Review of EPPIC research

Posted · Add Comment

The EPPIC study: methodology, results, and misrepresentation (Dr Melissa Raven, Psychiatric Epidemiologist)

Summary

The EPPIC study was a small, methodologically weak study that compared initial EPPIC treatment with initial treatment in an earlier specialised inpatient early psychosis intervention program in Melbourne, commencing in 1989. It provides moderately strong evidence that 12 months of EPPIC treatment was more effective in terms of some outcomes (but not some other important ones), and more cost-effective in terms of some outcomes, than several months of treatment in its immediate precursor specialised inpatient early intervention program (pre-EPPIC), followed by the balance of the 12 months in standard treatment in the public mental health system. It provides moderately strong evidence that 2 years of EPPIC treatment followed by approximately 6 years of standard treatment (if and as required) was more effective in terms of some outcomes (but not some other important ones), and more cost-effective in terms of some outcomes, than several months of pre-EPPIC treatment, followed by the balance of the 8 years in standard treatment (if and as required). It does not demonstrate that EPPIC is superior to late intervention, because patients in both groups received early intervention. It provides no evidence that EPPIC is superior to initial mainstream treatment, because all patients initially received specialist treatment. EPPIC has never been systematically compared with mainstream treatment, late or otherwise. However, the EPPIC study is widely misrepresented as providing very strong evidence that EPPIC is more effective and more cost-effective than standard late intervention. This misrepresentation has been widely propagated. Most problematically, it has been incorporated into government policy documents and has influenced Commonwealth Government funding. In addition to this misrepresentation, and the study’s methodological limitations, the EPPIC study’s relevance to 2011 and beyond is limited by its vintage (initial treatment occurred 18-22 years ago), and its generalisability is limited by the characteristics of the single catchment from which patients were drawn.

The EPPIC study

There is a key longitudinal study about the effectiveness and cost-effectiveness of EPPIC (Early Psychosis Prevention and Intervention Centre) treatment. It was conducted primarily in the 1990s. It has been successfully used to argue for greatly increased Commonwealth Government funding for EPPIC centres, and it is now being used to lobby state and territory governments to match that funding. Other studies of EPPIC have also been undertaken. However, the 1990s longitudinal study is by far the most important and the best known. It is referred to here as ‘the EPPIC study’.

The methodology and results of the EPPIC study have been published in four key journal articles (McGorry et al. 1996; McGorry & Edwards 1998; Mihalopoulos et al. 1999, 2009). The study was small and methodologically weak. It (specifically McGorry et al. 1996) was excluded from the Cochrane review of early psychosis intervention on the grounds that it was non-randomised and it used historical controls1.

The study was conducted at Royal Park Hospital (Parkville, Melbourne, Australia) commencing with patient recruitment in 1989. It was a small matched cohort study with historical controls:

  • 51 pre-EPPIC patients: initially treated at the Aubrey Lewis Unit between 1989 and 1992 (drawn from a larger sample of 200 patients)
  • 51 EPPIC patients: initially treated at the Early Psychosis Prevention and Intervention Centre between March and October 1993

Twelve-month clinical outcomes for the two samples were published in McGorry et al. (1996) and McGorry & Edwards (1998). Twelve-month economic outcomes were published in Mihalopoulos et al. (1999), followed by long-term economic, clinical, and social outcomes (approximately 7-8 years after diagnosis and initial treatment) in Mihalopoulos et al. (2009).

Strengths

    • Initial samples were matched on age, sex (65% male), diagnosis, marital status, and premorbid functioning2.
    • Likely to be representative samples of first-episode psychosis cases within the catchment area because of a lack of alternative services.
    • Long-term follow-up (approximately 7-8 years).

