Gender and the expression of schizophrenia

Gender and the expression of schizophrenia

J. psychiat. Res., Vol.22, No. 2, pp. 141-155, 1988. 0022-3956/88 $3.00+ .00 © 1988PergamonPresspie Printedin Great Britain. GENDER AND THE EXPRES...

1MB Sizes 0 Downloads 5 Views

J. psychiat. Res., Vol.22, No. 2, pp. 141-155, 1988.

0022-3956/88 $3.00+ .00 © 1988PergamonPresspie

Printedin Great Britain.




JILL M. GOLDSTEIN*$ and BgucE G. LINer *Section of Psychiatric Epidemiology and Genetics, Massachusetts Mental Health Center, Harvard Medical School and Psychiatry Service, Brockton-West Roxbury Veterans Administration Medical Center, and the Bigel Institute for Health Policy, Florence Heller Graduate School, Brandeis University; $Social Psychiatry Research Unit, New York State Psychiatric Institute, and the Department of Epidemiology, Columbia University, School of Public Health (Received 8 June 1987; revised 30 November 1987)

Summary--The expression of schizophrenia was examined in 169 DSM-III diagnosed schizophrenics. Restricted maximum likelihood factor analysis was used to test the invariance of the hypothesized symptom model across gender. Findings indicated that schizophrenic women not only expressed more impulsivity and affective symptomatology than did men, but their psychotic symptoms covaried consistently with the expression of impulsivity, anger and other affective symptomatology. Men's expression of schizophrenia covaried positively with withdrawal/isolation and an inability to function, suggesting a possible negative symptom pattern. Gender differences were not attributable to misclassification, differences in diagnostic subtypes, nor to selection. Results are discussed in light of their implications for understanding the heterogeneity of schizophrenia.


THERE is now considerable evidence suggesting that there are gender differences in schizophrenia regarding age at onset, course, premorbid functioning, expression of symptoms, family history and biological factors (GIT~L~a,T-I~um and KLEIN, 1969; GOLDSTEn'~, 1988; HL~ER et al., 1980; LEWINE, 1987; LORANGER, 1984; NASR~XL~, 1987; NY~A~ and JONSSON,1983; POGUE-GE~A~.and HARROW, 1984; S.~OK~XTGAS,1983; S E E ~ , 1985). However, much of the literature on schizophrenia assumes that the illness is similar in men and in women. The main critique of studies that men and women differ in their manifestation of schizophrenia, is that the differences are clue to the diagnostic inaccuracy of the DSM-II criteria of schizophrenia employed in these investigations. Specifically, it is argued that women are more likely than men to be misdiagnosed as schizophrenic using DSM-II criteria, and rediagnosed as affective disorders using DSM-III criteria (L~WlNE et al., 1984). Therefore, it is not that schizophrenic men and women differ, but that women who differ from schizophrenic men actually have affective disorders. This has consequences for the expression and course of the illness, since people with affective disorders have been found to have better prognoses than those with schizophrenic disorders (POPE et aL, 1980; TSUANO et aL, 1976), and certainly, given the nature of diagnostic criteria, to express the illness differently as well. The debate over the validity of gender differences in schizophrenia has important consequences for understanding the nature of the illness. That is, if there are significant SAddress communications to Dr Goldstein at the Psychiatry Service (l16A), Brockton-West Roxbury Veterans Administration Medical Center, 940 Belmont Street, Brockton, MA 02401. 141



gender differences in the expression and course of the illness, perhaps what is diagnosed as "schizophrenia" is expressed as different subtypes in men as opposed to in women. Although investigators have examined the contribution of non-symptom factors to the heterogeneity of schizophrenia, the explanatory domains have been primarily biological (F~oR-HENRY, 1974; JOHr~STO~-Eet al., 1976; KETYet al., 1978; MELTZERand STAr~, 1976; Wv.mBwRO~R et aL, 1983). There has been less work examining how characteristics such as gender have implications for understanding the nature of the disorder. The purpose of this study is to contribute to our understanding of the heterogeneity of schizophrenia by examining gender differences in the clinical expression of the illness. Findings have implications for the validity of earlier studies on gender differences that employed DSMII criteria of schizophrenia. Past literature Early work provided some evidence that schizophrenic men and women exhibited distinctly different patterns of symptomatology. The research designs in these studies consisted of survey techniques, systematic observations, and family interaction studies (CHEEK, 1964; LUCASet al., 1962; W•ICH, 1968). Taken individually each of these studies had methodological problems, such as sampling biases and diagnostic and measurement difficulties, although there were intriguing consistencies across them. Schizophrenic women were found to be more explosive, overtly hostile, physically active and dominating, seductive, sexually delusional, displaying more acting-out behavior, agitation, and more affective symptoms in general. Schizophrenic men were found to be less exhibitionistic, less hostile, less impulsive, more withdrawn, passive and isolated, symptoms that are now associated with a so-called negative symptom picture (Sommers, 1985). There is recent work that suggests gender differences in symptom expression showing a pattern of associations between gender, premorbid functioning and negative symptom expression. That is, negative symptoms have been found to significantly relate to poorer premorbid history, larger brain ventricles, poorer neurocognitive functioning, poorer neuroleptic response and poorer course for schizophrenia (ANDREASEr~and OLSEN, 1982; CROW, 1985; GOLDBERG, 1985; SOMMERS, 1985). Further, schizophrenic men have been found to have a significantly poorer premorbid picture (GrrTELMAN-KLE~and KLEIN, 1969) as well as poorer outcome than schizophrenic women (GoLDSTEIN, 1988; HUBER et aL, 1980; SALOKA~rOAS,1983). Finally, a few studies have directly suggested that men express more negative symptoms (LEwlNE and MELTZER, 1984; NASRALLaH,1987; POGUE-GEILE and HARROW, 1984), although the sample sizes were either small (LEwlNE and MELTZER, 1984; POGUE-GEn.E and HARROW, 1984) or biased towards poor prognostic patients (NAsRALLAH, 1987). TO summarize, a review of the literature suggests that men and women may differ in the clinical expression of schizophrenia. Since the differentiation of clinical phenomenology is an important criterion for validating a diagnosis (ROBINSand GuzE, 1970), an examination of gender differences may contribute to our understanding the heterogeneity of schizophrenia. Given the potential importance of the problem, our review of the literature suggests the need for two types of study. First and most obvious, we need new research focusing on gender differences that is



