Early intervention in first-episode psychosis - Springer Link

4 downloads 0 Views 263KB Size Report
Key words early psychosis – community development – duration of untreated psychosis – prevention – early intervention. Introduction. For a significant number ...
Soc Psychiatry Psychiatr Epidemiol (2004) 39 : 711–719

DOI 10.1007/s00127-004-0798-5

ORIGINAL PAPER

Helen Krstev · Steve Carbone · Susy M. Harrigan · Christina Curry · Kathryn Elkins · Patrick D. McGorry

Early intervention in first-episode psychosis The impact of a community development campaign

Accepted: 2 March 2004

■ Abstract Objective Substantial delays in providing access to treatment in first-episode psychosis have been well documented. The present study examines the impact of strategies aimed at improving access and reducing delays. Method A pilot community education campaign was conducted with the aim of reducing the duration of untreated psychosis (DUP) in a geographically defined intervention sector located in the northwestern region of Melbourne, Australia. Utilising a quasi-experimental design, a comparison sector with similar demographics was selected from another part of the north-western region.A mobile early detection team and the same treatment system served both sectors. Results While there was no significant difference between the mean DUP for intervention and comparison sectors, the distributional features of DUP between the two regions were significantly different. In the intervention sector, disproportionately more cases with very long DUP were detected. When a small number of outliers were removed, the mean and median DUP in the intervention sector was reduced. Conclusion These findings highlight the complexity of treatment access and delay and suggest that efforts to reduce DUP may have two effects, not one. Firstly, a different sample of cases is treated through the detection of hidden “long DUP” cases that otherwise may have remained untreated. Secondly, the DUP for the remainder may indeed be reduced. More research with larger samples and more potent campaign strategies is clearly required. It may also H. Krstev () · S. Carbone · S. M. Harrigan · C. Curry · K. Elkins · P. D. McGorry Early Psychosis Prevention & Intervention Centre Locked Bag 10 Parkville (Vic) 3052, Australia Tel.: +61-3/8346-8215 Fax: +61-3/9347-9099 E-Mail: [email protected]

■ Key words early psychosis – community development – duration of untreated psychosis – prevention – early intervention

Introduction For a significant number of people experiencing the first episode of a psychotic disorder, a considerable period of time elapses between the onset of their illness and their entry into appropriate treatment. A number of studies have reported an average period of untreated psychosis of the order of 12 months or more (Beiser et al. 1993; Helgason 1990; Johnstone et al. 1986; Loebel et al. 1992; McGorry 1995). The relationship of this delay to outcome remains controversial, largely due to the possibility of its confounding by the more ‘intrinsic’ prognostic features, that is, those features that are more closely associated with the course of the psychotic illness and are less amenable to change, such as age of onset, gender and premorbid adjustment (Verdoux et al. 2001). However, this is likely to be a partial rather than complete explanation (McGorry 2000). Nevertheless, all are agreed treatment for psychotic disorders should be offered on purely clinical grounds as soon as possible after definite onset (Lieberman and Fenton 2000). We are a long way from this objective. The reasons for these long treatment delays are poorly understood, but one review (Lincoln and McGorry 1995) suggests a plethora of interconnected factors are likely to operate. Factors such as age, gender, symptom patterns and severity, mental health literacy, help-seeking behaviour, stigma associated with mental illness, knowledge of access to services, acceptability of services and professional case recognition skills have all been put forward to explain treatment delay (Lincoln

SPPE 798

H. Krstev () · S. Carbone · S. M. Harrigan · C. Curry · K. Elkins · P. D. McGorry The University of Melbourne Parkville (Vic), Australia

be worth considering whether there is a safe and ethical way to undertake a RCT of early versus delayed antipsychotic treatment to perhaps settle the DUP debate once and for all.

712

and McGorry 1995). Certain characteristics such as male gender, prominence of negative symptoms and persecutory delusions have been linked to a greater likelihood of delay. If we wish our clients to benefit from the potential advantages of early intervention and if we wish to study causal relationships between early intervention and outcome, we must first find ways to actively reduce treatment delays. This strategy has been articulated clearly by McGlashan (McGlashan 1998) and studied carefully in the TIPS project which is ongoing (Johannessen et al. 2001). The evidence for a link between DUP and outcome has been the subject of major reviews (McGlashan 1999; Norman and Malla 2001; Bottlender and Möller 2003). The general consensus is that the association is robust, with longer DUP being associated with poorer outcome (Beiser et al. 1993; Drake et al. 2000; Helgason 1990; Larsen et al. 1996, 2000; Lewis 2002; Loebel et al. 1992; Norman and Malla 2001; Quinn et al. 2000; Scully et al. 1997; Wyatt 1991; Bottlender and Möller 2003); however, others have not found an association between DUP and outcome (Craig et al. 2000; Ho and Andreasen 2001). This may have been due to selection of samples with unusually poor outcome or who received poor quality care (Bottlender and Möller 2003). Loebel et al. (1992), in a prospective study, found that a longer duration of untreated psychosis (DUP) was significantly associated with longer time to remission, and poorer levels of remission and outcome. Prolonged DUP has also been linked to an increased risk of relapse (Crow et al. 1986) and more severe positive and negative symptoms (Haas et al. 1998). Other studies of clients with a first onset of psychosis have found significant correlations between shorter DUP and better outcome (Crow et al. 1986; Rabiner et al. 1986). It has, therefore, been suggested that acute psychotic symptoms may reflect an active morbid process, which, if not influenced by early treatment, could result in more lasting morbidity (McGlashan 1999). In addition to any putative biological entrenchment or toxicity of untreated disorder, delays in treatment inevitably lead to a number of psychosocial complications, which could increase mortality and contribute to impaired recovery and future outcomes (Bromet et al. 1996; Edwards et al. 1994; Humphreys et al. 1992; Johnstone et al. 1986; Lincoln et al. 1998). Although several studies have established a link between DUP and unfavourable outcome, the exact nature of this relationship is still poorly understood. Does DUP act as an independent variable in the outcome of psychosis or is it confounded by social factors and/or inherent illness characteristics such as gender and rate of onset of psychosis? (Norman and Malla 2001). Verdoux et al. (1998) found that DUP was independently predicted by the sociodemographic and clinical confounding variables of gender, low education level, poor global functioning, family history of psychiatric illness and more severe overall clinical state. A more recent paper,

