GENDER
DIFFERENCES IN SCHIZOPHRENIA OBSERVATIONS FROM
NORTHERN
FINLAND
SAMI
RÄSÄNEN
Departments
of Psychiatry and Public
Health
Science and General Practice
(
tässä osia em. tekijän tutkimuksesta, Jk)
(URL:
http://herkules.oulu.fi/issn03553221/)
Abstract
Using
three different schizophrenic populations from
Northern
Finland
,
gender differences in some socio demographic variables, age at onset,
incidence, treatment, outcome and deinstitutionalization of schizophrenia were
examined. The first study population comprises the
Northern
Finland
1966
Birth Cohort, which is an unselected, general population birth cohort. We
followed prospectively 11017 subjects from 16 to28 years of age by means of
the Finnish Hospital Discharge Register. From this study population gender
differences at the age of onset and incidence of schizophrenia were
calculated. The second study population was formed of 1525 patients who had
their first treatment episodes at the closed therapeutic community ward
situated at the Department of Psychiatry,
University
of
Oulu
during 1977 - 1993. Gender differences were assessed in relation to age at
first admission, some sociodemographic variables,
degree of active participation of the patients in individual, group, and
milieu therapy and institutional outcome of the patients with schizophrenia.
The third study population consisted of all the 253 long-stay psychiatric
inpatients treated for at least six months without a break during 1992 in the
Department of Psychiatry,
Oulu
University
Hospital
. From
this study population gender differences at the age of onset and in relation
to some sociodemographic and clinical variables
were studied. The placements after the last discharge and at the end of the
follow-up and factors predicting hospitalization after the follow-up were also
monitored. There were no statistically significant gender differences
regarding age at onset in any of these three different study populations. The
time lag between the first psychotic symptoms and the first psychiatric
hospitalization was minimal. In the Northern Finland 1966 Birth Cohort study
the annual incidence rate of DSM-III-R schizophrenia was relatively high, 7.9
per 10 000 in men and 4.4 in women by the age of 28. In men it was highest in
the age group of the 20-24 year-olds while in women the peak occurred earlier
in the age group of the 16-19 year-olds. In the Therapeutic community study
there were no statistically significant gender differences in the sociodemographic
variables, in the length of stay and in the number of treatment episodes in
this ward in any of the diagnostic groups. Differences with regard to male and
female participation in individual, group and milieu therapy and the
institutional outcome were minimal, some trends, however, favoring
females. In the long-stay patients study almost two-thirds of these patients
were men. Very few gender differences were found in relation to sociodemographic
and clinical characteristics or regarding the utilization of psychiatric
hospital care. About 70% of the long-stay patients were discharged during the
four year follow-up period and only 15% were able to live without continuous
support. Marital status (being not married), dwelling place (living in city),
absence of negative symptoms and severity of the illness were associated with
hospitalization at the end of the follow-up. Gender did not predict
hospitalization at the end of the follow-up period. The results of this study
indicate that there are probably different subgroups of schizophrenia in which
there are no gender differences regarding age at onset and in the clinical
picture of the disturbance or there are regional differences in the
manifestation of the illness. In
Finland
patients are hospitalized earlier after the onset of the first psychotic
symptoms than in many other countries. According to the
Northern
Finland
1966
Birth Cohort study the incidence of schizophrenia is higher among young men
than women and the total life-time incidence of schizophrenia may be smaller
in women. The results from the Therapeutic community study suggest that
therapeutic community treatment may level out the gender differences in the
treatment process and outcome. The long-stay patient study showed that
long-term patients are dependent on considerable support and that the most
seriously ill patients are in fact in hospital. Alternative residential
facilities have been a presupposition to the deinstitutionalization of the
long-stay patients.
Keywords:
age at onset, incidence, long-stay patients, deinstitutionalization
4.3.
Statistical methods
4.3.1.
Methods in the
Northern
Finland
1966 Birth Cohort study (II)
Gender
differences with regard to continuous and normally distributed variables were
assessed with Student’s t-test (Altman 1991). The ordinary chi-square test
was chosen to test for the statistical significance of differences in
frequency distributions in case of categorial
variables. Finally, male to female rate ratios with 95% confidence limits were
calculated for each age group (Ingelfinger et
al. 1994). The statistical software used was SPSS for Windows version 6.1
(Norusis 1994).
4.3.2.
Methods in the therapeutic community study (III)
The
readmission period (the number of days from discharge following the first
admission to rehospitalization) was evaluated in
all diagnostic groups. The groups of schizophrenia, schizophreniform
and schizoaffective were combined, and the readmission period of this combined
group was presented graphically. The data were analyzed using the SPSS for
Windows software. Cross-tabulation was used as the main tool for data
presentation and analysis. The statistical significance of factors associated
with gender was evaluated using the Chi-square test and the Mann- Whitney
U-test. The gender differences were evaluated by diagnostic groups. The
Kaplan-Meier method, as operationalized in the
Kaplan-Meier survival analysis procedure of the SPSS for Windows software (Norusis
1994), was used to obtain the estimated time to readmission of patients. These
curves (III: Figure 1) represent the proportion of readmitted patients at a
certain time point after discharge and are expressions of the time elapsed
before rehospitalization. The curves were
approximations for male and female patients in different diagnostic groups.
The statistical significance of the differences between gender groups was
determined by log-rank statistic.
4.3.3.
Methods in the long-stay patients study (IV,V)
In
the original paper IV, the Chi-square and Fisher’s exact test were used to
test statistical significances of differences in frequency distribution
between genders. For continuous variables, statistical significance of gender
differences were assessed with Student’s t-test or the non-parametric
Mann-Whitney U-test if the variables were not normally distributed. A
two-group discriminant function analysis was
performed to determine the most essential variables that discriminated male
and female long-stay psychiatric patients. In the original paper V,
cross-tabulation was used as the main tool for data presentation. Statistical
significance of differences between two independent groups were assessed with
the Chi-square test, Fisher’s exact test or Student’s t-test, whichever
appropriate. The logistic regression analysis was used to assess the
probability of 38 hospitalization at the end of the follow-up after the most
essential clinical and sociodemographic variables
were adjusted. In the original papers IV and V the primary software used to
analyze the data were SPSS and SAS for Windows (Norusis
1994). The level of significance was set at P0.05
in all statistical analyses.
5.
Results
5.1.
Age at onset of schizophrenia by gender (II, III, IV)
5.1.1.
