Identification of dementia using standard clinical assessments by primary care physicians in Japan

Aim The aim of the present study was to develop a way of identifying dementia using clinical assessments made by primary care physicians under the existing medical care system in Japan. Methods A total of 623 people aged ≥65 years underwent standard clinical assessments by primary care physicians under the long‐term‐care insurance program to determine their grade of activities of daily living related to dementia. To examine the validity of the diagnosis, neuropsychiatrists carried out further diagnosis of dementia for all the participants. We regarded the dementia patients who received care for disability under the long‐term care insurance program as having disabling dementia. Results Multivariable odds ratio (95% confidence interval) in single‐grade increments of the activity was 2.1 (1.7–2.5) for dementia and 2.8 (2.2–3.4) for disabling dementia. The grades ≥I and ≥IIa were near the upper‐left corner in the receiver operating characteristic curves. Setting the cut‐off point at grades ≥I or ≥IIa yielded a higher integrated discrimination improvement, suggesting a major improvement in reducing misclassification by using these cut‐off points. When we used grades ≥I as the cut‐off point, the sensitivity (95% confidence limits) was 65% (58–72%) and the specificity was 93% (91–96%) for dementia, and the corresponding values in grades ≥IIa were 54% (47–62%) and 96% (94–97%). The corresponding values for disabling dementia were 83% (76–90%), 92% (90–95%), 73% (65–80%) and 96% (94–97%). Conclusions Our findings suggest that selection of grades ≥I or ≥IIa as the cut‐off point would reduce instances of misclassification in the identification of dementia and disabling dementia. Geriatr Gerontol Int 2018; 18: 738–744