Géraud Dautzenberg

Chapter 7 158 7.1.3 Main findings of Section B: The MoCA validation in different old age psychiatry settings The results are described in detail in Chapters 3 and 4. First, if we compare our data with the (original) literature, using healthy controls as the reference group, excellent accuracy was found in our study, corresponding to the results described in the original study and other studies with a case-control design (mocatest.org; Davis et al., 2013; O’Caoimh, Timmons and Molloy, 2016; Carson, Leach and Murphy, 2018). At the cut-off based on the original study (MoCA<26), the sensitivity and specificity are high (95%) and good (73%), and it is comparable to the original study and many other studies with healthy volunteers as control subjects. For the more likely situation of determining whether the MoCA can be discriminatory among all referred patients to geriatric psychiatry who have and do not have cognitive impairment, the results are less favourable than in the originally reported study and case-control studies. The area under the curve (AUC) drops from excellent accuracy (0.93) to a fair to good accuracy (0.77) (Fischer, Bachmann and Jaeschke, 2003). Logically, the sensitivity remains the same and is excellent (> 95%). However, the specificity drops significantly to a questionable level to even below 40% at a cut-off of <26. If we further narrow down or concentrate the population to only referred patients in whom a cognitive disorder is suspected after baseline assessment, the patient groups to be analysed may increasingly resemble each other and it will become increasingly difficult for a test to distinguish the different aetiologies in this group. Again, the total MoCA scores were significantly different between the different study groups and were similar to those in the literature and our study on all referred patients. The AUC remained good, and the specificity remained similar for all referred patients and of moderate level; 47% at MoCA <26, rising to 73% at a MoCA score of <21. With these results, we conclude that of all the referred patients with MCI and mild dementia 95% have a MoCA of <26 (sensitivity), which is meaningful for clinical practice. The optimal cut-off values for dementia are <21 and for MCI <26. When a patient in the total referred patient group, that is, screening, has a negative MoCA (meaning a value of ≥ 21), it can be stated with 98% certainty that this person has no dementia (NPV). An MCI can be ruled out with 94% (NPV) certainty with a score of 26 or higher in this group (with this prevalence). In the cohort of suspected patients (i.e. triaging), 90% (PPV) of thosewith anMoCA score of <21will have cognitive impairment (MD+MCI), while 94% (NPV) with an MoCA score of ≥ 21 will not have dementia. This allows for a significant reduction (50%) in referrals in old-age psychiatric care through the MoCA by selecting those who do not need further referral to a memory clinic, even if they were suspected of cognitive impairment after the initial assessment. The PPV was too low for both situations to confirm a diagnosis (both MD and MCI) using only a MoCA in an old-age psychiatric setting.

RkJQdWJsaXNoZXIy MTk4NDMw