Worldwide-Schizophrenia Outpatient Health Outcomes (W-SOHO): Baseline Characteristics of Pan-regional Observational Data from More than 17,000 Patients*

J. Karagianis; D. Novick; J. Pecenak; J. M. Haro; M. Dossenbach; T. Treuer; W. Montgomery; R. Walton; A. J. Lowry


Int J Clin Pract. 2009;63(11):1578-1588. 

In This Article


The W-SOHO database provides us with unprecedented longitudinal observational data collected from more than 17,000 patients being treated for schizophrenia using the same recruitment criteria, study design, assessment tools and data collection form, facilitating comparisons and examination of disease course, social and functional outcomes. These are also recent data; the first patient was enrolled late in 2000. As such, representation of patients from 'developing/emerging' economies is very high in W-SOHO, despite some of the difficulties associated with conducting research outside of the usual western settings.[26] The importance of including such patients should not be underestimated; a majority of RCT data for schizophrenia are based on North American patients, who represent only 2% of the global schizophrenia population.[27] Furthermore, in a recent search of the ISI Web of Science database from 1992 to 2001, high-income countries (with 15% of the world's population) contributed 94% of the published mental health literature; only 6% of publications were from low or middle-income countries (which represent more than 85% of the world's population).[28] A similar trend was also evident in a review of the top six psychiatric journals (from 2002 to 2004), where only 3.7% of publications were from low or middle-income countries.[29] Moreover, restrictive trial design and stringent inclusion/exclusion criteria mean that patients in RCTs are not usually a very representative sample of patients with schizophrenia, one comparison revealing that 38–55% of patients in a psychiatric practice network would have failed selection for an RCT because of their medications and comorbidities.[30]

It appears that, despite being drawn from a variety of cultures and practice settings, the patients entered into W-SOHO were remarkably similar across regions in terms of demographic and clinical characteristics. That is not to say that there were not important differences noted in some measures. Patients from East Asian countries were distinct from those in other regional groups in terms of better symptomatology across all CGI-SCH domains (particularly cognitive), low previous use of depot typical agents, and less patient-reported sexual dysfunction. This may be because of a number of underlying factors. As 67% of the patients in the East Asian grouping are from Korea, regional results are heavily influenced by these sites. Care of patients with schizophrenia in long-stay medical facilities is a common practice in Korea; contributing factors include reimbursement and stigma.[31] Based on market share data, depot typical antipsychotics account for only 8% and 3% of days of therapy for patients receiving treatment for schizophrenia in Taiwan and Malaysia respectively, even lower rates of depot use are seen in Korea (< 1% of days of therapy) (data on file). Cultural sensitivities may account for the low reporting of sexual symptomatology, as the association between antipsychotic use and sexual dysfunction has been well established.[32,33]

Compared with other regions, the Northern European subgroup had a high prevalence of TD, yet this was not matched by a high prevalence of the use of typical antipsychotics, as expected. This may be because of several reasons, including the fact that the raters may have been better trained to recognise TD than raters in other regions and that intolerability to prior medications was a strong driver of medication change in this region, suggesting an emphasis on adverse events. It may be an artefact, but that is unlikely, given the correspondingly high prevalence of sexual dysfunction that might be expected under similar treatment conditions that give rise to TD.

The W-SOHO study included patients with a fairly broad distribution of illness severity, as judged by various proxies. The median duration of illness was 7 years (interquartile range 1–16 years), between 5.3% and 12.3% of patients had not previously been treated with an antipsychotic for schizophrenia, and 31.2–40.8% of patients had an inpatient psychiatric admission in the previous 6 months. With the exception of East Asian countries, at least 20% of patients had used a depot antipsychotic in the previous 6 months. Previous use of clozapine ranged from 3.1% to 10.8% of patients and the baseline CGI-SCH means ranged from 3.9 to 4.7. As such, the authors believe that they have captured a reasonable cross section of patients in this study.