conducted using current diagnostic criteria. Second and less obvious, we need to find studies conducted during the DSM-II era that have well-preserved clinical data so that we can rediagnose cases from such studies according to DSM-III criteria. Only this way will we know whether these studies have left us a legacy of interesting findings concerning gender differences in clinical expression or a mass of misleading data about misdiagnosed patients. Our study is of the second type. It reports results from a DSM-II era study which provided remarkably detailed clinical data that allowed us to rediagnose patients according to DSM-III ' criteria. Analytic approach For a systematic exploration of gender differences in the clinical expression of schizophrenia, it is necessary to begin at a more fundamental level than has some o f the past research. Previous work has examined whether men or women exhibit more of one type of symptom than another. We were also interested in whether the signs and symptoms o f schizophrenia cluster, or covary, in different ways in men as opposed to in women. If symptoms do in fact covary differently in men and women, it would suggest that the same behavioral symptom may have a different meaning for the different sexes, and this in turn may provide clues to etiological processes. The following hypotheses were tested in this study. Hol. Schizophrenia in women is expressed differently from schizophrenia in men. Schizophrenic women express more impulsivity, aggressiveness, anger, and other affective symptomatology than schizophrenic men. Schizophrenic men express more withdrawal, isolation, and passivity than women. HOE- In addition, the psychotic symptoms in schizophrenic men covary positively and more strongly with the expression o f passivity, isolation and withdrawal than they do in women. Psychotic symptoms in women covary positively and more strongly with impulsivity, aggressiveness and hostility, deviant sexual behavior, and affective symptomatology in general. Alternative Ho. Schizophrenia only looks like it is different in men vs women because of differences in diagnostic criteria. Women are more likely to be misdiagnosed as schizophrenic during the era of DSM-II criteria, and are more likely to be diagnosed as affective disorders. Thus it is not that schizophrenia in men and women differs, but that the women who differ from schizophrenic men actually have affective disorders. Sample and methods The data for the present analyses were originally collected from the early to mid 1970s for an NIMH-funded study called the " S y m p t o m Tolerance and Recidivism Study", by Drs Dolores Kreisman and Virginia Joy. The study was primarily conducted at a private teaching psychiatric hospital in New York, although a small proportion came from the state facility associated with this hospital. Patients were entered consecutively into the study during 1972-1973, if they met the following criteria: (1) must have had a DSM-II hospital diagnosis o f schizophrenia or acute schizophrenia on both admission and discharge; (2) must have been in the hospital less than six months and soon to be discharged;



(3) must have been returning to live with family of origin or spouse; (4) must not have had organicity, e.g. mental retardation, epilepsy, organic brain syndrome; (5) must not have had primary drug or alcohol abuse; (6) must have been between 18 and 45 yr old. Patients were rediagnosed by DSM-III criteria. Rediagnoses were made from microfilm of chart records and medical test reports. The records at this teaching hospital were excellent, allowing us to achieve our goal of rediagnosing patients from a DSM-II era study. They provided rich detailed histories of the patients' lives, the development and severity of their symptoms, and all medical and psychological tests, which allowed for the precision of our diagnoses (reported below). The diagnosis of schizoaffective disorder was included, because it was of interest as a possible subtype of schizophrenia. (Analyses were conducted with and without schizoaffectives.) Rediagnoses were conducted by one of the authors (J.G.) and a psychiatrist/diagnostician who was blind to study hypotheses. Since the author was not blind, a reliability study (reported below) was conducted by an expert diagnostician in the Biometrics Research Unit at the New York State Psychiatric Institute. All of the women~ a random subsample of the men, plus any man with a DSM-II diagnosis of schizoaffective disorder or latent schizophrenia were rediagnosed. This procedure was followed, since out of a random sample of 25 women and 25 men with diagnoses of DSMII schizophrenia, over a third of the women were misclassified as compared to approximately one tenth of the men. Out of the original 199 patients with DSM-II schizophrenia, acute schizophrenia or schizoaffective disorder, 169 (104 men and 65 women) met criteria for DSM-III schizophrenia, schizophreniform or schizoaffective disorders. Eight cases, half men and half women, were randomly selected within gender for the reliability study. The case charts were xeroxed, excluding names, medications or any other possible reference to the diagnosis. The information was given to two expert diagnosticians in the Biometrics Research Unit at the Psychiatric Institute, who were blind to the study hypotheses. The unweighted Kappa was good at 0.80 (FLEISS, 1983). There was only one disagreement between schizophrenia (the expert's diagnosis) and schizoaffective depressed (the researcher's diagnosis). In fact, the expert noted in her report that she could not rule out schizoaffective, depressed type.

