Journal of Alzheimer’s Disease 81 (2021) 1505–1527 1505 DOI 10.3233/JAD-201427 IOS Press The Behavioral and Psychological Symptoms of Dementia in Down Syndrome Scale (BPSD-DS II): Optimization and Further Validation1 Alain D. Dekkera,b,∗ , Aurora M. Ulgiatia,b , Henk Groenc , Vincent A. Boxelaard , Silvia Saccoe , Ségolène Falqueroe , Angelo Carfif , Antonella di Paolaf , Bessy Benejamg , Silvia Valldeneuh,i , Roelie Fopmab , Marjo Oosterikj , Marloes Hermelinkk , Gonny Beugelsdijkl , Mieke Schippersl , Hepie Henstram , Martine Scholten-Kuipern , Judith Willink-Voso , Lisa de Ruiterp , Liesbeth Willemsq , Anneke Loonstra-de Jongr , Antonia M.W. Coppuss,t , Marleen Tollenaereu,v , Juan Forteag,h,i , Graziano Onderw , Anne-Sophie Rebillate , Debby Van Dama,u and Peter P. De Deyna,u,v,∗ a Department of Neurology and Alzheimer Center, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands b Department of Practice-oriented Scientific Research (PWO), Alliade Care Group, Heerenveen, The Netherlands c Department of Epidemiology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands d Center for Information Technology, University of Groningen, Groningen, The Netherlands e Institut Jérôme Lejeune, Paris, France f Department of Geriatrics, Fondazione Policlinico Universitario Gemelli IRCCS, Rome, Italy g Barcelona Down Medical Center, Fundació Catalana Sı́ndrome de Down, Barcelona, Spain h Memory Unit and Biomedical Research Institute Sant Pau (IIB Sant Pau), Department of Neurology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain i Centro de Investigación Biomédica en Red en Enfermedades Neurodegenerativas (CIBERNED), Madrid, Spain j Aveleijn, Borne, The Netherlands k De Twentse Zorgcentra, Enschede, The Netherlands l Ipse de Bruggen, Nieuwveen/Nootdorp, The Netherlands m Nieuw Woelwijck, Sappemeer, The Netherlands n Philadelphia Zorg, Amersfoort, The Netherlands o Severinus, Veldhoven, The Netherlands p Sherpa, Baarn, The Netherlands q Sprank, Zwolle, The Netherlands r Vanboeijen, Assen, The Netherlands s Department of Primary and Community Care, Radboud University Medical Center, Nijmegen, The Netherlands t Dichterbij, Gennep, The Netherlands u Laboratory of Neurochemistry and Behavior, Department of Biomedical Sciences and Institute Born-Bunge, University of Antwerp, Antwerp, Belgium v Department of Neurology and Memory Clinic, Hospital Network Antwerp (ZNA) Middelheim and Hoge Beuken, Antwerp, Belgium w Department of Cardiovascular, Endocrine-metabolic Diseases and Aging, Istituto Superiore di Sanitá, Rome, Italy ∗ Correspondence to: Alain D. Dekker and Peter P. De Deyn, 1 This article received a correction notice (Erratum) with the Department of Neurology, University Medical Center Gronin- reference: 10.3233/JAD-219007, available at https://content.ios gen, P.O. Box 30.001, 9700 RB Groningen, The Netherlands. press.com/articles/journal-of-alzheimers-disease/jad219007. E-mails: a.d.dekker@umcg.nl; p.p.de.deyn@umcg.nl. ISSN 1387-2877 © 2021 – The authors. Published by IOS Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (CC BY-NC 4.0). 1506 A.D. Dekker et al. / BPSD in Down Syndrome Handling Associate Editor: Elizabeth Head Accepted 29 March 2021 Pre-press 4 May 2021 Abstract. Background: People with Down syndrome (DS) are at high risk to develop Alzheimer’s disease dementia (AD). Behavioral and psychological symptoms of dementia (BPSD) are common and may also serve as early signals for dementia. However, comprehensive evaluation scales for BPSD, adapted to DS, are lacking. Therefore, we previously developed the BPSD-DS scale to identify behavioral changes between the last six months and pre-existing life-long characteristic behavior. Objective: To optimize and further study the scale (discriminative ability and reliability) in a large representative DS study population. Methods: Optimization was based on item irrelevance and clinical experiences obtained in the initial study. Using the shortened and refined BPSD-DS II, informant interviews were conducted to evaluate 524 individuals with DS grouped according to dementia status: no dementia (DS, N = 292), questionable dementia (DS + Q, N = 119), and clinically diagnosed dementia (DS + AD, N = 113). Results: Comparing item change scores between groups revealed prominent changes in frequency and severity for anxious, sleep-related, irritable, restless/stereotypic, apathetic, depressive, and eating/drinking behavior. For most items, the proportion of individuals displaying an increased frequency was highest in DS + AD, intermediate in DS + Q, and