**4. Discussion**

The present data indicate that for adults with DS, variability in orientation, language, memory, visuospatial skills, praxis, mood, and social participation is largely explained by two underlying principal components. These two components seemed to differentiate cognitive from practical function (i.e., the ability to answer questions vs. the ability to carry out everyday tasks). Additionally, the use of the two-component model to categorize participants with respect to AD dementia status showed high classification accuracy. These findings support the further distillation of the modest-sized battery into a "short form" that can be easily administered in a primary care setting. Additionally, it takes less than an hour to administer to an informant who knows the patient with DS well, and requires minimal office space and test stimuli.

Results of the CART analysis also demonstrated that a small subset of the original battery—the cognitive subscale of the DLD (SCS) and the community subscale of the Vineland-II—were just as effective in classifying AD dementia status. However, the CART model exhibited better negative predictive value, in that fewer participants with dementia were misclassified as non-demented compared to the principal components model. A short battery based on the CART model is also quicker to administer and can in most cases be completed in less than 30 min.

A key finding is that the two contributory measures are not direct, objective measures of cognitive performance completed by the patient. Instead, they are informantbased scores of the patient's observed changes in cognitive abilities (DLD-SCS) and selfmanagement in community tasks (Vineland-II community). Unexpectedly, classification did not appreciably hinge on objective, performance-based neurocognitive measures. This highlights the critical component of informed caregiver ratings when screening for dementia in DS populations and provides some assurance that differential diagnosis of AD dementia is still possible when a patient's cognitive abilities cannot be directly assessed due to profound ID, limited cooperation, sensory impairments, or speech and language disorders.

Overall, the present data suggest that in clinical contexts with limited time and access to advanced training in test administration, the cognitive subscale of the DLD and community subscale of the Vineland-II, two widely available instruments, may suffice for screening and monitoring purposes. To be clear, we do not conclude that these two subscales constitute a comprehensive research or diagnostic battery, as definitive diagnosis should be based on longitudinal data. Nor is it the case that objective neurocognitive performance measures are redundant for diagnostic purposes. On the contrary, diagnostic criteria require objective neurocognitive assessment in order to make a firm diagnosis [33]. The present analysis was conducted for the specific aims of the study, namely identifying measures for resource-limited healthcare settings to encourage wide adoption of dementia screening among community DS practitioners. Prior efforts to use data reduction approaches to streamline a cognitive and behavioral battery for dementia in DS were focused primarily on developing a minimal comprehensive battery for research and specialty evaluation settings; thus, the resulting recommendations were not as relevant to primary care screening [38].

Furthermore, the present findings do not suggest that these two subscales represent an advancement in the early detection of AD dementia relative to more comprehensive test batteries. Instead, the benefit of adopting a minimal screening battery would enable more of the broader DS population to be evaluated, who may otherwise go unassessed. At the individual level, "early" detection is relative to the person's typical access to care, not the recommended standard of care. Given that nearly half of adults with DS do not receive regular screening for typical DS-associated health problems [11], it is reasonable to cast a wider net with "good enough" measures easily administered in primary care settings. Moreover, operating characteristics of the CART model align with a preference for high sensitivity (potential over-identification) over high specificity because the goal of screening is to provide support to this population.

Examination of the factor structure and of the item- and subscale-level operating characteristics in diagnostic batteries for dementia in DS is relatively new ground, and it is difficult to contextualize the present findings in the literature on the constructs measured. Broadly speaking, these data are in line with indications that adults with DS have reduced—but not absent—functional independence relative to other adults with intellectual disabilities [39], and dementia-related impairment in that domain may be captured by a community functioning measure such as the Vineland-II community subscale. Prior work using the Vineland-II to predict AD dementia in DS found that informant-rated receptive language skills, in addition to performance on a semantic verbal fluency task, were strong indicators of mild cognitive impairment in DS [40]. The present analysis instead examined individuals with and without AD dementia and found community management skills to be the most informative subscale of the Vineland-II. These findings are not contradictory, as in the present study it is likely that variability between participants cognitive functioning were captured by the DLD-SCS informant-based score, leaving more contextual community-based functioning to be best represented by the Vineland-II community subscale.

Beyond those discussed above, additional limitations of this study include the use of the SIB, DLD, and BPT along with the neurologic examination to determine consensus diagnosis. Our prior investigations have found that in 96% of cases, the final consensus diagnosis matched the neurologist's diagnosis that was formed independently of the SIB, BPT, and DLD scores. Still, discussion with the informant allows exposure to much of the same information captured by these instruments, and consideration of this information when forming a diagnosis is unavoidable. The eventual goal of both study cohorts is to substantiate consensus diagnoses with neuropathology at autopsy, allowing a more direct evaluation of the influence of potential criterion contamination. Additionally, the present study relied on informants who were very familiar with the participants with DS being rated, and in many cases, such a source of information cannot be found in practice.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/brainsci11091128/s1, Table S1: Principal Components Analysis Results.

**Author Contributions:** Conceptualization, J.P.H., L.M.K., K.L.V.P., C.L.H., E.D., E.H., I.T.L. and F.A.S.; methodology, J.P.H., L.M.K., K.L.V.P., C.L.H., E.H. and F.A.S.; software, K.L.V.P.; validation, K.L.V.P.; formal analysis, K.L.V.P.; investigation, J.P.H., L.M.K., K.L.V.P., C.L.H., E.D., E.H., I.T.L. and F.A.S.; resources, F.A.S. and E.H.; data curation, K.L.V.P. and E.D.; writing—original draft preparation, J.P.H., K.L.V.P. and L.M.K.; writing—review and editing, J.P.H., L.M.K., K.L.V.P., C.L.H., E.D., E.H., I.T.L. and F.A.S.; visualization, K.L.V.P.; supervision, J.P.H., L.M.K., K.L.V.P., C.L.H., E.D., E.H., I.T.L. and F.A.S.; project administration, J.P.H., L.M.K., K.L.V.P., C.L.H., E.D., E.H., I.T.L. and F.A.S.; funding acquisition, J.P.H., L.M.K., K.L.V.P., C.L.H., E.D., E.H., I.T.L. and F.A.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Institute on Aging, grant numbers P30 AG028383, 1T32AG057461, AG-21912, P50-16573, P30AG066519, and U01 AG051412, and the National Institute of Child Health and Human Development, grant number HDR01064993.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Boards of University of California, Irvine (HS#2002-2796; 08/07/2007) and University of Kentucky (#61423 2019; 09/03/2020).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are openly available in OSFHome at [DOI 10.17605/OSF.IO/EK3YH].

**Acknowledgments:** We thank the participants and their families/caregivers for participating in this study. We are grateful to the University of California, Irvine Institute for Clinical and Translational Science (UL1 RR031985) for providing resources in support of this project. The authors would like to thank Stacey Brothers, Katie McCarty, Roberta Davis, Allison Caban-Holt, and Amelia Anderson-Mooney for their assistance with data collection. We also want to thank our team of neurologists: Gregory Jicha, Donita Lightner, and William Robertson for their clinical expertise.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
