A Longitudinal Study of Symptom Dimensions in Schizophrenia: Prediction and

Patterns of Change
[Original Article]
Arndt, Stephan PhD; Andreasen, Nancy C. MD, PhD; Flaum, Michael MD, PharmD;
Miller, Del MD; Nopoulos, Peg MD
From the Mental Health Clinical Research Center (Drs Arndt, Andreasen, Flaum,
Miller, and Nopoulos) and the Departments of Preventive Medicine and
Environmental Health (Dr Arndt) and Psychiatry (Drs Andreasen, Flaum, Miller, and
Nopoulos), The University of Iowa Hospitals and Clinics, Iowa City.
Abstract
Background: Factor analytic studies have suggested that the symptoms of
schizophrenia may be divided into three uncorrelated dimensions. This study examines
the longitudinal course of the symptoms of schizophrenia using this three-dimensional
perspective.
Methods: The sample was composed primarily of neuroleptic-naive patients suffering
from schizophrenia. Subjects were studied in a prospective longitudinal design, with
comprehensive structured assessments at index, discharge, and 6-month intervals after
discharge over a 2-year period.
Results: Negative symptoms were already relatively prominent at the time of index
evaluation; they tended to remain stable throughout the follow-up period. The two
dimensions of positive symptoms, psychoticism and disorganization, although
prominent at index evaluation, declined over the course of the follow-up period and
tended to be less stable. A longitudinal factor analysis was conducted to determine
whether the changes in symptoms followed any consistent pattern. We observed that all
three groups of symptoms tended to change in unison and independently from one
another.
Conclusions: These results suggest that these three dimensions of psychopathology
show different patterns of exacerbation and remission during the course of
schizophrenia. This independent pattern of evolution suggests that these three
dimensions should be studied further with respect to response to treatment, cognitive
mechanisms, psychosocial correlates, and neural substrates.
(Arch Gen Psychiatry. 1995;52:352-360)
The Study of longitudinal course and outcome serves several purposes in the study of
psychopathology. First, the observation of the evolution of symptoms may help us to
identify homogeneous groups of patients, which may ultimately facilitate identification
of the fundamental mechanisms that produce the disease. Second, the manner in which
the symptoms change and evolve may in and of itself provide clues as to the
fundamental mechanisms; for example, symptoms that are relatively persistent and
stable are more likely to be caused by different neural mechanisms from symptoms that
wax and wane. Third, the study of longitudinal course may assist in the identification

of particular symptoms or patterns of symptoms that may have prognostic significance

and can aid clinicians in the prediction of outcome. Fourth, the study of longitudinal
course can provide important information about treatment efficacy, with the ultimate
goal of developing improved medications or psychosocial treatment programs.
Studies of the course of schizophrenia fall into several different natural groups. One
early group, typified by the work of Kraepelin et al [1] and Bleuler,[2] focused on the
detailed description of symptoms accompanied by careful longitudinal observation
whenever possible, with the goal of finding symptoms that were basic and fundamental
to the disorder; the construct of fundamental, or basic, was usually taken to refer to
symptoms that were more persistent and enduring, although Bleuler [2] and others
[3,4] also emphasized the importance of specificity. Since neuroleptics were not
available to affect course, patients were usually institutionalized for long periods of
time, and definitions were relatively narrow and clear, observations from that era
should be given careful attention.[1-13] Both Kraepelin et al [1] and Bleuler [2] were
especially impressed by the importance of fragmenting of thought, loss of volition, and
loss of emotional responsiveness. A second group of studies, typified by the work of
Vaillant [14] and Stephens et al,[15] emphasized the study of prognostic indicators and
identified predictors of good vs poor outcome, such as premorbid adjustment or intact
emotional responsiveness; these studies mark the beginning of the early neuroleptic era
and are characterized by a relatively broad conceptualization of schizophrenia. Other
studies have sought to measure treatment efficacy, comparing various medications,
psychotherapy, and psychosocial rehabilitation techniques.[16-24] Other studies from
the postneuroleptic era have examined the capacity of various specific symptom
groups, such as thought disorder or Schneiderian symptoms, to predict outcome.[2530] Still others,[31-54] represented by the Iowa 500,[41] the International Pilot Study
of Schizophrenia,[51] or the work of Ciompi,[39] have emphasized descriptive
psychopathology or epidemiological techniques and attempted to provide a clear and
accurate description of the pattern of symptoms and their relationship to psychosocial
disability and outcome. During the past decade, interests have returned to the early
formulations of Kraepelin et al [1] and Bleuler [2] and emphasized the study of
fundamental symptoms, now usually referred to as negative or deficit symptoms.
[29,55-70] One variant of this has been the study of the positive vsnegative dichotomy,
particularly the model of Crow,[71-73] which was an attempt to offer a heuristic
simplification to facilitate studies of nosology, pathophysiology, outcome, and
treatment.
While much has been learned from these studies, many issues are still debated. There
is still no consensus as to whether schizophrenia leads to deterioration, whether
particular symptoms respond to treatment, or whether any symptoms can be considered
enduring or fundamental. Longitudinal studies are difficult to do well, since
prospective studies require a long-term commitment with a slow return on the
investment, whereas retrospective studies have obvious inherent limitations, because
they are not based on direct observation of patients and may be handicapped by
reporting biases. Studies done in the postneuroleptic era can no longer observe the
natural history of untreated schizophrenia. Definitions of schizophrenia have also
changed over time; the patient samples studied in the early work of Vaillant [14] or
Stephens et al [15] are almost certainly quite different from those based on the recent
narrow DSM-III-R [74] and DSM-IV [75] definitions. Although astute observers have
worried since the time of Bleuler [2] that some symptoms may be secondary to others,
very few investigators have attempted to develop designs that will disentangle actual