lowest in DS. For various items within sections about anxious, sleep-related, irritable, apathetic, and depressive behaviors, the proportion of individuals showing an increased frequency was already substantial in DS + Q, suggesting that these changes may serve as early signals of AD in DS. Reliability data were promising. Conclusion: The optimized scale yields largely similar results as obtained with the initial version. Systematically evaluating BPSD in DS may increase understanding of changes among caregivers and (timely) adaptation of care/treatment. Keywords: Alzheimer’s disease, behavior, behavioral and psychological symptoms of dementia, dementia, Down syndrome, intellectual disabilities, neuropsychiatric symptoms, trisomy 21 INTRODUCTION at some point during their disease course [9, 10]. Various contributing factors have been identified to Down syndrome (DS, trisomy 21) is the most fre- explain the heterogeneity of these symptoms, among quent genetic cause of intellectual disability (ID), others, factors relating to the person with dementia, prevalent in approximately 1 in 900 live births [1, 2]. factors related to caregivers and environmental fac- People with DS are at high risk to develop dementia tors, and the interaction between these factors [11]. due to Alzheimer’s disease (AD). Despite substantial In the general population, BPSD are associated with variation between studies, prevalences rise strongly a reduced quality of life and earlier institutionaliza- from age 40 [3]. Indeed, by that age, virtually all peo- tion for people with dementia, and increased burden ple with DS have extensive AD-like pathology in the for caregivers [8, 12]. Additionally, BPSD are a key brain [4]. Nevertheless, onset of clinical symptoms reason for referral to specialist services [13]. varies substantially in time [5, 6]. Consequently, pre- Professional caregivers of people with DS + AD dicting and monitoring decline and onset of dementia found it particularly difficult to respond to the unpre- is a diagnostic challenge and of essence in daily care dictability of behavioral changes [14] and might and support for people with DS [7]. not seek intervention until the behavior becomes Behavioral alterations are very common in AD more difficult to manage [15, 16]. Moreover, changes in addition to decline in cognitive and functional may be perceived as disability-specific instead of skills. These so-called behavioral and psychological related to dementia (i.e., diagnostic overshadow- symptoms of dementia (BPSD), or neuropsychiatric ing) [14, 17]. Therefore, systematic evaluation of symptoms, are defined as “a heterogeneous range of BPSD is important to increase awareness, acceptance psychological reactions, psychiatric symptoms, and and understanding among family members and care behaviors resulting from the presence of dementia” professionals, which in turn may contribute to adapt- [8]. Nearly all people with dementia in the general ing care/support [15, 18]. Furthermore, BPSD are population experience at least one BPSD symptom reported in prodromal and early stages of dementia A.D. Dekker et al. / BPSD in Down Syndrome 1507 and might, as such, serve as early ‘alarm signals’ [19]. Translation For daily practice, anticipating the development of symptoms enables among others timely adaptation The optimized Dutch version (source document) of daily support, work and day-care and the living was forward and back translated into French, Italian, environment [20]. Importantly, BPSD can (partially) and Spanish by a certified professional translation be treated, either non-pharmacologically or pharma- company (DBF Communicatie B.V., Alphen aan cologically [17, 21–26]. den Rijn, The Netherlands). Translations were per- In the general population, BPSD are regularly formed according to the International Organization evaluated using validated scales such as the Neuro- for Standardization (ISO) standard for translation psychiatric Inventory [27, 28]. However, such scales services (ISO 17100:2015) and the company’s stan- are not adapted/validated for people with ID, includ- dard operating procedures based on this standard. ing DS. Scales commonly used to screen for or Firstly, the BPSD-DS II was translated into the target diagnose dementia in people with ID have a more languages by certified, experienced native-speaker limited focus on behavioral alterations [29]. Conse- translators and subsequently checked by experienced quently, BPSD have not been thoroughly evaluated in revisers. Next, back translations into Dutch were per- this