How many trials of Backward Digit Task are enough? Discriminating mild cognitive impairment subtypes on the BDT
Faculty Advisor Name
Bernice Marcopulos, PhD, ABPP
Department
Department of Graduate Psychology
Description
Objective: The Backward Digits Span Test (BDST) asks participants to repeat blocks of seven trials of 3-, 4- and 5-digit backward for a total of 21 trials with no discontinuation. Performance on this test clearly discriminates patients with mild cognitive impairment (MCI) from patients with dementia, and between MCI subtypes. However, work has yet to examine how many test trials are necessary to discriminate between MCI subtypes. This study investigated trial accuracy of the 5-span trial items of BDST across different subtypes of mild cognitive impairment.
Participants and Methods: Patients from Rowan University’s Memory Assessment Program (N = 136, age = 75.11 ± 6.64, education = 14.74 ± 2.65, female = 72%) were administered the BDST as part of a comprehensive neuropsychological evaluation. Patients were categorized into four clinical groups: non-MCI (N = 53), subtle cognitive impairment (SCI) (N = 18), amnestic MCI (N = 30), and mixed/dysexecutive MCI (N = 35). For group categorization, Edmonds et al., 2014 criteria was used for SCI and Jak et al., 2009 criteria was used for MCI subtypes. Hierarchical linear regression models were constructed examining the effect of MCI subtype on correct serial recall and any order recall, controlling for age, education, and sex.
Results: Using SERIAL order scoring, patients with mixed/dysexecutive MCI scored lower than all groups on all 5-span trials (p’s < .01). There was an effect of education on five-span trials 1, 6, and 7, with higher levels of education corresponding with better serial order scores (p’s < .01). Five span trials 4 (ηp2 =.120, .123, .116) and 7 (ηp2 = .114, .114, .141) showed the largest effect sizes. When using ANY order scoring, patients with mixed/dysexecutive MCI scored lower than all groups on 5-span trials 1 and 3 (p’s ≤ .01).
Conclusions: When using SERIAL order scoring, only four trials of 5-span are required to separate the mixed/dysexecutive MCI group from other groups. At the beginning and end of the 5-span trial block (i.e., trials 1, 6, and 7), an effect of education can be seen, with higher levels of education corresponding with higher trial accuracy. Education may be serving as a protective factor against testing fatigue. When using ANY order scoring, trials 1 or 3 of 5-span can separate the mixed/dysexecutive MCI group from the other groups. Future research should examine the optimal number of trials. These results emphasize the importance of multiple trials which are often not done with traditional discontinuation rules. This research should be replicated with a larger, more representative sample - as these results come from a predominantly white, female, and highly educated sample - and should be extended to other neuropsychological protocols and cognitive domains.
How many trials of Backward Digit Task are enough? Discriminating mild cognitive impairment subtypes on the BDT
Objective: The Backward Digits Span Test (BDST) asks participants to repeat blocks of seven trials of 3-, 4- and 5-digit backward for a total of 21 trials with no discontinuation. Performance on this test clearly discriminates patients with mild cognitive impairment (MCI) from patients with dementia, and between MCI subtypes. However, work has yet to examine how many test trials are necessary to discriminate between MCI subtypes. This study investigated trial accuracy of the 5-span trial items of BDST across different subtypes of mild cognitive impairment.
Participants and Methods: Patients from Rowan University’s Memory Assessment Program (N = 136, age = 75.11 ± 6.64, education = 14.74 ± 2.65, female = 72%) were administered the BDST as part of a comprehensive neuropsychological evaluation. Patients were categorized into four clinical groups: non-MCI (N = 53), subtle cognitive impairment (SCI) (N = 18), amnestic MCI (N = 30), and mixed/dysexecutive MCI (N = 35). For group categorization, Edmonds et al., 2014 criteria was used for SCI and Jak et al., 2009 criteria was used for MCI subtypes. Hierarchical linear regression models were constructed examining the effect of MCI subtype on correct serial recall and any order recall, controlling for age, education, and sex.
Results: Using SERIAL order scoring, patients with mixed/dysexecutive MCI scored lower than all groups on all 5-span trials (p’s < .01). There was an effect of education on five-span trials 1, 6, and 7, with higher levels of education corresponding with better serial order scores (p’s < .01). Five span trials 4 (ηp2 =.120, .123, .116) and 7 (ηp2 = .114, .114, .141) showed the largest effect sizes. When using ANY order scoring, patients with mixed/dysexecutive MCI scored lower than all groups on 5-span trials 1 and 3 (p’s ≤ .01).
Conclusions: When using SERIAL order scoring, only four trials of 5-span are required to separate the mixed/dysexecutive MCI group from other groups. At the beginning and end of the 5-span trial block (i.e., trials 1, 6, and 7), an effect of education can be seen, with higher levels of education corresponding with higher trial accuracy. Education may be serving as a protective factor against testing fatigue. When using ANY order scoring, trials 1 or 3 of 5-span can separate the mixed/dysexecutive MCI group from the other groups. Future research should examine the optimal number of trials. These results emphasize the importance of multiple trials which are often not done with traditional discontinuation rules. This research should be replicated with a larger, more representative sample - as these results come from a predominantly white, female, and highly educated sample - and should be extended to other neuropsychological protocols and cognitive domains.