primary from secondary symptoms. Finally, although many investigators pay lip
service to the possibility that schizophrenia may be heterogeneous, the practicalities of
assembling homogeneous samples are daunting. However, conclusions can vary widely
if symptoms are assessed either cross-sectionally or longitudinally in patient samples
that differ in terms of acuity, chronicity, hospitalization status, or phase of illness;
conclusions about characteristic course or pattern will inevitably vary widely if
samples of chronically institutionalized patients are compared with younger patients
treated primarily in outpatient settings.
See also pages 341 and 361
In the present investigation, we report on a study of the longitudinal course of the
symptoms of schizophrenia using a design and sample chosen to correct at least some
of these problems. To reduce the confounding effects of chronicity, we have recruited a
sample of patients who have been ill for less than 5 years and who are less than 30
years of age, with an emphasis on identifying patients at the time of first admission. To
reduce the confounding effects of neuroleptic treatment, we have emphasized the
recruitment of patients who have never received prior treatment with neuroleptics. To
have maximally accurate documentation of symptoms, we assess the patients with a
comprehensive interview at intake and follow up at regular intervals thereafter, using
instruments with established reliability and validity. To keep the sample relatively
homogeneous, this report is limited to those patients with a DSM-III-R diagnosis of
schizophrenia or schizophreniform disorder at intake.
The primary purpose of this report is to extend our previous investigation of the
interrelationships between positive and negative symptoms and obtain a longitudinal
perspective. Research investigating the factor structure of the symptoms of
schizophrenia has been remarkably consistent in showing that more than two factors
are needed to account for the interrelationships among the symptoms.[58,76-86] Most
of these studies have reported a three-factor solution. One is composed primarily of
negative symptoms, whereas positive symptoms appear to split into two independent
dimensions, ``psychotic'' and ``disorganized.'' Hallucinations and delusions make up
the former, while the latter typically include measures of disorganized speech,
disorganized behavior, and inappropriate affect. These three factor solutions also
appear consistently when different analytic methods are used.[87] Current research is
actively exploring the pathophysiological and neuropsychological correlates of these
three dimensions.[88,89]
The factor analytic studies published to date describe symptom interrelationships as
they occur in a sample at a cross-sectional point in time. The course of schizophrenia,
however, is typically characterized by chronicity and periods of relative exacerbations
and remissions. Thus, cross-sectional analyses are limited in their predictive power or
utility. Indeed, many of the most important questions about the symptoms of
schizophrenia require a longitudinal research strategy, preferably one that can take a
multivariate approach and will address patterns of change within individuals over time.
For example, strong support for the three-dimensional model would be provided if
individual symptoms that constitute a dimension covary or change together over time,
but independently of symptoms within other dimensions.
The study of first-episode or recent-onset patients is the ideal laboratory in which to
determine the course of these symptoms and their relationship to one another over
time. We have investigated the change, stability, and patterns of change in psychotic,

disorganized, and negative symptoms in such a sample. Specifically, we asked: Which

symptoms are present at onset or early in the course of schizophrenia? How much do
these symptoms change over time? How predictive (ie, stable) are the various
symptoms or patterns of symptoms? How do changes in psychotic, disorganized, and
negative symptoms correlate with one another (ie, do symptoms that cluster together
cross-sectionally covary together over time and independently of other symptom
dimensions)?
RESULTS
CHANGES IN SYMPTOM SEVERITY BASED ON ANALYSIS OF GROUP
MEANS
The first step in the data analysis was to compare symptom severity across the four
time periods to determine whether significant changes had occurred in any of the
specific symptoms or the three dimensions. Table 1 presents patients' symptom ratings
at index hospitalization, discharge, and 1- and 2-year follow-ups. These results provide
an index of the average degree of improvement or change in the various symptoms.

Table 1. Mean Symptom Ratings at Index

Hospitalization, Discharge, and 1- and 2-Year Followup Evaluations*
[Help with image viewing]
[Email Jumpstart To Image]
All symptoms showed an improvement from index hospitalization to discharge based
on paired t tests (P less than .01). Improvement, as assessed by comparison against
index hospitalization, continued throughout the follow-up period, with the exception of
two negative symptoms, anhedonia and avolition. These became nonsignificant at year
2, indicating that mean improvement in these two specific negative symptoms did not
persist throughout the follow-up period and these symptoms may have begun to
worsen or to progress by the time 2 years had elapsed.
The data are shown graphically in the Figure 1, which indicates that, while all
symptoms improved significantly, both positive symptom dimensions (ie, psychoticism
and disorganization) improved considerably more than negative symptoms. The
average severity score of positive symptoms at discharge was almost halved compared
with those at index hospitalization (from 3.15 to 1.75 for psychotic symptoms and
from 2.05 to 0.98 for disorganization symptoms), while the average severity of
negative symptoms showed only a modest improvement (from 3.08 to 2.6). Not only
do these data indicate that negative symptoms are prominent in untreated first-episode
and recent-onset patients, but they suggest that they tend to be more treatment resistant.
Although positive and negative symptoms showed significant differences between
follow-up and index ratings, there was no significant ongoing change in positive
symptoms compared with discharge ratings; negative symptoms showed some
improvement between discharge and follow-up at year 2, but the effect size is small.
Similar results were obtained when a repeated-measures analysis of variance was
applied to the 39 patients who were measured at index hospitalization, discharge, and
1- and 2-year follow-up.

Figure 1. Symptom severity at index hospitalization, hospital

discharge, and 1- and 2-year follow-up evaluations. Asterisk indicates
P less than .01 compared with symptoms at index hospitalization;
dagger; compared with symptoms at discharge. Symptom dimensions
[Help with image are defined in the ``Statistical Analysis'' subsection of the ``Methods''
viewing]
section of the text.
[Email Jumpstart
To Image]
PREDICTION OF CHANGES IN SYMPTOM SEVERITY BASED ON
CORRELATIONAL ANALYSES
While changes in means provide one type of indicator of the extent to which positive
and negative symptoms evolve over the course of time in a group of schizophrenic
patients, it is also important to determine whether patterns of change occur at the level
of the individual patient. Consequently, we also used correlational analysis to
determine the extent to which the pattern of symptoms at index evaluation were
predictive of the pattern at a later point in time. These results are shown in Table 2.
Almost all individual positive and negative symptoms and all three dimensions are
closely correlated between index and discharge evaluations, suggesting that patients
tend to maintain their relative pattern of symptoms even when the severity decreases.
After discharge, however, a different picture emerges. Correlations between index and
1- and 2-year follow-up evaluations tend to become nonsignificant, suggesting that the
index presentation is not predictive of long-term outcome. The correlations between
index and follow-up evaluations tend to remain significant for negative symptoms
throughout the 2-year follow-up period, however, suggesting that negative symptoms
tend to be more stable and traitlike. Since the negative symptoms tend to persist in the
presence of diminishing positive symptoms, they are also less likely to be secondary,
although the effects of akinesia cannot be excluded as an explanation.

