Prediction of psychosis: a follow-up study.

Background and significance

Prediction of psychosis

Psychosis is a serious psychiatric condition in which reality testing is
disturbed. This can be expressed through disturbed perception (hallucinations)
or disturbed thinking (delusions). The annual incidence rate of psychotic
disorders is approximately 3.0/10,0001, 2, 3 and the lifetime prevalence is
between 0.5 and 1.6 %4. First episode psychosis is a distressing and traumatic
experience for patients and families5. Two third of all patients have at least
one psychotic relapse and after each relapse 1 of 6 patients does not remit
from the episode6. More than 50 % of all patients need to be hospitalized
during their first psychotic episode and 17 % need hospitalization during
follow up1, 2. Psychotic disorders, particularly schizophrenia, are associated
with persistent social disability7, high rate of suicide (10 %)6, selfharm (26
%), agression to others (14 %)8, 2, low quality of life9, and high economic
burden10, 11. Longer duration of untreated psychosis is associated with poor
outcome12, 13. Since cost-effectiveness is likely to be maximal in early
intervention14 prediction and prevention of psychosis is of high social
importance.
In the light of prediction and prevention numerous studies have investigated
the correlation between signs and symptoms in early development and the risk of
developing psychotic symptoms. For example, in the British 1946 birth cohort
Jones15 found that delayed motor and speech development by the age of 2,
solitary play preference at age 4 and 6, feeling less socially confident at age
13, and being more anxious in social situations at age 15 were related to
subsequent risk for schizophrenia. The British 1958 birth cohort16, 17 revealed
associations between schizophrenia at adult age and over-reactive behavior in
boys at age 7 and 11 while girls at age 11 showed withdrawn and unforthcoming
behavior. Restlessness and hostility at age 7 were associated with affective
psychosis, however this correlation disappeared at age 11. In the North Finland
1966 birth cohort18 delayed milestones were most markedly found prior to a
psychotic disorder. Bearden19 found language problems (echolalie,
unintelligible speech, poor expressive language ability) and social
maladjustment at age 7 to predict schizophrenia in a large population cohort.
Malmberg20 found a relation between poor social functioning at age 18 and the
development of schizophrenia in a Swedish cohort study. Interesting in the
latter study was that accumulation of 4 different predictive variables was
associated with a linear increase in the risk for developing schizophrenia.
The studies here described were all investigating normal, predominantly
non-clinical populations. In contrast with these (birth cohort) studies, others
investigated subjects possibly at risk for developing psychosis. Yung21 focused
on a high-risk *prodromal* group and found an association between low-grade
psychotic symptoms, depression, and disorganization before onset of illness and
the risk of developing schizophrenia. Numerous studies investigated subjects
genetically at high risk for schizophrenia22, 23, 24, 25. For example,
Keshavan26 found associations between ADHD symptoms and psychosis-like features
in a group genetically at high risk for schizophrenia. In addition, in the New
York High-risk Project cognitive dysfunctions like verbal memory problems and
attention problems were associated with the risk of developing schizophrenia27.
Besides developmental signs and symptoms, environmental factors were also
investigated as for the correlation with psychosis at adult age. Early
environmental factors like stress during pregnancy, inconvenience of pregnancy,
and smoking during pregnancy were related to increased risk of psychosis at
adult age28. Other environmental factors like prenatal brain damage29, exposure
to obstetric complications30, and urban birth and upbringing31 were associated
with the development of schizophrenia. In addition, associations between early
trauma and psychosis were also demonstrated. For example, childhood
maltreatment and sexual abuse were related to the risk of psychosis at adult
age32, 33, 34 suggesting that trauma can cause a biological or psychological
vulnerability for the development of psychosis. This is supported by the
finding that life events increase sensitivity to daily stress and in that way
influence the risk of psychosis35.
In summary, developmental signs and symptoms and/or environmental factors like
exposure to urban environment, life events, and other stress related factors
possibly represent respectively influence the vulnerability of the brain and in
that way increase the risk for psychosis at adult age.