using a large sample size of 354 first-episode patients found that DUP predicted outcome even after adjusting for the effects of these types of variables (Harrigan et al. 2003). While it is possible that the association of ‘intrinsic’ prognostic variables may mean that DUP is a (partially) confounded ‘epiphenomenon’ in the prognostic process, it could also be that these variables influence prognosis by producing delays in treatment, e. g. male gender, low education level, etc. (McGorry 2000). This would mean that DUP would still be a valid target for intervention. If we believe DUP is a potentially malleable prognostic variable, what research methodologies are appropriate to explore the core questions? McGlashan (1998) argues that an ‘ideal’ study design would be a randomised control study, whereby individuals would be randomised into early versus delayed (neuroleptic) treatment. However ideal this design may be, McGlashan has noted that it poses an ethical dilemma and has suggested that the timing of treatment can only be manipulated experimentally by focusing on differential detection strategies. He proposed a quasi-experimental design consisting of two populations, one of which is exposed to an early detection strategy and one which is not (McGlashan 1996).A large-scale study using this design is in progress at three sites in Norway and Denmark (Johannessen et al. 2001). The early data from this study (TIPS) suggest that it is possible to reduce DUP through a combination of public education and mobile detection strategies (Larsen et al. 2001). The methodology is, however, more complex than originally realised. For a fuller discussion of the methodological limitations of this design, see McGorry (2000) and Friis et al. (1999). This paper describes a pilot project with a modest budget, utilising the quasi-experimental research design outlined above, carried out at the Early Psychosis Prevention and Intervention Centre (EPPIC), in Melbourne, Australia. This project attempted to further reduce DUP amongst young people, presenting with a first-episode psychotic illness. Similar to the TIPS Study, the major aims of the pilot study were: (a) to develop, implement and evaluate a comprehensive mental health literacy strategy that would promote the earlier recognition and treatment of first-episode psychosis within a defined geographic region with the aim of reducing DUP, and (b) to examine the impact of this DUP reduction (if achieved) upon clients’ recovery rates and levels (McGlashan 1996).

Subjects and methods To achieve a significant and widespread reduction in the DUP, a range of strategies need to be used in combination over a sufficient period of time. Increasing awareness, improving knowledge, changing attitudes, promoting skills in recognition, coupled with organisational changes to facilitate referral and assessment are all required in order to reduce delays. One segment of the EPPIC catchment area was chosen as the focus for a range of activities aimed at promoting earlier intervention

713 for young people with first-episode psychosis. The campaign lasted for 12 months. A second, demographically similar segment of the EPPIC catchment area was chosen as a non-intervention, comparison sector. The two sectors were of a similar size (population of 300,000 each) and were demographically comparable on gender, average personal incomes, family income, education, and employment characteristics. In addition to awareness campaigns, an essential aspect required for reducing delays is easy access to specialist support. A specialist multidisciplinary mobile early detection team (YAT) conducted all intake diagnostic assessments for entry into EPPIC either as an inpatient or on an outreach basis, usually in the individual’s own home. Specific objectives of the YAT team included early detection and the team operated across both the comparison and intervention areas. Any young person experiencing a significant and sustained change in their behaviour or personality was deemed to warrant assessment. The EPPIC service adopted a ‘low-threshold’ for willingness to assess young people with a broad range of presenting complaints given that symptoms associated with early psychosis overlap with a number of other mental health disorders, particularly in the early stage, and comorbidity is often present. A significant advantage of the EPPIC catchment area in terms of the methodology is that few, if any, patients eligible for treatment at EPPIC were receiving treatment at alternative clinics; thus, the sample is a true epidemiologically based study group. The reason for this is that there are very few private psychiatrists available in the catchment area, all of whom are known to rapidly refer new cases to the EPPIC service due to the comprehensive services offered at EPPIC, the close working relationship and their already high caseloads. Also, little, if any, leakage occurs to private facilities outside of the catchment area. ■ Intervention The sector chosen for the intervention activities is located in the North-West Melbourne metropolitan region and has a population of 300.000. This area is characterised by high socioeconomic disadvantage and ethnic diversity. The intervention consisted of a number of strategies, operating in parallel. Strategies were based on a variety of theoretical approaches drawn from community development theories and from the field of health promotion (Green and Kreuter 1991; Regier et al. 1988). Given the complex and multifactorial nature of delays in firstepisode psychosis, our approach was to focus on three key factors. These were: (a) promoting early help-seeking by symptomatic individuals and/or their family, (b) improving recognition of psychotic disorders and improving referral by primary care practitioners (e. g. general practitioners, school counsellors, youth workers, etc.), and (c) improving access to specialist psychiatric services and increasing willingness to access potential cases.As the EPPIC programme served both areas, this latter aspect of the intervention was common to both sectors and has been described above. The key target groups of the project were young people aged 16–30, parents with a child in this age range and primary care service providers such as general practitioners and youth workers. Help-seeking Due to budgetary restrictions, the level of intervention in this domain was restricted in reach. Activities focused on raising awareness about psychosis, increasing people’s knowledge about these conditions, in particular their typical presenting symptoms, changing attitudes that may inhibit help-seeking, as well as increasing people’s knowledge about services that the young person or family could contact for knowledge and assistance. These particular activities were designed around information obtained through a series of focus groups conducted before the project commenced to ascertain people’s knowledge, attitudes and beliefs in relation to help-seeking for mental illness. The majority of these activities were undertaken with secondary schools. A total of 12 sessions across six schools were undertaken.

