The Cambridge Cognitive Examination for Down's Syndrome (CAMCOG-DS) was developed to assess cognitive functioning and dementia-related cognitive decline in people with Down's Syndrome (DS). It has been translated into different languages and is often used in international studies. Although adapted for people with intellectual disabilities (ID), many tasks involve verbal responses and instructions are presented orally. Therefore, the administration for people with severe language deficits can be challenging. The aim of this retrospective data analysis is to examine the influence of language ability and reasoning on CAMCOG-DS performance. Study 1 examined the relationship between CAMCOG-DS, picture naming, single word comprehension and reasoning in adults with DS. Study 2 replicates and broadens the findings in a sample of children and adults with DS.

Study 1 included 40 adults with DS between 18 and 51 years (M = 28.6, SD = 8.4). 25 had a mild and 15 a moderate ID. CAMCOG-DS, the short form of the Boston Naming test (BNT), a test for single word comprehension from the Werdenfelser Testbatterie (WTB) and the Colored Progressive Matrices (CPM) were administered. Study 2 included 38 participants between 8 and 59 years (23 children, M = 11.4; 15 adults; M = 31.3). 3 had a borderline, 23 a mild, and 12 a moderate ID. The same tasks as in Study 1 were applied, but the CPM was replaced by its successor, the Raven's 2.

In Study 1, participants with mild ID performed better in all tasks than those with moderate ID (p < .05). Moderate relationships were found between CAMCOG-DS total score and the language tasks (r = .56 and r = .46), which remained significant when level of ID was controlled for. There was also a moderate relationship between CAMCOG-DS and reasoning (r = .46). Regression analysis showed that BNT performance predicted CAMCOG-DS performance (R2 = .31). In Study 2, those with mild ID, compared to those with moderate ID, performed better in all tasks (p < .05), however, regarding the CAMCOG-DS and language tasks, this effect was larger in adults than in children. Adults performed better than children in the CAMCOG-DS and BNT (p < .05), but not in single word comprehension or reasoning. As in Study 1, moderate to large correlations were revealed between CAMCOG-DS and language tasks and between CAMCOG-DS and reasoning (r > .52), remaining significant when age and ID level were controlled for. Regression analysis showed that both naming and reasoning but not single word comprehension or age predicted CAMCOG-DS performance (R2 = .69), however, performance was best predicted by naming (R2 = .65).

Our results suggest that language ability and reasoning relate to CAMCOG-DS performance, which is best predicted by BNT picture naming. This should be considered in CAMCOG-DS interpretation, as the capabilities of patients with lesser language ability might be underestimated. Future developments of dementia assessments for people with ID should include more nonverbal tasks.