Background. This study investigates the competence of patients with dementia, depression and schizophrenia to make treatment decisions. The outcome of an objective test instrument is presented and compared with clinical assessment of competence by the attending physician.

Method. The MacArthur Competence Assessment Tool-Treatment (MacCAT-T), a test instrument to assess abilities in different standards of competence, was administered to patients with diagnoses of dementia (N=31), depression (N=35) and schizophrenia (N=43). Statistical significance of group differences in the MacCAT-T results were tested with the chi-square test. The concordance of the test and clinical assessment of competence by the attending physician were evaluated by Cohen's kappa coefficient.

Results. Patients with dementia, as a group, showed significantly more often impaired performance than those with schizophrenia who were still more impaired than depressed patients. Patients were classified as impaired or not depending on the standards used. By combination of all standards substantially more patients were classified as impaired than by clinical assessment (67·7 v. 48·4% of patients with dementia, 20·0 v. 2·9% of patients with depression, 53·5 v. 18·4% of patients with schizophrenia).

Conclusions. Using different standards of competence the study showed substantial differences among patients with dementia, depression and schizophrenia. The high proportion of patients identified as incompetent raises several ethical questions, in particular, those referring to the selection of standards or the definition of cut-offs for incompetence. The discrepancy between clinical and formal evaluations points out the influence of the used procedure on competence judgements.