Increased recognition of psychotic disorders The second major focus was on activities aimed at improving skills, particularly in regard to recognition of early illness and enhancing channels for referral and assessment. Major recipients included GPs, schoolteachers and generalist youth workers. As an example, activities directed at general practitioners (GPs) included:  The establishment of an EPPIC/GP/Consumer reference group to improve links between GPs and EPPIC.  Visits to private practices of these GPs to provide information about EPPIC and basic information on psychosis.  Education workshops on early psychosis (three workshops with total attendance of approximately 90 GPs).  Advertising in the local GP newsletter (three advertisements in the one year).  GPs were the target of a mail-out of information about EPPIC and psychosis (twice in the one year).  Production and dissemination of the Early Psychosis Video targeted at GPs and sent to all 300 GPs in the intervention sector. Similar intensity of intervention occurred with school welfare workers, youth workers/counsellors, and other workers in day-to-day contact with young people. The intervention campaign commenced on 1st June 1996 and continued through to 31st May 1997. Recruitment of participants into the study commenced 2 months later, allowing an opportunity for the intervention to have made an impact on the community and continued for 12 months (8th August 1996–21st July 1997). ■ Participants The study was subject to ethics approval from the Royal Park Research and Ethics Committee and all participants gave written consent. Inclusion criteria required that all clients be aged 16–30, experiencing a first psychotic episode, have no prior history of treatment for a psychotic illness, be fluent in English, and have stable accommodation within the EPPIC catchment area. Clients with intellectual disability or a clear organic cause for psychosis were excluded from the study. Given that the study aimed to assess the impact of a community intervention, itinerant clients were also excluded from the study. A total of 98 consecutive-admission clients were recruited into the study, 58 from the comparison and 40 from the intervention region.A further 22 eligible clients refused to participate in the research either at intake (time 1) or recovery/stabilisation (time 2; within 3 months of service entry). Limited de-identified sociodemographic and clinical data on this group were made available to the research team, following approval given by the Ethics committee. Individuals who did not complete the time 2 assessment were excluded from the study as it is at this point that the final diagnosis and a substantial amount of the assessment is completed. The 98 participants were representative of first-episode clients admitted to EPPIC. There were no significant differences between the refusers and participants on any of the selected demographic variables (including age, gender, marital status, level of education and country of birth) and both groups were diagnostically similar. The mean DUP for the participant group was 278.6 days (SD = 459.4; median = 96.5 days) and the refuser group was 224.5 days (SD = 467.3; median = 68.0 days). An independent samples’ t-test on the log-transformed data revealed no significant differences between the refusers and participants on DUP (p = 0.90). The comparability of the intervention and comparison groups was examined on a range of sociodemographic, diagnostic and clinical variables (Table 1). There were no statistically significant differences between the groups on any of the sociodemographic, diagnostic, and clinical variables, with the exception that the comparison group scored significantly higher on the measure of baseline symptomatology (BPRS) than the intervention group. The time to remission for positive symptoms was also nearly 3 weeks higher in the comparison group. This may have been due to minor differences in the diagnostic mix or initial severity of illness between the two groups. In all other respects, however, the two groups were very similar.

714 Table 1 Demographic and illness-related variables of intervention and comparison samples

Comparison (n = 58)

Intervention (n = 40)

Age Mean (SD)

22.3 (3.4)

22.1 (3.4)

22.2 (3.4)

Gender % male % female

70.7 29.3

67.5 32.5

69.4 30.6

Marital status % never married % married

86.2 13.8

75.0 25.0

81.6 18.4

Education % secondary school only % post-secondary education

74.1 25.9

70.0 30.0

72.4 27.6

Country of birth % born in Australia % born overseas

74.1 25.9

77.5 22.5

75.5 24.5

Maximum daily dose of neuroleptics Mean (SD) Age of onset Mean (SD)

21.6 (3.5)

Duration of psychotic symptoms post treatment (days)a Mean (SD)

a

208.8 (140.4) median = 150.0

192.5 (172.9) median = 150.0 21.2 (3.1)

Total (n = 98)

202.2 (153.8) median = 150.0 21.5 (3.3)

68.8 (46.0) median = 65.5

50.6 (37.2) median = 47.5

61.4 (43.4) median = 61.0

DSM-IV diagnoses Schizophrenia Schizophreniform Schizoaffective disorder Delusional disorder Bipolar disorder Depressive disorder Psychotic disorder NOS

29 (50 %) 4 (6.9 %) 7 (12.1 %) 0 5 (8.6 %) 7 (12.1 %) 6 (10.3 %)

15 (37.5 %) 5 (12.5 %) 3 (7.5 %) 1 (2.5 %) 7 (17.5 %) 7 (17.5 %) 2 (5 %)

44 (44.9 %) 9 (9.2 %) 10 (10.2 %) 1 (1 %) 12 (12.2 %) 14 (14.3 %) 8 (8.2 %)

Baseline BPRSa Mean (SD)

67.0 (15.0)

60.9 (12.6)

64.4 (14.3)

Baseline SOFAS Mean (SD)

45.6 (14.7)

46.6 (13.2)

46.0 (14.0)

Baseline SANS Mean (SD)

