Escolar Documentos
Profissional Documentos
Cultura Documentos
DOI 10.1007/s00787-009-0018-7
ORIGINAL CONTRIBUTION
Received: 16 July 2008 / Accepted: 26 March 2009 / Published online: 21 April 2009
Springer-Verlag 2009
Abstract The objective of this study was to examine the Keywords Medication adherence Psychosis
predictors associated with adherence to atypical antipsy- First-episode Schizophrenia Mood disorders
chotic medication following discharge from hospital for Risk factors Children Adolescents
children and adolescents with first-episode psychosis. Sixty-
five children and adolescents, age 15.35 2.08 years, 59%
boys, who had participated in a longitudinal cohort study Introduction
examining relapse following first hospitalization for episode
of psychosis were included in this study. All those studied Children and adolescents are diagnosed with psychotic
were discharged on one of three atypical antipsychotics, disorders and mood disorders with psychotic features
risperidone, quetiapine, or olanzapine between January 1999 with increasing frequency. The average age of onset for
and October 2003. Time 1 data were collected retrospec- schizophrenia is between 15 and 25 years for males, and
tively from medical charts using a standardized question- 25 and 35 years for females [1]. Thirty-nine percent of
naire; time 2 data were obtained using questionnaire mailed males and 23% of females have their first-episode before
to participants parents a minimum of 2 years post-dis- the age of 19 years [2]. Prevalence rates for adolescents,
charge, mean 3.9 1.3 years. Variables examined as pre- 1518 years old, are estimated at 0.23% compared with 1%
dictors of adherence fell into broad categories of biological, in adults, 18 years and older [3]. Psychotic disorders are
social and treatment variables. Discharge on concurrent the primary reason for psychiatric hospitalizations, and
pharmacologic agent for affective symptoms in addition to adolescents often require a length of stay from 25 to
atypical antipsychotic, OR = 10.5 [95% confidence interval 45 days [4, 5]. Indirect, direct, and readmission costs
(CI) = 2.0653.19] was a strong predictor of medication associated with schizophrenia extend into the tens of bil-
adherence in adolescents. The results indicated that children lions of dollars annually [68]. Increasingly the rising costs
and adolescents discharged from their first hospitalization associated with non-adherence to medications are being
following a psychotic episode may be more likely to stay on recognized as the other drug problem [9]. The United
prescribed antipsychotic medication if they are prescribed States National Institute of Mental Health [10] has
concurrent medication for affective symptoms. emphasized the importance of investigating the problem of
non-adherence in psychiatric populations.
Children and adolescents with psychosis experience a
chronic episodic course characterized by relapses, impaired
social adjustment and poor educational and occupational
functioning. Early onset of psychotic disorders and mood
R. E. Gearing (&) disorders with psychotic features is associated with intel-
Columbia University, New York, USA lectual decline reflected in lowered IQ scores [11] and
e-mail: rg2372@columbia.edu
decreased quality of life marked by declines in daily
A. Charach functioning, employment-related problems, and social
The Hospital for Sick Children, Toronto, Canada deficits [1215]. Adolescent onset results in poorer
123
588 Eur Child Adolesc Psychiatry (2009) 18:587595
123
Eur Child Adolesc Psychiatry (2009) 18:587595 589
disorders) during index admission. Details to determine without medical supervision as recorded in the Information
diagnosis, inclusion, and exclusion criteria were extracted Update Profile Sheet. Where medication type was changed,
from the medical record. lowered or discontinued with the agreement and advice of
The sampling frame consisted of 229 youths admitted the physician, the participant was considered adherent.
