Géraud Dautzenberg

Summary and general discussion

In addition to ‘demonstrating’ the methodological shortcomings of other studies, it is important to reflect on the shortcomings of our own study. The goal of the studies in this section is to validate theMoCA for specific conditions that exist in geriatric psychiatry. Different conditions affect the reliability of a test because the population being tested is different. Of course, the most striking feature here is the advanced age of the referred patients (compared to general psychiatry). Frequent psychiatric complaints as a reason for referral and disorders are also a distinctive feature (versus non-psychiatric settings). In addition, widespread cognitive impairment is a factor that affects outcomes (prevalence). This last fact ensures that even within geriatric psychiatry, the population may also vary, and thus, it depends on the time of testing in the diagnostic process. Whether you test everyone, regardless of the referral reason, or whether you test only patients who are suspected of having cognitive symptoms after the initial interview, affects the population composition and thereby, the accuracy of the test. Therefore, we also examined the MoCA in two common settings or ‘the moment of assessment’ that occurs in old age psychiatry; in everyone referred to old age psychiatry (screening), or only in a population that was suspected of having cognitive problems (triaging). A methodological consideration was the creation of these two ‘different’ settings. One could argue that the second setting (triaging) was a simulation of the clinical memory setting. In retrospect, we selected patients who were referred to our memory clinic from the cohort of the first setting using patient records. Nevertheless, we think that this is a valid and efficient method, as the only inclusion criterion added was whether the patient was being referred to our memory clinic. By creating these settings, we attempted to make the study population resemble the clinical reality, but it will never match exactly. The study conditions will be (intentionally) rigidly defined through exclusion and inclusion criteria to clarify who is involved and increase comparability across groups. In addition, this is often used for eliminating other influences, for example, in many studies with the MoCA, psychiatric comorbidity was excluded. Clinical practice, on the contrary, is often fluid. This fact also reveals a limitation of our study. Pragmatismalso plays a role in the choice of cohort study design based on patients record research. If treatment, as usual, was to be deviated from and non-suspect patients had been offered a more extensive assessment, this had to be done on a voluntary basis. This would create two groups: those who participate in this follow-up study and those who do not. In addition, there would be a realistic chance of dropouts. There would also be unintended selection, and thus, a risk of bias. This is not often the case with a cohort design, but this also has disadvantages, including the previously mentioned rigid in- or out-classification by the total MoCA score without nuances. Again, as with using healthy

7

163