What am I measuring? Disentangling cognitive status from sensory impairment

Abstract

AbstractBackgroundSensory impairment in hearing or vision affects 55% of older adults aged 60+. Some studies have shown sensory impairment is associated with impaired cognitive test performance, however tests rely to varying degrees on hearing, vision, or both. We hypothesized that scores for cognitive tests whose administration or instructions depend on vision (e.g., Trail Making) or hearing (e.g., word‐list learning) are biased among those with vision or hearing impairment, respectively, after controlling for underlying cognitive performance.MethodWe leveraged cross‐sectional data from a comprehensive neuropsychological battery on N=658 men and women from the Baltimore Longitudinal Study of Aging (BLSA) and N=1,008 participants from the Atherosclerosis Risk in Communities study (EyeDOC ancillary study), all of whom had objective diagnostic hearing and vision assessed. We used item response theory methods to test for differential item functioning (DIF) by hearing (PTA>25 db in 4 speech frequencies) and visual acuity (20/40) impairment.ResultThe BLSA sample was 55.7% female and 66.3% white. The EyeDoc sample had a higher proportion of women (63.4%) but fewer white participants (54.3%). Overall, we found more evidence of DIF in the EyeDoc study, a sample with lower cognitive performance and a higher number of comorbidities, as compared to the BLSA. No DIF by visual impairment was observed in BLSA, but in EyeDoc we observed DIF by visual impairment for the Delayed Word Recall Task such that the task was harder for those with visual impairment. A larger number of tests showed DIF by hearing impairment in both BLSA and EyeDoc, including the Delayed Word Recall Task in EyeDoc and the Digit Symbol Substitution Task in BLSA.ConclusionWe identified DIF by hearing impairment for tests that do not rely on hearing and DIF by vision impairment for tests that do not rely on vision. Based on these findings, if bias is found in cognitive tests independent of characteristics of the test administration or instructions, cognitive load may be playing a larger role than previously acknowledged. Future investigation is warranted to determine how sensory impairment impacts completion of cognitive testing, as this study was limited to participants who successfully completed cognitive testing.