639 (Yes) Good
G16. Active Social Yes Yes Yes Yes Yes 0.509 (Yes) Very good
Avoidance
Positive
P1. Delusions Yes No Yes Yes Yes 0.513 (Yes) Good
P3. Hallucinatory Behavior No No Yes Yes No 0.585 (Yes) Weak
P5. Grandiosity Yes No Yes Yes Yes 0.608 (Yes) Good
P6. Suspiciousness/ No Yes Yes No No 0.426 (No) Weak
Persecution
N7. Stereotyped Thinking No No Yes No No 0.516 (Yes) Weak
G1. Somatic Concerns No No Yes No No 0.338 (No) Weak
G9. Unusual Thought Yes Yes Yes Yes Yes 0.761 (Yes) Very good
Content
G12. Lack of Judgment No No Yes Yes Yes 0.337 (No) Weak
and Insight

Biserial r = Biserial correlation.

range of discrimination was above the 95th percentile for all items.
Items that did not meet Criterion 1 and 3 are presented in Table 5.
For Criterion 2, all items, except Anxiety domain (Tension, Guilt
Feelings, Depression), and items from the Positive domain (Delusions,
Stereotyped Thinking, Hallucinatory Behavior, Somatic Concerns,
Grandiosity, Suspiciousness and Persecution, Lack of Judgment and
Insight), did not meet the requirements for a score of “Yes.”
3.2.2.2. Item Characteristic Curves. Figs. 6, 7, and 8 shows ICCs for items
for Chronic inpatients on Anxiety and Hostility domains, Negative and
Disorganized domains, and the Positive domain, respectively. For Cri-
terion 4, at least five options were selected (see y-axis to the highest
point on the ICC), for all items except the following: all items of the
Anxiety domain (Anxiety, Tension, Guilt Feelings, Depression), two
items of the Hostility domain (Excitement, Poor Impulse Control),
and three items of the Positive domain (Stereotyped Thinking, Suspi-
ciousness and Persecution, Somatic Concerns).
For Criterion 5 and 6, the slopes and biserial correlations of all
items are presented in Table 5.
3.2.3. Ambulatory outpatients
3.2.3.1. Option Characteristics Curves. Figs. 9, 10, and 11 show OCCs for
items for Ambulatory outpatients on Anxiety and Hostility domains, Neg-
ative and Disorganized domains, and the Positive domain, respectively.
Similar to First Episode and Chronic inpatients, the OCC for option 1 (ab-
sent) were clearly less likely to be rated than were other options at higher
severity scores. Option 7 (extreme) was also used infrequently; the range
of discrimination was above the 95th percentile for all items. Items that
did not meet Criterion 1, 2 and 3 are presented in Table 6 (See Table 7).
3.2.3.2. Item Characteristic Curves. Figs. 9, 10, and 11 show ICCs for items
for Ambulatory outpatients on Anxiety and Hostility domains, Negative
and Disorganized domains, and the Positive domain, respectively. For Cri-
terion 4, all items of the Anxiety domain, all items of the Hostility domain,
four items of the Disorganized domain (Difficulty in Abstract Thinking,
Disorientation, Mannerisms and Posturing, Poor Attention, Preoccupa-
tion) and four items of the Positive domain (Grandiosity, Suspiciousness
and Persecution, Unusual Thought Content and Active Social Avoidance)
were met.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 7

Table 6
Item selection for Ambulatory outpatients derived from the Item Response Analysis of the PANSS.

Criteria met (Yes/No) Overall rating

Based on OCCs Based on ICCs Biserial r

1 2 3 4 5 6

Anxiety
G2. Anxiety No No Yes No No 0.701 (Yes) Weak
G3. Guilt Feelings Yes Yes Yes No No 0.700 (Yes) Weak
G4. Tension No No Yes No No 0.715 (Yes) Weak
G5. Depression No No Yes No No 0.698 (Yes) Weak
Disorganized
P2. Conceptual Yes Yes Yes Yes Yes 0.861 (Yes) Very good
Disorganization
N5. Difficulty in No No Yes No Yes 0.688 (Yes) Weak
Abstract Thinking
G5. Mannerisms No No Yes No No 0.600 (Yes) Weak
and Posturing
G10. Disorientation No Yes Yes No No 0.536 (Yes) Weak
G11. Poor Attention No No Yes No No 0.539 (Yes) Weak
G13. Disturbance Yes Yes Yes Yes Yes 0.623 (Yes) Very good
of Volition
G15. Preoccupation No No Yes No No 0.498 (No) Weak
Hostility
P4. Excitement No No Yes No No 0.601 (Yes) Weak
P7. Hostility No No Yes No No 0.606 (Yes) Weak
G8. Uncooperativeness No No Yes No No 0.568 (Yes) Weak
G14. Poor Impulse No No Yes No No 0.369 (No) Weak
Control
Negative
N1. Blunted Affect No No Yes Yes Yes 0.513 (Yes) Good
N2. Emotional Yes Yes Yes Yes Yes 0.777 (Yes) Very good
Withdrawal
N3. Poor Rapport Yes Yes Yes Yes Yes 0.714 (Yes) Very good
N4. Passive/Apathetic Yes Yes Yes Yes Yes 0.723 (Yes) Very good
Social Withdrawal
N6. Lack of Spontaneity Yes Yes Yes Yes Yes 0.700 (Yes) Very good
and Flow of
Conversation
G7. Motor Retardation Yes Yes Yes Yes Yes 0.745 (Yes) Very good
G16. Active Social Yes Yes Yes No Yes 0.689 (Yes) Good
Avoidance
Positive
P1. Delusions Yes No Yes Yes Yes 0.756 (Yes) Good
P3. Hallucinatory Yes Yes Yes Yes Yes 0.751 (Yes) Very good
Behavior
P5. Grandiosity No Yes Yes No No 0.608 (Yes) Weak
P6. Suspiciousness/ No No Yes No No 0.512 (Yes) Weak
Persecution
N7. Stereotyped Thinking Yes No Yes Yes Yes 0.569 (Yes) Good
G1. Somatic Concerns Yes Yes Yes Yes Yes 0.541 (Yes) Very good
G9. Unusual Thought No No Yes No No 0.335 (No) Weak
Content
G12. Lack of Judgment Yes Yes Yes Yes Yes 0.506 (Yes) Very good
and Insight

Biserial r = Biserial correlation.