24.8 (14.1)

21.8 (12.2)

23.6 (13.4)

Intervention and comparison groups are significantly different at p ≤0.05

■ Statistical Analyses Differences between the comparison and intervention groups were assessed using a range of parametric and non-parametric analyses. Group differences on categorical variables were assessed using Pearson’s chi-square statistic, with exact tests used where necessary. Differences on continuous data were examined using independent samples t-tests, with probability values supported with Cohen’s d effect sizes where appropriate. Where variables were substantially skewed, appropriate mathematical transformations were applied in order to produce a more normal distribution and the resulting distribution carefully evaluated prior to applying t-tests to the transformed data. The non-parametric equivalent of the t-test, the MannWhitney U test, was also applied to the untransformed data where appropriate. ■ Measures All participants who gave written informed consent were assessed with the Royal Park Multidiagnostic Instrument for Psychosis (RPMIP) (McGorry et al. 1990 a, b), which allows a diagnosis to be for-

mulated according to a range of nosological systems, including DSMIV (American Psychiatric Association 1994). A specific illness duration interview, part of the RPMIP procedure, provided a reliable estimate of DUP. Formal interrater reliability has previously been established for the measurement of onset and duration symptoms in the RPMIP, with a mean κ of 0.79 (McGorry et al. 1990b). More details about the measurement of DUP can be found in Harrigan et al. (2003). Premorbid adjustment was measured using the Premorbid Adjustment Scale (PAS) (Cannon-Spoor et al. 1982) and the SoFAS (Goldman et al. 1992) was used as a measure of the individual’s present level of social and occupational functioning. Symptom measures were as follows: the Brief Psychiatric Rating Scale (BPRS) (Overall 1962; Lukoff et al. 1986), a 24-item semi-structured interview which allows an assessment of the degree of psychopathology, both observed (e. g. tension and excitement) and interview-elicited (e. g. unusual thought content); Scale for the Assessment of Negative Symptoms (SANS9 (Andreasen 1983) and the 13-item Beck Depression Inventory (BDI) (Beck et al. 1972).

715 ■ Procedure

Duration of untreated psychosis

The present study was part of a larger 12-month follow-up study. For the present study, clients were assessed on a range of psychopathological instruments at two time points: (i) treatment entry and (ii) recovery/stabilisation (usually completed within 3 months of entry). The diagnostic interview is commenced at treatment entry and completed by the recovery/stabilisation period on average some 11–12 weeks later. This procedure is regarded as more valid than a cross-sectional assessment of diagnosis as it allows the researcher to capture any marked changes in the individual’s clinical picture (McGorry et al. 1990a, b).

Results ■ Illness-related variables The means and standard deviations of the illness-related variables including duration of untreated psychosis and length of prodrome are presented in Table 2. Prodrome is defined as the earliest deviation from a person’s premorbid personality, behaviour or level of functioning prior to the onset of psychotic symptoms. The duration of prodrome is the length of time between the onset of prodrome and the onset of psychotic symptoms (see McGorry et al. 1990a, b for further details). Duration of untreated psychosis is defined as the length of time between the onset of psychosis and the commencement of treatment.

Duration of prodrome The duration of prodrome was highly positively skewed, as can be seen in Table 2. The mean prodromal duration for the comparison group was approximately 1 year and 3 months, but the median was only around 7 months; for the intervention group, prodromal duration was higher, with a mean of approximately 1 year and 8 months and a median of around 1 year. Because the means are so highly skewed by a relatively small number of outlying cases, the median is the most appropriate measure of central tendency. Despite transformation, this variable remained non-normally distributed, hence a MannWhitney U test was performed, the result of which indicated that the intervention and comparison groups were not significantly different (p = 0.107).

The distribution of the DUP data for the intervention and comparison groups was also highly skewed, as can be seen in Fig. 1 and Table 2. The mean DUP for the intervention group was just over 10 months, with a median of only 2 months (59 days), compared with a mean DUP of around 8 months for the comparison group, with a median of just over 3 months (104.5 days). A closer examination of Fig. 1 reveals that the DUP for the two groups had markedly different distributions. It appears that the distribution of DUP for the intervention group is weighted at the lower and higher end of the scatterplot, with fewer cases in the middle. Conversely, the distribution of the DUP for the comparison group is concentrated in the middle and lower region, with fewer outliers. Despite the descriptive differences apparent in the group distributions, there was no significant difference between the groups on log-transformed DUP (t = 0.589; df = 96; p = 0.557). This difference corresponded to a very small effect size (d = 0.12) as defined by Cohen (1992). A Mann-Whitney U test also revealed that there were no significant differences on the mean DUP rankings between the groups (Z = –0.784; p = 0.433). Nor were any differences apparent when the data were simplified according to the method described by Harrigan et al. (2003) in which the data were grouped into the following categories; 0–7 days; 8 days–4 weeks; over 4 weeks up to 3 months; over 3 months up to 1 year; over 1 year (exact p = 0.898). However, these methods were not sufficiently sensitive to detect the group distributional differences at the higher end (> 1 year) of the DUP spectrum. Therefore, the data were further simplified by

Table 2 Duration of untreated psychosis (DUP) and prodrome means (SDs) and medians for the intervention and comparison groups Comparison (N = 58)

Intervention (N = 40)

Total (N = 98)

DUP (days) Mean (SD) Median

254.4 (379.7) 104.5

313.8 (558.6) 59.0

278.6 (459.4) 96.5

Prodrome (days) Mean (SD) Median

460.7 (616.7) 207.5

612.0 (664.5) 355.0

522.5 (637.7) 258.5 Fig. 1 Distribution of DUP in control and experimental groups