Assessment of psychopathology Interviews assessing symptomatology were conducted with the patient and a family member by masters' level interviewers. Symptoms were assessed just prior to the index hospitalization when there likely would be the most variability. A short vignette illustrating the symptom was read to the patient and family member and the respondent was asked, "Was 'X' (or were you) anything like this just before entering into the hospital?" Ratings were based on a three point scale: (1) no/never; (2) sometimes; (3) often. Although, the range of symptom items did not cover all of the symptomatology relevant to schizophrenia, the items measured aspects of clinical impairment that were relevant to the comparison of schizophrenic men and women. Symptoms included auditory hallucinations, paranoia, isolation, withdrawal, agitation, anxiety, incoherence, grandiosity, impulsivity, inability to function, inappropriate affect and behavior, depressive mood, obsessive thought and



compulsions, somatic complaints, and dependency. For the analyses conducted in this study, family assessments of the patient's symptomatology were used. Family assessments of patient's symptomatic behavior have been found to be reliable and valid in a number of other studies (ELLsWORTHet al., 1968; GLAZERet al., 1980; HOGARTY, 1975; STRAUSSet al., 1978; SMALLet al., 1965). Analyses were also conducted using patient interviews assessing symptomatology and results were similar.

Sample characteristics There were 104 men (62%) and 65 (38%) women. The majority of the patients were young, mostly in their twenties with a mean age of 26 ± 6.6 yr, and never married (82%). The sample was a non-hispanic white (96%), mostly middle class group, who were more educated than is generally cited in the literature (3ceducation = 13.2 ± 2.2 yr). Thirty-five percent of the sample were working in the six months preceding the index hospitalization, only 6.5% were housewives, and 40% were unemployed in the labor market. They were mainly Jewish (47%) and Catholic (38.5%) reflecting the catchment area of the hospitals. Finally, the majority, 80%, were diagnosed as schizophrenia by DSM-III criteria, 9.5% schizophreniform, and 11% schizoaffective. In addition, the majority were in the early phase of their disorder, since 54.5% were experiencing their first or second hospitalization. Analytic procedure In order to test the study hypotheses, restricted maximum likelihood factor analysis was used as programmed in LISREL (Jo~SKOG and SOI~OM, 1981). Assuming that the data are multivariate normal, one may apply tests of significance to the factor model and its parameters. Confirmatory factor analysis attempts to explain the covariances among the observed variables in terms of prespecified latent variables and residuals which may be independent or correlated. One of the advantages of LISREL and our main reasons for using it is that it can be applied to testing the differences in models across groups (JoRESKOG, 1978; JORESKO6and GOLDBER6ER, 1972). For the purposes of this study, this means that the symptom model can be tested for schizophrenic men versus women. There are five tests for "factorial invariance" (JormSKO¢ and GOLDBERGER, 1972). These are tests for: (1) differences in covariance and correlation matrices; (2) differences in number of factors; (3) differences in factor loadings; (4) differences in residuals (variance unexplained by the factors); and (5) differences in the correlations between the factors. These tests will answer the following questions about symptom expression in schizophrenic men and women. Do men and women have the same number of factors generating the relationships among symptom variables? If so, does each symptom contribute similarly for men and women to defining specific symptom dimensions? In other words, if some of the variables share an underlying psychotic dimension, does each symptom relate to the dimension similarly for men and women? Or, do schizophrenic men and women express their psychosis differently? Finally, are there other causal processes, that although unspecified in the model are generating the symptom data, and that differ for men and



women? That is, do men and women express symptoms differently in other ways that the factors might not account for? These would be reflected as correlations among residuals using the LISREL program.


t-tests and chi-square statistics were used to test for differences between the characteristics of men versus women. Schizophrenic men and women did not significantly differ in terms of age, education, or number o f previous hospitalizations. There were more women who were either currently married or had been married, although the difference was just insignificant (26% of women vs 13.5°/0 of men, X2 = 3.50, d f = 1, P = 0.06). Schizophrenic men and women significantly differed in terms of their occupational status as well as diagnosis. Not surprisingly, there were significantly more men than women with blue collar jobs (X2=9.56, d f = 1, P = 0 . 0 0 2 ) , and more women were housewives, although the number was small (N= 11, 17%). Although men were more likely to be unemployed in the six months preceding hospitalization, 410/0 ( N = 4 3 ) vs 37% (N=24), the difference was not statistically significant. Finally, there was a significant difference between men and women on the types of schizophrenic diagnoses that they received (X2 = 16.23, d f = 4, P < 0.003). The difference was due to the increased percentage of women with schizophreniform or schizoaffective diagnoses. Approximately 6% ( N = 6) of the men vs 15.4% ( N = 10) of the women received a schizophreniform diagnosis, and 5% (N= 5) of the men received schizoaffective diagnoses vs 20% ( N = 13) of the women. Hypotheses regarding gender differences predict that if the expression of schizophrenia is the same for men and women, there will be no significant differences in the amount o f symptomatology or the covariation o f symptoms.