Table 2. Correlations of Index Hospitalization Ratings

With Discharge, 1- and 2-Year Follow-up Ratings*

[Help with image viewing]

[Email Jumpstart To Image]
PATTERNS OF INTRAINDIVIDUAL COURSE USING LONGITUDINAL
RATINGS

Our final analysis explored the patterns of symptoms during the follow-up period
available for each subject. This analysis used the weekly ratings to determine whether
the changes in symptom pattern observed using the cross-sectional measures reflected
a systematic and independent variation in the various individual symptoms. In this
analysis, we examined the intercorrelations of the symptoms over time to determine
whether any underlying factor structure was present.
The variance-covariance matrix for each patient over the weekly periods reflects how
symptoms covary over time. Factor analysis of an individual's covariation over time
has been referred to in the literature as the P-technique.[96] These individual matrices
were pooled across individuals, and the results were used in a principal components
analysis. Three factors met our criteria for inclusion, an analog of the eigenvaluegreater-than-one rule for covariance matrices. Factor loadings after varimax rotation
are shown in Table 3.

Table 3. Patterns of Intrasubject Change in Symptom

Severity Over the Weekly Periods: Loadings for Three
Factors of Intraindividual (Variance-Covariance)
Variation*
[Help with image viewing]
[Email Jumpstart To Image]
The first factor correlates highly with the negative symptoms, although attentional
impairment also loads on the second factor. This suggests that the negative symptoms
tend to exacerbate and remit in unison. The second factor appears as a disorganized
dimension, with high loadings for Bizarre (Disorganized) Behavior, Positive Thought
Disorder, and Inappropriate Affect. Hallucinations and Delusions almost equally define
the third factor. The loadings presented in Table 3 are based on an orthogonal rotation
(varimax), so that these factors are statistically independent (uncorrelated). While
symptoms within a factor tend to change together, changes occurring in one set of
symptoms are not correlated with other symptoms. When we allowed the factors to
become correlated using a promax rotation, the only change was a minor positive
correlation between the negative and psychotic factors (r equals .31), showing less than
a 10% overlap of the factor's variance.
There is still the possibility that symptoms may not co-occur simultaneously, but the
relationships are lagged in time. Some symptoms could precede others. For example,
positive symptoms might be antecedents of negative symptoms or vice versa. An
increase in positive symptoms during 1 week might correlate with negative symptoms
increasing the following week. As a consequence, correlations between symptoms that
were lagged an appropriate amount of time would be larger than the simultaneous
correlations. We generated lagged correlations to investigate this possibility, lagging by
1-week intervals up to 4 weeks and then monthly intervals up to 4 months. There was
no evidence of any time-lag structure to these data; all correlations tended to lower
with more distal time points. This suggests that symptom clusters tended not to

antecede other symptom clusters and provides further evidence that the clusters were
independent.
COMMENT
This study provides support for conceptualizing the symptoms of schizophrenia as
composed of three dimensions rather than two. The independence of the psychotic,
disorganized, and negative dimensions previously demonstrated by a series of crosssectional studies using factor analysis [77-79,81-86] has now been confirmed by our
study using longitudinal factor analysis. The availability of detailed fine-grained
ratings in a prospective longitudinal study of schizophrenic patients early in the course
of illness has permitted us to observe that the evolution of the symptoms of
schizophrenia over time occurs in three separate clusters. Positive symptoms break
down into two dimensions that are uncorrelated with one another in their longitudinal
progression, while negative symptoms form a third uncorrelated dimension. These
three dimensions show different patterns in the way they exacerbate and remit during
the course of schizophrenia.
Since patients in this study were all assessed during an acute hospitalization and 55
prior to receiving any neuroleptics, this sample provides us with an opportunity to
observe the course of the symptoms of schizophrenia and the three dimensions
independently of the effects of chronicity. The study demonstrates clearly that negative
symptoms are present early in the illness. Further, when these patients are treated
during a course of hospitalization, both negative symptoms and the two types of
positive symptoms improve significantly. The improvement in positive symptoms is
substantially greater than the improvement in negative symptoms, however, with a
much larger effect size (more than 1 SD, compared with 0.5 or less). These findings
suggest that negative symptoms tend to be less responsive to treatment, although they
do not suggest that negative symptoms are necessarily treatment refractory. Since
negative symptoms do not improve to the same degree as positive symptoms, these
findings decrease the likelihood that the persistent negative symptoms are secondary to
the positive symptoms; they could, of course, be in part due to neuroleptic treatment, a
possibility that cannot be clearly deconfounded in the present design.
After discharge from index hospitalization, both positive and negative symptoms
remain relatively stable overall, and by discharge, patients appear to have settled into a
pattern. Mean scores for the two types of positive symptoms and negative symptoms
remain nearly the same from the time of initial discharge through the 2-year follow-up
period.
Correlations between index ratings and summary scores at discharge are relatively high
for positive and negative symptoms. Thereafter, the pattern of correlations shifts and
becomes less predictive. Negative symptoms, both as a dimension and as individual
symptoms, retain high correlations throughout the follow-up period, suggesting that
these symptoms are relatively stable; nevertheless, the correlations decline over time.
The pattern for positive symptoms is more spotty. For the psychotic dimension, there is
no significant correlation between index and year 1 evaluations, but a significant
correlation reappears at year 2, probably indicating that a subset of patients has
relapsed during this time period, with a reemergence of the symptoms that they
originally manifested at index evaluation. Among the disorganized symptoms, only
inappropriate affect remains consistently correlated throughout the follow-up period,
indicating that this particular symptom is also relatively stable. The disorganized

