First-episode psychosis patients recruited into treatment via early detection teams versus ordinary pathways: course, outcome and health service use during first 2 years


Dr Jan Olav Johannessen, Division of Psychiatry, Stavanger University Hospital, PO Box 1163 Hillevaag, N-4095 Stavanger, Norway. Email:


Aim:  Within an early detection sector, to compare the 1- and 2-year course and outcome of first-episode psychosis patients coming into the treatment system via active outreach detection teams (DTs) versus those achieving help via ordinary referral channels (not-DT).

Methods:  Longitudinal, comparative study of two parallel consecutive samples using structured clinical interview for the DSM-IV, Positive and Negative Syndrome Scale Score, Global Assessment of Functioning Scale and Premorbid Assessment of Functioning Scale.

Results:  The DT group had significantly better functioning at baseline, but this was reversed after 3 months. At 2 years the groups had similar outcome. The DT group developed a more serious diagnostic pattern, had more cases of schizophrenia, and was more frequently treated on an outpatient basis only.

Conclusions:  The DTs recruited more chronic patients with poorer prognostic features, but fewer symptoms and better functioning at baseline. At 2 years the DT-patients did as well as the not-DT patients. They recovered more slowly, but given sufficient time, responded as well to therapy as the not-DT group.


The development of psychiatric health services for early intervention in first-episode psychosis (FEP) is based on the fact that duration of untreated psychosis (DUP) is very long and associated with a more severe and chronic course;1 and the need for developing more phase-specific treatments and streamed care that can improve outcome.2,3

Whether DUP has a causal impact on the person’s long-term prognosis has not yet been clearly demonstrated. Even if one presumed that it made no difference to the long-term prognosis whether one intervenes early or late in the development of a FEP disorder, it represents an additional and in most cases almost intolerable extra burden to be psychotic without receiving adequate help. People in active psychosis are vulnerable to initiating irrational, unpredictable and often permanently damaging behaviours, both to themselves and others. The reasons for treatment delay include ignorance, denial, stigma, a lack of motivation, absence of information about early signs of psychosis and lack of access to adequate treatment. The optimal way to organize psychiatric health services in order to achieve early intervention in FEP has not yet been demonstrated, as also outlined in a recent Cochrane review by Marshall and Lockwood.4

We here address this question from a health-care sector that has established an early detection (ED) system consisting of public education about early signs and symptoms of psychosis, and the possibility of direct and easy access to care through active, outreach detection teams (DTs). This is a low-threshold service, which offers an alternative route into treatment, including self-referral, compared with the ordinary referral system. These DTs do not provide treatment; treatment is provided after admission to a standardized programme within ordinary service systems. Patients still could approach the treatment system through ordinary channels, as before, that is, via referral from the general practitioners (not-DT). Details of the programme have been described elsewhere.5

The purpose of this part of the study is not to evaluate the overall effects of ED systems with DTs on DUP and outcome, but to investigate the difference between patients recruited through the DTs and patients coming into treatment via ordinary referral systems.

In a previous publication from this study of FEP, we reported on the differences at admission and 3 months of acute treatment between patients coming into the treatment system via DTs versus patients coming into treatment via the ordinary referral systems (not-DT).6

The DT patients were younger at the start of treatment (22.5 vs. 26.3 years) compared with not-DT patients. The DT teams recruited more male patients with longer DUPs (53.4 vs. 31.9 weeks (mean), 18.0 vs. 4.0 weeks (median)), and with higher scores on drug abuse. At admission there were more cases of psychosis not otherwise specified (NOS) in the DT sample and a trend for more affective cases in the not-DT sample. Measured by the Global Assessment of Functioning scale (GAF), the DT group was better functioning and had fewer symptoms. On the Positive and Negative Symptom Scale (PANSS) there were no significant differences between the two groups on positive and negative symptoms.

At 3 months follow up this was reversed; the not-DT patients had lower symptom levels measured by GAF and PANSS (general and total), and the same level of functioning as the DT group.

