A longitudinal study of late-life psychosis and incident dementia and the potential effects of race and cognition

Abstract

Later-life psychotic symptoms are meaningful and are associated with adverse outcomes. Psychosis is an important domain in mild behavioural impairment (MBI), a syndrome that incorporates later-life emergent and persistent neuropsychiatric symptoms (NPS) in dementia-free individuals into dementia prognostication. However, MBI-psychosis-associated risk and its interaction with race has not been well quantified. Here we determined risk of incident dementia in dementia-free participants with MBI-psychosis and effect modification by race as an important factor in assessing the risk of psychosis. Data for participants with normal cognition (NC) or mild cognitive impairment (MCI) from the National Alzheimer Coordinating Centre were used. Participants with neurodevelopmental, neurological and/or longstanding psychiatric disorders were excluded. MBI-psychosis was defined by persistence of delusions and hallucinations across two consecutive visits. Kaplan–Meier curves of ten-year dementia-free survival were generated for MBI-psychosis versus no NPS before dementia diagnosis. Cox proportional hazard models were implemented to assess relative incidence rates, adjusted for cognitive status, age, sex, education, race and APOE-ε4 status. Interaction terms were included for relevant demographic variables. Similar secondary analyses utilized MBI-no-psychosis as reference. The sample consisted of 3,704 no-NPS (age = 72.8 ± 9.9; 62.7% female; 13.4% MCI) and 66 MBI-psychosis (age = 75.2 ± 9.8; 53% female; 72.7% MCI) participants. For MBI-psychosis, in reference to no NPS, the hazard ratio (HR) for incident dementia was 3.76 (CI: 2.53–5.58, p < 0.001), while for conventionally captured psychosis, the HR was 1.92 (CI: 1.58–2.33, p < 0.001). Interaction analyses revealed that in NC, those with MBI-psychosis had a 9.96-fold greater incidence of dementia than those with no NPS (CI: 3.65–27.22, p < 0.001). In MCI, the MBI-psychosis-associated dementia incidence was 3.38-fold greater than no-NPS (CI: 2.22–5.15, p < 0.001). Furthermore, MBI-psychosis-associated dementia incidence in Black participants was 7.44-fold greater than no NPS (CI: 3.54–15.65, p < 0.001), while in white participants, it was 3.18-fold greater (CI: 1.94–5.2, p < 0.001). In a secondary analysis, compared with MBI-no-psychosis (n = 2,260), MBI-psychosis had a 2.47-fold greater incidence of dementia (CI: 1.69–3.59, p < 0.001). Although psychosis is an infrequently endorsed MBI domain, when present it is associated with substantial risk for dementia. HRs differed between cognitive strata, and these differences were significantly greater when MBI-psychosis emerged in NC as opposed to MCI, emphasizing the importance of cognitive assessment at the time of symptom emergence. In addition, the relationship between MBI-psychosis and incident dementia was stronger in Black participants than in white participants. The emergence of persistent psychotic symptoms in older adults is clinically meaningful, and MBI-psychosis identifies a high-risk group for precision medicine approaches to dementia prevention.