dimension as a whole remains poorly correlated with the index measure throughout the
2-year follow-up period; these results suggest that this particular dimension may be the
most variable and unstable among the three. Reasons for this instability will require
further study.
Interpretation of the results of this study is subject to limitations. First, all patients
within the study were treated after the initial index assessment. Thus, the study does
not provide us with information about the natural history of schizophrenia per se, but
rather its course in the context of contemporary treatment approaches. This study
design is naturalistic with respect to treatment, and all of the subjects were treated in a
noncontrolled manner.
A second limitation of this study was that the weekly ratings used for the longitudinal
factor analysis were determined retrospectively at the time of each 6-month follow-up
evaluation. This method of acquiring data may have dampened changes in symptom
variability over time. However, multiple sources, ie, informants, subjects, hospitals,
and social workers, were used to construct the ratings, thereby increasing the diversity
of information available and reducing the likelihood of dampening.
Rater effects also may have dampened the variability of changes during the period,
since the same interviewer evaluated the patient's condition during the 2-year followup period whenever possible. This could produce a ``halo effect'' that could enhance
the likelihood of finding longitudinal correlations. However, we believe that the
advantages of interviewer continuity outweigh those of independent blind assessments.
That is, as raters get to know the patient over time, building rapport with the patient
and informants, they are more likely to pick up real, meaningful changes in the
symptom profile and ability to function. Also, variability stemming from interrater
differences in ratings is lowered. The use of informants (ie, family members) and other
sources of information (eg, records from the referring psychiatrist following up the
patient after discharge) also reduces the likelihood of halo effect.
In summary, these findings provide strong support for three separate dimensions of
psychopathology in schizophrenia. The fact that the various positive and negative
symptoms display a pattern of evolution that is independent suggests that they could
reflect separate processes or even different neurobiological substrates. This study
provides impetus for additional investigations using various types of neurobiological
measures, as well as the exploration of the effects of treatments on these three
dimensions of psychopathology.
Accepted for publication August 8, 1994.
Supported in part by grants MH31593, MH40856, MH46011, and MHCRC 43271
from the National Institute of Mental Health (NIMH), Rockville, Md, The Nellie Ball
Trust Fund, Iowa State Bank and Trust Company (trustee), Iowa City (Dr Andreasen);
and Research Scientist Award MH00625 from the NIMH (Dr Andreasen).
Reprint requests to 2911 JPP MHCRC, University of Iowa Hospitals and Clinics, 200
Hawkins Dr, Iowa City, IA 52242-1057 (Dr Arndt).
METHODS
SUBJECTS

Subjects underwent evaluation with the Iowa Prospective Longitudinal Study of

recent-onset psychoses. Methods for selecting the overall population have been
previously described in detail.[90] Briefly, subjects are recruited from consecutive
admissions to the Psychiatry Inpatient Services at the University of Iowa Hospitals and
Clinics, Iowa City. All patients who meet entry criteria and consent to participate are
included. This longitudinal study emphasizes first-episode, recent-onset, and
neuroleptic-naive patients; its goal is to acquire a cohort in whom symptoms, course,
and outcome can be evaluated with a prospective design and minimal confounding of
prior morbidity. Subjects are recruited into the study who are experiencing their first
episode of psychosis or hospitalization or have experienced their first psychiatric
hospitalization within the past 5 years and are less than 30 years of age. This report
focuses on a subset of these patients who met DSM-III-R criteria for schizophrenia or
schizophreniform disorder at index assessment.
The patients in this subset were typically young, acutely ill, and predominantly
neuroleptic naive. The average age in the sample was 24.18 years (SD equals 5.2), and
the median time since onset of the first psychotic symptom was 8.3 months. All but 11
of the subjects (85%) had never received neuroleptics before entering the study. Most
subjects were male (77%) and never married (91%).
Sixty-five patients with a diagnosis of schizophrenia or schizophreniform disorder
have been enrolled in this study and followed up for at least 6 months. These subjects
have provided the sample for this report. This study is ongoing, and therefore the
sample size for each of the follow-up periods varies. While 65 subjects have been
followed up for 6 months, 53 have been followed up for 1 year, and 39 patients for 2
years. These decreasing sample sizes reflect study duration rather than subject attrition.
Dropouts have been minimal. A total of four patients were unavailable for follow-up
from this sample during the follow-up period: two patients committed suicide, one
refused further participation, and one later received a diagnosis of another disorder that
made her ineligible for the study. The available data for these four patients were
maintained to reduce the bias of attrition. The 39 patients who have been followed up
for 2 years are not different from the patients who were ascertained later, and therefore
had shorter periods of observation, on any of the demographic or clinical variables (eg,
age at onset, age at entrance into the study, gender, education, and severity of positive
and negative symptoms at index evaluation). In over 40 comparisons, only one
significant difference was found between the 39 subjects with 2-year follow-up data
and those with shorter follow-up intervals. Consequently, the data for all 65 subjects
were included in the statistical analyses.
INDEX ASSESSMENT
During the index hospitalization, all subjects underwent evaluation with the
Comprehensive Assessment of Symptoms and History (CASH)[57,91] and the
PSYCH-BASE.[92] Embedded within the CASH are expanded versions of the Scale
for the Assessment of Negative Symptoms (SANS)[93] and the Scale for the
Assessment of Positive Symptoms (SAPS),[94] rated at the subject's worst severity
during the index period. All possible sources of information, including hospital records
and interviews with informants, the patient, nurses, and social workers were used in
completing the CASH, as elaborated in the companion article.[95] The DSM-III-R
diagnosis was established by consensus of the clinical research team, using all
available sources of information, as previously described. Treatment was not a
controlled variable in this study. Consequently, during this index hospitalization,

patients were treated using ``clinician's choice'' medications. Treatment typically

