Aging
The effect of informant characteristics on reports of impairment in cognition and daily functioning in India: Findings from a multiple informant, within-person design Emma Nichols* Emma Nichols Nichols Nichols Nichols Nichols Nichols Nichols Nichols University of Southern California
Informant reports of respondents’ cognitive and everyday functioning are commonly used in studies of older adults to inform dementia diagnoses. However, characteristics of the informant may influence reporting, leading to bias. While existing literature suggests this source of bias may be important, evidence comes from between-person comparisons adjusting for respondent characteristics, and residual confounding due to links between informant and respondent characteristics is a concern. We used multiple informant data from four informant scales in the Longitudinal Aging Study in India – Diagnostic Assessment of Dementia (LASI-DAD) innovation sample (2024-2025; N=858) to estimate the within-person effect of informant characteristics using linear regression models with fixed effects for each respondent. Respondents were on average 66.7 years old (inter-quartile range = 62-70); 56.7% of the sample were women. Included informants were most commonly spouses or partners (25.8%), children (44.9%), or children-in-law (15.5%), who lived with the respondent (88.0%), and provided care to the respondent (73.3%). A large proportion of respondents (10.8%-21.5% across scales) had at least a 1 standard deviation difference in scale scores between informants. However, findings on the effects of informant characteristics were less consistent, with some evidence of differences by caregiving status, relationship with the informant, and frequency of contact. Effects were larger among older adults and were different across informant scales (Figure 1). Findings highlight the variability in scores across informants, but evidence on systematic differences by informant characteristics was inconsistent across scales and weaker compared to prior evidence from between-persons designs. Bias may be a concern predominantly for older respondents with less knowledgeable informants, such as distant relatives, non-relatives, or those who see respondents infrequently.