Age at onset in the
Northern
Finland
1966 Birth Cohort study (II)
There
were 89 DSM-III-R schizophrenia cases (58 men = 65.2% and 31 women = 34.8%) of
which one man and one woman were traced through outpatient search. Mean age at
onset (first psychotic symptoms) in men was 21.4 (SD 3.4) and in women 21.2
(SD 3.5) (Student’s t-test with equal variances p = 0.756). The peak of
onset in men lay in the 20-24 year age group whereas in women it occurred in
the 16-19 year age group. There were no statistically significant gender
difference in age at onset. Table 1 in the original paper II presents the
distributions of age at onset (first psychotic symptoms) in male and female
schizophrenia patients in three age groups. The mean age at first
hospitalization due to any psychiatric disorder in men (n=57) was 21.5 (SD
4.0) and in women (n=30) 21.4 (SD 3.9) (Student’s t-test p=0.874). The first
psychotic symptoms appeared slightly earlier than the first hospitalization.
The mean time difference between first psychotic symptoms and first
psychiatric hospital admission was 0.05 years (SD 2.4) for men and 0.20 (1.5)
for women (Student’s t-test p=0.757).
5.1.2.
Age at first treatment in the therapeutic community study (III)
The
mean age for males at the time of their first treatment episode in this ward
was 30.4 years (SD 10.4) and for females it was 31.2 (SD 10.6). The mean age
of patients with schizophrenia was lower, being 27.4 years in men (SD 7.4) and
27.8 in women (SD 7.5) and in patients with diagnoses of schizophreniform/schizoaffective
disorders 26.5 years in men (SD 7.2) and 27.4 in women (SD 7.7). Of the men
with schizophrenia 34 %, and of the women 31%, were under 23 years of age at
the onset of the illness. In the group of men with schizophreniform/schizoaffective
disorders 38% and in the women 37% were 40 under 23 years old. There were no
statistically significant differences between men and women in age
distribution in the groups of schizophrenia and schizophreniform/schizoaffective
disorders.
5.1.3.
Age at onset in the long-stay patients study (IV)
The
mean and median ages as well as standard deviations of long-stay patients at
five different points in time are presented in Table 3. There were no
statistically significant differences between men and women in any of these
age variables.
Table
3. Mean and median ages as well as standard deviations of long-stay patients
at five different points in time .Definition
of age at onset Males Females All mean median SD mean median SD mean
median SD First sign of the psychiatric symptoms
22.2
20.0 10.1 24.3 21.0 11.8 23.0 20.0 10.8
Manifestation
of the first psychotic symptoms
24.0
21.0 9.7 25.4 22.0 11.8 24.5 21.0 10.5
First
contact with health care for the psychiatric reasons
23.5
21.0 10.2 24.7 21.0 11.4 24.0 21.0 10.6
First
contact with specialist care
23.9
21.0 10.4 25.3 22.0 11.9 24.4 21.0 11.0
First
psychiatric hospital admission
24.6
21.9 10.0 26.5 22.3 13.0 25.3 22.1 11.3
Mean/median
time lag in years between the first sign of psychiatric symptoms and the first
psychiatric hospitalization was 2.5/1.0 (SD 5.1) overall, 2.6/1.0 (SD 4.4) for
males and 2.2/1.0 (SD 6.0) for females. Statistically significant gender
differences were not found. Mean/median time lag in years between the first
psychotic symptoms and the first psychiatric hospitalization was 0.06/0.0 (SD
3.5) overall, 0.04/0.0 (SD 3.3 ) in males and 0.09/0.0 (SD 3.7) in females.
Again, no statistically significant differences were observed. There were 32
patients (13.2% in all, of which 20=13.1% were male and 12=13.5% were female)
whose first psychotic symptoms appeared after the first hospitalization. 41
5.2.
Incidence of schizophrenia by gender in the
Northern
Finland
1966 Birth Cohort (II)
The
cumulative incidence was two times as high among men as among women until the
age of 28. Fig. 1 displays the cumulative incidence (%) of DSM-III-R
schizophrenia in males and females.
Fig.
1. The cumulative incidence (%) of DSM-III-R schizophrenia by gender up to
the
age of 28 in the
Northern
Finland
1966 Birth Cohort.
The
annual incidence in men was highest in the age group of the 20-24 year-olds
while in women the peak occurred in the age group of the 16-19 year-olds (II:
Table 1). Until the end of the follow-up period (at the age of 28) there was a
decreasing trend in the incidence rates for women. Male to female rate ratios
differed statistically significantly only in the age group of the 20-24
year-olds.
0,0
0,2
0,4
0,6
0,8
1,0
1,2
16
17 18 19 20 21 22 23 24 25 26 27 28
Age
at onset of schizophrenia Cumulative incidence % Males Females 42
5.3.
Treatment and outcome of schizophrenia by gender in the therapeutic community
study (III)
There
were no statistically significant (p < 0.05) differences between men and
women regarding age distribution, education, dwelling place and way of
admission of any diagnostic group. In the group with schizophrenia, however,
the general trend was that the men’s educational levels were lower than
those of the women and that most of the men were living in small localities
(under 10 000 inhabitants). Distribution of some first episode (at the ward
under study) patient variables in the therapeutic community ward are presented
in the original paper III:
Table
1. Women participated more actively than men in individual therapy of the
diagnostic group of mood disorders (very active males 60.2% and females 72.7%,
p = 0.033), as well as in group therapy in the patient group of schizophreniform/schizoaffective
disorders (very active males 34.8% and females 46.7%, p = 0.037). There were
no statistically significant differences regarding the participation in milieu
therapy of any diagnostic group. In the groups of schizophrenia and psychosis
generally, there was a trend towards women participating more actively in
milieu therapy than men. A difference in the institutional outcome was
established between men and women in the group of schizophreniform/schizoaffective
disorders (p = 0.038). A positive outcome was more common among women and a
negative outcome among men. A conflicting outcome (limitations in goal
attainment) was equally common in both gender groups. A similar trend was
observed also in the group with schizophrenia, but not in other diagnostic
groups. Data on the distributions of men and women in the participation of
psychosocial therapies, length of stay and outcome according to the five
diagnostic groups are presented in the original paper III:
Table
2. No statistically significant gender differences were found with regard to
the length of stay in the different diagnostic groups when assessed as
quartiles or when evaluating the means, medians and maxima (III: Table 3). In
the groups with schizophrenia, the lowest number of patients who had a short
length of stay were encountered. Correspondingly, in those same groups there
were the largest number of patients with very long lengths of stay when
compared with other diagnostic groups. There were no significant gender
differences in the number of treatment episodes at the ward under study in any
of the diagnostic groups. The number of episodes was highest in the group of
schizophrenia. Original paper III: Table 3 presents also the means, medians
and maxima of the number of episodes at the ward in all diagnostic groups. The
rate of readmission was evaluated in all diagnostic groups. There were no
significant gender differences in any of these groups. Among the combined
group of schizophrenia, schizophreniform and
schizoaffective patients, no gender differences were detected at the time to
readmission (III: Fig. 1). In the diagnostic groups of schizophrenia, the
proportion of male and female patients did not change during the second and
third treatment periods, but the gender differences diminished. During the
second treatment period, the gender difference came out as a more negative
outcome among males with schizophreniform/schizoaffective
disorders: positive outcome male 32 / female 34 (61.5 % / 85.0 %),
controversial 12/6 (23.1 % / 15.0 %), negative 8/0 (15.4 % / 0.0 %) (p=0.013).