Based on these findings, we formed the hypothesis that ‘early intervention systems that include outreach case-seeking structures (DTs) with easy access to treatment seem to recruit a younger but more chronically disordered subgroup of patients with FEP, with a poorer functional and symptomatic long-term outcome’.6

The purpose of this paper is to explore further the course and outcome at 1 and 2 years. We address the following questions:

  • 1Are the DT patients more chronically disordered, with a more protracted recovery process and poorer outcome at 2 years compared with the not-DT patients?
  • 2Do the DT patients have different patterns of health service consumption from the not-DT group during the first 2 years of treatment?


Setting and study population

In this study the sample consists of patients consecutively recruited to first treatment in the ED sectors (Rogaland County) of the TIPS-study during 1997–2000. The TIPS project (Early identification and treatment of psychosis) is a four-site prospective clinical trial in Norway and Denmark, designed to investigate the effect of the timing of treatment in FEP. Two health-care sectors (Stavanger and Haugesund, comprising Rogaland County, Norway) are experimental and have developed a system for ED, aimed at reducing DUP. Two other sectors (Ullevaal, Norway and Roskilde, Denmark) are comparison sectors and rely on existing referral systems for FEP (No-ED). The overall TIPS-study ultimately compares early detected with usual detected patients. The experimental sector (Rogaland county, ED sector) is characterized by a comprehensive education and detection system designed to enhance the knowledge about early signs of psychosis within the general public, schools and among health professionals.

The major target groups for information and education were the general public, families and friends, schools (teachers and students) and health professionals. The public information campaigns (ICs) aimed at: (i) raising the public’s awareness of signs and symptoms of psychiatric disorders, especially psychosis, and reducing stigma connected to psychiatric disorders in general and to psychosis and psychiatric services in special (to change the public’s attitudes); (ii) raising the public awareness of the early signs of serious psychiatric disorder (mental health literacy) and of the importance of seeking help early, especially among young people; and (iii) informing repeatedly about the available help, and especially about the low-threshold DTs.

Early DTs have been established in the experimental sectors to lower the threshold of entry to the specialized psychiatric services, and to recruit appropriate patients as early in the illness course as possible. The DTs consist of psychiatrists, psychologists, psychiatric nurses and social workers. They are on call from 8 a.m. till 4 p.m. Monday to Friday. The teams are highly mobile and work with an active outreach attitude.

First, the DT made an initial assessment over the telephone, and the first decision was whether there was a psychiatric problem or not. This decision was made from a clinical judgement only, based on a semi-structured interview guide that included prodromal signs according to DSM III-R, and PANSS. Based on this first interview, the DTs concluded whether it was psychiatry/not psychiatry. If ‘yes’, the possible further conclusions were: drug abuse only, life crises, depression, prodromal case, not psychosis but definitely psychiatry, psychosis, other.

Second, if it were a possible psychiatric case, the DTs would meet the patient or/and the referring people wherever was convenient: at home, in a general practitioner’s office, at school, or at the DTs’ office. The patient would be met by the DT within maximum 24 h after first contact. In most cases the assessment was carried out within only a few hours. The DT carried out a preliminary evaluation using PANSS and a GAF assessment. All DT members were trained raters in these manuals.

Third, when the DT had established that a psychosis case could not be ruled out, the patient was referred to a more comprehensive, scientifically based evaluation by the assessment team (AT). The AT consisted of scientifically trained psychiatrists or the equivalent, and they would decide, through an extensive assessment including structured clinical interview for the DSM-IV, whether or not the patient was suffering from a first psychotic episode. Patients meeting the inclusion criteria entered the study and received the standard treatment protocol. Patients who did not fulfil the criteria for a psychotic disorder, was referred to treatment based on their presenting symptoms and needs, and ‘at risk’ cases was contacted each 3 months to control for psychosis development.