consisted of neuroleptics (N equals 66), but also included antidepressants (N equals 7),
lithium (N equals 4), antianxiety agents (N equals 3), and antiparkinsonians (N equals
35).
FOLLOW-UP EVALUATIONS
Every 6 months after discharge from the index hospitalization, the subject underwent
reevaluation by an interviewer who made initial contact with the patient during the
index hospitalization and participated in the completion of the consensus CASH at the
time of discharge. Follow-up evaluations employed versions of the CASH and
PSYCH-BASE that have been modified for longitudinal follow-up (the CASH-UP and
PSYCH-UP). These instruments contain time lines that record global ratings of each
positive and negative symptom at 1-week intervals covering the previous 6 months.
The global ratings include: Hallucinations, Delusions, Positive Formal Thought
Disorder, Bizarre (Disorganized) Behavior, Inappropriate Affect, Alogia, Anhedonia,
Avolition, Affective Flattening, and Attention.
The data obtained using these instruments are both current (ie, describing the past
week) and retrospective (ie, summarizing the past 6 months). These data were obtained
at each of the 6-month follow-up periods by interviewing the subject concerning any
changes in medication, hospitalizations, or exacerbations in symptoms that had
occurred during the prior 6 months. Information was also obtained from an informant
and medical records and used to establish a consensus rating, as described below. To
reconstruct the retrospective weekly time line data, the PSYCH-UP is employed.
Informants and patients are asked to identify any specific time period when a change
occurred and describe the symptoms present, leading to a rating of level of severity at
the peak time for each symptom, as well as the time and degree to which the symptom
began to worsen or improve.
The CASH-UP covers symptoms during the past week, which provides a crosssectional assessment of psychopathology in a series of 6-month ``snapshots.'' Since the
assessments are almost always obtained while the subject is an outpatient (rather than
occurring at the time of hospital readmission during an exacerbation), they provide an
unbiased estimate of the clinical status of the patient during the follow-up interval.
That is, the time of assessment is not linked in any way to exacerbations or remissions.
Because they describe current status at the time of assessment, they are very accurate;
when analyzed over a 2-year time period, they provide the most valid repeatedmeasure data for longitudinal assessment.
The design of the follow-up evaluation has incorporated a systematic study of the
effects of informant on the assessment of psychopathology; at the time of each 6month follow-up, both the patient and an informant (most often the patient's mother,
but potentially a sibling, spouse, or companion) underwent separate evaluation by two
different interviewers. The same interviewer maintained contact with the patient or
informant to ensure continuity, sustain rapport, and reduce dropout rate; these two
different interviewers were, however, blind to one another's ratings. After these
independent blind evaluations were completed, the two interviewers shared their
information and established a consensus rating. In addition to the interviews, other
sources of information (eg, medical records) also were used to establish consensus
ratings. An analysis of the effects of informant on symptom reporting will be presented

elsewhere. In this investigation, we have employed the ``gold standard,'' or consensus

ratings, in our analyses.
The reliability of the various symptom measures analyzed in this study has been
carefully assessed; in general, reliability (determined by intraclass correlation
coefficients) is good (.5 to .7) to excellent (.8 or greater), with higher estimates for
interrater measures than for test-retest measures, as would be expected.[91] All
interviewers participating in this follow-up study attend weekly calibration meetings
coordinated by the Assessment and Training Core of the Iowa Mental Health Clinical
Research Center, Iowa City. The assessment and training program developed by this
core has extensive methods in place to ensure continuing reliability to and prevent
``rater drift,'' including regular calibration checks on all instruments and raters. This
program ensures that the high levels of reliability are maintained during the follow-up
period. In addition, since this study also involves retrospective reporting of symptoms,
the validity of such reporting is also being evaluated as a separate arm of the
investigation and will be reported elsewhere. In general, the results indicate acceptable
accuracy of assessment when all sources of information are used, as in the present
report.
During the follow-up period, the majority of patients received some type of treatment
that was prescribed by their primary physician. Thus, treatment in this study is
naturalistic rather than controlled. The mainstay of treatment was neuroleptics, which
the majority of the patients received throughout the follow-up period.
STATISTICAL ANALYSIS
One set of analyses addressed the questions of frequency of symptom categories at
onset or early in the illness (ie, the extent to which negative symptoms may be
primary) and the extent to which symptoms improve or worsen and to which some
types of symptoms are persistent or enduring. Data from four evaluation periods
(index, discharge, 1-year, and 2-year) were used to examine the change in symptoms
over the 2-year time period. Means, intersubject SDs, and paired t tests were used for
many of the descriptive analyses comparing index hospitalization or discharge ratings
with the 1- and 2-year follow-up ratings.
To evaluate the issue of stability and predictability, we also conducted a ``longitudinal
factor analysis,'' in which the unit of analysis was the individual patient and time of
measurement was the sample. The major purpose of this analysis was to determine
whether groups of symptoms can be identified that show similar patterns of change
over time. Correlations between symptoms over time were calculated by determining
the intrasubject variance and covariances. These correlations (covariances) were
generated for each subject separately, reflecting the intrasubject associations among the
symptoms. The summary measures of association (covariances) were averaged across
individuals according to the number of observation times and analyzed, using principal
components analysis. The effective sample size for this analysis is very large (N equals
7098), since the correlations were based on the total number of observations for all
subjects; statistically, it is therefore of a very optimal size for factor analysis. To
determine the number of factors, we used an eigenvalue criterion for covariances
analogous to the eigenvalue-greater-than-one rule for factoring correlations. A factor
was included if the eigenvalue (variance accounted for) was greater than the average
symptom variance (mean diagonal entry in the pooled intrasubject variance-covariance
matrix). The extracted factors were analyzed with varimax rotation.

To examine the utility of the study of individual symptoms vs the three-dimensional

model, we used two different approaches to exploring interrelationships between
symptoms: global ratings for specific SANS/SAPS symptoms (eg, Delusions, Affective
Flattening), and sums of global ratings that reflect the three dimensions of
psychopathology suggested by our own cross-sectional factor analyses and those of
others. In these analyses, global ratings of Delusions and Hallucinations were summed
to characterize the dimension of psychoticism; Positive Formal Thought Disorder
(disorganized speech), Bizarre (Disorganized) Behavior, and Inappropriate Affect were
summed to measure disorganization; and Alogia, Affective Flattening, Avolition,
Anhedonia, and Attention were summed to measure negative symptoms. The position
of Attention with respect to other negative and positive symptoms is controversial,
showing a relationship with some aspects of both in our own work and that of others. It
was grouped with the negative symptoms here primarily for the sake of continuity and
simplicity, anticipating that the longitudinal analyses might shed further light on its
relationship.
REFERENCES
1. Kraepelin E, Barclay RM, Robertson GM. Dementia Praecox and Paraphrenia.
Edinburgh, Scotland: E & S Livingstone; 1919. [Context Link]
2. Bleuler E; Zinkin J, trans. Dementia Praecox of the Group of Schizophrenias 1911.
New York, NY: International Universities Press; 1950. [Context Link]
3. Schneider K; Hamilton MW, trans. Clinical Psychopathology. New York, NY: Grune
& Stratton; 1959. [Context Link]
4. Schneider K. Primary and secondary symptoms in schizophrenia. In: Hirsch SR,
Shepherds M, eds. Themes and Variations in European Psychiatry. Littleton, Mass:
John Wright-PSG Inc; 1974:40-46. [Context Link]
5. Huber G, Gross G, Schuttler R. A long-term follow-up study of schizophrenia:
psychiatric course of illness and prognosis. Acta Psychiatr Scand. 1975;52:49-57.
Bibliographic Links [Context Link]
6. Kleist K. Schizophrenic symptoms and cerebrapathology. J Ment Sci. 1960;
106:246-255. Bibliographic Links [Context Link]
7. Langfeldt G. Schizophrenia: diagnosis and prognosis. Behav Sci. 1969;14:173-182.
Bibliographic Links [Context Link]
8. Leonhard K. The question of prognosis in schizophrenia. Int J Psychiatry. 1966;
2:633-635. Bibliographic Links [Context Link]
9. Fish FJ. Schizophrenia. Littleton, Mass: John Wright-PSG Inc; 1962. [Context Link]
10. Astrup C, Noreik K. Functional Psychoses: Diagnostic and Prognostic Models.
Springfield, III: Charles C Thomas Publishers; 1966. [Context Link]
11. Babigian HM, Gardner EH, Miles HC, Romano J. Diagnostic consistency and
change in a follow-up study of 1215 patients. Am J Psychiatry. 1965;121:895-901.
Bibliographic Links [Context Link]
12. Fish FJ. The classification of schizophrenia: the views of Kleist and his coworkers.
J Ment Sci. 1957;103:443-463. Bibliographic Links [Context Link]

