A Crosswalk of the IQCODE and AD8 Proxy Dementia Screening Tools

ABSTRACT Introduction Since many hospitalized older adults have delirium, assessment of baseline cognitive function often requires proxy‐based measures. While the 16‐item Informant Questionnaire on Cognitive Decline in Elderly (IQCODE) specifies cutoff scores for mild cognitive impairment (MCI) and dementia, the 8‐item Informant Interview to Differentiate Aging and Dementia (AD8) provides only a single cutoff for “cognitive impairment,” limiting its usefulness and precision. Our aim was to develop a crosswalk between the IQCODE and AD8 to identify cutoff scores on the AD8 for MCI and dementia. Methods The IQCODE and AD8 were administered to proxies of hospitalized older adults. We used two methods: equipercentile, which matches percentiles between IQCODE and AD8 score distributions, and item response theory (IRT), which matches estimated latent traits of the two measures, to create a crosswalk of equivalent scores and identify AD8 cutoff scores for MCI and dementia. Results A total of 264 hospitalized older adults (mean age 79.5 years, 55% women) and their proxies were recruited. Equipercentile and IRT methods yielded a crosswalk linking AD8 scores from 0 to 8 with IQCODE scores ranging from 1.0 to 5.0. An AD8 score of 2, corresponding to an IQCODE score of 3.2, was identified as the cutoff for MCI, while an AD8 score of 4, corresponding to an IQCODE score of 3.5, was identified as the cutoff for dementia. Discussion Our crosswalk equates AD8 and IQCODE scores and provides cutoff values for MCI and dementia for the AD8, enhancing its clinical utility in acute care. By establishing empirically derived thresholds, this study strengthens comparability across proxy‐based cognitive measures and expands the AD8's usefulness for distinguishing MCI from dementia. In acute settings, where delirium and time constraints limit direct testing, the AD8 offers a brief, reliable option for measuring baseline cognition and supporting early recognition of cognitive decline.