Mean levels o f s y m p t o m s in men as opposed to women

Table 1 presents t-tests o f the mean differences between men and women on individual symptom items. The significance o f the multivariate T score, or Hotellings T 2, ( F = 2.1) ( d f = 17, 151) P < 0.01), demonstrated that there were significant differences in the mean levels o f symptoms expressed by men compared to women. The multivariate T protects against Type I error and allows an examination of gender differences in individual symptom items. As is evident from the Table, women exhibited significantly more paranoia ( T = - 2 . 0 4 , P = 0 . 0 4 ) , impulsivity, manifest as indiscretion ( T = - 2 . 2 3 , P = 0 . 0 3 ) and sexually inappropriate behavior ( T = - 2.19, P = 0.03), and depressive symptomatology, manifest as depressive mood ( T = - 2 . 0 5 , P = 0.04) and obsessive thinking and behavior ( T = - 3 . 2 7 , P = 0 . 0 0 1 ) . Men exhibited more withdrawal/isolation and an inability to function than women, but the differences did not reach significance. These findings are consistent with past literature describing schizophrenic women as more impulsive, sexually deviant, and depressive, and schizophrenic men as exhibiting more symptomatology resembling the deficit syndrome.

GENDER AND THE Exvm~ss~o~ OE ScmzovnREm_~



Withdrawal/Isolation Agitation Anger Paranoia Hallucinations Stubborn Obsessive Thinking/ Behavior Incoherence Indiscretion Depressive Mood Grandiosity Does Nothing Blames Others Somatic Complaints Dependency Inappropriate Affect/ Behavior Sexually Inappropriate

Women (N= 65)

2-tailed probability





t (df= 167)

2.18 1.98 1.52 1.93 1.78 1.82

.87 .91 .78 .87 .88 .85

1.98 2.16 1.71 2.21 1.89 1.86

.82 .89 .86 .87 .90 .86

1.42 - 1.27 - 1.48 - 2.04 -0.80 -0.28

.16 ns .14 .04* ns ns

1.96 1.59 1.32 2.29 1.45 2.35 1.86 1.62 2.18

.88 .82 .65 .84 .78 .87 .88 .83 .85

2.40 1.52 1.56 2.55 1.55 2.27 2.02 1.59 2.42

.82 .82 .78 .76 .80 .81 .89 .82 .82

- 3.27 0.55 -2.23 - 2.05 - 0.63 0.54 - 1.14 0.09 - 1.78

.001"* ns .03* .04* ns ns ns ns .08

1.26 1.09

.58 .37

1.39 1.29

.72 .67

- 1.21 -2.19

ns .03*

Hotelling's T=0.24; F=2.1 (df= 17, 151); P < 0.01.

The covariation o f symptoms in men as opposed to women T h e s e c o n d h y p o t h e s i s s y s t e m a t i c a l l y tested w h e t h e r s y m p t o m s c o v a r i e d d i f f e r e n t l y in m e n a n d in w o m e n . T h r e e f a c t o r s were specified o n t h e basis o f e x a m i n i n g t h e a v a i l a b l e items, the descriptive literature o n gender differences in s y m p t o m a t o l o g y , a n d a n e x p l o r a t o r y f a c t o r analysis t o see i f t h e h y p o t h e s i z e d f a c t o r s were p l a u s i b l e f o r t h e w h o l e s a m p l e . Specification o f the symptom model. F i g u r e 1 presents t h e s p e c i f i c a t i o n o f t h e m o d e l . F a c t o r 1, called " P s y c h o t i c " , i n c l u d e d b o t h p s y c h o t i c s y m p t o m s a n d t h o s e t h a t a r e associated with psychotic regression, such as impulsivity, socially i n a p p r o p r i a t e a n d bizarre behavior, a n d agitation. F a c t o r 2, called " D e p r e s s i v e " , included s y m p t o m s t h a t are typically a s s o c i a t e d with d e p r e s s i o n , such as depressive m o o d , s o m a t i c c o m p l a i n t s , obsessiveness, a n d passivity. F a c t o r 3, called " A n g e r " , included three items reflecting anger a n d defiance, as well as g r a n d i o s i t y . G r a n d i o s i t y was p l a c e d o n f a c t o r 3 in o r d e r t o a c c o u n t for its e x p r e s s i o n in s c h i z o a f f e c t i v e p a t i e n t s . A s c a n b e seen in Fig. 1, several residuals were a l l o w e d to c o v a r y . This was d o n e b e c a u s e we believed t h a t t h e c o m m o n f a c t o r s a l o n e w o u l d n o t a c c o u n t for all o f t h e i m p o r t a n t c o v a r i a t i o n a m o n g s y m p t o m s in c o m p a r i n g m e n a n d w o m e n . This was tested a n d f o u n d t r u e . I n p a r t i c u l a r it was believed t h a t t h e c o m m o n f a c t o r s w o u l d n o t a c c o u n t f o r t h e e x p r e s s i o n o f negative s y m p t o m s in m e n , since t h e s y m p t o m scale used d i d n o t explicitly measure them. I n o r d e r t o reflect this, c e r t a i n r e s i d u a l c o v a r i a n c e s were i n c l u d e d in t h e m o d e l . These c o v a r i a n c e s a m o n g residuals were i n c l u d e d a f t e r e x a m i n i n g t h e M o d i f i c a t i o n Indices p r o d u c e d b y L I S R E L . W h i l e this is n o t t h e ideal w a y to specify a m o d e l , its l i m i t e d use w h e n c o u p l e d with g u i d a n c e f r o m t h e o r y is j u s t i f i e d .