with first-episode psychosis across six participating
adolescent inpatient units during the index time period. Independent variables
Following ethics approval, the ethics protocol required that
all potential participants families be contacted by a clini- Potential predictors of medication adherence were drawn
cian who knew them prior to contact by research team. from variables previously evaluated as potential predictors
Letters were sent to introduce the families and young people of symptom relapse following hospitalization [47]. Vari-
to the study, followed by a telephone contact to solicit ables fell into broad biological, social, and treatment
participation. Fifty-eight (25%) declined participation. categories. Information documenting each variable was
Eighty-seven (38%) families returned the data collection extracted from the hospital record. Biological variables
questionnaire and consent materials, and were enrolled in included gender, age at admission, gradual versus sudden
the study. Another 84 (37%) agreed to participate by tele- onset, length of time in weeks illness remained untreated
phone, but did not return the study materials. prior to admission, presence of medical diagnosis, diag-
nostic group categorized broadly into primary psychotic
Data collection instruments disorders (including schizophrenia, schizophreniform dis-
order, brief psychotic disorder, psychotic disorder not
Time 1 (T1): Hospital records from the index admission otherwise specified) or affective psychotic disorders
were available for data extraction. The Chart Review Data (schizoaffective, depressive disorder with psychotic fea-
Instrument was developed to extract data following estab- tures or bipolar disorders), presence of negative symptoms
lished criteria [45, 46]. Variables of interest were identified in addition to positive symptoms, presence of major
from those previously investigated in the literature, and depressive disorder (MDD) by DSM IV criteria, presence
operationalized using input from clinical and scientific of mania by DSM IV criteria, history of alcohol use, and
experts in conjunction with a review of several hospital history of illicit drug use. Social variables included foreign-
cases. The instrument was supported by a set of standard born, number of weekly parent visits during hospital
protocols and guidelines developed to instruct and guide admission, decrease in social support prior to admission,
reviewers in the collection of the data. Two independent number of peer relationships, special education prior to
reviewers piloted the instrument in a subset of 11 partici- admission, documented presence of family criticism
pants hospital charts. Agreement for variables, Kappa toward patient (negative expressed emotion), family his-
coefficient or intraclass correlation coefficient (ICC) tory of depression, family history of bipolar disorder, and
depending on the level of measurement ranged between family history of schizophrenia.
0.74 and 1.00. The variable, discharge medication, had a Treatment variables included length of hospital stay in
Kappa of 1.00 [47]. days, and discharged from hospital with prescription for
Time 2 (T2): The Information Update Profile Sheet was atypical neuroleptic with or without concurrent medication
mailed to the parent or caregiver to be completed and for affective symptoms.
returned along with the consent materials for participating
in the study. The questionnaire collected follow-up infor- Statistical analysis
mation from parents (caregivers) covering a minimum of
2 years post-discharge from hospital. Parents were asked All variables were first summarized using descriptive sta-
to record the history of any medication changes for their tistics such as mean and standard deviations, medians and
son or daughter since the index hospitalization, names of ranges, counts and percentages, as appropriate. Univariate
medications, dose, frequency, start and stop dates, and associations between adherence to atypical neuroloptic
reasons for stopping medications. They also recorded medication and each of the other variables were investi-
whether their child or adolescent was currently on medi- gated. For this purpose the Chi-square and Fisher exact
cation(s) at the time of follow up. The instrument was tests were used for binary or categorical variables and
piloted in a subgroup of participants and parents were able univariate logistic regression for continuous variables.
to provide the necessary data without apparent difficulty or The effect of biological, social, and treatment variables
confusion. Wherever possible, data from the Information on adherence to medication was evaluated using logistic
Update Profile Sheet were compared to the adolescents regression. To maximize statistical power, candidate pre-
hospital and medical records to confirm accuracy. Youths dictors were identified through a preliminary analysis. First,
were non-adherent when they discontinued medication variables with too much missing data were eliminated.
123
590 Eur Child Adolesc Psychiatry (2009) 18:587595
Second, variables measuring similar underlying constructs time, a medication for mood symptoms, either an antide-
were collapsed, and finally variables with a P-value above pressant or a mood stabilizing agent. The statistical model
0.20 in the univariate analysis were eliminated. To avoid identified as most robust included two variables, discharge
multicollinearity in the final model, pair-wise correlations on concurrent medication for mood symptoms, odds ratio
were evaluated for the remaining variables. Finally, models (OR) = 10.5 (95% CI = 2.0653.19, P = 0.005) and
limited to two or three predictors at a time were explored in fewer weeks of untreated illness prior to hospital admis-
the multiple logistic model. Variables with the highest sion, OR = 0.98 (95% CI = 0.951.01, P = 0.1).
statistical significance were retained in the final model. All Although the variable duration of untreated illness does not
analyses were performed using SAS Institute software for formally reach statistical significance at a 0.05 alpha level,
Windows version 9.1 [48]. it does show a trend toward statistical significance
(P = 0.1), so it was retained in the model.
