Practical Assessment of Neuropsychiatric Symptoms: Updated Reliability, Validity, and Cutoffs for the Neuropsychiatric Inventory Questionnaire.

OBJECTIVES

To improve assessment of neuropsychiatric symptoms (NPS) by expanding the measurement properties of the Neuropsychiatric Inventory Questionnaire (NPI-Q).

DESIGN

Multicenter, longitudinal observational study.

SETTING

Several Alzheimer's Disease Research Centers (ADRCs).

PARTICIPANTS

Individuals (n = 45,274) who presented to an ADRC with a collateral and completed the NPI-Q.

MEASUREMENTS

The NPI-Q total severity score, four NPI-Q subscales, dementia stage, expert NPS rating, consensus rating of dementia syndrome, global cognitive screening, collateral rating of daily functioning, and self-rating of depression.

RESULTS

There was strong evidence of criterion validity with both dementia stage and expert NPS rating for the NPI-Q total severity index, which informed cutoffs and interpretive ranges. Furthermore, subscales had adequate classification of dementia syndromes and appropriate convergent relationships with cognition, daily functioning, and mood. There was good-to-excellent evidence of reliability for the NPI-Q total severity index over several years, and subscales had adequate-to-good reliability.

CONCLUSIONS

This is the first study to provide empirically established cutoffs, interpretive ranges, and evidence of reliability over a period longer than a month on the NPI-Q and its subscales. This will improve assessment of NPS in clinical and research contexts.

ARTICLE SUMMARY

Neuropsychiatric symptoms of neurodegeneration are increasingly understood as early disease markers with tremendous functional impact later in disease, but are often missed or misdiagnosed. The most common measure of these symptoms, the Neuropsychiatric Inventory Questionnaire (NPI-Q), does not have clinically actionable guidance, which this article provided. We established cutscores for several conditions and test-retest reliability over longer periods for the total score and subscales using a multicenter database.