AbstractObjectivesThis study investigates the stability of neuropsychiatric symptoms (NPS) assessed biweekly using the Neuropsychiatric Inventory (NPI) in a memory clinic population during a 6 week period.MethodsTwenty‐three spousal caregivers (mean [SD] age = 69.7 [8.8], 82.6% female) of 23 patients (43.5% had dementia) completed all assessments. The NPI was assessed four times during 6 weeks. We examined whether NPI domains were present during all four assessments, studied within‐person variation for each NPI domain, and calculated Spearman's correlations between subsequent time‐points. Furthermore, we associated repeated NPI assessments with repeated measures of caregiver burden to examine the clinical impact of changes in NPI scores over time.ResultsThe course of NPS was highly irregular according to the NPI, with only 35.8% of the NPI domains that were present at baseline persisted during all 6 weeks. We observed large within‐person variation in the presence of individual NPI domains (61.3%, range 37.5%–83.9%) and inconsistent correlations between NPI assessments (e.g., range rs = 0.20–0.57 for agitation, range rs = 0.29–0.59 for anxiety). Higher NPI total scores were related to higher caregiver burden (rs = 0.60, p < 0.001), but changes in NPI total scores were unrelated to changes in caregiver burden (rs = 0.16, p = 0.20).ConclusionsWe observed strong fluctuations in NPI scores within very short time windows raising the question whether this represents erratic symptoms and/or scores. Further studies are needed to investigate the origins of these fluctuations.