The period of inclusion was 1997–2000, with follow up planned at 3 months, 1, 2, 5 and 10 years.7 The criteria for inclusion were: (i) a first episode of a non-affective psychosis, that is, schizophrenia, schizophreniform, schizoaffective and delusional disorder, brief psychosis, affective disorder with mood-incongruent delusions and psychosis NOS; (ii) living in the catchment area; (iii) age 15–65 years; (iv) IQ >70; and (v) suffering from a first episode of psychosis. The exclusion criteria were a history of an earlier treated first psychosis, receiving adequate prior neuroleptic treatment, and an organic or substance-induced psychosis. Written informed consent was obtained from all included subjects. Three patients were deemed unable to give informed consent and consequently excluded.

The study was approved by the regional ethical research committee.

The population in the ED sector is 380 000. Over the 4 years, 203 FEP patients were identified. Of these 42 refused to participate and 161 patients were included, 56 in the DT group, and 105 in the not-DT group.

The proportion of refusers was significantly higher in the DT group (28.2% vs. 16.0%, P=0.005). The refusers in the DT group were younger than the not-DT group (24.3 years vs. 30.0, P=0.0028), thus demonstrating the same relationship between the two groups as for those included. No difference on gender distribution between the two groups of refusers was found. Regarding diagnosis, the numbers in each category are small, and therefore difficult to interpret. When refusers were collapsed into core versus not-core schizophrenia, there seemed to be a trend towards more schizophrenia spectrum (core) in the not-DT group (P =0.047). DUP was 26 weeks (median) for both groups of refusers. Overall, these characteristics of refusers in the total ED sample did not seem to have biased the research samples in a direction that should have substantial influence on the results.

Attrition rates did not differ between DT/not-DT groups during follow-up period.


The structured clinical interview for the DSM-IV8 was used for diagnostic purposes. Symptom levels were measured by the PANSS.9 Global functioning was measured by the GAF,10 the scores were split into symptom scores (GAFs) and function scores (GAFf) to improve psychometric properties. The scale varies from 0 (lowest) to 100 (highest).

Duration of untreated psychosis was measured as the time from the first onset of positive psychotic symptoms (the first week with PANSS score of four or more on Positive scale items one, three, five, six or General scale item nine) to the start of first adequate treatment of psychosis (i.e. admission to the study). Multiple sources, including personal interviews with patients and relatives, were used to ascertain the length of this period. Premorbid functioning was measured by the Premorbid Assessment of Functioning Scale.11 Drug and alcohol abuse was measured by the Clinician rating scale.12 Social functioning (number of friends and participation in meaningful activities) was measured with the Strauss–Carpenter scale.13

All raters were trained in the use of study instruments by rating pre-prepared case notes and audio/videotapes before entering the study ATs. The rating of essential variables such as diagnosis and DUP was made by consensus with experienced clinical researchers. Reliability for the PANSS scores was measured by the rating of videotaped interviews of first-episode patients by all raters. Reliability for diagnosis, GAF and DUP was measured by the rating of actual case notes by blind raters with long clinical research experience. Baseline inter-rater reliability was fair to good for all the above measures.14 Reliability at follow up was assessed by rating 31 randomly selected vignettes from 1- and 2-year follow up. These were rated by two experienced psychiatrists on the following variables: diagnosis, GAF symptoms, GAF function, alcohol and drug scores. For all dimensions the reliability was good. For diagnosis: kappa=0.81, for the other dimensions Intra Class Correlation (ICC) (1.1) were: GAF function: 0.86; GAF symptoms 0.91; Alcohol 0.75; Drugs 0.86.


Analyses were made with the statistical package SPSS (version 12.01). Mean values are reported with standard deviations in subsequent brackets, and median values are reported for variables with skewed distribution. T-test was used for comparison between groups and the chi-squared test and the Fischer’s exact test were used with dichotomized data. Non-parametric tests are used for data without normal distribution.