For Criterion 5 and 6, the slopes and biserial correlations of all
items are presented in Table 6.
4. Discussion
This study aimed at using Item Response Analysis to identify
which symptoms of the PANSS were Very Good, Good or Weak at
assessing illness severity and at representing domains in three
cross-sectionally different subgroups of patients: First Episode pa-
tients, Chronic Inpatients and Ambulatory Outpatients. The findings
confirm differences in symptom presentation and specific predomi-
nance of particular domains in each subgroup.
First episode patients were well represented by most negative
symptoms, most disorganized symptoms, most hostility symptoms
(except G8: Uncooperativeness), and all anxiety symptoms, (except
G4: Tension). As hypothesized, depression and anxiety symptoms
are well represented in first episode patients as are most negative
symptoms. This observation indicates that negative symptoms are al-
ready present at the early stage of the disorder. Similarly, in a longitu-
dinal study of symptoms, Arndt and colleagues (1995) found that the
negative symptoms were already prominent at the time of the pa-
tients' first episode and remained relatively stable throughout the
2 years in which the patients were followed. However, two key neg-
ative symptoms, N3: Poor Rapport and N4: Passive/Apathetic Social
Withdrawal are not well characterized in first episode patients in
our sample, possibly because these symptoms are not yet fully devel-
oped, particularly in the area of diminished relatedness. In contrast,
the item N1: Blunted Affect is scored as a Very Good negative symp-
tom item pointing to its early presentation in the disorder
manifesting a marked deficit in expressiveness. Studies of first-
episode patients demonstrate that flat affect is indeed present at the
onset of illness (Shtasel et al., 1992) and are debilitating and resistant
to intervention (Carpenter, 2004). Surprisingly, PANSS symptoms of
disorganization are also well represented at the early stage of the

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
8 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Table 7 domain. This confirms findings in the majority of chronic schizo-

Comparison of First Episode, Chronic Inpatients and Ambulatory Outpatients and item phrenia samples where the negative, disorganized and positive
functioning for each PANSS domain.
components accounted for most of the variance in factor analyses
First episode Chronic Ambulatory studies (Emsley et al., 2003; Fresan et al., 2005). Similar results
patients inpatients outpatients were also obtained for the hostility domain, where all symptoms
Overall rating Overall rating Overall rating are scored as Good or Very Good and reflect the fact that these pa-
Anxiety tients are in an inpatient setting. Not surprisingly, the domain of
G2. Anxiety Good Good Weak anxiety is not well represented, with all PANSS items being scored
G3. Guilt Feelings Good Weak Weak as Weak for this subgroup. This may reflect the loss of an affective
G4. Tension Weak Weak Weak range and emotional responsiveness as the disorder progresses. As
G5. Depression Very Good Weak Weak
Disorganized
observed in our findings and other studies (Emsley et al, 1999;
P2. Conceptual Disorganization Very Good Very Good Very good Koreen et al, 1993), anxiety and depressive symptoms are more
N5. Difficulty in Abstract Very Good Very Good Weak common in first episode schizophrenia, than in patients with
Thinking chronic schizophrenia. Alternatively, this group most likely is
G5. Mannerisms and Posturing Weak Very Good Weak
also treated with more rigorous antipsychotic regimens and may
G10. Disorientation Weak Very Good Weak
G11. Poor Attention Very Good Very Good Weak be therefore more subject to extrapyramidal side effects further
G13. Disturbance of Volition Very Good Very Good Very good reducing their range of emotional depth. Surprisingly, the domain
G15. Preoccupation Very Good Very Good Weak of positive symptoms receives an uneven representation in this
Hostility chronic inpatient group, with two key positive items being scored
P4. Excitement Good Good Weak
P7. Hostility Very Good Very Good Weak
as Weak: P3: Hallucinatory Behavior and G12: Lack of Judgment
G8. Uncooperativeness Weak Very Good Weak and Insight with only G9: Unusual Thought Content assessed as
G14. Poor Impulse Control Very Good Good Weak a Very Good item.
Negative The ambulatory subgroup shows a symptom profile which reflects
N1. Blunted Affect Very Good Very Good Good
this group's greater stability and community dwelling. The domains
N2. Emotional Withdrawal Very Good Very Good Very good
N3. Poor Rapport Weak Very Good Very good of anxiety, disorganization and hostility are not well represented by
N4. Passive/Apathetic Weak Very Good Very good the corresponding PANSS items supporting the possibility that the
Social Withdrawal PANSS does not measure these three domains very sensitively in am-
N6. Lack of Spontaneity and Very Good Very Good Very good bulatory outpatients. This finding may also be related to the fact that
Flow of Conversation
the PANSS was originally developed based on a chronic inpatient
G7. Motor Retardation Very Good Good Very good
G16. Active Social Avoidance Very Good Very Good Good sample (Kay et al., 1987), more akin to the present chronic inpatient
Positive sample. An exception is the item of P2: Conceptual Disorganization,
P1. Delusions Weak Good Good which is rated as Very Good and confirms the validity and importance
P3. Hallucinatory Behavior Good Weak Very good
of cognitive deficits in the symptomatic presentation of the disorder
P5. Grandiosity Weak Good Weak
P6. Suspiciousness/Persecution Weak Weak Weak for all three subgroups. In contrast to the domains of anxiety, disorga-
N7. Stereotyped Thinking Weak Weak Good nization and hostility, the negative symptom domain is well
G1. Somatic Concerns Weak Weak Very good represented in this subgroup by the PANSS as all items are scored as
G9. Unusual Thought Content Good Very Good Weak Very Good, further confirming the validity of the negative domain
G12. Lack of Judgment Weak Weak Very good
and its corresponding items. These results are consistent with those
and Insight
obtained in previous studies indicating the most frequently present
negative symptoms in outpatients with schizophrenia are N4: Passive
disorder pointing to the centrality of cognitive symptoms at all Apathetic Social Withdrawal, N2: Emotional Withdrawal, N3: Poor
stages, except for the items G10: Disorientation and G5: Manner- Rapport and N1: Blunted Affect (Bobes et al., 2010; Lewis and
isms/Posturing. It should be noted that G5: Mannerisms and Postur- Lieberman, 2008).
ing is considered a difficult item to score based on previous IRT The ambulatory subgroup is also found to have the positive do-
analysis (Santor et al., 2000) and is more likely to be endorsed by main well represented by three strong items, P3: Hallucinatory Be-
raters at higher severity levels. Further, both G10: Disorientation havior, G1: Somatic Concern and G12: Lack of Judgment and
and G5: Mannerisms and Posturing are perhaps rare in first epi- Insight, while the other items are not as strongly represented. The
sode patients and are more characteristic of chronic patients. Un- fact that pharmacological treatments are most effective for positive
expectedly, positive symptoms were not well represented in this symptoms (Feldman et al., 2003; Miller 2004), and possible psycho-
first episode group, particularly the symptoms of P1: Delusions, social interventions the outpatient subgroup may have been exposed
P6: Suspiciousness and Persecution and G12: Lack of Judgment to, could explain the higher severity of negative symptomatology.
and Insight. The only strong PANSS positive items for the first ep- Our results confirm that a majority of PANSS items are either
isode group are P3: Hallucinatory Behavior and G9: Unusual Very Good or Good at assessing the overall illness severity and psy-
Thought Content. Both these items represent the developing psy- chometrically robust for two of the three subgroups: chronic inpa-
chotic process, while the weaker positive items do not describe tients (73.33%; 22 out of 30 items) and first episodes (60.00%, 18
this early stage of the disorder. Other studies have shown that out of 30 items). For the ambulatory group, only 46.67% (14 out of
positive symptoms were found to be prominent at the onset of 30 items) were identified as Very Good or Good. One may expect a
the illness and declined over the course of the follow-up period higher number of Very Good items representative of chronic inpa-
(Arndt et al. 1995). As hypothesized, the hostility domain is well tients as the PANSS scale was originally designed for inpatients in
represented by both G14: Poor Impulse Control and P7: Hostility a New York State psychiatric center (Kay et al., 1987) and was also
in these first episode patients. initially validated on similar inpatient samples (see Kay et al.,
For chronic inpatients we found that the disorganized and neg- 2000 for a review). As expected, our present nonparametric IRT
ative symptom domains were the most robust domains, showing showed that the Negative Symptom items (particularly, N1: Blunted
Very Good psychometric properties for all items of the disorga- Affect, N2: Emotional Withdrawal, N6: Lack of Spontaneity and Flow
nized domain, and for 6 of the 7 items of the negative symptom of Conversation, G7: Motor Retardation and G16: Active Social
domain. The positive dimensions formed the third most robust Avoidance) showed good discriminative properties across all three

