Neuropsychiatric symptoms due to Alzheimer’s disease (AD) and mild cognitive impairment (MCI) can decrease quality of life for patients and increase caregiver burden. Better characterization of neuropsychiatric symptoms is needed to identify effective treatment targets. The current investigation leveraged the National Alzheimer’s Coordinating Center (NACC) Uniform Data Set (UDS) to examine the network structure of neuropsychiatric symptoms among symptomatic older adults with cognitive impairment.

The identified sample includes those from the NACC UDS (all versions) with complete data on the Neuropsychiatric Inventory Questionnaire (NPI-Q) at initial visit. The NPI-Q is an informant-based estimation of the presence and severity of neuropsychiatric symptoms (delusions, hallucinations, agitation or aggression, depression or dysphoria, anxiety, elation or euphoria, apathy or indifference, disinhibition, irritability or lability, motor disturbance, nighttime behaviors, appetite and eating problems). The following inclusionary criteria were applied for sample identification: age 50+; cognitive status of MCI or dementia; AD was the primary or contributing cause of observed impairment; and at least one symptom on the NPI-Q was endorsed. Participants were excluded if they endorsed “unknown” or “not available” on any NPI-Q items. The final sample (n = 12,507) consisted of older adults (Mage=73.94, SDage=9.41; 46.2% male, 53.8% female) who predominantly identified as non-Hispanic white (NHW) (74.5% NHW, 10.9% non-Hispanic Black, 8.5% other, 5.8% Hispanic white, .3% Hispanic Black). The majority of the sample met criteria for dementia (77.6% dementia, 22.4% MCI) and AD was the presumed primary etiology in 93.9%.

The eLasso method was used to estimate the binary network, wherein nodes represent NPI-Q variables and edges represent their pairwise dependency after controlling for all other symptom variables in the network. In other words, the network represents the conditional probability of an observed binary variable (e.g., presence/absence of delusions) given all other measured variables (e.g., presence/absence of all other NPI-Q symptoms) (Finnemann et al., 2021; van Borkulo et al., 2014). Strength centrality and expected influence were calculated to determine relative importance of each symptom variable in the network. Network accuracy was examined with methods recommended by Epskamp et al. (2018), including edge-weight accuracy, centrality stability, and difference tests.

Edge weights and node centrality (CS(cor=.7)=.75) were stable and interpretable. The network (M=.28) consisted of mostly positive edges and some negative edges. The strongest edges linked nodes within symptom domain (e.g., strong positive associations among externalizing symptoms). Disinhibition and agitation/aggression were the most central and influential symptoms in the network, respectively. Depression or dysphoria was the most frequently endorsed symptom, followed by anxiety, apathy or indifference, and irritability or lability.

Endorsed disinhibition and agitation yielded a higher probability of additional neuropsychiatric symptoms and influenced the activation, persistence, and remission of other neuropsychiatric symptoms within the network. Thus, interventions targeting these symptoms may lead to greater neuropsychiatric symptom improvement overall. Depression or dysphoria, while highly endorsed, was least influential in the network. This may suggest that depression and dysphoria are common, but not central neuropsychiatric features of AD pathology. Future work will compare neuropsychiatric symptom networks across racial and ethnic groups and between MCI and dementia.