Weaknesses

  • Non-randomised allocation.
  • Historical controls.
  • Small initial samples (51 patients in each group, with some attrition at 12-month follow-up.3)
  • Very small long-term follow-up samples (32 EPPIC patients, 33 pre-EPPIC patients), with incomplete data for some participants.
  • Higher proportion of males in pre-EPPIC long-term follow-up sample (69.7% versus 53.1%)4.
  • Initial samples were not matched on duration of untreated psychosis (DUP), despite the fact that ‘Much better levels and rates of recovery were seen with the shorter duration of untreated psychosis’ in the larger pre-EPPIC sample5. The mean DUP was higher in the pre-EPPIC sample, particularly in the schizophrenia subgroup, because of ‘a major reduction in the number of outliers with extremely long durations of untreated psychosis’ in the EPPIC sample6.
  • Single catchment area reduces generalisability.
  • Age of data: initial treatment and data collection occurred 18-22 years ago.

Results

At 12-month follow-up:

  • EPPIC treatment was cheaper ($16,964) than pre-EPPIC treatment ($24,074)7. This was largely because pre-EPPIC was a dedicated inpatient program8, so pre-EPPIC patients had much higher hospitalisation costs.
  • There was no significant difference on the Brief Psychiatric Rating Scale (BPRS), which reflects severe psychopathology9, or the Scale for the Assessment of Negative Symptoms (SANS)10.
  • There was a significant difference on the Quality of Life Scale (QLS), which was higher for EPPIC patients.11 However, QLS had not been assessed at entry to the study.12
  • Antipsychotic drug doses were significantly lower in the EPPIC group.13 However, this was at least partly a consequence of EPPIC’s ‘low-dose neuroleptic prescribing policy strategy’14, rather than being based on severity of symptoms.

At long-term follow-up, approximately 7-8 years after diagnosis and treatment initiation:

  • 32 EPPIC patients (63%) and 33 pre-EPPIC patients were re-assessed (65%).15
  • For some variables, data were collected for as few as 27 patients in each sample (53% of the initial samples).16
  • More pre-EPPIC patients were males (69.7% versus 53.1%)17, who traditionally have worse outcomes than females.18
  • Annual costs from the beginning of year 2 onwards were significantly lower for EPPIC patients ($3,445) than for pre-EPPIC patients ($9,503) – ‘one-third the cost’.19
  • However, annual costs from the beginning of year 1 to the end of year 7 (the mean follow-up for EPPIC patients) were half, not one third, as much for EPPIC patients ($11,265 versus $5,404)20. This was not reported by Mihalopoulos et al. (2009).
  • There was no significant difference on the Brief Psychiatric Rating Scale (BPRS) or the Scale for the Assessment of Negative Symptoms (SANS)21. However, BPRS positive symptom scores were significantly lower (better) for EPPIC patients.
  • There was no significant difference on the Quality of Life Scale, nor on any of the scales of the World Health Organization Quality of Life-BREF.
  • There was no significant difference in the number of patients on a range of measures of employment and welfare receipt.
  • There was no significant difference in terms of social/vocational recovery, but significantly more EPPIC patients achieved remission based on BPRS criteria and on BPRS + SANS criteria.
  • There was a significant difference in course pattern in the previous two years, with fewer EPPIC patients having an actively psychotic course.
  • There was no significant difference in prominence of negative symptoms over the previous two years.

In summary, outcomes for EPPIC patients were better at both 12 months and at 7-8 years after diagnosis and treatment initiation on some important measures. However, no significant differences were found for many other important measures.

Furthermore, some of the worse outcomes in the pre-EPPIC sample are likely to be attributable to the longer mean duration of untreated psychosis (DUP) and the markedly higher number of patients with extremely long DUPs.

In addition, significant differences found for two key measures on which EPPIC patients fared better – inpatient days and antipsychotic doses – were at least partly artefacts of the differences in pre-EPPIC and EPPIC treatment policies. Twelve-month economic costs were also biased by these treatment policies.

McGorry et al. (1996) EPPIC: An evolving system of early detection and optimal management

The most important readily accessible source of information about the EPPIC study is McGorry et al.’s (1996) long ‘EPPIC’ paper. It was excluded from the Cochrane review of early psychosis interventions because of its weak methodology (non-randomised allocation and historical controls).