716

dividing DUP into three categories: short DUP up to 1 year; long DUP > 1 year up to 3 years; very long DUP > 3 years. Table 3 indicates that there was a statistically significant association between region and DUP category (exact p = 0.005). The adjusted standardised residuals were examined and these indicated that the intervention group had more cases with a DUP of more than 3 years than would be expected by chance (15 %; + 2.1 asresid.) and fewer comparison cases (3.4 %; –2.1 asresid.) in this very high DUP category. Conversely, there were fewer cases in the intervention group with a DUP of 1–3 years (2.5 %; –2.6 asresid.) than expected and more cases in the comparison group (20.7 %; + 2.6 asresid.). The intervention and comparison groups appeared to have reasonably similar proportions of subjects in the DUP category of 1 year or less (intervention group: 82.5 % vs. comparison group: 75.9 %; ±0.8 asresid.). Duration of prodrome (PRO) was also divided into the same categories (see Table 3); however, there were no significant differences between groups (chi-square = 2.46; df = 2; p = 0.292). In a further strategy, we removed nine cases (9 % of the sample) with DUP of greater than 1,000 days, on the basis that the presence of a very few extreme values has the potential to exert excessive influence on the results, potentially presenting a distorted picture for the majority of cases. With these nine cases removed (three from the comparison group and six from the intervention group), the mean and median DUP for the intervention group was approximately half that of the comparison group, with a mean DUP for the intervention group of 91.7 days (SD = 113.3; median = 44 days) compared with a mean of 189.7 days for the comparison group (SD = 232.5; median = 100 days). An independent samples t-test on the log-transformed data gave a statistically borderline result (t = 1.76; df = 87; p = 0.082), with a small-to-moderate effect size of d = 0.39. A Mann-Whitney U test gave a similar result, bordering on statistical significance (Z = –1.896; p = 0.058). Finally, because the campaign operated within a discrete time period, a clearer picture of its direct impact might be seen in those who developed psychosis following the commencement of the intervention on 1st June 1996. The mean DUP of 54.5 days for the intervention sample (n = 28; SD = 74.3; median = 22 days) was shorter than the mean DUP for the comparison group of 79.7 days (n = 39; SD = 94.3; median = 31 days); however, independent samples’ t-tests on the log-transformed data Table 3 Distribution of duration of untreated psychosis (DUP) and duration of prodrome (PRO) for the intervention and comparison groups Comparison (n = 58)

Intervention (n = 40)

DUP

< 1 year 1–3 years > 3 years

75.9% 20.7% 3.4%

82.5% 2.5% 15.0%

PRO

< 1 year 1–3 years > 3 years

65.5% 19.0% 15.5%

50.0% 25.0% 25.0%

showed that this difference was not statistically significant (t = 0.937; df = 65; p = 0.352). The difference between the means of the log-transformed data equates to a small effect size (d = 0.23). A Mann-Whitney U test also revealed that there were no significant differences on the mean DUP rankings between the groups (Z = –1.13; p = 0.258).

■ Rate of first-presentation cases We tried to ascertain whether the campaign had an effect on the rate of treated first-presentation cases between the comparison and intervention groups by assessing the percentage change in rates of such cases in both regions for the 2 years prior to the intervention and the 2 years post intervention. However, the rate of treated first-presentation cases was difficult to measure with a high degree of accuracy, hence the findings should be interpreted with caution. Firstly, it appeared that irrespective of sector, the rate of first-presentation cases appeared to increase steadily by approximately 30 %, probably reflecting the intensive detection efforts by the EPPIC mobile community assessment team, coupled with firmer links being established with the primary care sector. Secondly, there appeared to be a suggestion that the rate of first-presentation cases may have increased more in the intervention region than in the comparison region.

Discussion The results of this pilot study reveal unexpected complexity that is probably at least partly due to the impact of efforts to reduce treatment delay in first-episode psychosis. For instance, there were significantly more cases in the intervention sector with a very long DUP (more than 3 years) compared with the comparison sector, which was an unexpected finding. This finding was consistent with the higher mean DUP of around 10 months found for the intervention group, and lower mean of 8 months for the comparison group, though this difference was not statistically significant. Conversely, the median DUP for the intervention group was lower (59 days) than the comparison group (104.5 days). Hence, it appears that, whilst the intervention changed the distribution of DUP, it did not appear to impact substantially on the reduction of DUP.When these complex results are considered along with the relatively low statistical power of the study to detect any but moderate-to-large effects, the impact of our campaign appears inconclusive. The findings of the present study highlight the complexity of McGlashan’s model (McGlashan 1996) for community intervention and suggest that enhanced detection actually might have two effects, not one. Our results seem to suggest that the additional efforts to reduce DUP were ineffective. Interestingly, a group of very long DUP cases, not seen in the comparison sector, were