To test the development over time, a mixed within-between subjects analysis of variance was used, with scores of PANSS and GAF at baseline, 3 months, 1 year and 2 years as time variables and DT as grouping variable. All tests are two-tailed. The P-value of statistical significance is 0.05. As noted in several other studies, DUP is not normally distributed, whereas its natural logarithm is. All analyses that include DUP are thus non-parametric when possible. In parametric analysis, the DUP has been transformed to its natural logarithm.


Although the DT group had a less serious symptom- and function-picture at admission measured by GAFs and GAFf, the not-DT group was doing better at 3 months’ follow up (Figs 1,2).

Figure 1.

Global Assessment of Functioning scale (GAF) function scores for detection team (DT; inline image) and not-DT (inline image) patients.

Figure 2.

Global Assessment of Functioning scale (GAF) symptom scores for detection team (DT; inline image) and not-DT (inline image) patients.

At 1- and 2-year follow up of clinical state, symptoms and functioning, however, these differences (except for drug abuse) were no longer statistically significant, and at 2 years they had completely disappeared (Table 1).

Table 1.  Patients’ characteristics at 1 and 2 years
n (DT/not-DT)DTNot-DTP-valuen (DT/not-DT)DTNot-DTP-value
  1. DT, detection team; GAFf, Global Assessment of Functioning scale function scores; GAFs, Global Assessment of Functioning scale symptom scores; NS, not significant; PANSS, Positive and Negative Symptom Scale; SD, standard deviation.

GAFs (SD)52/10049.70 (14.00)53.10 (17.00)NS49/9053.30 (16.40)54.50 (18.90)NS
GAFf (SD)52/10049.40 (14.00)52.70 (17.00)NS48/9053.20 (15.60)54.80 (18.60)NS
PANSS+ (SD)49/9312.90 (5.70)11.30 (5.50)NS47/8811.50 (5.00)11.80 (6.00)NS
PANSS– (SD)49/9312.90 (5.80)12.20 (6.20)NS47/8811.90 (5.30)11.80 (5.90)NS
PANSS gen (SD)49/9327.20 (9.60)24.30 (10.20)NS47/8823.10 (6.30)23.40 (7.80)NS
Alcohol (SD)52/981.90 (0.53)1.84 (0.51)NS51/912.00 (0.53)1.86 (0.51)NS
Drugs (SD)52/981.87 (1.14)1.50 (0.87)0.02956/1041.84 (0.22)1.42 (0.80)0.014
In remission (%)56/10433 (59)72 (69)NS51/9835 (69)73 (70)NS
In relapse (%)56/1045 (9)11 (11)NS51/987 (14)12 (12)NS

To test the longitudinal development, we used a mixed between-within subjects analysis of variance. The results are given in Table 2.

Table 2.  Results of a mixed between-within subject analysis of variance
 Time × group interactionMain effect of timeGroup effect
Wilks lambdaP-valueWilks lambdaP-valuePartial etaFP-valuePartial eta
  1. GAF, Global Assessment of Functioning scale.

Positive symptoms0.970.310.46<0.00050.540.080.770.001
Negative symptoms0.980.570.890.00300.110.770.380.003
General symptoms0.960.160.56<0.00050.430.360.550.003
GAF symptoms0.940.060.37<0.00050.630.130.720.001
GAF function0.930.040.45<0.00050.550.040.840.000

As seen from Table 2 there was a fairly strong main effect of time for negative symptoms and a very strong effect for all other variables. There was a significant interaction between group and time for GAF function, and a close to significant interaction for GAF symptoms.

We reran all the analyses with DUP (natural logarithm) as a covariate. In these analyses lnDUP had a significant interaction with time for PANSS positive symptoms and GAF symptoms and function. The significant interaction between group and GAF function disappeared, indicating that it, at least to a large extent, was mediated through DUP. A closer inspection of our data indicated less improvement over time for the patients with DUP ≥26 weeks, especially among the not-DT patients.