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx
subpopulations, and most negative items reflected the entire range
of severity (i.e., increases in symptom intensity correspond to in-
creases in illness severity), and that these items most closely ap-
proximate the “ideal” item illustrated in Fig. 1. In contrast, items of
the Positive Symptoms domain, P1: Delusions, P3: Hallucinatory Be-
havior, P7: Hostility, G1: Somatic Concern and G12: Lack of Judg-
ment and Insight, showed good approximation to the “ideal” item
presented in Fig. 1 only for the ambulatory outpatients. Other than
P3: Hallucinatory Behavior and G9: Unusual Thought Content,
most of the positive domain were not representative of first episode
patients. This was also found for the chronic inpatient group where
most positive symptoms with the exception of G9: Unusual Thought
Content was not well represented. These cross-sectional findings of
variance in sensitivity and possible validity of PANSS measured
symptoms may suggest the need for some revision of PANSS items
or a more selective use of PANSS symptoms dependent on the pa-
tient population at hand. For example, it appears that positive symp-
toms in first episode patients are not be measured with the
positive PANSS domain as are depression and anxiety symptoms
in chronic inpatients. The PANSS may be least adapted to reliably
assess the range of symptoms in ambulatory outpatients where
we found that the domains of anxiety, disorganization and hostil-
ity were not well represented.
4.1. Limitations
The current study of symptom variations among three subgroups
of schizophrenia patients is subject to limitations. First, the sample
was selected according to a criterion for participation in a clinical
trial, and may not accurately represent patients encountered in clini-
cal practice. Because of the differences in investigators and study in-
clusion criteria, study participants may have differed widely by
demographic characteristics, and the level of PANSS training may
have differed across studies. Second, Cella and Chang (2000) warned
of the possible limitations of using IRT methods to evaluate
healthcare measures since IRT methods were originally developed
for, and used with a fairly homogeneous educational assessment pop-
ulation. When applying these methods to more heterogeneous clini-
cal populations there may be limitations to obtain item-free
estimates of sample latent traits. Thirdly, some studies have
suggested that a follow-up of two or more years may be more appro-
priate when exploring symptoms in first episode patients (Mane et
al., 2009; Milev et al., 2005). Our study only looked at baseline data
and this may have been a limitation in terms of understanding
whether symptom presentations observed in this IRT will persist
at follow-up for each subgroup. A 12- or 24-week follow up may
help us better delineate the course of the three subgroups. Fourth-
ly, it should be noted that subjects in the first episode group did
not have any other Axis I disorders, while subjects in the other sub-
groups were not screened out for other Axis I disorders. The pres-
ence of other Axis I disorders can provide present a confounding
variable as the other disorders may present with additional symp-
toms. Finally, for IRT, and item score patterns, false positives may
arise, for example, from attributing a pattern with relatively many
scores of 1 (i.e., Absent) as a “Weak” item. It may be noted that,
in general, several different causes might lead to the same kind of
pattern. False positives may be reduced with the use of a two-stage
procedure suggested by Marco (1977), which was not incorporated
in the present study. Use of this procedure involves first estimating
item parameters, calculating bias indices, identifying and deleting bi-
ased items, then estimating abilities using the remaining unbiased
items. However, this procedure would be applicable in previous
PANSS analysis for reducing the number of PANSS items (e.g., Khan
et al., 2011). Although, the available dataset for this study was ade-
quate for conducting an IRT, the symptom structures presented in
this study should not be considered a gold standard for identifying
Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
9
psychopathology in each subgroup. Other studies will be needed to
determine if these symptom structures are replicated across different
samples and show medication group difference through sensitivity to
therapeutic change.
5. Conclusions
This study used an Item Response Analysis to identify psycho-
metrically valid and sensitive symptoms of the five PANSS domains
in three cross-sectionally different subgroups of patients: first epi-
sodes, chronic inpatients and ambulatory outpatients. The emerging
dimensional approach to classification and treatment of psychiatric
disorders calls for a better understanding of symptom-related varia-
tions at different stages of schizophrenia and for proper validation
of psychometric scales used for the evaluation of symptom dimen-
sions. This examination of the psychometric suitability of the
PANSS items and its five symptom domains has shown that PANSS
domains have different predominance patterns in three patient sub-
groups. First episode psychosis patients are characterized by strong
domains of anxiety, disorganization and negative symptoms. The
chronic inpatient group is well represented by the disorganization,
hostility and negative domains. Finally, patients in the ambulatory
group show only a predominance of the negative domain with
some representation of the positive domain. While this examination
of PANSS items of the five psychopathology domains cannot replace
the examination of the longitudinal trajectory of these domains
across the illness course, the three patient groups do represent dis-
tinct stages of the disorder and allow for the examination of vari-
ances in symptom presentation overtime and possible underlying
mechanisms contributing to symptom expression.
One of the implications of our results is that the study of differ-
ent dimensions within subgroups of schizophrenia may help to bet-
ter define symptom domains of subgroups of schizophrenia patients
with a psychometrically sound rating scale instead with clinical
criteria and to benefit symptom dimension identification of patients.
This identification may favor the design of treatment programs,
which could address specific patient needs where appropriate treat-
ments could be available.
Funding body agreements and policies
• In the past five years none of the authors have received reim-
bursements, fees, funding, or salary from an organization that
may in any way gain or lose financially from the publication of
this manuscript, either now or in the future.
• All authors indicate that they do not hold any stocks or shares in
an organization that may in any way gain or lose financially from
the publication of this manuscript, either now or in the future.
• All authors indicate that they do not hold or are currently apply-
ing for any patents relating to the content of the manuscript. No
author has received reimbursements, fees, funding, or salary
from an organization that holds or has applied for patents relat-
ing to the content of the manuscript.
• All authors have no competing funding interests.
• Non-financial competing interests: All authors confirm that they
have no non-financial competing interests (political, personal,
religious, ideological, academic, intellectual, commercial or any
other) to declare in relation to this manuscript.
Contributors
Dr. Harvey and Dr. Khan participated in the development of the
concept for the study. Dr. Harvey, Dr. Khan, Dr. Lindenmayer, and
Dr. Opler participated in the design of the study and helped to draft
the manuscript. Dr. Khan performed the statistical analysis and
drafted the manuscript. Dr. Kelley, Dr. White, Dr. Compton and Ms.
10 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx
Gao participated in the design, coordination and drafted the manu-
script. All authors read and approved the final manuscript.
Conflict of interest
Dr. Opler owns shares in ProPhase LLC, New York, NY. ProPhase LLC
has received funding from Otsuka Pharmaceutical Company, Ltd.,
Pfizer Pharmaceuticals, Hoffman-LaRoche Pharmaceutical, Biomarin
Pharmaceuticals Inc., Angelini Group, Janssen Pharmaceuticals, Inc.,
Shire Pharmaceuticals Public Limited Company, and Reviva Pharma-
ceuticals Inc. Dr. Opler, Dr. Lindenmayer and Dr. Harvey have grants
and/or funding from National Institute of Health (NIH). Dr. Opler
also has funding from the Weill-Cornell Medical College – Qatar Foun-
dation. Dr. Lindenmayer and Dr. Khan have consulted for Janssen
Pharmaceuticals, Inc. Dr. Opler and Dr. Khan have consulted for Reviva
Pharmaceuticals Inc. Dr. Lindenmayer also has grants and/or funding
from Pfizer Pharmaceuticals, Eli Lilly and Company, Dainippon
Sumitomo Pharma, Azur Pharma, Inc., Amgen Biopharmaceutical
Company, Hoffman-LaRoche Pharmaceutical, Otsuka Pharmaceutical
Company, Ltd., EnVivo Pharmaceuticals, AstraZeneca PLC, and
Neurocrine Biosciences. All other authors declare that they have no
conflicts of interest. All other co-authors do not have any conflicts of
interest to disclose.
Acknowledgments
The authors would like to acknowledge the staff at ProPhase LLC,
New York, New York and Manhattan Psychiatric Center, New York,