Need for further investigation

One of the problems of the precursors found in non-clinical populations is the
fact that they are in some degree predictive, nevertheless lack specificity
which is possibly due to the complex interaction between individual and
environmental factors. This implicates that active screening for (and thereupon
follow up or even treat) a group of high risk children and adolescents at this
point is not profitable enough.
However, children and adolescents with a non-psychotic psychiatric disorder are
already assessed in a psychiatric hospital, meaning no active exploration of
subjects is necessary. These patients could be at particularly high risk of
developing psychosis in adult life since they have an individual vulnerability
that leads to the non-psychotic psychiatric disorder and this disorder in
itself could be considered a serious stress related life event leading to all
sorts of (medical) interventions and a disruption of balance, possibly
effecting (brain) development.
Prognosis of specific child and adolescent psychiatric disorders were
investigated before36, 37. Stahlberg38 found that patients over age 18
diagnosed as autism spectrum disorder (ASD) and attention deficit hyperactivity
disorder (ADHD) had psychosis in 14,8% respectively 10% of all cases in a large
clinical case cohort. Kim-Cohen39 used a follow back strategy of a
prospective-longitudinal cohort to examine the mental health histories of
adults who met diagnostic criteria at age 26 for a wide range of psychiatric
disorders. They found that adult mania had been preceded by a juvenile
diagnosis, particularly conduct disorder(CD), oppositional defiant disorder
(ODD), and juvenile depression, before age 15 in 61,9 % of all cases. Adult
schizophreniform disorder was associated with a juvenile diagnosis,
particularly juvenile anxiety, depression, ADHD, CD, and ODD, before age 15 in
54% of all cases.
Little is known about the prognosis of specific signs and symptoms in child and
adolescent psychiatry disorders regardless of diagnosis concerning the
development of psychosis. Cannon40 found a model for later onset of
schizophrenia in a sample of child psychiatric clinic attenders with a
predictive power of 81.2%. The variables in this model were *suspiciousness or
sensitivity*, *relationship difficulties with peers*, *family psychiatric
history*, and *past psychiatric contact*. In the same study, a comparable model
was found for affective psychosis with a predictive power of 63,6% and the
variables *disturbance of eating* and *hysterical symptoms* (e.g.
depersonalisation, non-epileptic disturbance of consciousness, conversion
hysterical symptoms). However, since only adults with affective psychosis and
schizophrenia were investigated it is not clear whether these models
specifically predict psychosis or psychopathology in general. Moreover, some of
the patients were diagnosed with psychosis while still in the children*s
department. This may implicate that the variables found were prodromes rather
than precursors of psychosis.
In the proposed study we will investigate sign and symptom profiles as well as

Hypothesis

Children and adolescents with non-psychotic psychiatric disorders represent a
vulnerable group high prone for psychotic symptoms. In contrast with normal
populations more pronounced symptom and cognitive profiles will be found
predicting the onset of psychotic symptoms at adult age. In line with earlier
investigations we expect to find a relationship between development of
psychosis and social functioning, ADHD symptoms, suspiciousness and as for
cognitive functioning verbal memory problems, attention problems and problems
in executive functioning. We will also investigate the effects of treatment
expecting longer duration of treatment predicting higher risk of psychosis. We
expect to find a cumulative effect of the specific predictive signs and
symptoms as described previously20. Screening for symptom and cognitive
profiles in child and adolescent psychiatry patients could be of preventive
value. Interventions in the light of treatment of the non-psychotic psychiatric
disorder will influence the onset of psychotic symptoms and thus this will
possibly lead to adjustments of treatment in patients at high risk for
developing psychosis

Procedure

All children that visited the Department of Child and Adolescent Psychiatry of
the University Hospital Utrecht between 1984 and 2004, were diagnosed
according to the DSM criteria following a thorough child psychiatric
examination by a child psychiatrist. Judgment of problem behavior was supported
by several standardized questionnaires, including the Child Behavior
Checklist41, the Teacher Report Form42, and the Maudsley Inventory for problem
behavior43. In addition to the standardized procedure, about 50 % of the
children were psychologically assessed with a standardized battery, including
an intelligence test (most likely the WISC), a test for hand-eye coordination
(VMI), measurements of attention regulation (CPT), tests for verbal (15WT) and
visual memory (RVDLT) and a problem solving task (WCST).

Figure 1 presents an overview of the three different phases of this study based
on study design.

Figure 1: Research design
Measurement level Phase of research n = Groups Outcome variable Collection of
data Study design Statistics Powercalculation
All children who had psychiatric assessment between 1984 until 2004 Psychosis &
demographic characteristics 6500 Already a part in pilot study collected. The
rest has to be collected Cross-sectional Chi-square (frequencies) Not
necessary for descriptive statistics
Two subgroups based on psychosis measured by SPQ Behavioral measurement &
cognitive profiles 400 already collected from 1984 until 2004
Prospective design Logistic multiple regression Rule of thumb: per 20
patients, one variable in the regression model
2 Phase 3 400 Two subgroups based on psychosis measured by SPQ Cognitive
profiles & behavioral outcome To be collected Prospective longitudinal design /
cross-sectional Paired sample t-test/independent-sample t-test

Measurement level 1
The data on child age are already collected from 1984 and 2004. The data on
adult age will be collected by questionnaire.

Phase One
At phase one of this study, we will select children based on age and diagnosis.
Subjects with psychotic disorder at child age, will be excluded. All included
subjects will be sent a written request (see appendix) to participate in a
follow-up study that is designed to evaluate their functioning and well-being
after their visit to the psychiatric department several years ago. Joining this
request an informed consent and an inventory questionnaire will be closed in
with a return envelope with the main focus on current problems in general
functioning (including psychosis), treatment, level of education, civil status
and social functioning. The questionnaire will conclude with the question
whether or not the department may contact him/her for further research purposes.

