Underlying structure of the MBI‐C in pre‐symptomatic and prodromic states of dementia: A multidimensional scaling approach

Resumo

BackgroundThe Mild Behavioral Impairment Checklist (MBI-C) (Ismail et al., 2017) is a 34-item scale that evaluates Neuropsychiatric Symptoms (NPS) in pre-dementia states. Its underlying structure has been little studied and remains largely unknown. Factor analysis-based approaches may have difficulty finding stable relationships between NPS collected on a checklist to account for the large heterogeneity of manifestations. Thus, our objective was to analyze the underlying structure of the MBI-C using an alternative approach.MethodEighty-two MCI and 117 SCD older adults of the CompAS were recruited in primary care health centers. Asymptomatic participants were excluded from the analyses. A weighted model of multidimensional scaling (MDS) will be used. A two-step bidimensional weighted dichotomous MDS was performed. All items were included in the first step. Items closely associated with each dimension (1 SD above or below the mean) were selected in a second step to obtain the final model solution.ResultsThe results obtained in the two analyzes were similar in terms of the good fit of the models, the type of two-dimensional solution and the group weights. Model weights were also similar for the three diagnostic groups.The final model was built considering the 12 most relevant items selected in the first step analysis and showed optimal fit indices (stress-II = .59; D.A.F. = .94).Figure 1 shows the coordinates for the 12 selected items in a bidimensional solution. Dimension I (right-left) differentiate high and low emotional activation of NPS. Dimension II (top-down) distinguishes between high and low behavioral activation. The combination of both generates four quadrants, indicating symptoms of resistance (Q-I: low emotional and high behavioral activation), restlessness (Q-II: high emotional and behavioral activation), flattening (Q-III: low emotional and behavioral activation) and desolation (Q-IV: high emotional and low behavioral activation) (see Figure 1).ConclusionsThe results suggest that two dimensions underlie the most discriminant NPS included in scale (i.e., emotional and behavioral activation). These two dimensions seem to differentiate between four NPS states (resistance, restlessness, flattening, and desolation), which could be the most useful NPS in the determination of risk factors for predementia patients.