13. Fish FJ. Leonhard's classification of schizohrenia. J Ment Sci. 1958;104:943-971.

Bibliographic Links [Context Link]
14. Vaillant GE. Prospective prediction of schizophrenic remission. Arch Gen
Psychiatry. 1964;11:509-518. [Context Link]
15. Stephens JH, Astrup C, Mangrum JC. Prognostic factors in recovered and
deteriorated schizophrenics. Am J Psychiatry. 1966;122:1116-1121. Bibliographic
Links [Context Link]
16. Klein DF, Rosen B. Premorbid asocial adjustment and response to phenothiazine
treatment among schizophrenic inpatients. Arch Gen Psychiatry. 1973; 29:480-485.
Bibliographic Links [Context Link]
17. Pritchard M. Prognosis of schizophrenia before and after pharmacotherapy, I: short
term outcome. Br J Psychiatry. 1967;113:1345-1352. Bibliographic Links [Context
Link]
18. Pritchard M. Prognosis of schizophrenia before and after pharmacotherapy, II:
three-year follow-up. Br J Psychiatry. 1967;113:1353-1359. Bibliographic Links
[Context Link]
19. McGlashan TH. Intensive individual psychotherapy of schizophrenia: a review of
techniques. Arch Gen Psychiatry. 1983;40:909-920. Bibliographic Links [Context
Link]
20. Johnstone EC, Frith CD, Gold A, Stevens M. The outcome of severe acute
schizophrenia illnesses after one year. Br J Psychiatry. 1979;134:28-33. Bibliographic
Links [Context Link]
21. Goldberg SC, Frosch WA, Drossman AK, Schooler NR, Johnson GF. Prediction of
response to phenothiazines in schizophrenia: a cross-validation study. Arch Gen
Psychiatry. 1972;26:367-373. Bibliographic Links [Context Link]
22. Herz MI, Szymanski HV, Simon JC. Intermittent medication for stable
schizophrenic outpatients: an alternative to maintenance medication. Am J Psychiatry.
1982;139:918-922. Bibliographic Links [Context Link]
23. Kane JM, Rifkin A, Quitkin F, Nayak D, Saraf K, Ramos-Lorenzi JR, Klein DF,
Sachar EJ. Low dose fluphenazine decanoate and maintenance treatment in
schizophrenia. Psychiatry Res. 1979;1:341-348. Bibliographic Links [Context Link]
24. Falloon IRH, Liberman RP. Interactions between drug and psychosocial therapy in
schizophrenia. Schizophr Bull. 1983;9:543-554. [Context Link]
25. Silverstein ML, Harrow M. First-rank symptoms in the postacute schizophrenic: a
follow-up study. Am J Psychiatry. 1978;135:1481-1484. Bibliographic Links [Context
Link]
26. Harrow M, Bromet E, Quinlan D. Predictors of posthospital adjustment in
schizophrenia. J Nerv Ment Dis. 1974;158:25-36. Bibliographic Links [Context Link]
27. McGlashan TH, Carpenter WT. An investigation of the postpsychotic depressive
syndrome. Am J Psychiatry. 1976;133:14-19. Bibliographic Links [Context Link]

28. Strauss JS, Carpenter WT. Characteristic symptoms and outcome in schizophrenia.
Arch Gen Psychiatry. 1974;30:429-434. Bibliographic Links [Context Link]
29. Strauss JS, Carpenter WT, Bartko JJ. The diagnosis and understanding of
schizophrenia, III: speculations on the processes that underlie schizophrenia symptoms
and signs. Schizophr Bull. 1974;1:61-69. [Context Link]
30. Chandrasena R, Rodrigo A. Schneider's first rank symptoms: their prevalence and
diagnostic implications in an Asian population. Br J Psychiatry. 1979;135:348-351.
Bibliographic Links [Context Link]
31. Pfohl B, Winokur G. The evolution of symptoms in institutionalized
hebephrenic/catatonic schizophrenics. Br J Psychiatry. 1982;141:567-572.
Bibliographic Links [Context Link]
32. Pfohl B, Winokur G. The micropsychopathology of hebephrenic/catatonic
schizophrenia. J Nerv Ment Dis. 1983;171:296-300. [Context Link]
33. Carr VJ. Recovery from schizophrenia: a review of patterns of psychosis.
Schizophr Bull. 1983; 9:95-121. Bibliographic Links [Context Link]
34. Gittelman-Klein R, Klein DF. Premorbid asocial adjustment and prognosis in
schizophrenia. J Psychiatr Res. 1969;7:35-53. Bibliographic Links [Context Link]
35. Zigler E, Glick M, Marsh A. Premorbid social competence and outcome among
schizophrenic and nonschizophrenic patients. J Nerv Ment Dis. 1979;167:478-483.
Bibliographic Links [Context Link]
36. Bland RC, Parker JH, Orn H. Prognosis in schizophrenia: prognostic predictions
and outcome. Arch Gen Psychiatry. 1978;35:72-77. Bibliographic Links [Context
Link]
37. Bland RC, Orn H. 14-year outcome in early schizophrenia. Acta Psychiatr Scand.
1978;58:328-338. [Context Link]
38. Bridge TP, Cannon E, Wyatt RJ. Burned-out schizophrenia: evidence for age
effects on schizophrenic symptomatology. J Gerontol. 1978;33:835-839.
Bibliographic Links [Context Link]
39. Ciompi L. Three lectures on schizophrenia. Br J Psychiatry. 1980;136:413-420.
Bibliographic Links [Context Link]
40. Tsuang MT, Woolson RF. Excess mortality in schizophrenia and affective
disorders: do suicides and accidental deaths solely account for this excess? Arch Gen
Psychiatry. 1978;35:1181-1185. Bibliographic Links [Context Link]
41. Winokur G, Morrison J, Clancy J, Crowe R. Iowa 500: the clinical and genetic
distinction of hebephrenic and paranoid schizophrenia. J Nerv Ment Dis. 1974;159:1219. Bibliographic Links [Context Link]
42. Tsuang MT, Woolson RF, Fleming JA. Long-term outcome of major psychoses, I:
schizophrenia and affective disorders compared with psychiatrically symptom-free
surgical conditions. Arch Gen Psychiatry. 1979;36:1295-1301. Bibliographic Links
[Context Link]