After the primary collection of the questionnaires, first goal is to determine
prevalence of psychosis in a clinical population. Thereby, we will
epidemiologically evaluate social functioning, well-being, treatment, medicine
and drug history of this psychiatric population in comparison to a normal
population, based on norms of Centraal Bureau voor de Statistiek 51 and between
initial diagnostic categories.

The power calculation of phase one can be based on results of the pilot study,
in which we found a 4% prevalence of psychosis in a representative group
(n=150) of the total clinical population (n=6500).

Phase two
At phase two, objectified subjects with (n=200) and without psychosis (n=200)
will be compared on earlier collected data from child age. All participants who
returned their questionnaire and who gave permission for further research and
had psychosis or where not sure about having psychosis on adult age, will be
contacted to fill in a questionnaire about psychotic symptoms (SPQ, Schizotypal
Personality Questionnaire50) to objectify their complaints. Further on, after
collecting a positive score on the SPQ, permission will be asked to contact the
general practitioners, to find a confirmation for the psychosis in their
personal file. Next, the objectified adult subgroups will be compared
(psychosis versus no psychosis) on earlier collected data from child age. 200
non -psychotic adults (based on their negative SPQ score) will be selected to
be a comparison group, matched on age and gender.

Goal of phase 2 is the search for a positive prognostic prediction model and a
specific prediction rule to predict psychosis in adult life. Power calculation
will be based on the rule of thumb that per 20 patients, one variable can be
included in the regression model. In line with the previous mentioned pilot
study, we expect 264 subjects to return their questionnaire with positive signs
of psychosis in adult life. Hereby we*ll be allowed to include at most 13
variables in the prediction model. About 6 variables will be entered in the
regression model. These variables include psychiatric (diagnosis) and
neurocognitive data (executive functioning, verbal memory, attention,
visual-motor integration and intelligence) collected on child age.


Measurement level 2
Child-age data are already collected. Data on adult age will be collected by
re-inviting the patients at the department of Child and Adolescent Psychiatry
for psychiatric and neurocognitive screening.

At measurement level two, 200 patients with psychosis based on positive SPQ
scores and an age and gender matched sample of 200 patients without psychosis
(based on negative SPQ scores), will be invited at the hospital for psychiatric
and neuropsychological assessment or a home visit will be planned. After
written informed consent is obtained, subjects will be invited to fill out
several interviews. The Lancashire Quality of Life Profile (QOL)44 will be used
to measure quality of life. The WHO Disability Assessment Schedule (DAS) 45
will be used to assess social disability. Life events will be interviewed with
the structured Brown and Harris Life events and Difficulties Schedule (LEDS) 46
and family history will be obtained using the FH-RDC47. A slightly modified
version of the Life Chart instrument48 will be used to assess longitudinal
information on clinical and therapeutic outcome. Finally, the Structured
Clinical Interview for DSM IV (SCID) 49 will be used to establish diagnosis.
Further, neuropsychological tests for different cognitive domains as attention,
verbal and visual memory, intelligence, executive functioning and hand-eye
coordination will be done. Interviews and neuropsychological tests will be
conducted by trained interviewers and diagnoses will be made after the SCID
during consensus meetings that include a senior psychiatrist.

Phase 3
By collecting data on adult life, we are able to compare patients with and
without psychosis cross-sectional on specific data: Quality of Life, Social
Disability, Life events, Family history, Treatment and Cognitive functioning.
Stability of cognitive functioning underlying behavior can be measured over
time by comparing already collected cognitive data on child age and to be
collected cognitive data in adult life.

Based on our finding in our pilot study (n=150) 4% had psychosis in adult life.
In line with this, we*ll expect 264 subjects of our total adult population
(n=6600) to have psychosis in adult life.

Outcome variables
Phase One outcome variables will be the dichotomous variable psychosis or no
psychosis in adult life and the demographic characteristics of the total
clinical population. Outcome variable of Phase Two will be a positive
prognostic prediction model and specific prediction rule to predict psychosis
in adult life. Outcome variables of Phase Three will be the developmental
cognitive profiles of assets and deficits of patients with psychosis and
further outcome variables will be the comparison of patients with and without
psychosis on quality of life and social disability.

The outcome variables of these three different phases will be r

All patients will be contacted by means of a short questionnaire.

Then after selection of patients on earlier-mentioned criteria, a consult on
the department of Child- and Adolescent Psychiatry will take place or a
home-visit. Hereby, earlier mentioned questionnaires will be the workload will
be a consult to the department of Child- and Adolescent PSychiatry or a
home-visit, information will be collected from earlier mentioned questionnaires.

No dangers have been linked to this research.