Results
Discussion
Sixty-five (n = 65) child and adolescent participants
enrolled in this study were discharged on a single atypical Of the 65 children and adolescents who participated in the
antipsychotic for treatment of psychosis, 39 (60%) were study, only 16 (25%), 9 males and 7 females were non-
male, age at admission was 15.35 2.08 years, and 49 adherent to medication treatment during the follow-up
(75%) were adherent to medication (see Table 1 for further period. This finding does not correlate with the literature on
details). adults, which routinely reports high non-adherence rates
The 65 youths on atypical antipsychotics included 27 [49]. For instance, Favre et al.s [32] study found that 75%
(42%) children discharged on resperidone, 26 (40%) on of patients were non-adherent by 18 months post-dis-
olanzepine, and 12 (18%) on quetiapine. Thirty-two charge. Similarly, Olfsen et al.s [50] study found that 20%
(n = 32) (49%) were discharged on an additional medi- did not remain compliant at 3-month post-hospital dis-
cation targeting mood symptoms, 16 on antidepressants charge. Valenstein et al. [51] found that 3637% of patients
(Sertraline, Paroxetine, Fluoxetine and Venlafaxine), 15 on are poorly adherent to antipsychotic agents over time.
mood stabilizers (Lithium, Divalproex Sodium and Car- While it has been suggested in the literature that young
bamazepine), and 1 was on both an antidepressant and a people may have higher non-adherence rates than adults
mood stabilizer (see Tables 2, 3 for further details). These [52], the current study does not support this conclusion.
pharmacological agents were available and in common use Rather, the findings suggest that a substantial proportion of
in Canada during 1999 to 2003. children and adolescents is likely to remain adherent to
The following eight variables were considered for medication following hospitalization.
inclusion in the multivariate analysis: age at admission, The current study identifies two predictors of adherence
gender, decreased social support, length of untreated illness, to atypical antipsychotic medication following discharge
diagnostic group, family history of depression, presence of from hospital. Youths discharged on an atypical antipsy-
MDD, and participant discharged on atypical neuroleptic chotic agent and an additional medication for affective
with or without medication for affective symptoms. Cor- symptoms, either an antidepressant or a mood stabilizing
relations among variables were evaluated to identify agent were 10 times more likely to remain adherent than
potential multi-collinearity (see Table 3 for further details). those discharged on atypical antipsychotic agent without a
Three variables, presence of MDD, Diagnostic group, second medication for mood symptoms. This finding was
and discharged on medication for affective symptoms as strengthened when associated with shorter duration of
well as atypical neuroleptic, were found to be highly untreated illness prior to hospital admission. While these
associated with each other. The variable, depressive dis- results must be seen as exploratory, they are variables
order, was dropped from further analysis as it was highly related to the underlying condition and its treatment, and
correlated with diagnostic group, r = 0.76, P \ 0.0001. may reflect the presence of an unidentified and undertreated
Diagnostic group was also highly associated with discharge mood syndrome. Supporting this understanding of the data
on atypical neuroleptic with medication for affective is the observation that two variables, presence of major
symptoms, r = 0.64, P \ 0.0001. Several variables were depressive disorder and diagnosis of affective psychosis,
not found to be significant: age at admission, gender, were highly associated with discharge on concurrent med-
family history of depression, and decreased social support. ications. This is not surprising since the treating physician
In the multivariable model, the strongest predictor who identified an affective disorder would also be likely to
of adherence to medication following discharge was dis- prescribe treatment for the affective disorder. A possible
