Canadian Outcomes Study in Dementia: study methods and patient characteristics.

Objective: To describe the methods and patient characteristics of the Canadian Outcomes Study in Dementia (COSID).

Methods: COSID is a 3-year prospective study of dementia patients living in the community at the time of study registration. We assessed patients' cognition, behaviour, and functioning every 6 months, using the Modified Mini-Mental State Examination (3MS), the Neuropsychiatric Inventory (NPI), and the Functional Autonomy Measurement System (SMAF), respectively. We assessed caregivers, using the Zarit Burden Interview (ZBI). Additional information included the Global Deterioration Scale (GDS), patients' driving status, and clinical information including family history, dementia type, concomitant medications, and comorbid conditions. From the patient or caregiver, we collected details of inpatient and outpatient resources used by the patient and (or) caregiver.

Results: We enrolled 766 patients from 31 Canadian sites. Overall mean age was 76.8 years, and mean age of onset was 73.1 years. Of the total patients, 98% were white, 54% were women, and 84% were diagnosed with Alzheimer's disease. Mean baseline 3MS was 66.5, NPI was 9.5, and SMAF was 18.30. Of these patients, 48% reported a GDS score of 3 (that is, moderate), 16% reported a GDS score of 4 (that is, moderately severe), and the remaining 36% reported a GDS score of 1 or 2 (that is, mild or very mild). At baseline, 83% of patients received cholinesterase inhibitors, 46% received nonsteroidal antiinflammatory drugs, 39% received vitamin E, and 25% received antidepressants. Adult day care and home help were the largest cost factors in this population, with mean monthly costs of $65 and $64, respectively. We found interesting differences in the resources used among geographic regions and care settings.

Conclusions: COSID is already generating valuable information about treatment patterns, outcomes, and resource use in Canadian patients with dementia. As the data mature, it will be possible to build robust models on treatment effectiveness and costs of care.