who contributed towards the study by making substantial contribu-
tions to formatting and data review and were involved in reviewing
the manuscript for important intellectual content. No financial contri-
butions were involved in the data exchange.
References
American Psychiatric Association, 2013. Diagnostic and statistical manual of mental
disorders (5th ed., text revision). Author, Washington, DC.
Andreasen, N.C., Arndt, S., Alliger, R., Miller, D., Flaum, M., 1995. Symptoms of
schizophrenia: Methods, meanings and mechanisms. Arch. Gen. Psychiatry 52,
341–351.
Arndt, S., Andreasen, N.C., Flaum, M., Miller, D., Nopoulos, P., 1995. A longitudinal study
of symptom dimensions in schizophrenia: prediction and patterns of change. Arch.
Gen. Psychiatry 52 (5), 352–360.
Bartlett, M.S., 1954. A note on the multiplying factors for various chi square approxima-
tions. J. Roy. Stat. Society 16 (B), 296–298.
Bartels, S.J., Drake, R.E., Wallach, M.A., Freeman, D.H., 1991. Characteristic hostility in
schizophrenic outpatients. Schizophr. Bull. 17 (1), 163–171.
Bell, M.D., Lysaker, P.H., Beam-Goulet, J.L., Milstein, R.M., Lindenmayer, J.P., 1994. Five-
component model of schizophrenia: Assessing the factorial invariance of the Posi-
tive and Negative Syndrome Scale. Psychiatry Res. 52, 295–303.
Bengston, M., 2006. Types of Schizophrenia. 2006. Psych. Central. Retrieved on March
26, 2013, from http://psychcentral.com/lib/2006/types-of-schizophrenia/.
Bobes, J., Arango, C., Garcia-Garcia, M., Rejas, J., 2010. Prevalence of negative symptoms
in outpatients with schizophrenia spectrum disorders treated with antipsychotics
in routine clinical practice: Findings from the CLAMORS study. J. Clin. Psychiatry 71
(3), 280–286.
Bowie, C.R., Leung, W.W., Reichenberg, A., McClure, M.M., Patterson, T.L., Heaton, R.K., et
al., 2008. Predicting schizophrenia patients’ real-world behavior with specific neu-
ropsychological and functional capacity measures. Biol. Psychiatry 63, 505–511.
Buchanan, R.W., 2007. Persistent negative symptoms in schizophrenia: An overview.
Schizophr. Bull. 33 (4), 1013–1022.
Carpenter Jr., W.T., 2004. Clinical constructs and therapeutic discovery. Schizophr. Res.
72, 69–73.
Caton, C.L., Shrout, P.E., Eagle, P.F., Opler, L.A., Felix, A., Dominguez, B., 1994. Risk factors
or homelessness among schizophrenia men: A case control study. Am. J. Public
Health 84, 265–270.
Caton, C.L., Shrout, P.E., Dominguez, B., Eagle, P.F., Opler, L.A., Cournos, F., 1995. Risk
factors for homelessness among women in schizophrenia. Am. J. Public Health
85, 1153–1156.
Cella, D., Chang, C.H., 2000. A discussion of item response theory (IRT) and its applica-
tions in health status assessment. Med. Care 38 (9), 66–72.
Chemerinski, E., Reichenberg, A., Kirkpatrick, B., Bowie, C.R., Harvey, P.D., 2006. Three di-
mensions of clinical symptoms in elderly patients with schizophrenia: Prediction of
six-year cognitive and functional status. Schizophr. Res. 85 (1–3), 12–19.
Chen, C.Y., Liu, C.Y., Yang, Y.Y., 2001. Correlation of panic attacks and hostility in chron-
ic schizophrenia. Psychiatry Clin. Neurosci. 55, 383–387.
Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
Compton, M.T., Kelley, M.E., Ramsay, C.E., Pringle, M., Goulding, S.M., Esterberg, M.L.,
Stewart, T., Walker, E.F., 2009. Association of pre-onset cannabis, alcohol, and to-
bacco use with age at onset of prodrome and age at onset of psychosis in first-
episode patients. Am. J. Psychiatry 166 (11), 1251–1257.
Compton, M.T., Gordon, T.L., Goulding, S.M., Esterberg, M.L., Carter, T., Leiner, A.S.,
Weiss, P.S., Druss, B.G., Walker, E.F., Kaslow, N.J., 2011. Patient-level predictors
and clinical correlates of duration of untreated psychosis among hospitalized
first-episode patients. J. Clin. Psychiatry 72 (2), 225–232.
Davidson, M., Harvey, P.D., Powchik, P., Parrella, M., White, L., Knobler, H.Y., Losonczy,
M.F., Keefe, R.S., Katz, S., Frecska, E., 1995. Severity of symptoms in chronically in-
stitutionalized geriatric schizophrenic patients. Am. J. Psychiatry 152 (2), 197–207.
Emsley, R.A., Oosthuizen, P.P., Joubert, A.F., Hawkridge, S.M., Stein, D.J., 1999. Treatment
of schizophrenia in low-income countries. Int. J. Neuropsychopharmacol. 2, 321–325.
Emsley, R., Rabinowitz, J., Torreman, M., 2003. The factor structure for the Positive and
Negative Syndrome Scale (PANSS) in recent-onset psychosis. Schizophr. Res. 61
(1), 47–57.
Feldman, P.D., Kaiser, C.J., Kennedy, J.S., Sutton, V.K., Tran, P.V., Tollefson, G.D., 2003.
Comparison of risperidone and olanzapine in the control of negative symptoms
of chronic schizophrenia and related psychotic disorders in patients aged 50 to
65 years. .J Clin. Psychiatry 64 (9), 998–1004.
First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 1998. Structured Clinical Interview
for DSM-IV Axis I Disorders. Biometrics Research Department, New York State Psy-
chiatric Institute, New York.
Fresán, A., De la Fuente-Sandoval, C., Loyzaga, C., García-Anaya, M., Meyenberg, N.,
Nicolini, H., Apiquian, R., 2005. A forced five-dimensional factor analysis and con-
current validity of the Positive and Negative Syndrome Scale in Mexican schizo-
phrenia patients. Schizophr. Res. 72 (2–3), 123–129.
Green, A.I., Canuso, C., Brenner, M.J., Wijcik, J.D., 2003. Detection and management of
comorbidity in schizophrenia. Psychiatr. Clin. N. Am. 26, 115–139.
Green, M.F., Kern, R.S., Heaton, R.K., 2004. Longitudinal studies of cognition and functional
outcome in schizophrenia: implications for MATRICS. Schizophr. Res. 72, 41–51.
Härdle, W., 1990. Applied nonparametric regression. Chapman & Hall, London.
Harris, M.J., 1991. Deficit syndrome in older schizophrenic patients. Psychiatry Res. 39,
285–292.
Harvey, P.D., Green, M.F., Bowie, C., Loebel, A., 2006. The dimensions of clinical and cog-
nitive change in schizophrenia: Evidence for independence of improvements. Psy-
chopharmacology 187 (3), 356–363.
Harvey, P., Reichenberg, A., Bowie, C.R., Patterson, T.L., Heaton, R.K., 2010. The course of
neuropsychological performance and functional capacity in older patients with
schizophrenia: Influences of previous history of long term institutional stay. Biol.
Psychiatry 67 (10), 933–939.
Kaiser, H.F., 1970. A second-generation Little Jiffy. Psychometrika 35, 401–415.
Kaiser, H.F., 1974. An index of factorial simplicity. Psychometrika 39, 31–36.
Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale
(PANSS) for schizophrenia. Schizophr. Bull. 13 (2), 261–276.
Kay, S.R., Sevy, S., 1990. Pyramidical model of schizophrenia. Schizophr. Bull. 16 (3),
537–545.
Kay, S.R., Opler, L.A., Fiszbein, A., 2000. The Positive and Negative Syndrome Scale
(PANSS) Manual. Multi-Health Systems Inc., Toronto, ON.
Keefe, R.S., 2008. Should cognitive impairment be included in the diagnostic criteria for
schizophrenia. World Psychiatry 7, 22–28.
Kelley, M.E., White, L., Compton, M.T., Harvey, P.D., 2013. Subscale structure for the
Positive and Negative Syndrome Scale (PANSS): A proposed solution focused on
clinical validity. Psychiatry Res. 205 (1–2), 137–142.
Khan, A., Lewis, C., Lindenmayer, J.P., 2011. Use of non-parametric item response theo-
ry to develop a shortened version of the Positive and Negative Syndrome Scale
(PANSS). BMC Psychiatry 16 (11), 178.
Koreen, A.R., Siris, S.G., Chakos, M., Alvir, J., Mayerhoff, D., Lieberman, J., 1993. Depression
in first-episode schizophrenia. Am. J. Psychiatry 150, 1643–1648.
Kurtz, M.M., 2005. Symptoms versus neurocognitive skills as correlates of everyday
functioning in severe mental illness. Expert Rev. Neurother. 6, 47–56.
Leff, J., Trieman, N., 2000. Long-stay patients discharged from psychiatric hospitals. So-
cial and clinical outcomes after five years in the community. The TAPS Project 46.
Br. J. Psychiatry 176, 217–223.
Lewis, S., Lieberman, J., 2008. CATIE and CUtLASS: can we handle the truth? Br.
J. Psychiatry 192 (3), 161–163.
Lindenmayer, J.P., Kay, S.R., Friedman, C., 1986. Negative and positive schizophrenic
syndromes after the acute phase: A prospective follow-up. Compr Psychiatry 27,
276–286.
Mane, A., Falcon, C., Mateos, J.J., Fernandez-Egea, E., Horga, G., Lomena, F., Bargallo, N.,
Prats-Galino, A., Bernardo, M., Parellada, E., 2009. Progressive gray matter changes
in first episode schizophrenia: A 4-year longitudinal magnetic resonance study
using VBM. Schizophr. Res. 114 (1–3), 136–143.
Marco, G.L., 1977. Item characteristics curve solutions to three intractable testing prob-
lems. J. Educ. Meas. 14, 139–160.
Marder, S.R., Davis, J.M., Chouinard, G., 1997. The effects of risperidone on the five di-
mensions of schizophrenia derived by factor analysis: combined results of the
North American trials. J. Clin. Psychiatry 58 (12), 538–546.
Milev, P., Ho, B.C., Arndt, S., Andreasen, N.C., 2005. Predictive values of neurocognition
and negative symptoms on functional outcome in schizophrenia: A longitudinal
first-episode study with 7-year follow-up. Am. J. Psychiatry 162 (3), 495–506.
Miller, A.L., 2004. Combination treatments for schizophrenia. CNS Spectrums 9 (9),
19–23.
Moller, H.J., 2005. Occurrence and treatment of depressive comorbidity/
cosyndromality in schizophrenic psychoses: Conceptual and treatment issues.
World J. Biol. Psychiatry 6, 247–263.
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 11