charged on an atypical antipsychotic agent and at the same explanation for the current results is that where affective
123
Eur Child Adolesc Psychiatry (2009) 18:587595 591
Gender
Male 39 (60) 30 (77) 9 (23)
Female 26 (40) 19 (73) 7 (27)
Type of onset
Sudden 17 (27) 12 (71) 5 (29)
Gradual 47 (73) 37 (79) 10 (21)
Duration of untreated illness
Weeks 62 (100) 20.76 (28.25) 47 (76) 23.47 (31.34) 15 (24) 12.27 (12.12)
Maintenance of social support
Decreased 39 (61) 27 (69) 12 (31)
Unchanged 25 (39) 21 (84) 4 (16)
Substance use
Illicit drug use
No 39 (60) 28 (72) 11 (28)
Yes 26 (40) 21 (81) 5 (19)
Alcohol use
No 41 (63) 30 (73) 11 (27)
Yes 24 (37) 19 (79) 5 (21)
Age at first-episode hospital admission 65 (100) 15.35 (2.08) 49 (75) 15.33 (2.26) 16 (25) 15.42 (1.46)
Type of symptoms
Positive only 17 (27) 12 (71) 5 (29)
Positive and negative 47 (73) 36 (77) 11 (23)
Presence of mania
Absent 50 (83) 39 (78) 11 (22)
Present 10 (17) 9 (90) 1 (10)
Presence of MDD
Absent 38 (62) 27 (71) 11 (29)
Present 23 (38) 21 (91) 2 (9)
Family history of depression
No 45 (69) 31 (69) 14 (31)
Yes 20 (31) 18 (90) 2 (10)
Family history of schizophrenia
No 57 (88) 14 (25) 43 (75)
Yes 8 (12) 2 (25) 6 (75)
Family history of bipolar
No 58 (89) 13 (22) 45 (78)
Yes 7 (11) 3 (43) 4 (57)
Length of stay (Days) 65 (100) 26.85 (36.62) 49 (75) 23.55 (19.45) 16 (25) 23.81 (13.54)
symptoms co-occur with psychosis, a conservative The other factor potentially associated with improved
approach to diagnosing and treating a possible early-onset adherence was a shorter length of illness prior to hospital-
bipolar disorder may not be helpful and may contribute to ization. Although a modest effect, not quite reaching sta-
poor medication adherence and relapse. Indeed guidelines tistical significance, this result is concordant with the
[53] based on evidence drawn from the adult literature association in the literature of a shorter duration of untreated
recommend treatment starting with both antipsychotic and illness with adherence [40, 43, 5467]. Since a history of
mood stabilizing medications at the time of diagnosis for rapid onset of psychosis symptoms is classically linked with
early-onset bipolar disorder with psychosis. bipolar disorder and consequently provides greater clinical
123
592 Eur Child Adolesc Psychiatry (2009) 18:587595
Table 2 Medications
Atypical antipsychotic Additional medications Freq. Group % Total %
Olanzapine
Alone 8 30.8 12.3
Antidepressanta 6 23.1 9.2
b
Mood stabilizer 7 26.9 10.8
Mood stabilizerb and Benzodiazepinec 1 3.8 1.5
Mood stabilizerb and other medicationf 1 3.8 1.5
Typical antipsychoticd and Benzodiazepinec 2 7.7 3.1
Benzodiazepinec 1 3.8 1.5
Olanzapine subtotal 26 100 40
Quetiapine
Alone 3 25.0 4.6
Antidepressanta 4 33.3 6.2
Antidepressanta and mood stabilizerb 1 8.3 1.5
Mood stabilizerb, typical antipsychoticd and Benzodiazepinec 2 16.7 3.1
c
Benzodiazepine 1 8.3 1.5
Other medicationf 1 8.3 1.5
Quetiapine subtotal 12 100 18
Risperidone
Alone 14 51.9 21.5
a
Antidepressant 5 18.5 7.7
Antidepressanta, anticholinergice and Benzodiazepinec 1 3.7 1.5
Mood stabilizerb 2 7.4 3.1
Mood stabilizerb and anticholinergice 2 7.4 3.1
Typical antipsychoticd and anticholinergice 1 3.7 1.5
Anticholinergice 2 7.4 3.1
Risperidone subtotal 27 100 42
Total 65 100 100
a
Antidepressants include Sertraline, Paroxetine, Fluoxetine and Venlafaxine
b
Mood Stabilizers include Lithium, Divalproex Sodium and Carbamazepine
c
Benzodiazepines include Lorazepam and Clonazepam
d
Typical antipsychotic include Haloperidol, Perphenazine and Nozinan
e
Anticholinergics include Cogentin, Benztropine and Procyclidine
f
Other medications include Synthroid and Tri-cyclen
certainty in using that diagnosis, this second predictor sup- those with more complicated medication regimens, that is,
ports the suggestion that lack of clear diagnosis led to poor two rather than one medication, actually appear to have
adherence and subsequent relapse. Also, it is possible that improved adherence. However, more study is required to
those youths with the most puzzling symptom patterns and replicate this finding and to elucidate its significance.
