In the next decade the number of demented people is likely to increase. This has a financial impact on the resources allocated for health and social care. Classical methodologies for dealing with dementia include the historic model that has been criticised for its poor co-ordination between general practitioners, psychiatrists, social workers and the treatment services, and the needs-based approach that, although it involves a co-ordinating needs manager, does not prevent leapfrogging of patients. In order to overcome these problems we propose an annual screening of patients to take place in the needs-based models and compare this with a more sophisticated model in which after being screened patients are investigated for potential drug treatment. We also consider mix intervals for under 75 and over 75. Decision analysis is used to assess the different screening options. Assuming a 20% reduction by treatment of the cost of care we find optimal the investigation/treatment of all patients after screening. Sensitivity analysis shows that even an 8.5% reduction would make the needs-based models non-optimal.