Inevitably, schizophrenia also has an impact on functional outcomes. Given the diverse and dynamic factors involved, including treatment accessibility, variation in local resources, infrastructure and cultural practices, it is unsurprising that features such as housing vary across regions. While housing/employment status and level of social interaction can be used as a proxy for level of engagement in the community, self-sufficiency, etc., they are heavily influenced by factors outside of both the diagnosis and prognosis, and as such can be difficult to interpret. A more useful measure in terms of cross-regional comparison may be the EQ-VAS score, a single index measure of how good the patient considers their current health state to be. The results from W-SOHO show a patient population with significantly diminished self-rated valuation of health-related quality of life, well below the age-adjusted mean values found in the countries where population data are available. For example, non-smoking adults of a similar mean age (range 35–44 years) in the UK have a mean VAS score of 87, (which drops to 83 for those who smoke more than 20 cigarettes per day), this is 27 (or 23) points better than the highest-rated health state captured in W-SOHO.[34]

Although not tested for statistical significance, NE and LA had the highest rates of suicide attempts, with the lowest being in EA and North Africa/Middle East. It is possible that cultural sensitivities may contribute to the willingness to report a suicide attempt or to the taboo against making a suicide attempt. The highest level of recent attempts occurred in LA and North Africa/Middle East. Future analyses might examine associations with prescriptions for antidepressants or mood stabilisers, adherence to treatment, or background suicide rates in those countries. Suicidal tendencies are a well-known feature of schizophrenia; the fact that among the W-SOHO patients, at least one in five patients has attempted suicide at some point is a sobering reminder of the episodic and destructive nature of this illness. Recent meta analyses estimate that the risk of suicide for individuals with schizophrenia is 12 times greater than in the general population[35] and that 4.9% of people with schizophrenia will commit suicide, usually early in their illness.[36] Older estimates suggest that approximately 50% of people with schizophrenia have attempted suicide or experienced suicidal ideation.[37] Overall results showing that, on average, 5% (and as high as 9%) of patients in the study had attempted suicide in the previous 6 months, indicate that even for these relatively young, moderately ill outpatients who are currently receiving care, suicide risk is part of their active disease. Coupled with the symptom scores, these data demonstrate that patients in this study population are experiencing a significant burden of illness. This is particularly relevant when considering recent studies on mortality in schizophrenia, suggesting that antipsychotic treatment may increase certain types of mortality.[38] Suicide is one type of mortality that would be expected to decline with successful treatment. Any study that examines mortality should consider all sources of potential mortality, and both increase and reduction in mortality overall.

These data need to be interpreted in the context of the limitations of the study. Countries in the regional groupings were selected on a pragmatic basis and should not be considered representative samples; regional comparisons may not be valid outside the countries studied. For simplicity, the regional groupings are treated as such; however, it would be more accurate to say, e.g., 'East Asian countries' rather than 'EA'. Although all participating investigators were trained in the use of the study instruments and methodology, no assessment of inter-rater reliability was made across sites. In addition, blood sampling was not part of the study design, so no assessment of pre-existing or treatment-emergent metabolic events was available. Weight was measured at baseline and weight changes in patients during the study participation will be reported in separate analyses. The requirement for the initiation or change of antipsychotic medication at study entry may have prevented the inclusion of stable patients with good symptom control on their current therapy, and the need for informed consent provided by patients may have excluded severely unwell patients from participation. Intentional overrepresentation of olanzapine in the sample and open label use may contribute to bias in outcome assessments. However, for this introductory paper, we are strictly addressing baseline conditions before any new antipsychotics were given. Finally, the SOHO studies were not designed to test regional differences.

Despite the inherent variation in individuals; uncontrollable environmental, intra-ethnic and intra-societal factors and the prognostic diversity of schizophrenia, the simple observation remains that there are striking cross-regional similarities in baseline characteristics for most measures. By applying an observational methodology and using instruments that allow international and cross-cultural comparisons, this database allows us to explore the consistency of results gathered from across different regions, healthcare systems, etc. The combined study population also affords us the opportunity to explore less common events and subgroups with more precision than either of the source studies. Further analyses aimed at examining the differences in disease course across countries, and assessing how medication affects these outcomes, are planned; this global database will be used to analyse different subjects that cannot be covered by the two original separate datasets.