Boston Naming Test automatic credits inflate scores of nonaphasic mild dementia patients

The Boston Naming Test (BNT), a 60-item test of confrontation naming, may be administered either from Item 1 or Item 30, depending on assumptions of performance. If the BNT is administered from Item 30, 29 automatic credits are given for preceding items, allowing identical norms for either administration. We aimed to compare effects of automatic credits.

We compared effects of automatic credits in the Gothenburg Mild Cognitive Impairment Study, first between normal controls (n = 23) and patients (n = 259), and then between the same patients grouped by stage of impairment: subjective cognitive impairment (SCI, n = 75), mild cognitive impairment (MCI, n = 117), or mild dementia (n = 67).

Automatic credits added to all groups. Both administrations from Item 1 and those from Item 30 discriminated between controls (n = 23) and all patients (n = 259), as well as between the above stages of impairment. However, neither administration discriminated between normal controls and SCI patients. When earned scores were compared, with scores counted from Item 30 plus 29 automatic credits, mild dementia patients on average received a 3.4-credit boost. This equals 82% of the standard deviation of Tallberg's Swedish norms [Brain and Language, 94(1), 19-31 (2005)] or 117% of our normal controls' standard deviation.

In our homogenous material, administration of BNT from Item 30 distinguished between stages of deterioration as well as administration from Item 1. In line with recent literature, we also find BNT results skewed. Thus, for clinical accuracy, we recommend use of cumulative percentages, careful consideration of education and demographic factors, and, most importantly, never to mix forms of administrations with and without automatic credits. While BNT automatic credits diminish accuracy on all levels, they inflate scores significantly for nonaphasic mild dementia patients.