The sample consisted of patients with a clinical diagnosis of a schizophrenic or schizoaffective disorder (n = 86) consecutively admitted to the Department of Child and Adolescent Psychiatry at the University of Wuerzburg between 1990 and 2000. We also included patients with the diagnosis of a schizoaffective disorder, since previous studies showed no substantial differences in the outcome of this disorder compared to schizophrenic patients .
Specifically, we included former patients in our schizophrenia sample that met the following criteria:
They were consecutively admitted to our clinic and were treated as schizophrenia or schizoaffective patients according to the ICD-9- (295) or ICD-10-criteria (F20, F25) for more than one day.
They were younger than 18 years old and received their first inpatient treatment due to schizophrenia.
A retrospective diagnostic evaluation was carried out by two experienced clinicians checking the patients' records. For example, the clinical data from first admission were screened for symptoms like hallucinations, delusions, ideas of reference or social withdrawal to examine if the described symptoms referred to the ICD-9 or ICD-10-criteria. Based on a consensus analysis, all subjects stayed in the study.
To track the patients we used phone numbers and addresses found in the patients' records of their first admission. Their homes were spread over a large area including Bavaria, northern Baden-Wuerttemberg, and southern parts of Hessen and Thuringia. Those patients whose addresses could not be verified after intensive search were dismissed from the study. 37 of the 86 subjects systematically targeted for enrolment could not be traced because they and their families moved to unknown domiciles. A review of death records revealed one case of death of unknown reason. Therefore, we tried to include 48 former patients to our study. 21 subjects (43.7%) refused to participate (8 women/13 men). This left 27 patients (56.3%) in the study (8 women/19 men).
Each participant or their legal representative signed a consent form after having the study explained to them.
Variables and assessment instruments
To identify patients with the diagnosis of schizophrenia or schizoaffective disorder, two experienced clinicians reviewed the detailed hospital records from 1990 to 2000. For the assessment of characteristics of the first episode, ICD-9 and ICD-10 diagnoses for the first admission were taken from those records, as well as information about the anamnesis, symptomatology on admission, course of first inpatient treatment, family, and psychodiagnostic data (IQ and other test results).
After identifying the patients with schizophrenia or schizoaffective disorder (295, F20 or F25) diagnosis, the families or former patients were contacted in the next step by phone and were informed about the study. Then, we sent a cover letter, an information sheet and a questionnaire to the former patients (mail survey). The questionnaire consisted of several questions about psychopathological symptoms, further inpatient treatments, psychosocial functioning, and demographic characteristics like living situation, financial income, family situation, the state of education, and work. We also collected information about psychopathological and psychosocial outcome derived from semi-structured telephone interviews with patients and significant others before and after receiving the questionnaires. The semi-structured interview referred very closely to the items of the questionnaire.
For a standardized assessment of the schizophrenic symptoms and possible negative symptoms, two well established scales were included in the mail survey: the "Eppendorfer Schizophrenie-Inventar" (ESI) [21, 22] and the "Allgemeine Depressions-Skala" (ADS) .
In our study, we applied the ESI total score and the ADS total score for statistical analysis. The ESI is a relatively new questionnaire for self-assessment of pre-psychotic and psychotic disturbances in several cognitive and perceptual areas. It was designed for diagnostic, therapeutic control and research purposes and is well validated. Compared to a well established instrument like the Frankfurt Complaint Questionnaire (FCQ), the ESI showed superior results regarding reliability and diagnostic validity. Several studies initiated to evaluate the ESI showed correlations to neuropsychological, psychopathological and anamnestic variables . The total score ranges from 0 to 102. A cut-off value of 30 was established, as the authors found that only a small percentage (6.3%) of a non-schizophrenic sample had higher results. The ADS is a German scale for the assessment of depression, based on the "Center for Epidemiological Studies Depression Scale" (CES-D) . In the case of the ADS, we used the short version, consisting of 15 items with a maximal total score of 45 and a cut-off value of 18.
In summary, we followed a multidimensional approach with standardized instruments to assess the patients' present state of outcome. All scales show satisfying reliability and validity.
First, we checked the variables regarding their distribution by using the Shapiro-Wilk-test and by evaluating the skewness and kurtosis. To compare dropouts with patients we followed up, we used chi-square tests for the categorical variable sex and Mann-Whitney-U-tests for continuous variables (age at first admission, duration of first inpatient treatment, follow-up-interval).
A comparison of the mean ESI- and ADS-scores between genders was drawn, using the Mann-Whitney-U-tests. To analyze a possible association between ESI- and ADS-scores, a Pearson correlation was performed.
For an exploratory analysis of associations with explanatory variables, the outcome variable for the degree of schizophrenic psychopathology (ESI-Score) was dichotomized via a median split. In order to investigate predictive factors, a logistic regression analysis was performed.
The significance level was fixed at α = 0.05. All statistical calculations were performed with SPSS 13.0.