Psychometric properties of the Neuropsychiatric Inventory: a review

Neuropsychiatric symptoms (NPSs) cause a significant burden to individuals with memory disorders and their families. Insights into the clinical associations, neurobiology, and treatment of NPSs are largely dependent on informant questionnaires, such as the commonly used Neuropsychiatric Inventory (NPI). Like any scale, the utility of the NPI relies on its psychometric properties, but unlike many scales, the NPI faces unique challenges related to its skip-question and scoring formats.We examined and reviewed the psychometric properties of the NPI in a framework including psychometric properties pertinent for construct validation, and health-related outcome measurement in general. We found that aspects such as test-retest and inter-rater reliability are major strengths of the NPI in addition to its flexible and relatively quick administration. These properties are desired in clinical trials. However, it seems that the reported properties cover only some of the generally examined psychometric properties, representing perhaps necessary but not sufficient reliability and validity evidence for the NPI. There appear to be significant gaps in psychometric data, at least partially owing to small sample sizes in the studies that preclude more comprehensive analyses. Regarding construct validity, only one study examined structural validity with the NPI subquestions. Measurement error was not assessed in the reviewed studies. For future validation, we recommend using data from all subquestions, collecting larger samples, paying specific attention to construct validity and formulating hypotheses a priori. As the NPI is an outcome measure of interest in clinical trials, examining measurement error could be of practical importance.