During the third treatment period, there was a gender difference with regard
to the activity in milieu therapy in the group of 43 schizophreniform/schizoaffective
disorders, women being more active than men: very active male 6 / female 12
(23.1 % / 57.1 %), moderately active 12/4 (46.2 % / 19.0 %), passive 8/5 (30.8
% / 23.8 %) (p=0.044).
5.4.
Gender differences of long-stay patients (IV)
About
62% of the patients were males and 38% females. The mean/median age (standard
deviation) of all subjects at the end of the index hospitalization was
47.0/46.0 (15.5). The age of the men was 45.6/45.0 (14.4) and that of the
women 49.4/47.5 (16.9), resulting in a statistically marginal difference
between men and women with regard to age (P=0.069). Compared to women, men
were living more commonly outside the City of
Oulu
. Men
were also more commonly single, had fewer children, and lived more commonly in
a primary family than did women. In original paper IV: Table 1 presents the sociodemographic
characteristics of the patients by gender before the index hospitalization. Of
the patients 97% were on disability pension at the end of the index
hospitalization. Three of the patients were so young (under 16 years) that
they did not have the right to collect a disability pension.
About
40% of the patients had most of the time (≥75%) been in inpatient care
since their first admission to the psychiatric hospital. There were no
statistically significant gender differences in the number of admissions, in
the number of involuntary hospitalizations and in the proportion of treatment
days under psychiatric hospital care before the index hospitalization (IV:
Table 2). The diagnostic distribution did not differ significantly between
genders. There was a trend that women had more other functional psychoses and
mood disorders than men and men had more organic disorders than women. There
were no significant gender differences in CGI- and GAS-scores. Men had more
DSM-III-R negative symptoms than women. The mean of the total
Strauss-Carpenter score in men was lower (10.3, sd
4.7) than in women (11.6, sd 5.0) (t-test,
P=0.039). There was no statistically significant gender difference in the
doses of the antipsychotic medication as determined by analyses in the seven
categories and in the mean/median quantities of equivalents of chlorpromazine
(men 755/460, women 723/500) (IV: Table 3). A total of 51 patients had
depot-injection medication; 39 of them were men (24.8% of all men) and 12 of
them were women (12.5% of the women) (P=0.018). A discriminant
analysis between male and female long-stay psychiatric patients revealed that
a higher number of children (F=17,619, P<0.001), fewer negative symptoms
(F=5.402, P=0.021) and being married (F=6.305, P=0.013) were more typical for
females and served as the only significant factors for discriminating the
genders. The patients were homogeneous with regard to diagnoses. To ensure the
comparability with the previous studies concerning schizophrenia patients, all
analyses were repeated in the group of schizophrenic patients (n=203). The
results, however, remained unchanged. 44
5.5.
Deinstitutionalization of long-stay patients by gender (V)
During
the follow-up, about 90 % of all patients had had at least one
hospitalization. About one third of both men and women were less than 25% of
the follow-up period in hospital care and one third over 75% of the time.
There was no difference between men and women in the utilization of hospital
care. The placements after the last discharge and at the end of the follow-up
as well as the change of the placements between the last discharge and at the
end of the follow-up are presented graphically in Fig. 2. During the follow-up
there were only minor changes in the placements of the patients. About 90% of
the patients placed in residential facilities or at home did not change the
placement. If the patients were discharged to other institutions, 72% of them
stayed there and 25% of the patients died
a
Before index hospitalization
b
At the end of index hospitalization
c
Proportion of days in inpatient care between the first psychiatric
hospitalization and
the
beginning of index hospitalization
All
other variables (gender, number of children, way of living, age, positive and
depressive
symptoms, social functioning measured with the Strauss-Carpenter scale,
number
of involuntary hospitalizations, number of psychiatric hospitalization)
remained
non-significant.
6.
Discussion
6.1.
Main findings of the study (I, II, III, IV, V)
The
main findings in a review of gender differences in schizophrenia (I) were that
most studies have accepted that the lifetime risk of schizophrenia has been
equally common in both sexes, with men developing schizophrenia 3 - 4 years
earlier than women and women having more often a favorable
course of the disease than men. Males with schizophrenia furthermore tend to
have more central nervous system abnormalities than females. However, there
were also opposite findings, and thus the final conclusions at this point in
time, are difficult to formulate. In the
Northern
Finland
1966
Birth Cohort study (II) there were no significant gender differences in age at
the onset of schizophrenia. However, the peak of onset occurred earlier in
females than in males. The incidence rates of schizophrenia were relatively
high compared to earlier incidence studies and incidence among men was about
two times higher than in women. In the Therapeutic Community Study (III) only
minimal gender differences were observed in the sociodemographic
variables, with regard to participation in psychosocial therapies (individual,
group and milieu therapy) and in the institutional outcome of the
schizophrenia patients. In the long-stay patients study (IV,V) almost twothirds
of the long-stay patients were men. Some gender differences were found in sociodemographic
and clinical characteristics or in the utilization of psychiatric hospital
care. Long-term patients were dependent on considerable support, but only the
most seriously ill patients were left in hospital at the end of the follow-up.
Most of the patients were placed in alternative residential facilities. Gender
was not associated with hospitalization at the end of the follow-up period. 51
6.2.