717

detected in the intervention sector. To make sense of this unexpected finding, we consider that the impact of increased community education and more responsive early detection systems for incident cases are likely to be twofold. Firstly, there may be a reduction of treatment delays and DUP as expected, supported by the present results which illustrate that, when the outliers are removed, the average delay in treatment in the intervention sample appears to have been reduced. Secondly, individuals with protracted delays who may not otherwise have been identified may be detected and treated. This is supported by our finding that disproportionately more cases of very long DUP were found in the intervention sample.We assume that there is a finite level of untreated 1-year incidence of psychosis in any area, and this will be made up of two components; firstly, those who eventually do enter treatment (with a DUP > 1 year),and, secondly, those who will never receive treatment. The latter group (which constitutes the relatively inaccessible extreme of the total prevalence curve) has hitherto not been considered in methodological reviews (McGlashan 1996, 1999). Helgason (1990) has indicated in a special population, in Iceland, that substantial percentages of never-treated individuals can be undetected for prolonged periods of time. Furthermore, evidence is emerging from other sources (van Os et al. 2001) that psychotic symptoms have a wider distribution in the community than previously thought, so the boundaries of what constitutes “illness” are increasingly unclear even in psychosis, and certainly in other disorders (Regier 1998). In any case, with enhanced education and detection efforts, there may be an increase in the rates of first-presentation psychosis; that is, new cases which would never have been detected under prior conditions or “detection as usual” will be found (McGlashan 1996). Our attempt at examining these rates of first-presentation cases suggested a small increase in rate of referrals for the intervention sector over the comparison sector. Although this difference is marginal, it suggests that the rates of first-presentation cases in the intervention sector may have increased somewhat more than in the comparison sector. Regardless of sector, however, the apparent increase in referrals over the period 1996–1998 may be attributable to the greater access provided via YAT’s detection strategies, which operates in both sectors, or could also be due to higher rates coming into treatment due to recent increases in migrant groups (Harrison et al. 1997; Selten et al. 2001) or increasing levels of urbanisation in the area (Allardyce et al. 2001; Pedersen et al. 2001; van Os et al. 2001). The DUP of these new cases may be different from the detection-as-usual sample, either shorter (perhaps including some self-limiting brief psychoses) or longer. If the latter predominate, this will spuriously suggest that the DUP has not been reduced, but actually lengthened. This will be true even if there is an excess of cases with a mean DUP only moderately greater than the previous mean, and does not necessarily require many extremely

long DUP cases. In other words, it could be that greater efforts to reduce DUP may appear unsuccessful if the rates of first-presentation cases increase through the detection of long-neglected additional cases with long DUP. Some of these trends have been noted elsewhere (Malla and Norman 2002). The inherent severity of illness of such extra cases, if they existed, could also adversely influence outcomes for the sample. This issue has surfaced as a problem for TIPS research group (Friis et al. 1999). Conversely, if the extra cases already have a shorter DUP, this could lead to a false conclusion that the detection efforts have reduced the DUP for the original kind of case detected under usual conditions. The detection of particularly brief DUP cases is also likely to improve outcome, and wrongly lead to the conclusion that reducing DUP improves outcome. The rate of first-presentation cases therefore needs to be carefully measured on an annual basis to examine all of these possibilities, and to be clear whether changes in DUP are due to changes in the sample or true changes in DUP. Further interpretation problems may result if there is differential recruitment from the samples of first-presentation cases into research samples in intervention and comparison sectors in studies with this quasi-experimental design (McGorry 2000). Fortunately, in the present study, an analysis of the DUP of the research sample was not significantly different from the group that refused participation in research, though this may vary in other centres (Friis et al. 2001). Thus, there are two possible confounds of increasing rates of first-presentation cases. Firstly, there may be a change in DUP mix and it may prove difficult to partial out effects (though one could partition the sample and merely select those whose psychotic symptoms onset after the intervention was implemented). Secondly, a better-prognosis (or worse-prognosis) group may be recruited either via different DUP levels or other intrinsic severity features. These could be possible explanations for the differential outcome characteristics in developing countries versus developed countries in WHO studies (Eaton et al. 1998), as it is possible that better casefinding in developed countries results in poorer prognosis for those samples. However, it may also be the case that in developing countries with poorer resources only the more severely disabled will be detected and treated. The results of this study are not clear-cut and the sample size remains too small to provide ultimate clarification of these issues, though the sample itself appears to be a relatively representative first-episode psychosis group. Further limitations relate to the power of the intervention. It is difficult to calculate and garner the financial and social resources that would be needed to achieve widespread impact on the community.Although the campaign was relatively weak with limited resources, it did seem to show some effect, though primarily with respect to the distribution of DUP, rather than reduction of DUP. Secondly, we need to consider

718

whether we allowed enough ‘lag’ time for the intervention to seep into the community. It should be acknowledged that a large randomised controlled trial where individuals are randomised to delayed treatment versus immediate treatment is the ideal study design that would perhaps settle the DUP debate. Although the groups appeared relatively similar on selected demographic and clinical characteristics, other unmeasured baseline differences between the two groups may have been present. Secondly, in a community campaign of this nature, with the two geographical regions being so close in proximity, it is possible that those in the comparison group were exposed to the campaign. However, given the design of this targeted campaign, this would have been a minor factor, the leakage was a possibility, but unlikely to have been significant. Furthermore, the assessors were not blind to group membership and the possibility of confounding the study findings is, therefore, a further potential limitation.

Conclusion Although the limitations are manifold, it should be acknowledged that this is one of the first community campaigns of its kind to attempt to systematically reduce the duration of untreated psychosis. This pilot project allowed us to explore the magnitude of intervention (both in monetary terms and breadth of community campaigns) that is necessary to influence a complex and multifactorial phenomenon such as duration of untreated psychosis. Yet, despite the sense that the intervention was underpowered, the differential distribution of DUP across the two regions provided some intriguing clues for future research to explore further. The results from this study, although preliminary and on a small scale, suggest it may indeed be possible to manipulate the duration of untreated psychosis, therefore further enhancing currently prescribed early intervention treatment strategies. The key finding is that the impact of intervention is complex and reveals potential parallels with general medicine where earlier case detection methods may influence the prognostic mix in quite complex ways. Finally, though the RCT has been seen as ethically challenging, the complexity of the issue of DUP reduction and the limitations of the quasi-experimental design exposed by this study and by the early TIPS experiences (Friis et al. 2001; McGorry 2000) indicate it may be worth another look. This proposal is all the more compelling in the light of reductions in DUP in many centres and the likelihood that new types of patients and a different phase of illness is in fact being treated and the recent identification of the group with non-clinically significant psychosis (van Os et al. 2001). It is worth reconsidering whether there may be a safe and ethical way to do an RCT of early versus delayed antipsychotic treatment.