At admittance there were no differences in diagnosis between the two groups except for a higher proportion of psychotic disorder NOS in the DT group. At 1 year follow up there were fewer with affective disorder with mood-incongruent psychotic features in the DT group. At 2 years there were significantly more cases with a schizophrenia diagnosis in the DT group (Table 3).

Table 3.  Diagnostic distribution at 1 and 2 years
Diagnosis %DTNot-DTSignificant/NS
  • *

    P < 0.05;

  • **

    P < 0.005.

  • DT, detection team; NS, non-significant.

Schizophreniform disorder16002577NS
Schizoaffective disorder131915192121NS
Brief psychosis4441147NS
Delusional disorder744311NS
Affective disorder with mood-incongruent psychotic features964152018(B/E, C/F)*
Psychotic disorder not otherwise specified20159321(A/D, B/E, C/F)**

Therapeutically, the number of admissions to hospital care during the first and second year was lower in the DT group, as was the total length of stay in hospital, and 30% of the DT group was not admitted at all the first year (Table 4). The time from inclusion to start of medication was longer in the DT group. The DT group had fewer changes in the number of individual therapists. There were no differences in attrition rates or for participation in the standard treatment programme between the two groups.

Table 4.  Treatment during first 2 years
 During first yearDuring first 2 years
  1. DT, detection team; NS, non-significant; SD, standard deviation.

Per cent treated as outpatients only55/10430.96.70.00048/9825.02.00.000
No. admissions (SD)55/1041.07 (0.94)1.51 (1.01)0.00748/981.63 (1.47)2.37 (1.82)0.015
Duration of time as inpatient in weeks, mean (SD)55/10412.1 (15.3)19.9 (17.7)0.00447/9622.8 (30.3)32.7 (32.3)NS
No. therapists (SD)55/1041.33 (0.61)1.63 (0.80)0.00849/991.71 (0.91)2.02 (1.18)NS
Continuously psychotic (%)56/10418 (32)21 (20)NS51/989 (17.6)17 (17.3)NS
Involuntary admission, %56/1046873NS    
Time from inclusion to onset medical treatment, weeks, mean (SD)56/1040.87 (1.6)0.09 (0.8)0.034    


Strategies for achieving early intervention in FEP are strategies for changing the help-seeking behaviour in the population. In the TIPS ED and intervention project, an ED programme was created consisting of ICs about early psychosis and outreach early DTs. Both components undoubtedly contributed to the documented reduction in DUP.7 The relative contributions of the ICs versus the DTs have yet to be assessed, but this study demonstrates clearly that adding DTs to recruitment efforts draws a sample of FEP patients who differs somewhat in character and course from those coming to first evaluation and treatment via existing psychiatric service system structures.

Those accessing treatment via the DTs tended to be younger men who used/abused drugs and had been psychotic for some time with symptoms that were not highly affect laden. Despite this picture of apparent early chronicity, the patients in this sample were functioning moderately well, and many presented with symptomatic pictures atypical for acute onset FEP. They were psychotic but not disorganized or overwhelmed by their symptoms as were many study patients who came to treatment through the usual channels, for example the emergency room to the hospital. The better functioning and organization was reflected in their pattern of treatment utilization, which required fewer hospitalizations, less inpatient time and more outpatient treatment with fewer changes of therapists (perhaps reflecting a more robust capacity for relatedness). The difference in time to start of medication probably reflects the more acute and disorganized symptom picture in the not-DT group at presentation.

The initial course of this group’s response to treatment, on the other hand, again suggested greater chronicity. They were less ill at intake than the not-DT patients but by 3 months they had responded less completely to a standard treatment package than their more acutely ill and disorganized not-DT counterparts. Unlike the latter patients, however, the DT sample’s response to treatment did not plateau at 3 months. The DT patients continued to improve and by 2 years had eliminated the 3 months’ outcome difference.

