Informant characteristics are associated with the Clinical Dementia Rating Sum of Boxes scores in the Alzheimer's Disease patients participating in the National Alzheimer’s Coordinating Center Uniform Data Set

Abstract Background The Clinical Dementia Rating® Sum of Boxes (CDR®-SB) is used to stage dementia severity; it is one of the most common outcome measurements in Alzheimer's Disease (AD) research and clinical trials. The CDR®-SB requires an informant to provide input to stage a patient’s dementia severity. The effect of the informant’s characteristics on the CDR®-SB is unknown. We aimed to evaluate the effect of the informant’s sex, relationship to the patient, and frequency of contact on the CDR®-SB scores in patients with Alzheimer's Disease with mild cognitive impairment or dementia included in the National Alzheimer’s Coordinating Center Uniform Data Set (NACC-UDS). Methods We included all participants from the NACC-UDS that had AD as diagnosis, and information about the Mini-Mental State Examination or Montreal Cognitive Assessment scores, informant sex, relationship to patient and frequency of contact; we also analyzed the possible interaction between these characteristics on the CDR®-SB as the outcome. We performed a multilevel linear regression analysis. Results We included data from 20636 participants, totalling 47727 visits. Patients' age was 74.0 ± 9.4 years and 54.1% were females. Informant characteristics were mean age of 66.2 ± 13.2 years, 69.1% were females, and the relationship to patients was 60.5% spouse or partner, 26.7% children and 12.8% other relation. The CDR®-SB scores were 0.20 higher (CI 95%: 0.11 to 0.29) when the informant was female. When comparing to informant’s relationship with the baseline being spouse or partner, the CDR®-SB was 0.39 higher (CI 95%: 0.25 to 0.53) when the informant was the patient’s child and 0.18 lower (CI 95%: -0.35 to -0.01) if relationship was other. Regarding the frequency of contact, CDR®-SB scores were 0.38 higher (CI95%: 0.28 to 0.47) when contact was at least once a week, 0.65 higher (CI95%: 0.52 to 0.78) when contact was daily, and 0.57 higher (CI95%: 0.46 to 0.69) when informant was living with the patient, baseline was a frequency of less than once per week. Finally, the interaction between informant relationships other and female patients showed a 0.24 higher CDR®-SB score (CI95%: 0.03 to 0.46). Conclusions We found that the CDR®-SB scores are significantly modified by informant characteristics and frequency of contact in the NACC-UDS patients with AD diagnosis. These findings hold clinical significance as informant characteristics ideally should not impact the staging of AD patients, and any such effects could introduce bias into clinical evaluations in clinical trials. Future research endeavours should investigate strategies to address and mitigate the influence of these confounding variables.