한국에서 이용되고 있는 인지기능 평가 도구들의 예측치 - Bayes 공식에 의한 -

Predictive Values of Korean Cognitive Function Tests in Unselected Elderly Community Samples - Bayesian Analysis -

저자들은 기왕에 발표된 연구들을 이용하여 지역사회 노인들에서 인지기능 검사 도구들의 양성 및 음성 예측치를 계산했다. 치매 유병율에 따른 양성 및 음성 예측치의 계산은 Bayes 공식을 이용하였다. MMSEK에서 치매 유무의 구분점을 21/20으로 했을 경우, 양성 예측치는 치매 유병율이 2.5%일 때와 10.8%일 때 각각 72.3%와 92.5%이었고, 음성 예측치는 각각 99.5%와 97.8%였다. CIDI에서 57점 이하 를 치매로 정했을 때 치매의 유병율이 2.5%인 노인 집단에서는 양성 예측치와 음성 예측치가 각각 27.5% 와 99.8%이었고, 치매의 유병율이 10.8%인 집단에서는 양성 및 음성 예측치가 각각 64.2%와 99.1%이었다.

Objective：The authors calculated the positive and negative predictive values of the Korean version of minimenteal state examination(MMSEK) and Cognitive Impairment Diagnosing Instrument(CIDI) in order to estimate the reliability of these tests in large unselected populations. Methods：The data of the MMSEK were selected from Park et al’s study4). The data of the CIDI were selected from Park’s another study2). Calculation of the positive and negative predictive values was based on the Bayes theorem. Results：When the prevalence of dementia is 2.5% among the elderly people in a community, the MMSEK cutoff point of 20 for identification of dementia yielded 72.3% PV(+) and 99.5% PV(-). When it is 10.8%, the cutoff point for identification of dementia yielded 92.5% PV(+) and 97.8% PV(-). Using the CIDI cutoff point of 57, PV(+) was estimated as 27.5% and PV(-) as 99.8% when the prevalence of dementia is 2.5% among the elderly people in a community. PV(+) and PV(-) were computed as 64.2% and 99.1%, respectively, assuming that the prevalence of dementia is 10.8% in an elderly population based on the above CIDI cutoff point. Conclusion：When the MMSEK and CIDI are to be used in unselected elderly samples, it is important to know the predictive values of these tests related with prevalences.