population. Therefore, we developed the Behav- formed by different translators than those involved ioral and Psychological Symptoms of Dementia in in the original forward translation. Together with Down Syndrome (BPSD-DS) scale to evaluate behav- medical/psychological experts of various consortium ioral changes by comparing frequency and severity partners (see below), translations were additionally of behavioral items in the last six months with char- checked and refined. Checks for consistency were acteristic/typical behavior before decline occurred. performed before and after refinement. The manual Initial development of this systematic informant was only translated from Dutch to English, given the interview has been previously described in Journal fact that the interviewers were proficient in English. of Alzheimer’s Disease [30]. The initial version con- sisted of 83 behavioral items in 12 clinically defined Digitization sections. In 281 individuals with DS without demen- tia (DS), with questionable dementia (DS + Q) and To facilitate administration, improve data quality with dementia (DS + AD), behavioral changes were (completeness of entered data, guidance of interview- identified with promising results in terms of first ers), and facilitate data processing, a digital version of validity and reliability data [30]. Based on results and the optimized scale was developed in various rounds experiences obtained in the initial study, we aimed to of optimization and problem-solving. An online ver- optimize the scale (among others by shortening and sion of the BPSD-DS II was built and customized refining it for applicability in daily practice) and to in the web application Limesurvey. The system has further study this optimized version (including relia- been hosted and managed in a secured in-house IT bility and discriminative ability, such as confirmation environment at the Center for Information Tech- of previous results about item relevance) in a large nology of the University of Groningen (RUG) in representative DS study population. accordance with privacy and security policy, the Acceptable Use policy, ethical guidelines and the METHODS EU General Data Protection Regulation. The sys- tem provided a step-by-step guide for interviewers, Scale optimization automatically calculated change scores and enabled downloading PDF-reports after completion. A pseu- The initial BPSD-DS scale (83 items) was short- donymized SPSS-export of this database constituted ened based on item irrelevance in combination with the basis for analyses. multidisciplinary experiences gained in the initial study. An item was considered irrelevant if ≥ 85% of Multidisciplinary consortium DS + Q and DS + AD individuals showed unaltered scores (change = 0) for both frequency and severity, Expanding on the previously established consor- i.e., a lack of clinically relevant change in relation tium [30], a total of 17 Dutch care institutions and to dementia [30]. After shortening, the scale was 4 European expertise centers took part in the study. textually refined, and the manual was improved con- This broad, multidisciplinary consortium enabled the sidering practical experiences in the initial study. study of a large, representative study population of 1508 A.D. Dekker et al. / BPSD in Down Syndrome people with DS in a daily clinical practice setting. three aspects: 1) manageability of symptoms, 2) addi- The following Dutch care institutions participated, tional time required, and 3) emotional burden. Care providing care, diagnostics and therapy to people burden was scored on a four-point scale: 0 = none, with ID/DS throughout nearly the entire country: 1 = slight, 2 = moderate, or 3 = serious. Per section, Amerpoort, Aveleijn, Cosis, De Twentse Zorgcentra, the resulting care burden change (score for sub-item Dichterbij, Elver, Ipse de Bruggen, Nieuw Woelwijck, (a) – score for sub-item (b)) ranged from –3 to + 3. Philadelphia, Reinaerde, ’s Heeren Loo, Severinus, Sherpa, Sprank, Talant (part of Alliade Care Group), Interviewers Vanboeijen, and Zuidwester. In addition, University of Antwerp and its Flemish network of care institu- The BPSD-DS II was completed by conducting tions (Belgium), Institut Jérôme Lejeune (France), interviews with informants according to a detailed Policlinico Gemelli (Italy), and Hospital de la Santa manual. Interviews were performed by experienced Creu i Sant Pau (Spain) participated in administering clinicians, ID (neuro)psychologists (including behav- the scale. ioral therapists who studied special needs