Age at onset (II,III,IV)
The
Northern
Finland
1966
Birth Cohort study and the Therapeutic Community study did not detect any
gender difference with regard to age at onset of schizophrenia, which was
contradictory to findings in the numerous earlier studies (Angermeyer
& Kuhn 1988, Goldstein et al. 1989, Evenson
et al. 1993, Häfner et al. 1993, Susser
& Wanderling 1994, Maurer & Häfner
1995). In the long-stay patient study, the mean age at the onset of the
illness, using five different criteria, was between 22.2-24.0 years in men and
24.3-25.4 years in women. The results remained practically the same when only
the schizophrenia group was analysed. Age at onset in men was lower than that
of women, but the difference was about two years smaller than that found in
earlier studies on schizophrenia patients. This can be explained by the
selected study sample. Our study population represented the most chronic
schizophrenia patient group, which is probably more homogeneous with regard to
the course of the illness than the schizophrenia patient group in the broader
hospital population. In the
Northern
Finland
1966
Birth Cohort study, the probands were 28 years old
at the end of the present follow-up. This means that if there are more females
in new schizophrenia cases of the older age groups (Angermeyer
et al. 1989, Goldstein et al. 1989, Jablensky
1993) and a second peak of onset in women appears, as reported earlier (Häfner
et al. 1991, Castle et al. 1993), the gender difference with
regard to age at onset will probably appear in the future. Naturally, the
extent of the possible age difference will depend on the number of men and
women in the new schizophrenia cases of the older age groups. In the
Therapeutic Community study, the same age of males and females at their first
admissions to the ward can be explained by the selected study sample. It is
also possible that men and women were selected in different ways. Most of the
patients were selected for therapeutic community treatment by personal
interview or by discussing a patient’s medical records in a team before the
patient’s admission to the ward. For example, extensive age gaps and gender
disproportions were avoided, if possible, although different adult age groups
were admitted. Also, it was considered in the selection of the patients that
the patients benefit from the therapeutic community treatment (Isohanni
1983). In this study group there were also some patients who were not real
first timers in the psychiatric hospital, although the treatment period under
study was their first at this particular therapeutic community ward. Thus, the
findings of this study may not be generalisable
and applicable to other schizophrenic populations. In the
Northern
Finland
1966
Birth Cohort study the schizophrenia patients were divided into three age
groups according to the onset age at the first psychotic symptoms (16-19
years, 20-24 years and 25-28 years). This kind of grouping facilitated
comparisons with earlier studies. In our study a remarkable proportion (42%)
of the female schizophrenia patients had become schizophrenic before the age
of 20. This finding challenges earlier studies in which the peak of onset in
women has been reported to occur after 25 years of age (Eaton et al. 1988,
Goldstein et al. 1989, Häfner et al. 1991).
The finding may even indicate that the main peak of onset could be earlier in
women than in men. The peak of onset in men occurred in the age group of the
under 24 year-olds, which concurs with earlier studies (Eaton et al. 1988,
Goldstein et al. 1989). Naturally, this finding may be a chance
phenomenon and requires replication. 52 The onset age of schizophrenia has
been defined in several ways - from the first sign of mental disorder to the
first hospital admission with a diagnosis of schizophrenia (Häfner
et al. 1993). In spite of the different definitions, a
significant gender difference in age at onset has been found to exist (Häfner
et al. 1993, Maurer and Häfner 1995). In
the
Northern
Finland
1966
Birth Cohort study, the age at onset was defined in two ways and in the
long-stay patient study in five different ways. In the
Northern
Finland
1966
Birth Cohort study and in the long-stay patient study, there was practically
no time lag between the age at which the first psychotic symptoms appeared
according to the patients’ case notes and the age at first hospitalization
due to any psychiatric disorder. It was much smaller than in some other
studies where the first psychotic symptomatology
had been found to appear on average two years prior to the first
hospitalization with a diagnosis of schizophrenia (Häfner
et al. 1993a, McGorry et al. 1996).
The small difference noticed in this study may be due to incomplete
information, because our data for symptoms were collected from case records
retrospectively and some information may have been missing from the case
notes. In addition, the first psychiatric hospitalization may not necessarily
have been critical for the diagnosis of schizophrenia. It is also possible
that in
Finland
the
patients are hospitalized earlier than in many other countries after the
manifestation of the first psychotic symptoms. In
Finland
good
social security and relatively small localities (people know each other) have
led to the welldeveloped practice of taking care
of the patients and, therefore, this time lag may be smaller than in many
other countries.
6.3.
Incidence (II)
In
the
Northern
Finland
1966
Birth Cohort study, the incidence of schizophrenia was higher in men than in
women. The male to female rate ratio ranged from 1.3 to 2.3 in three different
age groups. These ratios were similar to ratios observed in previous studies
and thus supported the findings that incidence among young men was higher than
in young women (Sartorius et al. 1986, McCovern
& Cope 1987, Goldstein et al. 1989, Castle et al. 1993).
Opposite to our hypothesis and also to earlier studies, the male to female
rate ratio remained relatively high in the oldest age group, that of the 25-28
year-olds. There was also a trend that, regarding women, the number of new
schizophrenia cases decreased with age. Based on earlier studies we expected a
smaller difference of the ratios as well as an increase in the incidence of
schizophrenia in women over time, because women have been found to exhibit a
peak of onset after an age of 25 and men before the age of 24 (Eaton et
al. 1988, Goldstein et al. 1989, Häfner
et al. 1991). This study does not cover the incidence of schizophrenia
in age groups older than 28 years of age and therefore a final conclusion of
the total life-time incidence can not yet be formulated. However, the results
of this study suggest that the number of female schizophrenia cases may not
reach the number of male cases. Such a conclusion would support the finding of
a higher prevalence of schizophrenia in males (2.4%) than in females (1.8%) in
Northern
Finland
,
reported in the Mini-Finland Health Survey (Lehtinen
et al. 1991). The same life-time incidence in men and women of this
study population would require a remarkable second peak of onset 53 in women.
Such a big second peak does not seem to be probable, because until the age of
28 only 35% of all schizophrenia cases involved women. As far as we know,
prospective age cohort studies like this study of the incidence of
schizophrenia have not previously been performed and a comparison of the
incidence rates of this study with earlier studies may be difficult. Sampling
differences may lead to misleading conclusions concerning the differences in
incidence rates between different studies. For example, hospital
first-admission rates, which have been used in many earlier incidence studies
may be unreliable, because not all patients with schizophrenia are so
diagnosed at the time of their first hospital admission and some patients
recorded as first admissions can be re-admissions. Diagnostic procedures and
methods of calculation of incidence statistics may also vary between studies.