therefore most difficult to diagnose disorders and also most It is important to note the limitations of this preliminary
likely to require multiple hospitalizations. study. The retrospective component of the research design
Although we had anticipated that increased social sup- using data from health records could result in potential bias
port would be associated with improved adherence, this from the way information was recorded, or from lack of the
hypothesis was not supported by the data. This may be completeness and inaccuracy of charts. Another limitation
because our population was biased toward including those was the lack of a semi-structured diagnostic interview
with relatively high levels of family support reducing the across all settings. Similarly, the retrospective follow-up
variability in this potential predictor. Likewise, we relied on parental memory about the treatment over the
hypothesized that those young people with more complex period of follow up, 2 years or more. An additional limi-
disorders and complicated treatment regimens would show tation of the study was the low response rate. Only 38% of
poorer adherence. The current results may suggest that the eligible subjects in the sample frame participated,
123
Eur Child Adolesc Psychiatry (2009) 18:587595 593
Pearson correlation r r r r r r r r
Prob [ |r| under H0 P P P P P P P P
Rho = 0
Number of observations n n n n n n n n
Maintenance of social 0.205 -0.147 -0.185 -0.051 0.12 -0.058 -0.22
support 0.11 0.25 0.14 0.69 0.34 0.66 0.08
61 64 64 64 64 60 64
Duration of untreated -0.145 -0.211 0.128 0.06 -0.097 0.214
illness 0.26 0.1 0.32 0.64 0.47 0.09
62 62 62 62 58 62
Psychosis diagnostic 0.638 -0.045 -0.194 0.758 0.227
group \0.0001 0.72 0.12 \0.0001 0.07
65 65 65 61 65
Prescribed mood -0.21 -0.226 0.52 0.298
stabilizer or 0.09 0.07 \0.0001 0.02
antidepressant
65 65 61 65
Age at admission 0.268 -0.034 -0.106
0.03 0.79 0.4
65 61 65
Gender -0.135 -0.204
0.3 0.1
61 65
Depression 0.249
0.05
61
Family history of
depression
leading to potential bias because participants were often here is hypothesis-generating, and limited by lack of sta-
those still connected to treatment and who had volunteered tistical power. Further investigation is required to replicate
to participate. Response and attrition rates have remained a the findings and clarify the significance. The findings may
recognized limitation in first-episode schizophrenia and reflect difficulties in obtaining an accurate diagnosis of
adult studies on treatment adherence with psychotic dis- early-onset bipolar disorder in the context of the initial
orders [49, 68]. The response rate reported here, while low, psychotic presentation in unselected community popula-
is within the range reported in other studies using compa- tions. Where patients are hard to diagnose, practitioners
rable samples [50, 69, 70]. As in many exploratory studies, sometimes use a conservative approach to prescribe med-
the sample size for this analysis was small and lack of ication, using one agent initially rather than using two.
power likely contributed to modest effects of variables Hard to diagnose patients understandably may relapse and
such as duration of untreated illness. require additional hospitalizations, both for further diag-
nostic clarification as well as treatment. However, active
treatment of affective symptoms at the time of initial
Clinical implications psychotic presentation using concurrent agents may
improve treatment adherence and relapse rate compared
The strongest predictor of atypical antipsychotic medica- with using an atypical antipsychotic alone. Further research
tion adherence in youth following hospitalization for first- is required to clarify the best intervention approach for
episode psychosis was discharged on concurrent medica- youth with first-episode psychosis when diagnostic clarity
tion for affective symptoms. The information presented is elusive.
123
594 Eur Child Adolesc Psychiatry (2009) 18:587595
123
Eur Child Adolesc Psychiatry (2009) 18:587595 595
43. Johnstone EC, Macmillan JF, Frith CD, Benn DK, Crow TJ schizophrenic and schizoaffective patients. Eur Arch Psychiatry
(1990) Further investigation of the predictors of outcome Clin Neurosci 252(5):226231
following first schizophrenic episodes. [see comment]. Br J 57. Crow TJ, Done DJ (1986) Age of onset of schizophrenia in
Psychiatry 157:182189 siblings: a test of the contagion hypothesis. Psychiatry Res