education (in Dutch: orthopedagogiek)), psychological assis- Scoring tants or researchers. Interviewers all received an instruction session about the procedure and the digi- Frequency and severity were scored for each scale tal system and were able to practice with the system item. To identify behavioral changes over time and in advance. Interviewers adhered to the manual and account for pre-existing behavior, frequency and the sequence of sections and items. severity were scored for two periods of time: (a) last six months and (b) typical/characteristic behav- Informants ior before any deterioration occurred. Frequency was scored on a five-point scale: 0 = never or once only, Interviews were conducted with at least one key 1 = less than once a month, 2 = monthly, not weekly, informant of the person with DS, such as a caregiver 3 = weekly, not daily, or 4 = daily or continuously. The working in a day-care center/residential facility or a resulting frequency change score (score for sub-item family member. Informants had to be able to provide (a) – score for sub-item (b)) is a measure of behav- an accurate description of the behavior in the last six ioral change over time and ranged from –4 to + 4. months as well as the typical/characteristic behav- Severity was considered from the perspective of the ior in the past before decline occurred. Additional person with DS and based on two aspects: personal key informants were recruited, especially in cases suffering and degree of impact on daily life. Severity where a single informant could not provide answers was scored on a four-point scale: 0 = none, 1 = slight, to all questions and both time periods. If informants 2 = moderate, or 3 = serious. The resulting severity were not able to describe the typical/characteristic change score (score for sub-item (a) – score for sub- behavior and this became apparent during the inter- item (b)) is a measure of behavioral change over time view, this resulted in exclusion. In case of multiple and ranged from –3 to + 3. informants, they were interviewed in a single ses- Depending on the residential circumstances, infor- sion (not separately). Interviews were conducted in mant(s) may not always be aware of the person’s absence of the person with DS. Prior to the inter- sleep behavior. The answer option ‘unknown’ (?) view, set-up and scoring system were explained to the was therefore provided to items in section 2 about informant(s). In case of disagreement between infor- sleep problems. In addition, for some items the mants, consensus on the score was reached during the answer option ‘not applicable’ (N/A) was provided. interview. If an informant did not understand an item, Depending on the person’s physical disability and/or the interviewer provided clarification. If informants freedom-restricting measures (items 1.2, 1.4, 2.3, 5.2, evidently exaggerated or trivialized symptoms, the and 5.4, see Results section) and verbal (in)abilities interviewer addressed this. Informants were asked to (items 3.3, 5.5, 6.1, and 10.2, see Results section), the give short, succinct answers, and were reminded to do interviewer could select the answer ‘not applicable’ so if they gave long-winded or anecdotal responses. if a symptom could not occur. For each person with DS, the BPSD-DS II was admin- Finally, care(giver) burden was evaluated in each istered once-only, apart from a number for individuals section from the perspective of caregivers/family for whom a retest interview was conducted (test-retest members. Care burden score per section was based on reliability). A.D. Dekker et al. / BPSD in Down Syndrome 1509 Ethics exclusion if the person functioned normally (e.g., due to effective treatment). ID level, medical conditions, The Medical Ethics Review Board of the Univer- and medication use were based on (medical) records, sity Medical Center Groningen (UMCG) evaluated and if necessary, inquiries with involved care insti- the study protocol (no. 2018/220) and concluded that tutions, clinicians, and ID (neuro)psychologist. After the Dutch Medical Research Human Subjects Act did selection, an information letter with consent form was not apply. The study was registered in the UMCG sent. Except for a few individuals with DS capable to Research Register (no. 201800252) and compliant provide consent themselves (adapted informed con- with the EU General Data Protection Regulation sent form with pictograms), written informed consent and standards for medical research in humans rec- was generally obtained from legal representatives ommended