.~ Indiscretion • Hallucinations

9 q

E 1 (~ 2

...~. Grandiosity


~ 3 ~k






~ 5







E 8

~ -Inappropriate



2 , 12

Affect/Behavior nappropriateness




,(~ 9

~ " ' ~Depressive Mood



Isolation/Withdrawal Does Nothing

~ - , - ~ 11 4~/" 4t ~12~

Somatic Complaints



• Obsessive Behavior


E14 615



Blames Others








FiG. 1. Specification of the measurement model for expression of symptoms. 02,12: example of a residual covariance allowed to correlate, i.e. Hallucinations and Does Nothing (inability to function). Other residual covariancesallowed to correlatewere: (Hallucinations,Blame); (Paranoia, Obsessive); (Agitation, Does Nothing); (Inappropriate, Anger); (DePressive, Obsessive); (Indiscretion, Isolation). El, E2, and E3 can be correlated. It was believed that freeing the residual covariances that reflected the expression o f negative symptoms, would be significant for m e n but not for women. For example, the residual covariance o f Hallucinations and the symptom called, Does Nothing, was freed since it might indicate whether psychosis in men covaried with a possible indicator o f a loss o f drive or will (Does Nothing). The residual covariance o f Hallucinations and Blame was freed, since impulsivity and angry outbursts have been associated with schizophrenic w o m e n ' s expression o f psychosis (CrmsLER, 1972; WEICH, 1968). Finally to capture the hypothesized differences between men and w o m e n more fully, Indiscretion and Isolation, Agitation and Does Nothing, and Inappropriate Affect and Anger were freed. With the model specified ( J o ~ s K o G and GOLDBERGER, 1972), the question o f interest turned to whether there were gender differences in the covariation o f the patterning o f symptomatology. As discussed previously, there are several tests o f the invariance hypothesis. (1) Test of the invariance ofcovariance matrices. First, the equality of covariance matrices o f the symptom items for men versus w o m e n was tested. The hypothesis that they were equal was rejected (x2=251.44 ( d f = 1 9 0 ) , P < 0 . 0 0 2 ) . This means that there were



significant differences in how symptoms covaried among men compared to women. The next set of tests identified more precisely the sources of those differences. (2) Test of the specified factor pattern. The second test involved a determination of whether the three factors and residual covariances, specified in Fig. 1, could be applied to both men and women. That is, were the same number of factors and residual covariances, i.e. factor pattern, generating the relationships between symptom items among men compared to among women? A X~ goodness-of-fit statistic tests whether there were significant differences in the factor pattern for men versus women. The ~kWaS 240.77 with 216 degrees of freedom, at P < 0.119, suggesting that the factor pattern, or form of the model, was invariant across gender. Thus, while the men and women in this sample differed in symptom expression (test 1), they do not appear to differ extensively in the structural relations among these symptoms, since the form of model in Fig. 1 fits both sexes. This suggests that the reason for gender differences lies in the magnitude of the coefficients associated with the symptom covariances in Fig. 1. Differences may be located in the magnitude of the factor loadings, the residuals or both. (3) Test of the invariance of the loadings. Table 2 presents the unstandardized maximum likelihood estimates for the matrices of the loadings in men as compared to women. As is standard practice, one variable for each factor was chosen for the scale metric: Indiscretion for F1, Dependency for F2, and Anger for F3. The magnitude of the loadings within group are in relation to the variable chosen for the metric. One can examine the loadings within group to see which items were significant for men and which were significant for women. As can be seen in Table 2, sexually inappropriate behavior did not significantly load on the Psychotic factor for men. The other items on this factor generally were significant and of comparable magnitude, as well as all items on the Depressive and Anger factors. On the other hand, for women, Grandiosity loaded significantly on Anger but not on Psychotic and Depressive Mood, and Isolation did not have significant loadings on the Depressive factor. A test was conducted of whether the factor loadings of men versus women were significantly different from each other. In order to do this, parameter estimates of the factor loadings for women are constrained to equal men's. This constrained model produces an overall goodness-of-fit statistic or Xa. In order to test whether the loadings are significantly different from each other, the chi-square from the unconstrained model is compared to the chi-square of the constrained model, by subtracting the former from the latter k~. If the resulting ~ is significantly greater than its degrees of freedom, the hypothesis of invariance across gender is rejected. That is, the constrained model would not have fit the data as well as the unconstrained model, thereby suggesting that the loadings of men and women differed. Results showed a ~ of 25.44 with 15 degrees of freedom, that was significant at P = 0.05. Thus, men's and women's loadings on the factors significantly differed. In order to understand which parameter estimates were significantly different from each other, LISREL provides "Modification Indices" (MIs) (JoRESKOGand SORBOM, 1981). The modification indices provide insights into which items cannot be accounted for by constraining the women's loadings to equal men's. Results showed that symptom items