43. Waxler NE. Is outcome for schizohrenia better in nonindustrial societies? J Nerv
Ment Dis. 1979;167:144-158. Bibliographic Links [Context Link]
44. Watt DC, Szulecka TK. The effect of sex, marriage and age at first admission on
the hospitalization of schizophrenics during 2 years following discharge. Psychol Med.
1979;9:529-539. Bibliographic Links [Context Link]
45. Parnas J, Schulsinger F, Teasdale TW, Schulsinger H, Feldman PM, Mednick SA.
Perinatal complications and clinical outcome within the schizophrenia spectrum. Br J
Psychiatry. 1982;140:416-420. Bibliographic Links [Context Link]
46. Summers F, Hersh S. Psychiatric chronicity and diagnosis. Schizophr Bull.
1983;9:122-133. Bibliographic Links [Context Link]
47. McGlashan TH. The Chestnut Lodge Follow-Up Study, I: follow-up methodology
and study sample. Arch Gen Psychiatry. 1984;41:573-585. [Context Link]
48. McGlashan TH. The Chestnut Lodge Follow-Up Study, II: long-term outcome of
schizophrenia and the affective disorders. Arch Gen Psychiatry. 1984;41:586-601.
Bibliographic Links [Context Link]
49. McGlashan TH. The prediction of outcome in chronic schizophrenia, IV: the
Chestnut Lodge Follow-Up Study. Arch Gen Psychiatry. 1986;43:167-176.
Bibliographic Links [Context Link]
50. McGlashan TH. Schizotypal personality disorder: Chestnut Lodge Follow-Up
Study, VI: long-term follow-up perspectives. Arch Gen Psychiatry. 1986;43:329-334.
Bibliographic Links [Context Link]
51. Wing JK, Cooper JE, Sartorius N. The Measurement and Classification of
Psychiatric Symptoms. New York, NY: Cambridge University Press; 1974. [Context
Link]
52. Fenton WS, McGlashan TH. Natural history of schizophrenia subtypes, I:
Iongitudinal study of paranoid, hebephrenic, and undifferentiated schizophrenia. Arch
Gen Psychiatry. 1991;48:969-977. Bibliographic Links [Context Link]
53. Harding CM, Strauss JS. The course of schizophrenia: an evolving concept. In:
Alpert M, ed. Controversies in Schizophrenia: Changes and Constancies. New York,
NY: Guilford Press; 1985. [Context Link]
54. Harding CM, Zubin J, Strauss JS. Chronicity in schizophrenia: fact, partial fact, or
artifact? Hosp Community Psychiatry. 1987;38:477-486. [Context Link]
55. Andreasen NC. Negative symptoms in schizophrenia: definition and reliability.
Arch Gen Psychiatry. 1982;39:784-788. Bibliographic Links [Context Link]
56. Andreasen NC, Olsen S. Negative v positive schizophrenia: definition and
validation. Arch Gen Psychiatry. 1982;39:789-794. Bibliographic Links [Context
Link]
57. Andreasen NC. Comprehensive Assessment of Symptoms and History (CASH).
Iowa City, Iowa: The University of Iowa; 1985. [Context Link]
58. Andreasen NC. Can Schizophrenia Be Localized in the Brain? Washington, DC:
American Psychiatric Press, Inc; 1986. [Context Link]

59. Andreasen NC. Method for assessing positive and negative symptoms. In:
Andreasen NC, ed. Positive and Negative Symptoms and Syndromes: Modern
Problems in Pharmacopsychiatry. New York, NY: S Karger AG; 1990:73-88. [Context
Link]
60. Andreasen NC, Flaum M. Schizophrenia: the characteristic symptoms. Schizophr
Bull. 1990;17:27-49. [Context Link]
61. Andreasen NC, Flaum M, Arndt S, Alliger R, Swayze VW. Positive and negative
symptoms: assessment and validity. In: Marneros A, Andreasen NC, Tsuang MD, eds.
Negative Versus Positive Schizophrenia. Heidelberg, Germany: Springer-Verlag Berlin;
1991:28-51. [Context Link]
62. Andreasen NC. Thought, language, and communication disorders, I: clinical
assessment, definition of terms, and evaluation of their reliability. Arch Gen Psychiatry.
1979;36:1315-1321. Bibliographic Links [Context Link]
63. Andreasen NC. Thought, language, and communication disorders, II: diagnostic
significance. Arch Gen Psychiatry. 1979;36:1325-1330. Bibliographic Links [Context
Link]
64. Johnstone EC, Owens DGC, Firth CD, Crow TJ. The relative stability of positive
and negative features in chronic schizophrenia. Br J Psychiatry. 1986;150:60-64.
Bibliographic Links [Context Link]
65. Carpenter WT, Kirkpatrick B. The heterogeneity of the long-term course of
schizophrenia. Schizophr Bull. 1988;14:645-659. Bibliographic Links [Context Link]
66. Buchanan RW, Kirkpatrick B, Heinrichs DW, Carpenter WT. Clinical correlates of
the deficit syndrome of schizophrenia. Am J Psychiatry. 1990;147:290-294.
Bibliographic Links [Context Link]
67. Carpenter WT Jr, Heinrichs DW, Alphs LD. Treatment of negative symptoms.
Schizophr Bull. 1985;11:440-452. Bibliographic Links [Context Link]
68. McGlashan TH, Fenton WT. The positive-negative distinction in schizophrenia:
review of natural history validators. Arch Gen Psychiatry. 1992;49:63-72.
Bibliographic Links [Context Link]
69. Pogue-Geile MF, Harrow M. Negative symptoms in schizophrenia: their
longitudinal course and prognostic importance. Schizophr Bull. 1985;11:427-439.
Bibliographic Links [Context Link]
70. Sommers AS. `Negative symptoms': conceptual and methodological problems.
Schizophr Bull. 1985;11:364-373. Bibliographic Links [Context Link]
71. Crow TJ. Molecular pathology of schizophrenia: more than one disease process?
BMJ. 1980;280:66-68. Bibliographic Links [Context Link]
72. Crow TJ. Positive and negative schizophrenic symptoms and the role of dopamine.
Br J Psychiatry. 1980;137:383-386. Bibliographic Links [Context Link]
73. Crow TJ. The two syndrome concept: origins and current status. Schizophr Bull.
1985;11:471-486. Bibliographic Links [Context Link]