by the Declaration of Helsinki. Local (proxy consent). Consent was provided for evaluation institutional review committees gave their approval, of behavioral changes using the BPSD-DS II scale and whenever applicable. In Flanders, Belgium, the Insti- processing/analysis of the coded data for this study. tutional Review Boards of the Hospital Network Antwerp (no. 5058) and the University Hospital Dementia diagnosis Antwerp/University of Antwerp (no. 17/50/566) app- roved the study. In France, the study was authorized Three diagnostic study groups were distinguished under research standard MR-004 by the Commis- in this cross-sectional study: 1) DS without demen- sion Nationale de l’Informatique et des Libertés (no. tia (DS), 2) DS with questionable dementia (DS + Q), 2214487 v 0). In Italy, the study was approved by i.e., (slight) deterioration that is suggestive of demen- the Ethical Committee of the Universitá Cattolica tia, but does not (yet) clearly meet the diagnostic del Sacro Cuore (no. 2731520). In Spain, the study criteria, and 3) DS with a clinical diagnosis of was approved by the Sant Pau Ethics Committee and dementia (DS + AD). People with DS were assigned reported to the Minister of Justice according to the to one of the three study groups on the basis of Spanish law for research in people with ID. expert clinical judgement by clinicians and/or ID (neuro)psychologists at the participating care institu- Study population, recruitment, and consent tions. Clinical diagnosis of dementia in people with DS is valid and reliable [16, 31]. This judgement was To ensure a representative study population, par- generally based on routine multidisciplinary clini- ticipants were recruited through the aforementioned cal evaluation, informant interview(s), information care institutions (various backgrounds, regions, living from medical records, and general dementia crite- situations) based on inclusion and exclusion cri- ria [32, 33]. People with DS were not subjected to teria. Inclusion criteria: phenotypical diagnosis of new dementia assessments. The diagnosis, and thus DS, aged ≥ 30 years, and a stable dose of psy- assignment to the three study groups, was established choactive medications (if any). Exclusion criteria: in advance and not based on outcomes of the BPSD- profound ID, long-term admission to hospital in the DS II. past six months, bed-ridden or in terminal care (e.g., end-stage dementia), presence of a confirmed cere- Validity brovascular accident, and absence of at least one informant able to describe the person’s behavior in This study builds further on the initial develop- the past six months and typical/characteristic behav- ment process in which face and content validity of ior in the past. People who faced a recent life event, the BPSD-DS were ensured by deducing items from e.g., moving home or death of a family member, with literature (review provided in [29]), multidisciplinary continued impact on behavior were excluded, accord- experiences in daily practice and consecutive feed- ing to clinical judgement. Furthermore, individuals back rounds with experts from different disciplines were excluded who presented behavioral changes [30]. Validity was strengthened here by substan- that, according to clinical judgement, were due to tially optimizing the scale (BPSD-DS II) based on another condition (comorbidity), if diagnosed, for test results and multidisciplinary clinical feedback example (un)treated depression, epilepsy, hypothy- obtained in the initial study. Like in the initial study, roidism, vitamin B12 deficiency, hearing problems, concurrent validity could not be established due to vision problems, sleep apnea, and chronic pain. A the lack of a (gold) standard instrument for behav- diagnosis of such a comorbidity per se did not result in ioral changes in DS or ID to compare the BPSD-DS 1510 A.D. Dekker et al. / BPSD in Down Syndrome II with. Finally, discriminative ability of the BPSD- within 4 weeks. However, due to practical difficul- DS II in relation to the status of dementia was assessed ties, this was extended in the course of the study. IRR (see below). and TRR may be calculated using percent agreement, correlations or Cohen’s kappa [35]. Because of the Discriminative ability categorical nature of the item scores in the BPSD-DS II, we have used percent agreement instead of correla- In the context of discriminative ability, 1) item tion, since correlation is the recommended measure (ir)relevance, 2) total scale scores, and 3) sensitivity, of agreement for results on a continuous