353 – Contributed Oral Poster Presentations: Section on Statistics in Epidemiology
Performance of Five Equating Methods in Assessing Cognitive Impairment in Delirium Research
Jamey Guess
Beth Israel Deaconess Medical Center
Long H. Ngo
Harvard Medical School
Edward Marcantonio
Beth Israel Deaconess Medical Center
The Mini-Mental State Examination (MMSE) is well accepted for cognitive assessment, while the newer publicly available Montreal Cognitive Assessment (MoCA) is growing in popularity. We examined several methods to obtain equivalence scores for these 2 assessments, both scored 0-30, 30 best. Mean, linear, equipercentile, circle-arc, and item response theory (IRT) equating procedures were used to derive MMSE-equivalent scores from MOCA scores in a study of 199 hospitalized older adults. All methods gave closed-form equating equations for point estimation. Standard errors for the circle-arc method were obtained via bootstrapping. The estimated equivalence scores for all 5 methods were similar. The equipercentile method provided the best agreement with observed MMSE scores. In terms of precision, although the distributions of standard errors varied widely across the five methods, in the range of scores that most patients have (from 15 to 30), the standard errors were similar and small (within 1 point). Thus, for our data and from this evaluation of the 5 methods, equipercentile equating was the method of choice.