Cognitive domains and their assessment
When describing the cognitive profile in PD-D there are at least two implicit assumptions that should be elaborated: (1) the existence of distinct cognitive functions or domains, and (2) that cognitive functions may be quantified and compared between groups of individuals. Assumption (1) has been extensively discussed in the cognitive neuropsychology literature. Although most researchers would probably agree that cognitive domains are separable, there is still some disagreement about the interrelationship between them, and the terminology may be inconsistent and confusing between studies.
For instance, ‘executive functions’ is a term for processes that regulate all goal-directed behaviour [6]. This implies that these ‘executive functions’ should exert a global effect on tests that were designed to assess other cognitive domains if those tests require goal-directed behaviour, such as planning, monitoring, effort, and flexibility. This is problematic, however, considering that the ‘double dissociation’ method has often been used in order to analytically identify independent cognitive functions [7]. According to this method, it should be demonstrated that the hypothetical cognitive function A could be impaired in some patients while cognitive function B could be intact, and there should be patients in whom function B is impaired while function A is intact. However, as executive and attentional deficits, by definition, should have modulatory effects on a wide range of cognitive functions, the ‘double dissociation’ method does not always apply. Rather, it has been demonstrated that brain pathology in fronto-subcortical areas leads to a global deficit in all types of tests that require controlled attention. This is of particular relevance to PD-D, as it has been proposed that cognitive deficits in PD, in general, are caused by disruption of fronto-subcortical loops [3]. In the dual -syndrome hypothesis of Kehagia and colleagues [2], it is proposed that such a fronto-striatal dysfunction mediates a dysexecutive syndrome seen in PD without dementia, while PD-D is characterized by more widespread memory and visuospatial deficits.
Another important issue concerns phenomena such as alertness and arousal. In DLB and PD-D, attentional ‘fluctuation’ is a feature of the cognitive profile. Such fluctuation could possibly be related to changes in arousal or alertness, this could exert a global effect on other cognitive and behavioural functions [8]. Thus, when interpreting patterns of cognitive dysfunction in PD-D one should be mindful of the possible influence of executive dysfunction and of reduced arousal/ alertness. I will return to this issue later and discuss the term ‘bradyphrenia’ as related to deficits in arousal and alertness.
Assumption (2) (i.e. that it is possible to quantify and compare cognitive functions across groups) is more straightforward. Cognitive measures are constructed according to statistical psychometric principles and can be subjected to relative comparisons between groups. The fundamental ‘gold standard’ for describing a cognitive profile in a patient group is comparison with a healthy control group matching the patients on background variables that are known to affect cognition, such as age, gender, and education. Further, cognitive comparisons can be carried between patient groups.
Average effect size: a meta-analytic approach
The simplest way of comparing cognitive functions between different groups is to simply arrange them ordinally (e.g. group A are worse than group B), as in qualitative reviews. This strategy was employed in two reviews on cognitive functions in PD-D and DLB [9, 10]. The problem with this is that it is very difficult to draw an overall conclusion regarding the literature as a whole. Another method is to compare standardized effect sizes to identify differences in standardized group mean values (Cohen’s d), a strategy used in quantitative meta-analysis. In this strategy, the differences of the means are divided by the pooled standard deviations of the groups, enabling a comparison of the magnitude of differences in various cognitive domains, in this case between PD-D and the comparison groups. In the present review this will be done for studies on cognition in PD-D that reported means, standard deviations, and group sizes. Other studies will be described in a qualitative manner. The cognitive profile of patients with PD-D as compared with healthy control subjects (HC) will be described, representing the cognitive deficits in PD-D relative to healthy ageing. Further, the cognitive profile of PD-D compared with other neurodegenerative dementias, i.e. PD-D versus Alzheimer’s disease (AD), and PD-D versus DLB, will be presented. A fixed-effects meta-analysis of the average standardized difference between DLB and PD-D on attention and executive functions, memory, and visuospatial functions will also be presented, as the distinction between DLB and PD-D remains contested.