cq O g~





o m o o~






that contributed significantly to the overall difference between men and women were Sexually Inappropriate Behavior and two depressive symptoms, Depressive Mood and Somatic Complaints. Findings suggested that the covariance of Depressive Mood and Somatic Complaints with the Psychotic factor was not accounted for when women's loadings were constrained to equal men's. Thus, depressive symptomatology was more highly related to women's expression of psychosis than to men's. In addition, women and men significantly differed in their expression of socially inappropriate sexual behavior. Findings suggested that when women psychotically regress they expressed sexually inappropriate behavior while this was not as typical a pattern for the men. (4) Differences in the residuals. Since the factor pattern was not invariant across gender, it was not meaningful to test whether the residual estimates were significantly different for men versus women, nor whether the correlations between factors were invariant (J o ~s Ko 6 and GOr~DBEROER, 1972). The reason for this is that when the loadings significantly vary in magnitude between two groups, it means that the factors are not defined in the same way in the different groups. As a result, one cannot meaningfully compare correlations between factors or assess how much covariance among items the factors explain. However, a test was conducted in order to examine whether negative symptoms correlated more highly with psychosis in men than in women, since it was not meaningful to test the residual covariances in the model that we hypothesized would shed light on this question. The depressive factor was broken up into two factors: the two possible negative symptoms, i.e. Isolation/Withdrawal and Does Nothing, and the four depressive ones, i.e. Depressive Mood, Obsessiveness, Somatic Complaints, and Dependency. The factors were scored using unit weighting and taking an average of the items. The four-item depression factor was then partiaUed from the negative symptom factor and correlated with the psychotic factor for men and women separately. The hypothesis was that the negative symptom factor would correlate higher with psychosis for men, and the depressive factor would correlate higher with psychosis for women, given that schizophrenic women expressed more affective symptomatology. Findings showed that the correlation of negative symptoms with psychosis, partialled for depression, was higher among schizophrenic men than women (for men: beta= 0.10 vs for women: beta= -0.09). The correlation of depressive symptoms with psychosis, partialled for negative symptoms, was higher among schizophrenic women than men (for women: beta = 0.49 vs for men: beta = 0.34). Thus, findings suggested that the expression of negative symptomatology may be more common with psychosis in schizophrenic men and affective symptoms may be more common with psychosis among schizophrenic women, although further work is needed in order to test this.

Summary and comment The analysis of the symptom data suggests that schizophrenic men and women differed in their expression of the illness, both in terms of the amount of certain symptoms expressed as well as the organization of the symptom patterns. Women expressed higher mean levels of paranoia, impulsivity, manifest as indiscretion and sexually inappropriate behavior, and depressive symptomatology, manifest as depressive mood and obsessive thinking and



behavior, as compared to men. Schizophrenic men exhibited more withdrawal and isolation and a greater inability to function, although the differences were not significant. In terms of the organization of symptoms, findings showed that the form of the model was similar for men and women, but the magnitude of certain covariances for men and women significantly differed. This was attributed to the expression of deviant sexual behavior and to depressive symptomatology, such as somatic complaints. Findings also suggested that men and women may differ in their expression of negative symptoms, although this must be explored further. Among men, the higher correlation of the negative symptoms with psychosis, partialled for depressive ones, supports this interpretation. Significant gender differences in the mean levels and covariation of symptoms remained when analyses were conducted excluding the multiply hospitalized cases in the sample in order to control for the chronicity of the patient's illness, and excluding the schizoaffective patients. The results of this study are coherent with previous work that has suggested that schizophrenic women express more impulsivity, explosivity, sexually deviant behavior, and more affective symptomatology, and that the expression of negative symptomatology may be more common among schizophrenic men. Previous work has been criticized as being artifacts of the diagnostic inaccuracy of DSM-II criteria for schizophrenia. However, findings from this study support the validity of past research, for a group of schizophrenic patients diagnosed by DSM-III criteria. More importantly, our work extends the findings by showing that these symptoms covary differently with psychotic regression in men versus women. Since almost all of the women were rediagnosed by DSM-III criteria, the misclassification of affective disorders among women should not be a problem. If misclassification existed, it would be a problem among the men, given that a subsample were rediagnosed. However gender differences would be attenuated if affective disorders were a problem among the men. Further, analyses were conducted excluding schizoaffective cases, and with first and second admission patients only. Thus, findings could not be attributed to diagnostic misclassification, differences in diagnostic subtypes, nor to possible biases arising from differences in the chronicity of the patient's illness. Previous work has found that approximately 9% of schizophrenic women onset after 45 years as compared to few, if any, schizophrenic men (LEw-m-~, 1980; ROSENrHAt, 1970). These late onset women have been found to express more paranoia than men or early onset women (Fom~a~STand HAY, 1972). Thus, although this study did not include these women, symptomatic differenes between men and women would most likely be enhanced rather than attenuated by including them. In addition, one might claim that schizophrenic patients returning to live in a family setting may be different than the universe of schizophrenic patients or may differentially return to families depending on gender. However, there is evidence from a number of studies that families are relied on heavily to take care of their relatives (DOLL, 1976; GOLDSTEn,r and CATON, 1983; GOTDMAN, 1982). In fact it has been demonstrated that most patients return to their families while in the early stages of their illness ( N ~ _ n and JONSSON, 1983), and it is usually after several hospitalizations that families may separate from them. Therefore, it is not likely that the requirement for a family setting has selected an unusual



