Close Relationships and Cohabitation Improve the Prognostic Utility of Informant-Rated Symptoms for Predicting Cognitive Decline

Olivia Horn, The Ohio State University, Columbus, United States
Andreana Benitez, Medical University of South Carolina, Charleston, United States
Stephanie Aghamoosa, Medical University of South Carolina, Charleston, United States


Most large-scale observational studies and clinical trials in Alzheimer’s disease and related dementias (ADRD) require that a study partner provide ratings of participants’ functioning. Informant-rated changes in cognition, mood/behavior, and daily activities are considered gold standard outcomes, but previous work suggests that the accuracy of these ratings may be impacted by informant type (e.g., spousal status, gender). The selection of accurate informants is particularly important in studies of preclinical AD given the challenges of predicting cognitive change in initially unimpaired participants. Thus, this study investigated whether study partner demographics, frequency of contact, and relations to the participant moderate how well their ratings predict cognitive decline in preclinical AD.

Participants and Methods:

Data were drawn from the National Alzheimer’s Coordinator Center (NACC) June 2023 data freeze. We included 2164 older adults over age 50 (range: 51-97 years, M = 69.6 years, 78.2% White, 64.1% female) who were determined by clinical consensus to be cognitively unimpaired (CU) at baseline. Each participant’s study partner completed the Functional Activities Questionnaire (FAQ), Neuropsychiatric Inventory Questionnaire (NPI-Q), and Clinical Dementia Rating (CDR) at baseline. Participants completed the NACC Neuropsychological Battery v3.0 at the baseline study visit and the first follow-up visit (M = 1.26 years). The primary outcome was change in a general composite score (memory, visual, language, attention, and speed/executive scores) calculated using normative equations adjusting for age, sex, and years of education. Linear regressions were conducted to determine whether the informant-rated measures (FAQ, NPI-Q, and CDR sum of boxes [CDR-SB]) predict change in cognition and if these effects were moderated by study partner characteristics: age, sex, years of education, relationship to participant, cohabitation, frequency of visits, frequency of phone contact, and how long they have known the participant. All models included participants’ baseline age and follow-up interval as covariates.


Decline in cognition was predicted by higher baseline NPI-Q scores and CDR-SB, but not FAQ scores (likely due to limited range). These effects were not moderated by informant age, sex, or education. Prediction of cognitive decline from FAQ, NPI-Q, and CDR-SB was more accurate for study partners who cohabitated with participants than those who did not. Similarly, prediction of cognitive decline from FAQ and NPI-Q scores was most accurate for spouses/partners/companions compared to children, friends/neighbors/acquaintances, or “Other.” Effects were not moderated by frequency of contact (in-person or phone) or how long the study partner had known the participant.


Study partners who are spouses/partners/companions or cohabitants provide ratings of participants’ cognitive and functional changes that best predict cognitive decline over 1.5 years. It is difficult to disentangle the relative impacts of relationship and cohabitation given the high degree of overlap between the two (>95%). Nonetheless, our results suggest that informants who have close interpersonal relationships and/or the experience of living with participants may be best able to detect subtle changes in CU individuals and provide ratings that have better prognostic utility when predicting cognitive change. These findings may guide the selection of individuals as study partners in ADRD studies.

Category: Aging

Keyword 1: cognitive functioning
Keyword 2: everyday functioning
Keyword 3: dementia - Alzheimer's disease