It outlines the background to the Early Psychosis Prevention and Intervention Centre, including the establishment of the Aubrey Lewis Clinical Research Unit Centre, the site of the pre-EPPIC early intervention program that immediately preceded the EPPIC program. It describes the larger sample (N = 200) of pre-EPPIC patients that the sample of 51 pre-EPPIC patients was drawn from for the EPPIC study (as historical controls for the 51 EPPIC patients). Then it outlines the study methodology and results.

The discussion that follows briefly acknowledges the limitations of the non-randomised historical controls22. It also acknowledges that pre-EPPIC treatment was different from standard care in mainstream psychiatric services23, and that treatment did not generally occur earlier in the EPPIC program.24

It misleadingly refers to ‘the significant reduction in negative symptoms’ as ‘especially encouraging’25, despite the fact that this significant reduction occurred only when 3-month results for the pre-EPPIC patients were compared with 6-month results for EPPIC patients.26

It also disingenuously discusses the lower hospitalisation rates and lower antipsychotic doses in the EPPIC sample without acknowledging either the fact that pre-EPPIC was a dedicated inpatient program27 or the fact that EPPIC had a ‘low-dose neuroleptic prescribing policy strategy’.28

McGorry & Edwards (1998) The feasibility and effectiveness of early intervention in psychotic disorders: The Australian experience

This paper provides much of the same information about the study as McGorry et al. (1996), but more concisely.

Mihalopoulos et al. (1999) Is phase-specific, community-oriented treatment of early psychosis an economically viable method of improving outcome?

Like McGorry et al. (1996) and McGorry & Edwards (1998), this paper analyses data about the 51 EPPIC and 51 pre-EPPIC patients over the first year of treatment. However, it focuses primarily on economic analysis.

Mihalopoulos et al. (1999) compared the cost of the first 12 months of EPPIC treatment with several months of treatment in the pre-EPPIC control program (a specialised inpatient early intervention program run by McGorry as the immediate precursor to EPPIC)29, followed by the balance of the 12 months in standard treatment in the public mental health system.

The analysis provides moderately strong evidence that 12 months of EPPIC treatment was more effective in terms of some outcomes, and more cost-effective, than several months of pre-EPPIC treatment, followed by the balance of the 12 months in standard treatment. However, there were no significant differences on a number of important outcomes, including severity of psychopathology.30

The average 12-month cost for EPPIC patients was $16,96431, compared with $24,074 for pre-EPPIC patients.32

A major reason for the difference in cost was the fact that pre-EPPIC was a dedicated inpatient program, so patients were automatically hospitalised, for an average of 80 days at a cost of $21,386 (compared with 42 days at a cost of $11,298), regardless of clinical need. McGorry and colleagues acknowledged that some EPPIC patients who were not hospitalised would have been hospitalised in the pre-EPPIC program33. This biased the economic analysis in favour of EPPIC.

Mihalopoulos et al. (2009) Is early intervention in psychosis cost-effective over the long term?

Mihalopoulos et al. (2009) calculated the cost of approximately two years of EPPIC treatment, followed (if and as required) by standard treatment in the public mental health system for up to about 6 years. They compared it with the cost of several months of treatment in the pre-EPPIC control program (a specialised inpatient early intervention program run by McGorry as the immediate precursor to EPPIC), followed (if and as required) by standard treatment in the public mental health system for up to about 7.5 years.

The analysis provides moderately strong evidence that 2 years of EPPIC treatment followed by approximately 6 years of standard treatment was more effective in terms of some outcomes, and more cost-effective, than several months of treatment in pre-EPPIC, followed by the balance of the 8 years in standard treatment. However, there were no significant differences on a number of important outcomes, including severity of psychopathology, quality of life, and social/vocational recovery.34

Harris et al. (2008). Impact of a specialized early psychosis treatment programme on suicide. Retrospective cohort study

This paper reports another study, with a weak study design, which compared patients in the mainstream Victorian mental health system, comparing data from the Victorian Psychiatric Case Register with EPPIC and pre-EPPIC data combined35 (reflecting the fact that pre-EPPIC was an early intervention program).

