The three main findings from this study are (a) generic risperidone should be used as the first line drug treatment for schizophrenia if the cost is less than 10 baht per 2 mg tablet; (b) combining risperidone with family interventions will substantially increase health gain with lower hospitalization cost; and (c) clozapine could be a second line medication for patients with high severity who fail to respond to risperidone.
The key recommendation of this study is that risperidone should be included in the national drug list while olanzapine should not (Table ). Clozapine should be reserved for severe patients who do not respond to risperidone as it may have superior benefits in treatment-resistant schizophrenia and in preventing suicidal behaviour [33
]. However, the poor compliance with the Clozapine Patient Monitoring Service program in Thailand [47
] means that clozapine cannot be recommended until monitoring has improved.
The stigma of schizophrenia could be a significant barrier to family interventions [48
]. More importantly, due to a lack of mental health resources and high travel costs, poor households and people living in rural areas may have limited access. A new stream of long term government funding would be required to provide family interventions across the country and make these services accessible long-term.
This study has several limitations. First, the studies included in our meta-analyses were generally from Western countries not Thailand or other Asian countries. Second, the course of disease varies considerably between patients. Modelling the average cost and impact on the average case of disease could lead to erroneous conclusions [49
]. The alternative is to use a microsimulation approach; however, that would require more clinical and epidemiological information on individuals than we had available. Third, the methods used for translating the effect sizes into a change in DW need the assumption that the effect sizes from trials can be directly applied to general health status estimated using a generic measurement. Normally, generic measurements are less relevant to schizophrenia's symptoms by comparison with specific ones as BPRS used in this study. However, this is the only method allowing us to clarify DWs in patients with different treatments. An advantage of using this method is that we could include Thai data on the individual patients into the analysis. We would content that it is not necessarily so that departure from the assumed linear relationship between symptom change and disability change would lead to over-estimation of health benefits in DALYs. Fourth, the assumption that patients remain on the same treatment throughout the course of their illness is in contrast to the common clinical practice of switching patients from one treatment to another treatment based on individual medication responses [50
]. However, a large double-blinded randomized clinical trial in the US suggested no difference in rates of improvement between patients with atypical antipsychotics switching to a new medication and patients staying on their initial treatment [51
]. Finally, we have not incorporated the costs of treating diseases related to increased body weight due to unavailability of local disease costing data. Because there is only a small difference in weight gain between typicals and risperidone, this limitation is unlikely to affect the findings.
Our study also has several major strengths. First, the information used for assessing costs and effectiveness of each intervention is documented and transparent, and from Thai sources wherever possible. Second, to our knowledge, there is no previous study including time and travel costs in the cost estimations. Our findings show that these are a significant burden on patients and families, particularly for clozapine and family interventions. Third, we clearly address the issue of uncertainty in the data, presenting our results as a range within which the value is expected to fall. Fourth, we have had regular contact with local policy makers over the five-year span of our work. As a result of this, the food and drug administration in Thailand has already allocated 40 million baht to add generic risperidone to the national essential drug list.
This is the first study of its kind in Thailand. Studies elsewhere have modeled cost-effectiveness of different atypical antipsychotics, with fewer studies addressing both drug and non-drug interventions. Our results are in line with previous research. A study conducted in Slovenia which measured effectiveness as the percentage of patients in remission found that risperidone was more cost-effective than haloperidone and olanzapine for schizophrenia [52
]. Studies in Canada, Brazil and Australia consistently found that using risperidone rather than olanzapine would reduce the medical costs of schizophrenia [22
]. In contrast, a number of studies using the method developed by the WHO-CHOICE project suggested using typical antipsychotics in combination with psychosocial interventions [55
]. However, the authors discussed that if generic forms of atypical medication would become available, the findings could change [55
] as our sensitivity analysis indicates. Costs associated with drug treatments are a major driver of the results of cost-effectiveness analysis and limit the applicability of our conclusions to other countries.
A combination of psychosocial interventions and medications has been highly recommended as a successful treatment package for schizophrenia [16
]. Our results confirm that this package not only provides more health gain to patients, but also could help the government to reduce hospitalization cost by as much as 40% compared to typicals (Table ). The hospitalization cost is generally considered to be the most expensive component of direct costs [3