Orlando, M., Thissen, D., 2000. Likelihood-based item-fit indices for dichotomous item
response theory models. Appl. Psychol. Measure 24, 50–64.
Petersen, M.A., 2004. Book review: Introduction to nonparametric item response theo-
ry. Qual. Life Res. 14, 1201–1202.
Putnam, K.M., Harvey, P.D., Parrella, M., White, L., Kincaid, M., Powchik, P., Davidson,
M., 1996. Symptom stability in geriatric chronic schizophrenic inpatients: A one-
year follow-up study. Biol. Psychiatry 39 (2), 92–99.
Ramsay JO. 2000. TESTGRAF: A computer program for nonparametric analysis of testing
data. Unpublished manuscript, McGill University [Available online from ftp://ego.
psych.mcgill.ca/pub/ramsay/testgraf].
Reckase, M.D., 1979. Unifactor latent trait models applied to multifactor tests: Results
and implications. J. Educ. Stat. 4 (3), 207–230.
Roseman, A.S., Kasckow, J., Fellows, I., Osatuke, K., Patterson, T.L., Mohamed, S., Zisook,
S., 2008. Insight, quality of life, and functional capacity in middle-aged and older
adults with. Int. J. Geriatr. Psychiatry 23 (7), 760–765.
Santor, D.A., Ascher-Svanum, H., Lindenmayer, J.P., Obenchain, R.L., 2000. Item response
analysis of the Positive and Negative Syndrome Scale. BMC Psychiatry 15, 7–66.
Shtasel, D.L., Gur, R.E., Gallacher, F., Heimberg, C., Cannon, T.D., Gur, R.C., 1992. Phe-
nomenology and functioning in first episode schizophrenia. Schizophr. Bull. 18,
449–462.
Sijtsma, K., Molenaar, I.W., 2000. Introduction to nonparametric item response theory.
Sage, Thousand Oaks, CA.
Sijtsma, K., Emons, W.H., Bouwmeester, S., Nyklícek, I., Roorda, L.D., 2008. Nonpara-
metric IRT analysis of Quality-of-Life Scales and its application to the World
Health Organization Quality-of-Life Scale (WHOQOL-Bref). Qual. Life Res. 17,
275–290.
White, L., Harvey, P.D., Opler, L., Lindenmayer, J.P., 1997a. Empirical assessment of the
factorial structure of clinical symptoms in schizophrenia. A multisite, multimodel
evaluation of the factorial structure of the Positive and Negative Syndrome Scale.
The PANSS Study Group. Psychopathology 30 (5), 263–274.
White, L., Parrella, M., McCrystal-Simon, J., Harvey, P.D., Masiar, S.J., Davidson, M.,
1997b. Characteristics of elderly psychiatric patients retained in a state hospital
during downsizing: A prospective study with replication. Int. J. Geriatr. Psychiatry
12 (4), 474–480.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
12 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 3. Left to right: Option Characteristic Curves (graphs shows 7 overlapping lines). Item 1: Anxiety Domain, Anxiety; Item 2: Anxiety Domain, Guilt Feelings; Item 3: Anxiety
Domain, Tension; Item 4: Anxiety Domain, Depression. Item Characteristic Curve (graph shows one line with green intersecting lines). Item 1: Anxiety Domain, Anxiety; Item 2:
Anxiety Domain, Guilt Feelings; Item 3: Anxiety Domain, Tension; Item 4: Anxiety Domain, Depression.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 13