For example, diagnostic criteria may have been changed during the study period
or different diagnostic criteria were used in different studies. (Kendell
et al. 1993). The
diagnostic criteria used have been found to be related to the incidence rates
and to the apparent gender differences in incidence (Regier
et al. 1998). It has been suggested that the more stringent the
diagnostic criteria for schizophrenia are, the more females are excluded (Lewine
et al. 1984, Lewis 1992, Castle et al. 1993, Navarro et al. 1996,
Haas & Castle 1997). Especially the age criteria are important, because
late onset cases are more common in women. In this study, diagnoses were based
on DSM-III-R schizophrenia criteria, which are rather stringent, but not as
stringent as, for example, DSM-III criteria used in some earlier reports.
Furthermore, there is a tendency to diagnose women with schizophrenia at a
later age than men (Bardenstein & McGlashan
1990), but in this study the careful validation process has led to a reduction
of this potential bias. However, compared with many other incidence studies,
the annual incidence rates of DSM-III-R schizophrenia in this study over the
whole 13 year follow-up period were relatively high (7.9 per 10 000 in men and
4.4 in women). Cooper et al. (1987) reported incidence rates of DSM-III
schizophrenia in men of 1.7 per 10 000 and in women of 0.7 per 10 000, while Iacono
and Beiser (1992) provided respective figures of
0.7 and 0.2. In the study of Nicole et al. (1992), the incidence of
DSM-III-R schizophrenia was 4 per 10 000 in men and 2.2 per 10 000 in women.
The main reason for the rather high incidence rates of schizophrenia observed
in the present study may be the young age of the study population. Thus far,
the cohort has been followed through the high risk years for the morbidity to
schizophrenia. Presumably the annual incidence rates will decrease as time
goes on and the follow-up continues. One explanation for the high rates may
also be the earlier finding of a higher prevalence of schizophrenia in
Northern
Finland
rather than in other parts of
Finland
(Lehtinen
et al. 1990, Hovatta et al. 1997). A
sizeable reduction of first admissions with schizophrenia has been
demonstrated in
Scotland
,
England
,
Wales
,
Denmark
, and
New
Zealand
(Geddes
et al. 1993, Kendell et al. 1993)
and it has been discussed whether the incidence of schizophrenia has, indeed,
been decreasing (Kendell et al. 1993). The
incidence of schizophrenia has also been reported to decline in
Finland
(Suvisaari
et al. 1999). In
Finland
there
are only few studies reporting incidence rates of schizophrenia and
methodological discrepancies make it difficult to compare the incidence rates
of different studies with each other (Kendell et
al. 1993). However, in spite of the major differences between the studies,
the high annual incidence rates in our study compared with earlier DSM-III
schizophrenia findings of incidences in Finland (1.6-2.8 per 10 000
population) (National Board of Health in Finland 1988; 54 Pakaslahti
1992) and all psychotic disorders (20 per 10 000) (Lehtinen
1996) do not support the view that the incidence of schizophrenia is actually
decreasing in Finland.
6.4.
Treatment and outcome (III)
In
the therapeutic community study, there were no statistically significant
differences between men and women regarding the sociodemographic
factors (age, education, dwelling place). However, in the schizophrenia group
the educational level of men was somewhat lower than that of women. A higher
proportion of the men with a diagnosis of schizophrenia and schizophreniform/schizoaffective
disorders were living in sparsely inhabited localities, compared with the
women. These two trends may indicate a poorer social functioning of male
patients. Men with a psychiatric illness are commonly staying in their home
districts with their parents. Involuntary admissions into a mental hospital
were equally common for both genders. A similar finding has been described
earlier in another Finnish sample (Kaltiala-Heino
1995), but it does not support the expectation often presented that women
would be more inclined to apply for psychiatric help than men (Seeman
1986). The same rates of involuntary admission in men and women into this
acute, closed, therapeutic community ward can be explained by the selected
study sample. Besides, there were equal proportions of male and female nurses
amongst the staff, which in practice made it impossible to treat the violent
male patients who were difficult to treat. During the day shift, three out of
five staff members were women and two were men. The hospital attendants and
the ward secretary were all women. During the night shift there were two
nurses, one of them was always a man. There were no major gender differences
among patients with a diagnosis of schizophrenia regarding participation in
psychosocial therapies. This group of patients had the most severe clinical
profile; their clinical pictures were very uniform and they participated in
their treatments in the same way. The women in the group of schizophreniform/schizoaffective
disorders participated significantly more actively in group therapy and, to
some extent, also in milieu therapy. On the other hand, there were no
statistically significant differences in the level of activity in individual
therapy between men and women in any other diagnostic group apart from mood
disorders. However, this shows that women are more actively involved in their
treatment than men. Statistically significant gender differences regarding the
length of stay, number of episodes and readmission rates were not found in
this study. These findings are at odds with various other studies suggesting
that female patients with schizophrenia have shorter and fewer hospital
treatment stays than men (Salokangas 1983, Seeman
1986). The finding of this study can be explained by the selection of the
patients, a process that probably homogenized the study population. It was
also very clearly shown that the more severe the diagnosed disorder which the
patient had (non psychotic vs. psychotic), the longer was the stay in
hospital. The same result has also been obtained previously in a study on the
same sample (Nieminen et al. 1994). The
women with schizophreniform/schizoaffective
disorders had better institutional outcomes than the men, but this was not
seen in the case of schizophrenia or in the other 55 diagnostic groups.
Previous studies have shown that there are gender differences in outcome
especially among the patients with schizophrenic disorders (Bardenstein
& McGlashan 1990, McGlashan
& Bardenstein 1990, Mueser
et al. 1990). At least in part, the better outcome of women may be due
to women having more remissions and their psychosocial functioning being
better than that of men (Pakaslahti 1992). The
better outcome in the diagnostic group of schizophreniform/schizoaffective
disorders can also be partly explained by the finding that women were more
seriously committed to their treatment in this diagnostic group than men.
However, the minor gender differences are indicative of the realization of the
treatment goals of a milieu-therapy-oriented ward in the treatment of severe
disorders. The gender differences of the patients in this study sample were
found to be very minor also during the second and third treatment periods.
This is due to the even more stringent selection of the patients at the
beginning of the second and third treatment periods than at the first. The
patients who had responded positively to the therapeutic community treatment
during their first treatment period were admitted to the same ward at rehospitalization
and the patients who did not respond at all were more probably admitted to
another ward. On the other hand, the clinical picture of the patients with
several treatment periods was probably more severe. Thus, the patients with
several treatment periods formed the more homogenous group of the patients
with a severe clinical picture.