The DT group had a diagnostic distribution at 1 and 2 years that traditionally is considered to reflect a more serious illness development, with more patients developing schizophrenia. It thus seemed that a subgroup of the DT sample is characterized by a longer phase of illness before getting into treatment, a longer phase of recovery and a slower response to treatment. This is illustrated in Figure 3.

Figure 3.

Longer duration of untreated psychosis (DUP; median) and longer time to recovery. GAF, Global Assessment of Functioning scale. (inline image) Detection team; (inline image) not detection team.

It is a promising finding that the DT sample reached the same level of functioning after 2 years as the not-DT group, in spite of their apparently more serious diagnostic distribution, and slower recovery process during the first 3 months. Our hypothesis that they should have a more chronic, symptomatic and functional development thus is rejected; however, the diagnostic distribution reflected a more serious illness. Gelber et al. demonstrated that patients with an established schizophrenia diagnosis at 1-year follow up had longer duration of untreated illness (the sum of DUP and the prodromal period) than patients with other FEP diagnosis.15 We did not measure duration of untreated illness, but our DT group with a longer DUP also had more cases of schizophrenia. The fact that the clinical state is similar for the two groups at 2-year follow up might reflect the fact that they are both offered a comprehensive treatment. This could be an indication that the DTs serve their purpose in relation to this younger group of patients with a more subtle onset, more substance abuse and more often treated on an outpatient basis, a group that has been found to be more likely to be non-adherent with medication and to not attend appointments.16

Drake and Lewis state that there is a curvilinear relationship between DUP and outcome, such that the improvement in outcome is much greater if one moves from 6 weeks down to 3 weeks compared with 6 months down to 3 months.17 They add, ‘this implies that the maximum benefit of early intervention services will be gained only by shifting patients into the shortest part of the DUP range’ (p. 149). The DTs recruit younger men with a protracted recovery course, the reasons for which are obscure at this point. If Drake and Lewis are right, however, reducing the delay of treatment for this subgroup may be a special challenge and have a special payoff.

A few studies have examined the effects of different health service configurations on patients with early psychosis. Craig et al. investigated whether a specialist team could achieve better outcomes than existing services for people with early non-affective psychotic disorders.18 This randomized controlled trial demonstrated that such a team was superior to standard care for maintaining contact with professionals and for reducing hospital readmissions. Our DTs are different from the Craig et al.’s teams in that they do not provide treatment and we have no data to suggest that the DTs reduce readmissions once contact with the treatment system is established.

The importance of the finding that a significantly higher part of patients recruited into treatment via the DTs are treated on an outpatient basis only, is difficult to estimate at this point. If it is a result of the ED strategies, and implies that these patients would otherwise have been admitted to inpatient treatment at a later stage, the importance for both patients and society are vast, both in relation to the individual’s suffering, and to the costs connected to treatment. Only a few other studies have reported what proportion of a FEP population should be expected to be treated on an inpatient basis only. Wade et al. found that 76.9% of FEP patients were admitted to an inpatient unit during the first 3 months of treatment.19 Inpatient admission was associated with a diagnosis of affective psychosis and more severe behavioural and functional disturbance but not positive symptoms. The authors claim that the substantial proportion of young patients with FEP being admitted to hospital emphasizes the need for youth-friendly treatment environments and practices. In a study from Nottingham, UK, 19.3% of FEP patients were never admitted during the 3 years of follow up; negative symptoms and a longer DUP had an increased risk of late admission.20