Fig. 3. (continued) Left to right: Option Characteristic Curves. Item 1: Hostility Domain, Excitement; Item 2: Hostility Domain, Hostility; Item 3: Hostility Domain,
Uncooperativeness; Item 4: Hostility Domain, Poor Impulse Control. Item Characteristic Curve. Item 1: Hostility Domain, Excitement; Item 2: Hostility Domain, Hostility; Item 3:
Hostility Domain, Uncooperativeness; Item 4: Hostility Domain, Poor Impulse Control.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
14 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 4. Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 15

Fig. 4. Left to right: Option Characteristic Curves. Item 1:. Negative Domain,: Blunted Affect; Item 2: Negative Domain, Emotional Withdrawal; Item 3: Negative Domain, Poor Rapport; Item 4:
Negative Domain, Passive Apathetic Social Withdrawal; Item 5: Negative Domain, Lack of Spontaneity and Flow of Conversation; Item 6: Negative Domain, Motor Retardation; Item 7: Negative
Domain, Active Social Withdrawal. Item Characteristic Curves. Item 1: Negative Domain, Blunted Affect; Item 2: Negative Domain, Emotional Withdrawal; Item 3: Negative Domain, Poor Rap-
port; Item 4: Negative Domain, Passive Apathetic Social Withdrawal; Item 5: Negative Domain, Lack of Spontaneity and Flow of Conversation; Item 6: Negative Domain, Motor Retardation; Item
7: Negative Domain, Active Social Withdrawal.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
16 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 4. (continued). Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 17