74. American Psychiatric Association. Diagnostic and Statistical Manual of Mental

Disorders, Revised Third Edition. Washington, DC: American Psychiatric Association;
1987. [Context Link]
75. American Psychiatric Association. Diagnostic and Statistical Manual of Mental
Disorders, Fourth Edition. Washington, DC: American Psychiatric Association; 1994.
[Context Link]
76. Moscarelli M, Maffei C, Cesana BM. An international perspective on assessment
of negative and positive symptoms in schizophrenia. Am J Psychiatry. 1987;144:15951598. Bibliographic Links [Context Link]
77. Arndt S, Alliger RJ, Andreasen NC. The distinction of positive and negative
symptoms: the failure of a two-dimensional model. Br J Psychiatry. 1991;158:317-322.
Bibliographic Links [Context Link]
78. Miller DD, Arndt S, Andreasen NC. Alogia, attentional impairment, and
inappropriate affect: their status in the dimensions of schizophrenia. Comp Psychiatry.
1993;34:221-226. Bibliographic Links [Context Link]
79. Bilder RM, Mukherjee S, Rieder RO, Pandurangi AK. Symptomatic and
neuropsychological components of defect states. Schizophr Bull. 1985;11:409-491.
Bibliographic Links [Context Link]
80. Kulhara P, Kota SK, Joseph S. Positive and negative subtypes of schizophrenia: a
study from India. Acta Psychiatr Scand. 1986;74:353-359. Bibliographic Links
[Context Link]
81. Lenzenweger MF, Dworkin RH, Wethington E. Models of positive and negative
symptoms in schizophrenia: an empirical evaluation of latent structures. J Abnorm
Psychol. 1989;98:62-70. Bibliographic Links [Context Link]
82. Schuldberg D, Quinlan DM, Morgenstern H, Glazer W. Positive and negative
symptoms in chronic psychiatric outpatients: reliability, stability, and factor structure:
psychological assessment. J Consult Clin Psychol. 1990;2:262-268. [Context Link]
83. Gur RE, Mozley D, Resnick SM, Levick S, Erwin R, Saykin AJ, Gur RC. Relations
amoung clinical scales in schizophrenia. Am J Psychiatry. 1991;148:472-478.
Bibliographic Links [Context Link]
84. Minas IH, Stuart GW, Klimidis S, Jackson HJ, Singh BS, Copolov DL. Positive
and negtive symptoms in the psychoses: multidimensional scaling of SAPS and SANS
items. Schizophr Res. 1992;8:143-156. Bibliographic Links [Context Link]
85. Brown KW, White T. Syndromes of chronic schizophrenia and some clinical
correlates. Br J Psychiatry. 1992;161:317-322. Bibliographic Links [Context Link]
86. Peralta V, de Leon J, Cuesta MJ. Are there more than two syndromes in
schizophrenia? a critique of the positive-negative dichotomy. Br J Psychiatry.
1992;161:335-343. Bibliographic Links [Context Link]
87. Gorsuch RL. Factor Analysis. Philadelphia, Pa: WB Saunders Co; 1974. [Context
Link]

88. Liddle PF, Friston KJ, Frith CD, Frackowiak RSJ. Cerebral blood flow and mental
processes in schizophrenia. J R Soc Med. 1992;85:224-226. Bibliographic Links
[Context Link]
89. Andreasen NC, Rezai K, Alliger R, Swayze VW, Flaum M, Kirchner P, Cohen G,
O'Leary DS. Hypofrontality in neuroleptic-naive and chronic schizophenic patients:
assessment with xenon 133 single-photon emission computed tomography and the
Tower of London. Arch Gen Psychiatry. 1992;49:943-958. Bibliographic Links
[Context Link]
90. Flaum M, Andreasen NC, Arndt S. The Iowa Prospective Longitudinal Study of
recent-onset psychoses: methodological and statistical strategies. Schizophr Bull.
1992;18:481-490. [Context Link]
91. Andreasen NC, Flaum M, Arndt S. The Comprehensive Assessment of Symptoms
and History (CASH): an instrument for assessing diagnosis and psychopathology. Arch
Gen Psychiatry. 1992;49:615-623. Bibliographic Links [Context Link]
92. Andreasen NC. PSYCH-BASE. Iowa City, Iowa: The University of Iowa; 1989.
[Context Link]
93. Andreasen NC. The Scale for the Assessment of Negative Symptoms (SANS).
Iowa City, Iowa: The University of Iowa; 1983. [Context Link]
94. Andreasen NC. The Scale for the Assessment of Positive Symptoms (SAPS). Iowa
City, Iowa: The University of Iowa; 1984. [Context Link]
95. Andreasen NC, Arndt S, Alliger A, Miller D, Flaum M. Symptoms of
schizophrenia: methods, meanings, and mechanisms. Arch Gen Psychiatry. 1995;
52:341-351. Texto Completo OVID Bibliographic Links [Context Link]
96. Cattell RB. The structuring of change by P-technique and incremental R-technique.
In: Harris CW, ed. Problems of Measuring Change. Madison, Wis: The University of
Wisconsin Press; 1963. [Context Link]