■ Acknowledgments The authors would like to acknowledge the young people who participated in this study, Prof. Thomas McGlashan and Dr. Jan Olav Johannessen for their ideas and support, Birsen Fitch for helping to conduct this study, and the Victorian Health Promotion Foundation for funding this project.

References 1. Allardyce J, Boydell J,Van Os J, Morrison G, Castle D, Murray RM, McCreadie RG (2001) Comparison of the incidence of schizophrenia in rural Dumfries and Galloway and urban Camberwell. Brit J Psychiatry 179:335–339 2. American Psychiatric Association (1994) Diagnostic and Statistical Manual of Mental Disorders. 4th Edition, APA, Washington D.C. 3. Andreasen N (1983) Schedule for the Assessment of Negative Symptoms. University of Iowa, Iowa City 4. Beck AT, Beck RW (1972) Screening depressed patients in family practice. A rapid technique. Postgrad Med 52:81–85 5. Beiser MD, Erickson D, Fleming JA, Iacono WG (1993) Establishing the onset of psychotic illness.Am J Psychiatry 150:1349–1354 6. Bottlender R, Moller HJ (2003) The impact of the duration of untreated psychosis on short- and long-term outcome in schizophrenia. Curr Opin Psychiatry 16(Suppl 2):S39–S43 7. Bromet EJ, Jandorf L, Fennig S, Lavelle J, Kovasznay B, Ram R, Tanenberg-Karant M, Craig T (1996) The Suffolk County Mental Health Project: demographic, pre-morbid and clinical correlates of 6-month outcome. Psychol Med 26:953–962 8. Cannon-Spoor HE, Potkin SG, Wyatt RJ (1982) Measurement of premorbid adjustment in chronic schizophrenia. Schizophr Bull 8:470–484 9. Cohen, J (1992) A power primer. Psychol Bull 112:155–159 10. Craig TE, Bromet EJ, Fennig S, Tanenberg-Karant M, Lavelle J, Galambos N (2000) Is there an association between duration of untreated psychosis and 24-month clinical outcome in a first-admission series? Am J Psychiatry 157:60–66 11. Crow T, MacMillan JF, Johnson AL, Johnstone EC (1986) A randomized controlled trial of prophylactic neuroleptic treatment. Brit J Psychiatry 148:120–127 12. Drake RJ, Haley CJ, Akhtar S, Lewis SW (2000) Causes and consequences of duration of untreated psychosis in schizophrenia. Brit J Psychiatry 177:511–515 13. Eaton WW, Thara R, Federman E, Tien A (1998) Remission and relapse in schizophrenia: The Madras longitudinal study. J Nerv Ment Dis 186:357–363 14. Edwards J, Francey SM, McGorry PD, Jackson HJ (1994) Early psychosis prevention and intervention: Evolution of a comprehensive community-based specialised service. Behaviour Change 11:223–233 15. Friis S, Melle IS, Larsen TK, Johannesen JO Simonsen E, Vaglum PJW, McGlashan TH (2001) Does duration of untreated psychosis (DUP) bias schizophrenia study samples? Schizophr Res 49(Suppl 1–2):260 16. Friis S, Melle I, Haahr U, Larsen TK, Johannessen JO, Simonsen E, Vaglum PJW, McGlashan T (1999) Threats to validity in outcome studies of early intervention in schizophrenia. Schizophr Res 36:340 17. Goldman HH, Skodol AE, Lave TR (1992) Revising axis V for DSM-IV: A review of measures of social functioning. Am J Psychiatry 149:1148–1156 18. Green LW, Kreuter MW (1991) Health Promotion Planning: An Educational and Environmental Approach. Second ed, Mountain View, Mayfield 19. Haas G, Garratt LS, Sweeney JA (1998) Delay to antipsychotic medication in schizophrenia: impact on symptomatology and clinical course of illness. J Psychiatric Res 32:151–159 20. Harrigan S, McGorry PD, Krstev H (2003) Does treatment delay in first-episode psychosis really matter? Psychol Med 33:97–110