Fig. 4. (continued) Left to right: Option Characteristic Curves. Item 1: Disorganized Domain, Conceptual Disorganization; Item 2: Disorganized Domain, Difficulty in Abstract Think-
ing; Item 3: Disorganized Domain, Mannerisms and Posturing; Item 4: Disorganized Domain, Disorientation; Item 5: Disorganized Domain, Poor Attention; Item 6: Disorganized
Domain, Disturbance of Volition; Item 7: Disorganized Domain, Preoccupation. Item Characteristic Curves. Item 1: Disorganized Domain, Conceptual Disorganization; Item 2: Dis-
organized Domain, Difficulty in Abstract Thinking; Item 3: Disorganized Domain, Mannerisms and Posturing; Item 4: Disorganized Domain, Disorientation; Item 5: Disorganized
Domain, Poor Attention; Item 6: Disorganized Domain, Disturbance of Volition; Item 7: Disorganized Domain, Preoccupation.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
18 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 5. Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 19

Fig. 5. Left to right: Option Characteristics Curves. Item 1: Positive Domain, Delusions; Item 2: Positive Domain, Hallucinatory Behavior; Item 3: Positive Domain, Grandiosity; Item
4: Positive Domain, Suspiciousness/Persecution; Item 5: Positive Domain, Stereotyped Thinking; Item 6: Positive Domain, Somatic Concerns; Item 7: Positive Domain, Unusual
Thought Content; Item 8: Positive Domain, Lack of Judgment and Insight. Item Characteristics Curves. Item 1: Positive Domain, Delusions; Item 2: Positive Domain, Hallucinatory
Behavior; Item 3: Positive Domain, Grandiosity; Item 4:; Positive Domain, Suspiciousness/Persecution; Item 5: Positive Domain, Stereotyped Thinking; Item 6: Positive Domain,
Somatic Concerns; Item 7: Positive Domain, Unusual Thought Content; Item 8: Positive Domain, Lack of Judgment and Insight.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
20 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 6. Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 21

Fig. 6. Left to right: Option Characteristic Curves (graphs shows 7 overlapping lines). Item 1: Anxiety Domain, Anxiety; Item 2: Anxiety Domain, Guilt Feelings; Item 3: Anxiety
Domain, Tension; Item 4: Anxiety Domain, Depression. Item Characteristic Curves (graphs show one line with green intersecting lines). Item 1: Anxiety Domain, Anxiety; Item
2: Anxiety Domain, Guilt Feelings; Item 3: Anxiety Domain, Tension; Item 4: Anxiety Domain, Depression.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
22 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 6. (continued). Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 23

Fig. 6. (continued) Left to right: Option Characteristic Curves. Item 1: Hostility Domain, Excitement; Item 2: Hostility Domain, Hostility; Item 3: Hostility Domain,
Uncooperativeness; Item 4: Hostility Domain, Poor Impulse Control. Item Characteristic Curves. Item 1. Hostility Domain, Excitement; Item 2: Hostility Domain, Hostility; Item
3: Hostility Domain, Uncooperativeness; Item 4: Hostility Domain, Poor Impulse Control.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
24 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 7. Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 25

Fig. 7. Left to right: Option Characteristic Curves. Item 1: Negative Domain, Blunted Affect; Item 2: Negative Domain, Emotional Withdrawal; Item 3: Negative Domain, Poor Rap-
port; Item 4: Negative Domain, Passive Apathetic Social Withdrawal; Item 5: Negative Domain, Lack of Spontaneity and Flow of Conversation; Item 6: Negative Domain, Motor Re-
tardation; Item 7: Negative Domain, Active Social Withdrawal. Item Characteristic Curves: Item 1: Negative Domain, Blunted Affect; Item 2: Negative Domain, Emotional
Withdrawal; Item 3: Negative Domain, Poor Rapport; Item 4: Negative Domain, Passive Apathetic Social Withdrawal; Item 5: Negative Domain, Lack of Spontaneity and Flow of
Conversation; Item 6: Negative Domain, Motor Retardation; Item 7: Negative Domain, Active Social Withdrawal.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
26 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 7. (continued). Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 27