6.5.
Gender differences of long-stay patients (IV)
In
the long-stay patient study, women were more commonly married, more likely to
have had children and lived alone or with a secondary family than men. All
these factors are indicators of the women’s better ability to cope socially
even when suffering from schizophrenia; this has also been shown in earlier
studies (Salokangas 1996). Women lived more
commonly in the city of
Oulu
,
which reflects better social functioning in
Northern
Finland
,
because many people have to move to the city because of work, study or spouse.
On the other hand, there was no gender difference in relation to social class
or education. The difference between men and women regarding social
functioning seemed toremain stable at the end of
the index hospitalization, as
measured by the Strauss-Carpenter scale. It seems that in our study population
the better social functioning of women did not protect them from long-stay
hospitalization. Men exhibited significantly more negative symptoms than
women, which is compatible with other studies (Salokangas
& Stengård 1990, Gur
et al. 1996, Salokangas et al. 1997).
There were no significant gender differences with regard to positive symptoms.
There were also no gender differences in the number of hospitalizations,
proportion of treatment days and number of involuntary hospitalizations after
the first admission to the psychiatric hospital. There were, however,
differences between men and women in CGI- and GAS-scores at the end of the
index hospitalization. All these variables provide a good measure of the
severity of the illness. Almost all patients had had more than one
hospitalization and at least one involuntary hospitalization before the index
hospitalization. Over half of the patients had spent at least 50% of the days
between the first psychiatric admission and the index hospitalization in
inpatient care. In practice this 56 has meant years of hospitalization because
the difference between mean/median age of first contact for psychiatric care
as marked on the patients’ case notes and mean age at index hospitalization
of all subjects was 22.6 years. The patients’ psychopathological state,
assessed by CGI scores, was rated as at least moderately ill in almost every
patient and severely or extremely ill in over half of the patients. The low
GAS scores confirmed these results. Almost every patient displayed a
remarkable deficiency in social functioning according to measurements on the
Strauss- Carpenter-scale. Almost all were on a disability pension.
Collectively, all these results indicated that the long-stay patients were
seriously ill. Mean daily doses of antipsychotic medication were relatively
high in the long-stay patient study group, also being a reflection of the
severity of the illness in these patients. Doses were surprisingly high in
females and, therefore, no differences between men and women were noticed. The
finding is contradictory to many earlier studies in which the doses of
antipsychotic medication have been reported to be lower in females (Andia
et al. 1995, Baldessarini et al. 1995).
On the other hand, the gender difference in medication has been found to
become smaller in older patient groups and especially following menopause (Seeman
1983). In this study, the mean age of the women was close to menopause, which
might explain the similarity of medication in male and female patients. Men
had significantly more neuroleptic
depot-injections than women, which could be a consequence of the men’s
poorer ability to take care of their medication and their greater lack of
compliance (Lewis 1992). Our study population represented a selected group of
the most severely ill, long-stay psychiatric patients, of whom 80% were
diagnosed as schizophrenic. About 10 % of our sample were patients with
organic disorders, which demonstrates that these patients along with the
schizophrenia patients in this catchment area are
also most frequently included in the long-stay patient group (Lamb 1993). Only
a few patients could be assigned to the diagnostic groups of personality
disorders, mood disorders or other functional psychoses. The lack of gender
differences within the various diagnoses, may be a reflection of the
homogeneity of the long-stay patient group. The hospital diagnoses used can be
regarded as reliable, because every patient had been in the ward for at least
six months and in addition a great majority of the patients had already
experienced years of earlier hospitalizations. The possibility of false
positive diagnoses is very unlikely (Pakaslahti
1986, Isohanni et al. 1997).
JNE….
6.7.
Methodological concerns (I, II, III, IV, V)
This
thesis is based on three different study populations, which enabled analyses
of gender differences in patients representing either different phases of the
illness or different subgroups of patients with the same diagnosis. To study
clinically homogenous samples has been suggested to be important in future
gender research (Harding & Hall 1997). This study allows comparisons of
the results from selected study populations with the results from earlier
studies representing the whole spectrum of schizophrenia. There are, however,
no earlier studies of gender differences in schizophrenia from the therapeutic
community and, considering everyday clinical problems, the studies from the
therapeutic community ward and the long-stay patients are clinically
important. In a review article of gender differences in schizophrenia (I),
mostly studies involving the whole schizophrenic population are reviewed. To
include this article as a separate part of this thesis was legitimate because
the theme of gender differences in schizophrenia is very extensive and covers
practically everything under the topic of schizophrenia. This practice
provided better possibilities to focus in the summary of this study on gender
differences that were examined in the other original papers (II-V) and also
gave a wider perspective to the differences in schizophrenia between men and
women.
6.7.1.
Strengths and limitations of the
Northern
Finland
1966 Birth Cohort study (II)
The
major strength of the Northern Finland 1966 Birth Cohort is the large study
population - over 12 000 individuals. The study population is an unselected
epidemiologically intact general population sample, representing 96% of all
births in the region in one year and thus the results can be generalized to
the population in
Northern
Finland
. The
second strength is that psychiatric diagnoses in this study can be regarded as
reliable (Isohanni et al. 1997). Diagnostic
codes appeared to transfer reliably from case records (Keskimäki
& Aro 1991, Mähönen
1993) into the FHDR and therefore probably only few, if any, of the
psychiatric hospital treated cases were lost (Isohanni
et al. 1997). In this study all psychotic outpatients of the age group
were identified to diminish any bias. All psychiatric diagnoses were also
validated carefully using the criteria of DSM-IIIR, which allows comparison of
the results with the results from other methodologically valid epidemiological
studies. The third strength of this study is the very reliable information of
the age at first hospitalization. The first limitation of this study is that
the information of the time point when the first psychotic symptoms appeared
was collected from the case notes. Although it is the basic data that a
physician and other staff should gather through psychiatric interviews and
report them in the case notes, it is possible that in some cases this
information is inexact. The amount of the documented information may vary
between different physicians, hospitals and years. However, the information
from the case notes may even be more reliable or at least as reliable as more
retrospective interviews. The number of recalled events decreases steadily
over time, which is called the ”fall-off” effect (Maurer & Häfner
61 1995). Also, several studies have reported a lower number of positive
findings in the retest situation, which is called ”retest artefact”
(Maurer & Häfner 1995). In the study of
Maurer and Häfner (1995), different sources of
information of age at onset of schizophrenic disorders were compared. There
were no differences in age at onset defined as first positive symptom, age at
first treatment or first admission for schizophrenia between the ratings of a
research psychiatrist on the basis of medical case records and the results of
structured interviews of the patients or close relatives.