44. Parker G, Hadzi-Pavlovic D (1995) The capacity of a measure of 18(2):107117
disability (the LSP) to predict hospital readmission in those with 58. Edwards P, Roberts I, Clarke M et al (2002) Increasing response
schizophrenia. Psychol Med 25(1):157163 rates to postal questionnaires: Systematic review. BMJ
45. Findley TW, Daum MC (1989) Research in physical medicine 324(7347):11831185
and rehabilitation. III. The chart review or how to use clinical 59. Harrigan SM, McGorry PD, Krstev H (2003) Does treatment
data for exploratory retrospective studies. Am J Phys Med delay in first-episode psychosis really matter? Psychol Med
Rehabil 68(3):150157 33(1):97110
46. Gearing RE, Mian IA, Barber J, Ickowicz A (2006) A method- 60. Keshavan MS, Haas G, Miewald J et al (2003) Prolonged
ology for conducting retrospective chart review research in child untreated illness duration from prodromal onset predicts outcome
and adolescent psychiatry. J Can Acad Child Adolesc Psychiatry in first episode psychoses. [see comment]. Schizophr Bull
15(3):126134 29(4):757769
47. Gearing RE (2007) Developing a risk-model of time to first 61. Loebel AD, Lieberman JA, Alvir JM, Mayerhoff DI, Geisler SH,
relapse for children and adolescents with primary psychotic Szymanski SR (1992) Duration of psychosis and outcome in first-
disorders or mood disorders with psychotic features [Ph.D.], episode schizophrenia. Am J Psychiatry 149(9):11831188
University of Toronto 62. McGorry PD, Harrigan SM, Amminger P, Norman R, Malla A
48. SAS II (2002) SAS 9.1.3 Help and Documentation. SAS Institute (2001) Untreated initial psychosis. [comment]. Am J Psychiatry
Inc, Cary 158(7):11611163
49. Bergen J, Hunt G, Armitage P, Bashir M (1998) Six-month 63. Oosthuizen P, Emsley RA, Keyter N, Niehaus DJ, Koen L (2005)
outcome following a relapse of schizophrenia. Aus N Z J Psy- Duration of untreated psychosis and outcome in first-episode
chiatry 32(6):815822 psychosis. Perspective from a developing country. Acta Psychiatr
50. Olfson M, Mechanic D, Hansell S, Boyer CA, Walkup J, Weiden Scand 111(3):214219
PJ (2000) Predicting medication noncompliance after hospital 64. Szymanski SR, Cannon TD, Gallacher F, Erwin RJ, Gur RE
discharge among patients with schizophrenia. Psychiatr Serv (1996) Course of treatment response in first-episode and chronic
51(2):216222 schizophrenia. Am J Psychiatry 153(4):519525
51. Valenstein M, Ganoczy D, McCarthy JF, Myra Kim H, Lee TA, 65. Ucok A, Polat A, Genc A, Cakiotar S, Turan N (2004) Duration
Blow FC (2006) Antipsychotic adherence over time among of untreated psychosis may predict acute treatment response in
patients receiving treatment for schizophrenia: a retrospective first-episode schizophrenia. J Psychiatr Res 38(2):163168
review. J Clin Psychiatry 67(10):15421550 66. Wyatt RJ, Damiani LM, Henter ID (1998) First-episode schizo-
52. Birchwood M, Todd P, Jackson C (1998) Early intervention in phrenia. Early intervention and medication discontinuation in the
psychosis. The critical period hypothesis. Br J Psychiatry Suppl context of course and treatment. Br J Psychiatry Suppl 172(33):
172(33):5359 7783
53. Kowatch RA, Fristad M, Birmaher B, Wagner KD, Findling RL, 67. Yung AR, Phillips LJ, McGorry PD et al (1998) Prediction of
Hellander M (2005) Treatment guidelines for children and psychosis. A step towards indicated prevention of schizophrenia.
adolescents with bipolar disorder. J Am Acad Child Adolesc [see comment]. Br J Psychiatry Suppl 172(33):1420
Psychiatry 44(3):213235 68. Birchwood M (1992) Early intervention in schizophrenia: The-
54. Amminger GP, Edwards J, Brewer WJ, Harrigan S, McGorry PD oretical background and clinical strategies. Br J Clin Psychol
(2002) Duration of untreated psychosis and cognitive deteriora- 31(Pt 3):257278
tion in first-episode schizophrenia. Schizophr Res 54(3):223230 69. Olfson M, Mechanic D, Boyer CA, Hansell S (1998) Linking
55. Beiser M, Erickson D, Fleming JA, Iacono WG (1993) Estab- inpatients with schizophrenia to outpatient care. Psychiatr Serv
lishing the onset of psychotic illness. Am J Psychiatry 49(7):911917
150(9):13491354 70. Pfeiffer SI (1989) Follow-up of children and adolescents treated
56. Bottlender R, Sato T, Jager M, Groll C, Strauss A, Moller HJ in psychiatric facilities: a methodology review. Psychiatr Hosp
(2002) The impact of duration of untreated psychosis and 20(1):1520
premorbid functioning on outcome of first inpatient treatment in
123