Fig. 7. (continued) Left to right: Option Characteristic Curves. Item 1: Disorganized Domain, Conceptual Disorganization; Item 2: Disorganized Domain, Difficulty in Abstract Thinking;
Item 3: Disorganized Domain, Mannerisms and Posturing; Item 4: Disorganized Domain, Disorientation; Item 5: Disorganized Domain, Poor Attention; Item 6: Disorganized Domain,
Disturbance of Volition; Item 7: Disorganized Domain, Preoccupation. Item Characteristic Curves. Item 1: Disorganized Domain, Conceptual Disorganization; Item 2: Disorganized
Domain, Difficulty in Abstract Thinking; Item 3: Disorganized Domain, Mannerisms and Posturing; Item 4: Disorganized Domain, Disorientation; Item 5: Disorganized Domain,
Poor Attention; Item 6: Disorganized Domain, Disturbance of Volition; Item 7: Disorganized Domain, Preoccupation.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
28 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 8. Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 29

Fig. 8. Left to right: Option Characteristics Curves. Item 1: Positive Domain, Delusions; Item 2: Positive Domain, Hallucinatory Behavior; Item 3: Positive Domain, Grandiosity; Item 4: Positive
Domain, Suspiciousness/Persecution; Item 5: Positive Domain, Stereotyped Thinking; Item 6: Positive Domain, Somatic Concerns; Item 7: Positive Domain, Unusual Thought Content; Item 8:
Positive Domain, Lack of Judgment and Insight. Item Characteristics Curves. Item 1: Positive Domain, Delusions; Item 2: Positive Domain, Hallucinatory Behavior; Item 3: Positive Domain, Gran-
diosity; Item 4:; Positive Domain, Suspiciousness/Persecution; Item 5: Positive Domain, Stereotyped Thinking; Item 6: Positive Domain, Somatic Concerns; Item 7: Positive Domain, Unusual
Thought Content; Item 8: Positive Domain, Lack of Judgment and Insight.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
30 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 9. Left to right: Option Characteristic Curves (graphs shows 7 overlapping lines). Item 1: Anxiety Domain, Anxiety; Item 2: Anxiety Domain, Guilt Feelings; Item 3: Anxiety
Domain, Tension; Item 4: Anxiety Domain; Depression. Item Characteristic Curves (graphs show one line with green intersecting lines). Item 1: Anxiety Domain, Anxiety; Item
2: Anxiety Domain, Guilt Feelings; Item 3: Anxiety Domain, Tension; Item 4: Anxiety Domain, Depression.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 31

Fig. 9. (continued) Left to right: Option Characteristic Curves. Item 1: Hostility Domain, Excitement; Item 2: Hostility Domain, Hostility; Item 3: Hostility Domain,
Uncooperativeness; Item 4: Hostility Domain, Poor Impulse Control. Item Characteristic Curves: Item 1: Hostility Domain,: Excitement; Item 2: Hostility Domain, Hostility; Item
3: Hostility Domain, Uncooperativeness; Item 4: Hostility Domain, Poor Impulse Control.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
32 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 10. Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 33

Fig. 10. Left to right: Option Characteristic Curves. Item 1: Negative Domain: Blunted Affect; Item 2: Negative Domain, Emotional Withdrawal; Item 3: Negative Domain, Poor Rap-
port; Item 4: Negative Domain, Passive Apathetic Social Withdrawal; Item 5: Negative Domain, Lack of Spontaneity and Flow of Conversation; Item 6: Negative Domain, Motor Re-
tardation; Item 7: Negative Domain: Active Social Withdrawal. Item Characteristic Curves.: Item 1: Negative Domain, Blunted Affect; Item 2: Negative Domain, Emotional
Withdrawal; Item 3: Negative Domain, Poor Rapport; Item 4: Negative Domain, Passive Apathetic Social Withdrawal; Item 5: Negative Domain, Lack of Spontaneity and Flow of
Conversation; Item 6: Negative Domain, Motor Retardation; Item 7: Negative Domain, Active Social Withdrawal.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
34 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 10. (continued). Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 35

Fig. 10. (continued) Left to right: Option Characteristic Curves. Item 1: Disorganized Domain, Conceptual Disorganization; Item 2: Disorganized Domain, Difficulty in Abstract Thinking; Item
3: Disorganized Domain, Mannerisms and Posturing; Item 4: Disorganized Domain, Disorientation; Item 5: Disorganized Domain, Poor Attention; Item 6: Disorganized Domain, Disturbance
of Volition; Item 7: Disorganized Domain, Preoccupation. Item Characteristic Curves. Item 1: Disorganized Domain, Conceptual Disorganization; Item 2: Disorganized Domain, Difficulty in
Abstract Thinking; Item 3: Disorganized Domain, Mannerisms and Posturing; Item 4: Disorganized Domain, Disorientation; Item 5: Disorganized Domain, Poor Attention; Item 6: Disorga-
nized Domain, Disturbance of Volition; Item 7: Disorganized Domain, Preoccupation.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
36 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx

Fig. 11. Complete caption for this figure is provided on the next page.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002
 A. Khan et al. / Schizophrenia Research: Cognition xxx (2014) xxx–xxx 37

Fig. 11. Option Characteristics Curves. Item 1: Positive Domain, Delusions; Item 2: Positive Domain, Hallucinatory Behavior; Item 3: Positive Domain, Grandiosity; Item 4; Positive
Domain, Suspiciousness/Persecution; Item 5: Positive Domain, Stereotyped Thinking; Item 6: Positive Domain, Somatic Concerns; Item 7: Positive Domain, Unusual Thought Con-
tent; Item 8: Positive Domain, Lack of Judgment and Insight. Item Characteristics Curves. Item 1: Positive Domain, Delusions; Item 2: Positive Domain, Hallucinatory Behavior; Item
3: Positive Domain, Grandiosity; Item 4: Positive Domain, Suspiciousness/Persecution; Item 5: Positive Domain, Stereotyped Thinking; Item 6: Positive Domain: Somatic Concerns;
Item 7: Positive Domain, Unusual Thought Content; Item 8: Positive Domain, Lack of Judgment and Insight.

Please cite this article as: Khan, A., et al., The evolution of illness phases in schizophrenia: A non-parametric item response analysis of the Positive
and Negative Syndrome Sc..., Schizophr. Res. Cog. (2014), http://dx.doi.org/10.1016/j.scog.2014.01.002