The shortened version of a web‐based cognitive test tool, cCOG, retains good accuracy in detecting mild cognitive impairment and dementia

Abstract

AbstractBackgroundcCOG, a web‐based cognitive test tool, has been shown to detect mild cognitive impairment (MCI) and dementia accurately. However, the test now takes approximately 25 minutes to complete, which could be burdensome for patients, especially with cognitive impairment. We assumed that shortening certain subtasks leads to a decrease in total test duration and therefore increases the user‐friendliness of cCOG. In this study, we examined whether shortening these subtasks influences the discriminative value of cCOG.MethodFrom the PredictND and Amsterdam Dementia Cohort (ADC) cohort, we included cCOG results of 424 participants (n = 234 cognitively normal (CN), n = 84 MCI, n = 106 dementia) who were on average 70±9 years and 54% were female. cCOG consists of five subtasks: episodic memory learning, reaction time, modified trail‐making A and B and episodic memory recall. Scores for all subtasks except episodic memory recall were abbreviated into shorter versions, e.g. by cutting the number of items/trials. A global cognitive score (GCS) was computed using a linear regression model based on scores of all subtasks for both the original as the short version. To investigate the diagnostic accuracy of both versions, we calculated pairwise area under the curve (AUC) comparisons for differentiating MCI and dementia from CN.ResultShortening of subtasks led to an average decrease of 4.7 minutes of the total test time (20.1%) for the PredictND cohort (n = 314), with even 5.7 minutes decrease in the dementia group (Figure 1). The accuracy of the GCS based on the short version for differentiating MCI from CN was lower than based on the original version though still considered excellent (AUC = 0.87 [0.83‐0.92] original; AUC = 0.86 [0.81‐0.90] short; Figure 2). For differentiating dementia from CN, the short version performed comparable to the original version (AUC = 0.93 [0.89‐0.96] original; AUC = 0.92 [0.88‐0.96] short; Figure 3).ConclusionShortening subtasks decreased the test duration of cCOG with approximately 5 minutes, while the diagnostic accuracy was preserved. An additional pilot study is needed to test the effect of shortening subtasks on episodic memory recall and global cognitive score. Further development of a user‐friendly and accurate web‐based cognitive test tool is crucial to keep dementia care accessible and affordable.