In the TIPS project two major elements have been part of our early intervention strategies: low threshold/easy access, as represented by the DTs, and extensive public ICs. The relative contribution of the two components to the reduction of DUP is still unclear, and in an ongoing study in Rogaland, Norway, we try to investigate this in more detail. The OPUS study in Copenhagen established a form of DTs but no education campaigns but did not reduce DUP.21 Krstev et al. demonstrated that adding public information to DTs in two Australian counties reduced DUP only when some outliers with very long DUP were removed from the sample.22 They concluded that early intervention may have two effects: first, unmasking long DUP cases that would otherwise remain anonymous, and second, the DUP for the rest may be reduced. Our findings lend some support to the theory that DTs recruit a different sample, although at this point we have only indirect indications that these cases would otherwise be ‘hidden’, or detected later in the illness course, with longer DUPs. In a study of the influence of age on outcome of psychological treatments in FEP patients, Haddock et al. found that younger participants were more difficult to engage in therapy.23 This finding lends some support to the importance of designing our service systems to the specific needs of patient subgroups, such as represented by the DT sample.

Thus it appears that the DT element of the TIPS ED package selects a sample of FEP patients with a different presentation, treatment engagement and treatment response. We are unaware of any similar findings in other studies and therefore regard this as an observation worthy of note and further study. Of special interest, for example, is whether this sample will continue to improve with treatment as opposed to the not-DT sample, which appears to have plateaued in its response. Further follow up is clearly indicated. It is also unclear whether such a sample is unique to treatment programmes with DTs and whether such patients would not enter networks without them, at least at this stage and state of their illness. This question can only be answered with further studies such as this that offer patients with first psychosis more than one pathway to help.

Clinical implications

  • • In a sector with active information directed at early intervention, active outreach DTs recruit male patients with longer DUP, less ill at presentation, who recover more slowly, but reach the same level of functioning as patients recruited via ordinary pathways, after 2 years
  • • Patients recruited via DTs are more often treated on an outpatient basis only
  • • Patients recruited via DTs more often develop a more serious diagnostic picture


  • • The study does not offer any final explanation why the two groups of patients (DT group and not-DT group) recover at a different pace
  • • The study does not explain which factors contribute most to improvement within or between the two groups
  • • The study does not offer final conclusions of the relative importance of low-threshold strategies as represented by the DTs, versus the importance of targeted information strategies


This paper is part of the TIPS project with the following research group: Thomas McGlashan, MD (PI), Department of Psychiatry, Yale University, New Haven, CT, USA; Per Vaglum, MD (PI), Department of Behavioral Sciences in Medicine, University of Oslo, Norway; Svein Friis, MD, Institute of Psychiatry, Ullevål University Hospital, Norway; Ulrik Haahr, MD, Mid-sector Roskilde Psychiatric University Hospital Fjorden, Denmark; Jan Olav Johannessen, MD, Division of Psychiatry, Stavanger University Hospital, Norway; Tor K. Larsen, MD, Division of Psychiatry, Stavanger University Hospital, Norway; Ingrid Melle, MD, Institute of Psychiatry, Ullevål University Hospital, Norway; Stein Opjordsmoen, MD, Institute of Psychiatry, Ullevål University Hospital, Norway; Bjørn Rishovd Rund, PhD, Institute of Psychology, University of Oslo, Norway; Erik Simonsen, MD, Mid-sector, Roskilde Psychiatric University Hospital Fjorden, Denmark. Funding was received from the Norwegian National Research Council (No. 133897/320 and No. 154642/320), the Norwegian Department of Health and Social Affairs, the National Council for Mental Health/Health and Rehabilitation (No. 1997/41 and No. 2002/306), Rogaland County and Oslo County (Drs Vaglum, Johannessen, Friis, Larsen, Melle, Opjordsmoen). This paper was also funded by the Theodore and Vada Stanley Foundation, the Regional Health Research Foundation for Eastern Region, Denmark; Roskilde County, Denmark, Helsefonden Lundbeck Pharma, Eli Lilly and Janssen-Cilag Pharmaceuticals (Drs Simonsen and Haahr). Support was also provided by a National Alliance for Research on Schizophrenia and Depression (NARSAD) Distinguished Investigator Award and NIMH Grant MH-01654 (Dr McGlashan) and a NARSAD Young Investigator Award (Dr Larsen).