EPPIC/pre-EPPIC patients had more inpatient treatment days, more community treatment days, and more total treatment days than mainstream patients36, contrary to claims that EPPIC reduces demands on the mental health system.

Furthermore, EPPIC/pre-EPPIC patients had a higher suicide risk after four and a half years; however, they had a lower suicide rate in the first three years37. Overall, there was no significant difference in suicide rates.

Misrepresentations of EPPIC and the EPPIC study

EPPIC is routinely misrepresented as the most evidence-based approach in the mental health field38, despite its exclusion from the Cochrane review. Although there is a substantial body of evidence about EPPIC, the evidence about its effectiveness and cost-effectiveness is very weak. Most importantly, the EPPIC study is widely misrepresented as a study comparing early specialist intervention with late mainstream intervention. Also the one-third cost from years two onwards is misrepresented as applying to the whole period from treatment initiation, exaggerating the cost-effectiveness of EPPIC.

Consequently, it is widely believed that the EPPIC study conclusively demonstrates that EPPIC is highly effective and extremely cost-effective compared with mainstream treatment.

The EPPIC study did not compare EPPIC with late intervention, and it did not compare it with initial mainstream treatment. It compared EPPIC with its immediate precursor treatment program, which also provided early intervention.

However, the descriptions of the pre-EPPIC control intervention and patients have morphed over time, becoming increasingly inaccurate and misleading.

McGorry et al. (1996) accurately reported that EPPIC patients did not generally receive earlier treatment than pre-EPPIC patients:

  • ‘we believe that the improved short-term outcomes we have demonstrated derive largely from more phase-specific and intensive treatment than from earlier provision of treatment, since the latter does not seem to have occurred to a widespread extent apart from a subsample of “outliers.”39

Mihalopoulos et al. (1999) acknowledged the early psychosis focus of the pre-EPPIC program:

  • ‘the Aubrey Lewis Unit (ALU), which was a specialist in-patient research ward of the Royal Park Psychiatric Hospital, with a focus on early psychosis40

Mihalopoulos et al. (2009) acknowledged the early psychosis focus of the pre-EPPIC program in the body of the paper, but misleadingly referred in the abstract (which is much more likely to be read) to initial treatment as generic and standard:

  • ‘The historical controls were sourced from a prior study investigating depression in early psychosis41
  • ‘a specialist inpatient research ward with a focus on early psychosis42
  • ‘a matched cohort of 33 participants initially treated by generic mental health services43
  • ‘Specialised early psychosis programs can deliver a higher recovery rate at one-third the cost of standard public mental health services44

Most importantly, despite his acknowledgement in 1996 that the pre-EPPIC program provided early intervention, McGorry now routinely refers to it as ‘normal late intervention’ or ‘generic late intervention in the standard system’. For example, on ABC Lateline he claimed:

‘We published a paper last year showing that the cost of care over the next eight years after diagnosis dropped to one third for those patients who have received early intervention compared to those who’ve received generic late intervention in the standard system. So actually, it’s very economically irresponsible not to role [sic] this out in Australia.’45

As well as misrepresenting the nature of the pre-EPPIC program, this also misrepresents the cost.

Misrepresentations of EPPIC have been a feature of submissions to governments, and in some cases have been incorporated into government policy documents. For example, in a submission to the 2010 Senate inquiry into the COAG reforms relating to health and hospitals, McGorry grossly exaggerated the strength of the evidence for EPPIC, claiming that ‘EPPIC is demonstrably the most evidenced [sic] based model in the spectrum of mental health care’46. He also misrepresented the nature and cost of the pre-EPPIC program and the impact on suicide rates:

We were able to develop an effective set of evidence based programs that reduced delay in diagnosis, improved outcomes, reduced suicide rates and improved recovery rates…. the evidence has accumulated steadily over the last 15 years or so that this is a highly effective and highly cost-effective program. In fact, if you compare patients that are treated in standard psychiatric care with patients that go through these streamed early psychosis programs, the costs are three times as much over an eight-year period if patients just go through the normal late intervention system. So it is actually irresponsible of Australian governments, state and federal, to not roll this out.47

Part of the misrepresentation was incorporated into the Senate report:

‘5.46 The committee also notes that there is overwhelming evidence that the benefits of providing increased access for young people far outweigh the additional cost to the health budget. The reduction of disability, the improvement in survival rates, the decrease in use of services and improved outcomes point to the high cost effectiveness of early intervention programs. Professor McGorry stated that the costs are three times as much over an eight-year period if a person goes through the normal late intervention system.’48

This demonstrates the effectiveness not of EPPIC but of the misrepresentation of the evidence about EPPIC.

 

FOOTNOTES

  1. Marshall & Rathbone (2011, p. 81)
  2. McGorry et al. (1996, p. 315)
  3. Mihalopoulos et al. (1999, p. 51)
  4. Mihalopoulos et al. (2009, p. 914, table 2)
  5. McGorry et al. (1996, p. 314)
  6. McGorry et al. (1996, pp. 317, 322)
  7. Mihalopoulos et al. (1999, p. 50)
  8. McGorry et al. (1996, pp. 309, 321); Mihalopoulos et al. (1999, p. 48)
  9. McGorry et al. (1996, p. 319)
  10. McGorry et al. (1996, p. 318, table 5)
  11. McGorry et al. (1996, p. 318, table 5)
  12. McGorry et al. (1996, p. 318, table 5); Mihalopoulos et al. (1999, p 52, table 2)
  13. McGorry et al. (1996, p. 318, table 5)
  14. McGorry & Edwards (1998, p. S49)
  15. Mihalopoulos et al. (2009, p. 909)
  16. Mihalopoulos et al. (2009, p. 914, table 2)
  17. Mihalopoulos et al. (2009, p. 914, table 2)
  18. Grossman et al. (2008)
  19. Mihalopoulos et al. (2009, p. 909)
  20. Calculated from Mihalopoulos et al. (2009, p. 915, table 3) and Mihalopoulos et al. (1999, p. 50)
  21. Mihalopoulos et al. (2009, p. 914, table 2)
  22. McGorry et al. (1996, p. 321)
  23. McGorry et al. (1996, p. 321)
  24. McGorry et al. (1996, p. 322)
  25. McGorry et al. (1996, p. 321)
  26. McGorry et al. (1996, p. 318, table 5)
  27. McGorry et al. (1996, p. 321)
  28. McGorry & Edwards (1998, p. S49)
  29. McGorry et al. (1996, p. 309)
  30. McGorry et al. (1996, p. 318, table 5)
  31. Mihalopoulos et al. (1999, p. 51)
  32. Mihalopoulos et al. (1999, p. 52)
  33. McGorry et al. (1996, p. 317); McGorry & Edwards (1998, p. S49)
  34. Mihalopoulos et al. (2009, p. 914, table 2)
  35. Harris et al. (2008, p. 12)
  36. Harris et al. (2008, p. 16, table 1)
  37. Harris et al. (2008, p. 15, figure 1; p. 16, 3rd last paragraph; figure 4 (last line))
  38. McGorry, P. (2010a)
  39. McGorry et al. (1996, p. 322)
  40. Mihalopoulos et al. (1999, p. 48)
  41. Mihalopoulos et al. (2009, p. 910)
  42. Mihalopoulos et al. (2009, p. 910)
  43. Mihalopoulos et al. (2009, p. 909 [abstract])
  44. Mihalopoulos et al. (2009, p. 909 [abstract])
  45. Australian Broadcasting Corporation (2010)
  46. McGorry (2010c, p. 2)
  47. McGorry (2010a, p. 12)
  48. Senate Finance and Public Administration References Committee (2010, pp. 86-97); Greens Senator Rachel Siewert endorsed this statement in her minority report (2010, pp. 117-118)