scale. We specificity and predictive values were analyzed. First did not use Cohen’s kappa, because this measure is of all, we aimed to confirm relevance of behavioral affected by the prevalence or, as in our case, clustering items in the optimized BPSD-DS II, that is, to what of agreement in one of the categories, and is therefore extent did behavioral items change in relation to the less reliable for items where change is rare and both status of dementia. Frequency change and severity assessments agree on this. For items with low varia- change were compared between the three groups for tions in the change scores, low values of kappa may every individual item. Resembling the initial study not necessarily reflect low rates of overall agreement [30], an item was considered irrelevant if ≥ 85% of [36]. Although the total scale score could be regarded DS + Q and DS + AD individuals showed unaltered as continuous, we also applied percent agreement scores (change = 0) for both frequency change and here to be able to evaluate reliability both for per- severity change (cross-sectional data), i.e., a lack of fect agreement as well as with margins of plus or clinically relevant change in relation to dementia. minus 1, 2, or 3 points. Moreover, percent agreement That is, applicability/implementation of the scale in is considered a safe reliability measure in situations daily clinical practice would be hampered if specific where raters (interviewers here) are well trained and items need to be scored (at the cost of time), while guessing the answer is unlikely [35]. Finally, internal such items only change for a very limited number consistency was determined per item and for the total of individuals in relation to dementia. Secondly, fre- scale by calculating Cronbach’s alphas for frequency quency change and severity change scores for the change and severity change scores. total scale were calculated as the sum of individ- ual item change scores for frequency and severity, Data processing and quality control respectively. To evaluate discriminative ability of the BPSD-DS II, sum scores for the total scale were Each completed scale, including those administe- compared between the three groups. Thirdly, we red for reliability testing, was thoroughly checked for calculated sensitivity, specificity, positive predictive any lack of clarity, missing data, inclusion/exclusion values, and negative predictive values [34] for dif- criteria, and compliance with the instructed method ferent cut-off (total scale) scores on the BPSD-DS and rules described in the manual, including rules II. ROC analyses were performed using total scale regarding the interviewer, informants and scoring frequency change scores (all items). (see above). In particular, rules concerning the possi- bility to answer ‘not applicable’ were double checked Reliability and scores adapted, if necessary. That is, ‘not appli- cable’ was only allowed if a person was not verbal, Reliability was studied by evaluating interrater physically impaired, or freedom-limiting measures reliability (IRR) and test-retest reliability (TRR) for were in place. If required, the involved interview- a subset of individuals with DS. For IRR, the same ers were consulted, and issues were solved through interview was scored by two interviewers blinded consensus. to each other’s scores. For this purpose, the first interviewer conducted the interview, and a second interviewer was present as a ‘fly on the wall’ (not Statistics involved in the interview). IRR assessments were performed in various combinations of interviewers For population characteristics (Table 1), Pearson’s in multiple centers. For TRR, a second interview by chi-squared tests were applied to compare categori- the same interviewer and with the same informant(s) cal data between groups. ANOVA tests were used to was conducted within 1–7 weeks after the first inter- compare normally distributed continuous data (age view. Originally, we aimed to have a retest conducted and IQ-scores) between groups. A.D. Dekker et al. / BPSD in Down Syndrome 1511 In the context of discriminative ability, anal- severity change as dependent variable. The effect of ysis focused on item (ir)relevance of individual sex (male/female) and depression (presence/absence) items. ‘Unknown’ and ‘not applicable’ answers were were studied using Mann-Whitney U tests on the treated as ‘missing values’. To compare individual entire study group of 524 participants (regardless of item scores (frequency change and severity change) dementia status). Finally, the