he
second limitation is the possibility, that some schizophrenic cases have not
yet or may never come to medical attention. This bias, however, can not be
significant, because almost all the schizophrenic patients in
Finland
and
also in many other developed countries will be admitted to a mental hospital
at a relatively early phase of a disturbance (Häfner
et al. 1993, Maurer & Häfner 1995, Isohanni
et al. 1997). The third limitation of this study is the fact that the
first psychiatric hospitalisation could have been due to any psychiatric
disorder prior to manifestation of schizophrenia. The fourth limitation of
this study is the relatively young age of the study population. The probands
were 28 years old at the end of the follow-up, thus representing only the highrisk
years for the morbidity to schizophrenia and not the whole spectrum of
schizophrenia. Late-onset cases for schizophrenia are still lacking from the
study population (Castle et al. 1995, Häfner
et al. 1998). However, there are no other birth cohorts than the
Northern
Finland
1966
Birth Cohort, in which the morbidity to schizophrenia has been followed for
this long. The fifth limitation of this study is the possibility that
prospective cohort studies may be inapt when rare diseases are studied.
Because of the low annual incidence rate it is difficult to gather enough
subjects to identify the true differences, because of limited statistical
power. Also, when interpreting and generalizing the results of this study one
should keep in mind that the prevalence of schizophrenia may be higher in
Northern
Finland
than
in other parts of
Finland
(Lehtinen
et al. 1990, Lehtinen et al. 1991, Hovatta
et al. 1997).
6.7.2.
Strengths and limitations of the therapeutic
community
study (III)
The
principal strength of the Therapeutic community study is the high number of
subjects, 1525 patients, representing all first treatment episodes during 17
years at the therapeutic community ward. The second strength of this study is
the fact that the median length of hospitalization was relatively long, 39
days in all patients and 62 days in schizophrenia patients (Nieminen
et al. 1994). Long treatment periods give sufficient time to observe
the patients and consider the diagnoses, the degree of active participation
and the outcome properly. The third strength of this study is that the degree
of active participation and the institutional outcome were assessed at the
time of discharge by 3-5 team members, alwaysincluding
the personal doctor as well as the nurse. Also the interrater
reliability of the classifications was tested and it was shown to be
satisfactory (Nieminen 1996).62 The first
limitation of this study population from the therapeutic community ward is
that it does not represent a random hospital population. Most of the patients
in the therapeutic community unit were selected by personal interview or by
discussing a patient’s medical records in the team before a patient’s
admission to the ward. However, different adult age groups were admitted and
at a rough estimate a total of 30-50% of the patients were admitted in
emergency situations during the evening or night shift, and many of them
involuntarily. This diminished to a certain degree the selectivity of the
study population (Nieminen 1996). The second
limitation of this study is that the diagnoses of this study were not
validated. They were clinical and determined by the psychiatrists working at
the ward. However, they were probably more reliable than clinical diagnoses
generally, because they were mainly based on the decision made by a
psychiatrist (Matti Isohanni)
with long work experience (1972 - 1989) in the ward. Furthermore, he has later
tested his diagnostic reliability and it was found to be very satisfactory (Tienari
et al. in press, Isohanni et al. 1997).
There have also been changes in the diagnostic practice during the study
period. For example, at the beginning of the study period, the diagnoses were
not based on the criteria-based diagnostic classification and in uncertain
cases giving the diagnosis of schizophrenia was avoided. Thus, false positive
schizophrenia diagnoses were not likely. 63
6.7.3.
Strengths and limitations of the long-stay patients study (IV,V)
The
first strength of this study is that the study population represents all
long-stay psychiatric patients (n=253) in the region in 1992, because at that
time there were no other psychiatric hospitals in the catchment
area. Thus the study population can be considered as representative regarding
similar patient groups in general. The second strength of this study is that
the case records included detailed information on the patients, because the
patients had been for such a long time in the ward. Also the majority of them
had had several admissions, which meant practically years of hospitalization
and copious written case notes. Thus, the collected data and the clinical
hospital diagnoses used in this study can be regarded as reliable although the
diagnoses of this study were not validated. False positive schizophrenia
diagnoses are not likely, because in earlier diagnostic studies in
Finland
, only
false negative diagnoses of schizophrenia have been found to be remarkable (Pakaslahti
1986, Isohanni et al. 1997). The third
strength of this study is that the interrater
reliability regarding assessments of CGI, GAS, DSM-III-R symptoms and
Strauss-Carpenter scale was satisfactory between researchers. The fourth
strength is that we were able to complete the data concerning the placements
of the patients after the last discharge by interviewing the personnel of the
last inpatient ward. The first limitation of this study is that the data of
the long-stay patients, including symptoms, were collected retrospectively
from case records. Due to this procedure, some information may be lacking.
However, the data collection from the case records diminish the ”fall-off”
effect and ”retest artefact” (Maurer & Häfner
1995) compared to more retrospective interviews. These phenomena are described
and discussed in more detail earlier in chapter 6.7.1. The second limitation
in this study is that the time between the last discharge and the end of the
follow-up may differ in theory from one day to three years. This means that
further follow-ups are needed to draw final conclusions about the success of
the deinstitutionalization at this area. However, one third of the patients
were less than 25% and two thirds of the patients were less than 75% of the
follow-up period in hospital care.
8.
Summary
8.1.
Background and aims of the study
During
the last decade schizophrenia research has emphasized the importance of gender
differences. The purpose of the present study was to analyze gender
differences with regard to some important sociodemographic
variables, age at onset, incidence, treatment, outcome and
deinstitutionalization in three different, mainly schizophrenic populations
from
Northern
Finland
.
8.2.
