Which Mild Behavioral Impairment‐Checklist (MBI‐C) items discriminate between cognitive diagnoses at baseline and follow‐up? Results from the Compostela Ageing Study (CompAS)

Abstract

AbstractBackgroundThe Mild Behavioral Impairment Checklist (MBI‐C) is a 34 Neuropsychiatric Symptoms (NPS) checklist for the evaluation of pre‐dementia states. Our aim was to analyze which MBI‐C symptoms could best discriminate between participants with different cognitive outcomes both at baseline and at diagnosis time (after 50‐75 months).MethodA total of 179 participants from the Compostela Aging Study (CompAS) were included in the analyses, 108 identified as Subjective Cognitive Complainers (SCC), 47 diagnosed as Mild Cognitive Impairment (MCI) and 24 diagnosed as dementia. NPS were assessed using the validated Spanish version of the MBI‐C (Mallo et al., 2018) at baseline and 50‐75 months follow‐up. Discriminant analyses were carried out at baseline and follow‐up, being the dependent variable the diagnosis at the follow‐up and the independent variable the MBI‐C scores.ResultThe results of the discriminant analysis at baseline indicated a poor adjustment of the model to the data (Wilk’s lambda = .821; p<.001). At baseline, three items (item 9 “discouraged”, 22 “hoard objects”, and 19 “stubborn/rigid”) discriminate 65% of cases, mostly SCCs (97.2%) (Table 1).The results of the discriminant analysis at follow‐up indicated a much better adjustment of the model to the data (Wilk’s lambda = .468; p<.001) (Table 1).The variance was explained almost exclusively by the first function obtained (53%) while the second only reached 7%. At follow‐up, seven items discriminate between groups (item 4 “unmotivated”, 25 “less sensible”, 26 “talk to strangers”, 21 “food no tasty”, 24 “control difficulties”, 5 “less emotional”, 28 “no social judgement”). These items correctly discriminated the SCC (99%) and the 66.7% of dementia patients, but only 6.8% of the MCI (Figure 1).ConclusionModel fit at baseline was poor, and almost exclusively discriminate SCC. The discrimination power increased as the MBI‐C measure were close to the diagnosis, especially for dementia.Distal prediction of dementia nor predementia states using NPS was not successful though some MBI‐C items showed slight better performance at baseline than in follow‐up.