effect of ID level was between the three groups, Kruskal-Wallis tests were studied on the entire study group comparing total used. Statistical analysis was conducted using orig- scale scores (one-way ANOVA) between groups with inal underlying frequency change scores (–4 to + 4), mild, moderate ID, and severe ID. severity change scores (–3 to + 3), and care bur- p-values below 0.05 were regarded statistically sig- den change scores (–3 to + 3). In Figs. 3–9 and nificant. However, for analyses of frequency change Supplementary Material Figures 1–6, however, a and severity change per item and care burden change simplified graphical representation of changes is per section, the level of significance was adjusted provided in which the changes were simplified to according to the procedure described by Benjamini ‘decrease’, ‘unaltered’, and ‘increase’. Secondly, fre- & Hochberg [37] to account for multiple testing. quency change and severity change scores for the After applying this procedure original p-values below total scale were calculated as the sum of individual 0.020 were regarded statistically significant. Data item change scores for frequency and severity, respec- were analyzed using SPSS version 26 (IBM Corp.) tively, and were compared between groups using and Stata/SE version 14.2 (StataCorp LP). Kruskal-Wallis tests. Thirdly, using total scale fre- quency change scores (all items), ROC analyses were performed. In addition to sensitivity and specificity, RESULTS positive and negative predictive values were calcu- lated as well [34]. Scale optimization Concerning reliability, IRR and TRR were cal- culated as percent agreement for frequency change Based on prior results and clinical experiences scores and severity change scores per item and for obtained in the initial study [30], the BPSD-DS scale the total scale. Total scale scores for frequency change was optimized. A total of 23 items was consid- could range from –208 to + 208 points and for sever- ered irrelevant, including a large number of items ity change from –156 to + 156 points. Given the fact relating to delusions, hallucinations, and disinhib- that the BPSD-DS II consists of 52 items and the ited behavior. These items were removed with the fact that scoring behavior is not absolute, IRR and exception for the item about pica, which was main- TRR were determined for perfect agreement (identi- tained based on multidisciplinary expert opinion. The cal scores) as well as with margins on the total score scale was finally reduced from 83 to 52 items (Fig. 1). ranging from –1 to + 1 point, from –2 to + 2 points and The scoring procedure was optimized as well: simi- from –3 to + 3 points between both raters or between lar to the initial study, frequency and severity were test and retest. Those margins are comparable with, scored per item for the last six months and the for example, one item among 52 being scored dif- characteristic/typical behavior. However, care bur- ferently. IRR and TRR were also calculated for care den, previously scored per item for the last six months burden change scores (11 items) and the total scale only, was now scored per section (instead of per item) care burden change score which could range from –33 and for both periods of time. to + 33 points, again for perfect agreement and with The BPSD-DS II consisted of two parts. Part A margins. To determine the internal consistency, Cron- covered general information about interviewer, infor- bach’s alphas were calculated for frequency change mant(s), and the person with DS, including items and severity change scores per item and for the total about differential diagnostics. Part B covered 52 scale. items subdivided in 11 clinically defined sections: Additional analyses were performed to evaluate the 1) anxious behavior, 2) sleeping problems, 3) irri- effect of age, sex, ID level, and presence of depres- table behavior, 4) obstinate behavior, 5) restless & sion on total scale scores for frequency change and stereotypic behavior, 6) aggressive behavior, 7) apa- severity change. To assess the effect of age, a linear thetic behavior, 8) depressive behavior, 9) psychotic regression analysis was performed within the study behavior, 10) disinhibited behavior, and 11) eating group without dementia (DS) using age as indepen- & drinking behavior. Items per section are described dent variable and total scale frequency change or below. The aim of these sections was not to diagnose, 1512 Table 1 Characteristics of the three diagnostic study groups A: Participants’ characteristics DS (N = 292) DS + Q (N = 119) DS + AD (N = 113) p Sex (% male) 51.7 48.7 61.1 0.135 Mean age ± SD (min.