sample, since most of the patients in this study are in the early stages of the course of disorder, and findings are consistent for first and second admission patients alone. Further, findings from this study are coherent with previous work that did n o t require family setting as a criterion for sample selection. Other possible critiques of this work may arise from the use of family assessments of symptomatology. However, a number of studies (listed previously) have found that family assessments of symptomatology are reliable and valid. Further, analyses were conducted using patient interviews and results were similar. Finally, findings from this study are coherent with previous work that used clinical ratings systematically collected by trained researchers (CHEEK, 1964; LEWINE, 1987; LUCASet al., 1968). Thus, findings in this study were most likely not due to family's perception of deviance from sex role behavior. Finally, one might claim that findings in this study are due to selection into hospital treatment, with males being hospitalized for negative symptomatology and females being hospitalized for impulsivity, sexual deviance, and depression symptomatology. While we cannot fully rule out this selection argument, it is unlikely for the following reasons. Past research has shown that, at most, 20% of schizophrenics are not hospitalized in the early stages of the disorder (LINK and Dom~W~ND, 1980). Thus, in order to negate findings in this study, the 20% excluded would have to consist of female schizophrenics expressing negative symptomatology and males expressing impulsivity and sexual deviance. This is unlikely given that schizophrenic men are more likely to be hospitalized involuntarily, brought in by the police, for assaultive and violent, impulsive behavior than schizophrenic women (HoLLn~GSrmADand R~DLICH, 1985; LII~SKY, 1970). Further, analyses examining family attitudes and tolerance towards their schizophrenic male or female relatives showed that, in general, sons were more likely to be rehospitalized and left in the hospital longer over a ten-year period than daughters, regardless of the type of symptoms expressed (GOLDS~m and KRv.isui~r, 1988). Thus, research suggests that findings on the differences in the expression of schizophrenia in men and women in this study are not likely due to selection into treatment. Further, if findings were due to selection, then variations in selection into treatment, such as first admission cases versus multiply hospitalized cases, should produce different results. However, there were no significant differences in the ways that schizophrenic men and women differentially expressed the illness among first versus multiple admission cases. Future research should examine gender differences in schizophrenia among cases obtained in a community study in order to fully rule out the selection argument. In addition, given that this study was a secondary analysis, our results need further replication, in particular since gender differences in this sample may be in part due to differences in the variability of some of the symptoms for men and women. However, we had a sizeable sample, who were diagnosed by current criteria, and in the early stages of their disorder to control for institutional effects. Even given the limitations of the symptoms we examined, the effect of gender on the clinical expression of schizophrenia was strong enough to demonstrate significant and meaningful results. These findings have broader implications for considering gender in understanding the nature of schizophrenia and its related disorders. For example, given that women's expression of schizophrenia covaries positively with affective symptomatology, the



distinctions between schizoaffective disorder and schizophrenia may in part be related to gender. In addition, the meaning of certain symptoms may differ for men and women. That is, the expression of isolation and withdrawal, passivity, and dependency may reflect a depressive syndrome in schizophrenic women and a negative syndrome in schizophrenic men. Thus, hypotheses should be tested regarding possible differences in etiological determinants of the symptoms. Given that the differentiation of clinical phenomenology is an important criterion for diagnostic understanding (ROBINSand GUZE, 1970), the differences between phenomenology expressed by schizophrenic men and women have implications for understanding the heterogeneity of the illness. The methodology in this study may provide an illustration of how to extend our understanding of clinical phenomenology among diagnostic groups. If a diagnostic entity is homogeneous, its indicators should show "invariance validity". Acknowledgements--The authors would like to express their appreciation to Drs Patricia Cohen, Patrick Shrout, Elmer Struening and the Ming Tsuang for their helpful comments on this work; Dr Dolores Kreisman for her generosity in offering her data for this study, and Donna Kantarges for her help in the preparation of the manuscript. Support for this research was provided by a grant from the National Institute of Mental Health, MH 09165-01.

REFERENCES A~D~ASE~, N. and OLSE~, S. (1982) Negative v. positive schizophrenia: definition and validation. Archs gen. Psychiat. 39, 789-794. Cn~ET¢, F. (1964) A serendipitious finding: sex roles and schizophrenia. J. Abnorm. $oc. Psyehol. 69, 392-400. C~SLER, P. (1972) Women and Madness. Doubleday, Garden City, NY. CRow, T. (1985) The two-syndrome concept: origins and current status. Schizophr. Bull. 11, 471-486. DOLL, W. (1976) Family coping with the mentally ill: an unanticipated problem of deinstitutionalization. Hosp. Com. Psychiat. 27, 183-185. ET-LSWORTH,R., FOSa~R, L., CI~DERS, B., G~B~RT, A. and KROE~rER, D. (1968) Hospital and community adjustment as perceived by psychiatric patients, their families and staff. J. consult, clin. Psychol. 32, Monogr. Suppl. FL~ISS, J. (1983) Statistical Methods for Rates and Proportions, 2nd Edn. John Wiley and Sons, New York. F~oR-HE~Y, P. (1974) Psychosis, neurosis, and epilepsy: developmental and gender-related effects and their aetiological contribution. Br. J. Psychol. 124, 144-150. FOREST, A. D. and HAY, A. J. (1971) Sex differences and the schizophrenic experience. Actapsychiat. scand. 47, 137-149. GrrT~r~a~-Kr~n% R. and KLan, D. (1969) Premorbid asocial adjustment and prognosis in schizophrenia. J. psychiat. Res. 7, 35-53. GLAZER,W., AARONSON,H., PRusorr, B. and W ~ s , D. (1980) Assessment of social adjustment in chronic ambulatory schizophrenics. J. nerv. ment Dis. 168, 493-497. GOLDB~Ra, S. (1985) Negative and deficit symptoms in schizophrenia do respond to neuroleptics. Schizophr. Bull. VII, 453-456. G O L D ~ , H. (1982) Integrating health and mental health services: historical obstacles and opportunities. Am. J. Psychiat. 139, 616. GOLDSa~r~, J. (1988) Gender differences in the course of schizophrenia. Am. J. Psychiat. (in press). GOLDSa'E~, J. and CATO~, C. (1983) The effects of the community environment on chronic psychiatric patients. Psychol. Med. 13, 193-199. G O r D S ~ , J. and K R ~ I S ~ , D. (1988) Gender family environment, and schizophrenia. Psychol. Med. (in press). HOa~TY, G. (1975) Informant ratings of community adjustment. In Psychotherapy Change Measures (Edited by WASKOW, R. and P~a~o~, M.). U.S. Government Printing Office, DHEW Publication #ADM 74-120. H~ER, G., GRoss, G., S c m y r ~ R , R. and Ln~z, M. (1980) Longitudinal studies of schizophrenic patients. Schizophr. Bull. 6, 592-605.