719 21. Harrison G, Glazebrook C, Brewin J, Cantwell R, Dalkin T, Fox R, Jones P, Medley I (1997) Increased incidence of psychotic disorders in African Caribbean migrants to the United Kingdom. Psychol Med 27:799–806 22. Helgason L (1990) Twenty years’ follow-up of first-presentation for schizophrenia: What could have been prevented? Acta Psychiatr Scand 81:231–235 23. Ho BC,Andreasen NC, Flaum M, Nopoulos P, Miller D (2000) Untreated initial psychosis: its relation to quality of life and symptom remission in first-episode schizophrenia. Am J Psychiatry 157:808–815 24. Humphreys MS, Johnstone EC, MacMillan JF, Taylor PJ (1992) Dangerous behaviour preceding first admissions for schizophrenia (comment). Brit J Psychiatry 161:501–505 25. Johannessen JO, McGlashan TH, Larsen TK, Horneland M, Joa I, Mardal S, Kvebaek R, Friis S, Melle I, Opjordsmoen S, Simonsen E, Ulrik H, Vaglum P (2001) Early detection strategies for untreated first-episode psychosis. Schizophr Res 51:39–46 26. Johnstone EC, Crow TJ, Johnson AL, MacMillan JF (1986) The Northwick Park study of first-episodes of schizophrenia. I. Presentation of the illness and problems relating to admission. Brit J Psychiatry 148:115–120 27. Larsen TK, McGlashan TH, Johannessen JO, Friis S, Guldberg C, Haahr U, Horneland M, Melle I, Moe LC, Opjordsmoen S, Simonsen E,Vaglum P (2001) Shortened duration of untreated first episode of psychosis: changes in patient characteristics at treatment. Am J Psychiatry 158:1917–1919 28. Larsen T, Moe LC, Vibe-Hansen L, Johannessen JO (2000) Premorbid functioning versus duration of untreated psychosis in 1 year outcome in first-episode psychosis. Schizophr Res 45:1–9 29. Larsen T, McGlashan TH, Moe LC (1996) First-episode schizophrenia: I Early course parameters. Schizophr Bull 22:241–256 30. Lewis S (2002) Integrating individuals psychosocial treatments in early psychosis. Acta Psychiatr Scand 106(Suppl 414):S32 31. Lieberman J, Fenton W (2000) Delayed detection of psychosis: Causes, consequences, and effect on public health. Am J Psychiatry 157:1727–1730 32. Lincoln CV, McGorry PD (1995) Who cares? Pathways to psychiatric care for young people experiencing a first episode of psychosis. Psychiatr Serv 46:1166–1171 33. Lincoln CV, Harrigan S, McGorry PD (1998) Understanding the topography of the early psychosis pathways. Brit J Psychiatry 172(Suppl 33):21–25 34. Loebel A, Lieberman JA, Alvir JM, Mayerhoff DI, Geisler SH, Szymanski SR (1992) Duration of psychosis and outcome in firstepisode schizophrenia. Am J Psychiatry 149:1183–1188 35. Lukoff D, Neuchterlein KH, Ventura J (1986) Manual for the Expanded Brief Psychiatric Rating Scale.Schizophr Bull 12:594–602 36. Malla AK, Norman RMG (2002) Early intervention in schizophrenia and related disorders: advantages and pitfalls. Curr Opin Psychiatry 15:17–23 37. McGlashan T (1998) Early detection and intervention of schizophrenia: rationale and research. Brit J Psychiatry 172(Suppl 33):3–6 38. McGlashan T (1999) Duration of untreated psychosis in firstepisode schizophrenia: Marker or determinant of course? Biol Psychiatry 46:899–907 39. McGlashan T (1996) Early detection and intervention with schizophrenia: Research. Schizophr Bull 22:327–345

40. McGorry PD (1995) A treatment-relevant classification of psychotic disorders. Aust NZJ Psychiatry 29:555–558 41. McGorry PD (2000) Evaluating the importance of reducing the duration of untreated psychosis. Aust NZJ Psychiatry 34(Suppl): 145–149 42. McGorry PD, Copolov D, Singh B (1990a) Royal Park Multidiagnostic Instrument for Psychosis: Part 1. Rationale and review. Schizophr Bull 16:501–515 43. McGorry PD, Singh BS, Copolov DL, Kaplan I, Dossetor CR, van Riel RJ (1990b) The Royal Park Multidiagnostic Instrument for Psychosis: Part II. Development, reliability and validity. Schizophr Bull 16:517–536 44. McGorry PD, Copolov DL, Singh BS (1989) The validity of the assessment of psychopathology in the psychoses. Aust NZJ Psychiatry 23:469–82 45. Norman RM, Malla AK (2001) Duration of untreated psychosis: A critical examination of the concept and its importance. Psychol Med 31:381–400 46. Overall JE,Gorham DR (1962) The Brief Psychiatric Rating Scale. Psychol Rep 10:799–102 47. Pedersen CB, Mortensen, PB (2001) Family history, place and season of birth as risk factors for schizophrenia in Denmark: a replication and analysis. Brit J Psychiatry 179:46–52 48. Quinn J, Moran M, Lane A, Kinsella A, Waddington J (2000) Long-term adaptive life functioning in relation to initiation of treatment with antipsychotics over the lifetime trajectory of schizophrenia. Biol Psych 48:163–166 49. Rabiner CJ, Wegner JT, Kane JM (1986) Outcome study of firstepisode psychosis. I: Relapse rates after 1 year. Am J Psychiatry 143:1155–1158 50. Regier D (1988) The NIMH depression awareness recognition and treatment program: Structure, aims and scientific basis. JAMA 145:1351–1357 51. Scully P, Coakley G, Kinsella A, Waddington J (1997) Psychopathology, executive (frontal) and general cognitive impairment in relation to duration of initially untreated versus subsequently treated psychosis in chronic schizophrenia. Psychol Med 27:1303–1310 52. Selten JP,Veen N, Feller W, Blom JD, Schols D, Camoenie W, Oolders J, van der Velden M, Hoek HW, Rivero VM, van der Graaf Y, Kahn R (2001) Incidence of psychotic disorders in immigrant groups to The Netherlands. Brit J Psychiatry 178:367–372 53. van Os J, Hanssen M, Bijl RV, Vollebergh W (2001) Prevalence of psychotic disorder and community level of psychotic symptoms: An urban-rural comparison. Arch Gen Psychiatry 58:663–668 54. Verdoux H, Liraud F, Bergey C,Assens F,Abalan F, van Os J (2001) Is the association between duration of untreated psychosis and outcome confounded? A two year follow-up study of first-admitted patients. Schizophr Res 49:231–241 55. Verdoux H, Bergey C, Assens F, Abalan F, Gonzales B, Pauillac P, Fournet O, Liraud F, Beaussier J, Gaussares C, Etchegaray B, Bourgeois M, van Os J (1998) Prediction of duration of psychosis before first admission. Eur Psychiatry 13:346–352 56. Wyatt R (1991) Neuroleptics and the natural course of schizophrenia. Schizophr Bull 17:325–351