Reliability and validity of the Chinese version of the Neuropsychiatric Inventory Questionnaire (NPI-Q) in patients with Aizheimer's disease / 中国心理卫生杂志

ABSTRACT

Objective: To develop the Chinese version of the Neuropsychiatric Inventory Questionnaire (NPI-Q) and evaluate its reliability and validity. Methods: Three investigators assessed 10 dementia cases together with the NPI-Q to evaluate the inter-rater reliability. Eighty-six dementia cases and 30 healthy controls were assessed with the Behavioral Pathology in Alzheimer's Disease (BEHAVE-AD), the Brief Psychiatric Rating Scale (BPRS) and NPI-Q to evaluate the criterion validity. A subgroup of 30 dementia cases was tested with the NPI-Q again 24 h later to evaluate the test-retest reliability. Results: (1) Internal consistency. Cronbach's a coefficient of the severity subscale was 0.589, mean inter-item correlation coefficient of the severity subscale was 0.107. Cronbach's α coeffi-cient of the distress subscale was 0.612, and the mean inter-item correlation coefficient of the distress subscale was 0.114. (2) Inter-rater reliability. Intraclass correlation coefficient (ICC) of the severity subscale was 0.97, and ICC of the distress subscale was 0.94. (3) Test-retest reliability. The test-retest correlation coefficient of the severi-ty subscale was 0.89, and the test-retest correlation coefficient of the distress subscale was 0.86. (4) Criterion va-lidity. The severity subscale total score was correlated significantly with BEHAVE-AD (r=0.70, P <0.001) and BPRS(r=0.40, P<0.001) total score; the distress subscale total score was correlated significantly with the global ratings in part 2 of BEHAVE-AD (r=0.76, P<0.001) . (5) The severity subscale total score in demen-tia group was significantly higher than that in control group (median 10 vs.1.5, P <0.001). The distress sub-scale total score in dementia group was significantly higher than that in control group (median 10 vs 0; P< 0.001) . (6) Exploratory factor analysis identified 4 common factors, including psychosis, abnormal behavior,discontrol and affective symptoms, which explained 58.3% of total variance. All of the items loaded were between 0.596 and 0.803 on their conrresponding factor. Conclusion: The reliability and validity of the Chinese version of NPI-Q meet the psychometrics properties, and deserves to conduct further study in larger samples.