References

Australian Broadcasting Corporation. (2010, March 11). Mental health system in crisis: McGorry. Lateline. http://www.abc.net.au/lateline/content/2010/s2843609.htm

Grossman, L. S., Harrow, M., Rosen, C., Faull, R., & Strauss, G. P. (2008, November-December). Sex differences in schizophrenia and other psychotic disorders: a 20-year longitudinal study of psychosis and recovery. Comprehensive Psychiatry, 49(6), 523-529. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2592560/?tool=pubmed

Harris, M. G., Burgess, P. M., Chant, D. C., Pirkis, J. E., & McGorry, P. D. (2008, February). Impact of a specialized early psychosis treatment programme on suicide: Retrospective cohort study. Early Intervention in Psychiatry, 2(1), 11-21. http://onlinelibrary.wiley.com/doi/10.1111/j.1751-7893.2007.00050.x/abstract

Marshall, M., & Rathbone, J. (2011) Early intervention for psychosis. Cochrane Database of Systematic Reviews. Issue 6. Art. No.: CD004718. DOI: 10.1002/14651858.CD004718.pub3. http://www.onlinelibrary.wiley.com/doi/10.1002/14651858.CD004718.pub3/pdf

McGorry, P. (2010a, March 4). Council of Australian Governments reforms relating to health and hospitals. Proof Committee Hansard. Canberra: Senate Finance and Public Administration References Committee. http://www.aph.gov.au/hansard/senate/commttee/S13184.pdf

McGorry, P. (2010b, February 6-7). Mental health needs early care: health system. The Australian. http://www.theaustralian.com.au/news/health-science/mental-health-needs-early-care-health-system/story-e6frg8y6-1225826908208

McGorry, P. (2010c). “Still waiting for mental health reform”: Submission to The inquiry into the Council of Australian Governments reforms relating to health and hospitals by The Senate Standing Committee on Finance and Public Administration. https://senate.aph.gov.au/submissions/comittees/viewdocument.aspx?id=e2473cc2-7f01-4773-bda1-f3e0cfe8c8e4

McGorry, P. D. & Edwards, J (1998). The feasibility and effectiveness of early intervention in psychotic disorders: The Australian experience. International Clinical Psychopharmacology, 13(suppl. 1), S47–S52. http://journals.lww.com/intclinpsychopharm/abstract/1998/01001/the_feasibility_and_effectiveness_of_early.8.aspx

McGorry, P. D., Edwards, J., Mihalopoulos, C., Harrigan, S. M, & Jackson, H. J. (1996). EPPIC: An evolving system of early detection and optimal management. Schizophrenia Bulletin, 22(2), 305-326. http://schizophreniabulletin.oxfordjournals.org/content/22/2/305.full.pdf+html

Medew, J. (2010, February 22). McGorry urges mental health overhaul. Sydney Morning Herald. http://www.smh.com.au/lifestyle/wellbeing/mcgorry-urges-mental-health-overhaul-20100221-onzf.html

Mihalopoulos, C., McGorry, P. D., & Carter, R. C. (1999, July). Is phase-specific, community-oriented treatment of early psychosis an economically viable method of improving outcome? Acta Psychiatrica Scandinavica, 100(1), 47-55. http://onlinelibrary.wiley.com/doi/10.1111/j.1600-0447.1999.tb10913.x/abstract [Copies of this paper available are on request: melissa.raven@flinders.edu.au]

Mihalopoulos, C., Harris, M., Henry, L., Harrigan, S., & McGorry, P. (2009, September). Is early intervention in psychosis cost-effective over the long term? Schizophrenia Bulletin, 35(5), 909-918. http://schizophreniabulletin.oxfordjournals.org/content/35/5/909.long

Senate Finance and Public Administration References Committee (2010). Council of Australian Governments reforms relating to health and hospitals. Canberra: The Senate. http://www.aph.gov.au/senate/committee/fapa_ctte/coag_health_reforms/report/report.pdf