Material and methods
The
first study population was the
Northern
Finland
1966
Birth Cohort, an unselected, general population birth cohort based upon 12068
pregnant women in the provinces ofLapland and O ulu
with an expected delivery date during 1966. Their 12058 live-born children
represent 96.3% of all births in the region. We followed prospectively 11017
subjects (males: 5636 = 51.2%; females: 5381 = 48.8%) from the age of 16 up to
the age of 28 by means of the Finnish Hospital Discharge Register. All the
case notes of the 387 subjects hospitalized due to the psychiatric disorder
were obtained. The diagnoses were validated for the DSM-III-R criteria
resulting in 89 (males: 58 = 65.2%; females: 31 = 34.8%) DSM-III-R
schizophrenia cases. From this study population gender differences with regard
to age at onset and incidence of schizophrenia were calculated. The second
study population was derived from the closed therapeutic community ward
situated at the Department of Psychiatry,
University
of
Oulu
. A
total of 1525 patients who had their first treatment episodes at the ward
between 1. Jan., 1977 and 2. July 1993 formed the study population. Of them,
51 % (n = 784) were males and 49 % (n = 741) females. 85 males and 64 females
had been diagnosed as schizophrenics, 112 males and 90 females had been
diagnosed to suffer from schizophreniform or
schizoaffective psychoses. Age at first admission, the degree of active
participation of the patients in individual, group, and milieu therapy and the
institutional outcome of these patients were assessed. 69
The
third study population was formed of 253 long-stay psychiatric inpatients
treated for at least six months without a break during 1992 (index
hospitalization) in the Department of Psychiatry,
Oulu
University
Hospital
. The
patients were identified from the computerized patient register and data were
collected from this register and from case records at the end of the index
hospitalization and at the end of the four year follow-up (1992-1995). About
80% of the patients had a diagnosis of schizophrenia and because of the
homogeneity of the patients, they were analyzed as a single group with regard
to the diagnosis. From this study population gender differences regarding age
at onset and other sociodemographic and clinical
variables were studied. Also the placement after the last discharge and at the
end of the follow-up as well as factors predicting hospitalization after the
follow-up were studied.
8.3.
Results and conclusions
In
the
Northern
Finland
1966
Birth Cohort study the mean age at onset of schizophrenia using two different
criteria was between 21.4-21.5 in men and 21.2-21.4 in women. The peak of
onset in men lay in the 20-24 year age group whereas in women it occurred in
the 16-19 year age group. The time difference between first psychotic symptoms
and first psychiatric hospital admission was minimal for men and women. In the
therapeutic community study the mean age of the patients with schizophrenia at
their first admission to this ward was 27.4 years in men (SD 7.4) and 27.8 in
women (SD 7.5) and patients with the diagnoses of schizophreniform/schizoaffective
disorders 26.5 years in men (SD 7.2) and 27.4 in women (SD 7.7). In the
long-stay patients study, mean age at onset of illness using five different
criteria was between 22.2-24.0 years in men and 24.3-25.4 years in women.
Mean/median time lag in years between the first sign of psychiatric symptoms
and the first psychiatric hospitalization was 2.5/1.0 (SD 5.1) overall,
2.6/1.0 (SD 4.4) for males and 2.2/1.0 (SD 6.0) for females. Mean/median time
lag in years between the first psychotic symptoms and the first psychiatric
hospitalization was minimal in both males and females. There were no gender
differences regarding age at onset in any of these three different studies.
However, the peak of onset of females was lower than that of men. These
results are contradictory to numerous earlier clinical studies of age at onset
and indicate that there is no gender difference with regard to age at onset.
The cohort studied is presently about halfway through the period of risk for
schizophrenia and the other study populations did not represent random
hospital populations. This suggests the possibility that within the
schizophrenic patient group there are homogeneous subgroups with regard to the
onset of illness. The minimal time lag between the first psychotic symptoms
and the first psychiatric hospitalization indicates that in
Finland
the
patients are hospitalised earlier than in many other countries after the
manifestation of their first psychotic symptoms. In the Northern Finland 1966
Birth Cohort study the annual incidence rate of DSM-IIIR schizophrenia was
high: 7.9 out of 10 000 in men and 4.4 in women by the age of 28 years. In men
it was highest in the age group of the 20-24 year-olds while in women the peak
occurred in the age group of the 16-19 year-olds. There was a decreasing trend
in the incidence rates for women of the older age groups. The higher incidence
rates in men 70 support the findings of earlier studies that incidence among
young men is higher than in women. The high annual incidence rates of
DSM-III-R schizophrenia in this study over the whole 13 year follow-up period
may be due to the fact that the cohort has been followed through the high risk
years for the morbidity to schizophrenia and a final conclusion of the total
life-time incidence can not yet be formulated. However, the high rates do not
support the view that the incidence of schizophrenia is decreasing in
Finland
.
Incidence among men was almost two times higher than that in women. It
suggests that the number of female schizophrenia cases will never reach the
number of male cases, because an identical life-time incidence in this study
population of men and women would require a remarkable second peak of onset in
the women. In the Therapeutic community study there were no gender differences
regarding the sociodemographic variables, the
length of stay and the number of treatment episodes in this ward in any
diagnostic group. Differences in the degree to which males and females
participated in the psychosocial therapies (individual, group and milieu
therapy) and differences with regard to the institutional outcome were
minimal, some trend favoring females. These
minimal gender differences indicate the achievement of the treatment goals, or
achieving intergender equality and balance. These
goals are especially important for schizophrenic males because of their more
severe clinical profile and poorer prognoses. The therapeutic community model
may help psychotic male patients in approaching their age- and gender-specific
life span challenges. In the long-stay patients study almost two-thirds of the
long-stay patients were men. Very few gender differences were found in sociodemographic
and clinical characteristics or in the utilization of psychiatric hospital
care. The female patients had a better ability to adjust socially. It would
seem that men are overrepresented amongst long-stay inpatients because of
their poorer outcome. The observed similarity between male and female patients
could be a result of the severe nature of the illness in the patient group.
Patients have had to pass through several filters through health care contacts
before they reach this most intensive of the health services. At each filter,
clinical characteristics will become constrained and thus differences between
the sexes will be attenuated. The homogeneity of men and women can also be
partly iatrogenic - caused by the traditional custodial care and by the high
doses of medication. About 70% of the long-stay patients were discharged
during the four year follow-up and only 15% were able to live without
continuous support. The discharge did not cause homelessness. Marital status,
dwelling place, absence of negative symptoms and severity of the illness were
associated with hospitalization at the end of the follow-up. Gender was not
associated with hospitalization at the end of the follow-up period. Our study
showed that long-term patients are dependent on considerable support and that
the most seriously ill patients are, in fact, in hospitals. Alternative
residential facilities have to be regarded as pre-requisites for the
deinstitutionalization process of the long-stay patients to be successful.