JOHNSTONE,E., CROW, T., ZmTH, C., HUSBAND,J. and KR~Er, I. (1976) Cerebral ventricular size and cognitive impairment in chronic schizophrenia. Lancet ii, 92-926. JORESKOC, K. (1978) Statistical analysis of covariance and correlation matrices. Psychometrika 43, 443-477. JORESKOO,K. and GOU)BEROER,A. (1972) Factor analysis by generalized least squares. Psychometrika 43, 443-477. JORESKOO, K., SORBOM,D. and LISRE~, V. (1981) User's Guide. National Educational Resources, Chicago. KETY, S., ROSENTrL~,D., WENDER, P., SCmn.SmOER,F. and JACOESON,B. (1978) Mental illness in the biological and adoptive families of adopted individuals who become schizophrenic: prevalence of mental illness and other characteristics. In The Nature of Schizophrenia: New Approaches to Research and Treatment (Edited by WYNNE, L., CROMWELL,R. and MATTHYSSE,S.), pp. 25-37. Wiley, New York. KOL~aCOWSKA,T., Wn-LtAMS,A., ARDEEN,M., R~WLEY, M., JAMBOR,K., GELDER,M. and MANDELEROTE,B. (1985) Schizophrenia with good and poor outcomes. I. Early clinical features, response to neuroleptics, and signs of organic dysfunction. Br. J. Psychiat. 146, 229-246. LEwn,m, R. (1980) Sex differences in age of onset and first hospitalization in schizophrenia. Am. £ Orthopsychiat. 50, 316-322. LEWlNE, R. (1987) Gender and the group of schizophrenias. In Behavioral Medicine for Women (Edited by BLECrn~AN, E. and BROWNELL,K.) LEWlNE, R., BURBACa,D. and M~LTZER, H. (1984) Effect of diagnostic criteria on the ratio of male to female schizophrenic patients. Am. J. Psychiat. 141, 84-87. LEWm~, R. and MELTZER, H. (1984) Negative symptoms and platelet monoamine oxidase activity in male schizophrenic patients. Psychiat. Res. 12, 99-109. LINK, B. and DOHRBNW~ND,B. (1980) Formulation of hypotheses about the ratio of untreated to treated cases in the true prevalence studies of functional psychoses in adults in the U.S. In Mental Illness in the United States: Epidemiologic Estimates (Edited by Dom~NWEND et al.). Praeger Press, New York. LINSKY, A. (1970) Who shall be excluded? The influence of personal attitudes in the community rejection of the mentally ill. Soc. Psychiat. 5, 166-171. LORANGER,A. (1984) Sex differences in age at onset of schizophrenia. Archs gen. Psychiat. 41, 157-161. LocAs, C., SAmSEURY,P. and COLLIns, J. (1962) Social and clinical study of delusions in schizophrenia. Br. J. Psychiat. 108, 747-758. MBLTZF.R,H. and STAHL,S. (1976) The dopamine hypothesis of schizophrenia: a review. Schizophr. Bull. 2, 19-76. NASRALL~, H. (1987) Gender differences in the etiology of schizophrenia. Paper presented at American Psychiatric Association Meetings, Chicago, IL. NYMAN, A. K. and JOHNSSON,H. (1983) Differential evaluation of outcome in schizophrenia. Acta psychiat. scand. 68, 458-475. POGUE-GE~E, M. and HARROW, M. (1984) Negative and positive symptoms in schizophrenia and depression: a follow-up. Schizophr. Bull. 10, 371-387. POPE, H., Lmmsra, J., COttON, B. and AXELROD,D. (1980) Schizoaffective disorder: an invalid diagnosis? A comparison of schizoaffective disorder, schizophrenia, and affective disorder. Am. J. Psychiat. 137, 921-927. ROBINS, E. and GUZE, S. (1970) Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. Am. J. Psychiat. 126, 107-111. ROSENTHAL,D. (1970) Genetic Theory and Abnormal Behavior. McGraw-Hill, New York. SALOKANOAS,R. K. R. (1983) Prognostic implications of the sex of schizophrenic patients. Br. J. Psychiat. 142, 145-151. SEE~AN, M. (1985) Sex and schizophrenia. Can. J. Psychiat. 30, 313-315. SMALL,J., SMALL,I. and GONZALES,C. (1965) The contribution of the informant in psychiatric evaluation. Int. J. Neuropsychiat. 1, 446-451. Som~mRS,A. (1985) Negative symptoms: conceptual and methodological problems. Schizophr. Bull. 11, 364-379. STRAUSS,J., CARPENTER,W. and NASRALLAH,A. (1978) How reliable is the psychiatric history? Comp. Psychiat. 19, 213-219. TSOANO, M., DE~WSEY,T. and RAVSC~mR,F. (1976) A study of "atypical schizophrenia": comparison with schizophrenia and affective disorder by sex, age of admission, precipitant, outcome, and family history. Archs gen. Psychiat. 33, 1157-1160. WBICH, M. (1968) Behavioral differences between groups of acutely psychotic (schizophrenic) males and females. Psychiat. Quart. 42, 107-122. WEINEEROER,D., WAGNER,R. and WYATT,R. (1983) Neuropathological studies of schizophrenia: a selective review. Schizophr. Bull. 9, 194-212.