-max.) 47.4 ± 9.0 (30.0–73.0) 53.8 ± 7.7 (33.4–74.0) 57.3 ± 6.4 (37.2–70.0) < 0.001 ID level (baseline): mild; moderate; severe (%) 11.3; 66.8; 21.9 16.0; 57.1; 26.9 8.0; 62.8; 29.2 0.160 IQ-score available (%) (mean IQ-score ± SD) 21.2 (42.3 ± 11.6) 22.7 (44.2 ± 9.4) 15.9 (40.4 ± 11.9) 0.387; 0.543 Living independently; with family; assisted facilities; other (%) 1.7; 12.7; 85.6; 0.0 0.0; 12.6; 86.6; 0.8 0.0; 10.6; 86.7; 2.7 0.066 Attending day-care (%) 93.5 89.9 82.3 0.003 Type DS: full trisomy; translocation; 37.0; 0.3; 0.3; 62.3 39.5; 0.8; 0.8; 58.8 45.1; 0.9; 0.9; 53.1 0.706 mosaicism; unknown (%) Depression: treated; untreated (%) 2.7; 0.0 3.4; 0.0 8.0; 0.0 0.051 Epilepsy: treated; untreated (%) 3.8; 0.3 9.2; 1.7 20.4; 2.7 < 0.001 Hypothyroidism: treated; untreated (%) 41.4; 1.0 46.2; 0.8 56.6; 0.9 0.106 Vitamin B12 deficiency: treated; untreated (%) 4.1; 0.0 8.4; 0.0 5.3; 0.0 0.214 Hearing problems: treated; untreated (%) 28.8; 15.1 34.5; 19.3 38.1; 25.7 0.006 A.D. Dekker et al. / BPSD in Down Syndrome Vision problems: treated; untreated (%) 58.9; 19.2 59.7; 22.7 58.4; 26.5 0.355 Sleep apnea: treated; untreated (%) 4.1; 2.4 4.2; 5.0 6.2; 4.4 0.536 Chronic pain: treated; untreated (%) 10.6; 3.1 13.4; 5.0 18.6; 2.7 0.211 Physical disability (%) 7.9 7.6 8.8 0.929 Dependent on wheelchair (%) 0.3 1.7 5.3 0.003 Swallowing problems (%) 10.6 14.3 21.2 0.020 Dental problems causing eating/drinking problems (%) 10.3 13.4 14.2 0.458 Verbal communication: able; no longer; never (%) 93.2; 0.3; 6.5 90.8; 1.7; 7.6 90.3; 5.3; 4.4 0.013 Any psychoactive medication use (%) 18.8 33.6 54.9 < 0.001 - Antiepileptics (N03A, %) 5.1 12.6 25.7 < 0.001 - Antipsychotics (N05A, %) 8.6 13.4 23.9 < 0.001 - Anxiolytics (N05B, %) 0.7 2.5 8.8 < 0.001 - Hypnotics and sedatives (N05C, %) 1.0 0.8 3.5 0.142 - Antidepressants (N06A, %) 7.9 18.5 24.8 < 0.001 - Antidementia (N06D, %) 0.0 1.7 5.3 < 0.001 - Opiods (N02A, %) 0.0 0.0 2.7 0.004 Number of psychoactive medications (% N = 0; 1; 2; 3; 4; 5) 81.2; 14.0; 3.8; 0.3; 0.7; 0.0 66.4; 19.3; 10.9; 0.0; 3.4; 0.0 45.1; 23.9; 19.5; 5.3; 4.4; 1.8 < 0.001 B: Informants’ characteristics DS (N = 534 informants) DS + Q (N = 217 informants) DS + AD (N = 203 informants) p Informants per participant (% N = 1; N = 2 or N = 3) 35.6; 47.3; 17.1 34.5; 48.7; 16.8 36.3; 47.8; 15.9 0.997 Sex (% men) 21.6 19.3 15.9 0.436 Caregiver: day-care; residential; nurse; family; other (%) 6.8; 53.2; 0.2; 36.0; 3.8 7.4; 54.8; 0.9; 33.6; 3.2 4.4; 57.1; 0.5; 34.0; 3.9 0.779 Years knowing participant (% < 2; 2–10; 10–20; > 20 years) 6.6; 37.0; 14.0; 42.5 5.5; 36.9; 18.9; 38.7 11.3; 38.4; 12.8; 37.4 0.116 Hours per week with participant (% < 10; 10–20; > 20 hours) 27.9; 32.3; 39.8 29.0; 19.8; 51.2 33.5; 19.7; 46.8 < 0.001 ID level refers to the highest level of functioning (baseline) before dementia-related decline occurred. Dependence on a wheelchair was defined as requiring a wheelchair not only outdoors for longer distances, but also indoors. If the person with DS does not need a wheelchair indoors, they may show most behavioral items related to physical activity in the BPSD-DS II, such as avoiding places, wandering, etc. The presence of (un)treated comorbidities did not result in evident behavioral changes. The fact that a condition had sometimes not been treated (no need yet, lack of/poor compliance, no treatment possible) was not the cause of behavioral changes in the BPSD-DS II scale, according to clinical judgment. Pearson’s chi-squared tests were applied to compare (categorical) data between groups. ANOVA was used to compare normally distributed continuous data (age and IQ-scores) between groups. A Kruskal-Wallis test was used to compare the number of psychoactive medications between groups (categorial, not normally distributed). p-values are reported with three decimals. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia; ID, intellectual disabilities; IQ, intelligence quotient; SD, standard deviation. A.D. Dekker et al. / BPSD in Down Syndrome 1513 Fig. 2. Schematic overview of included and excluded interviews and the three study groups. BPSD-DS II, Behavioral and Psycho- logical Symptoms of Dementia in Down Syndrome II scale; CVA, cerebrovascular accident; DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Fig. 1. Schematic overview of optimization of the BPSD-DS scale. Down syndrome with diagnosed AD dementia; ID, intellectual disability. e.g., anxiety or aggression as an entity as such, but simply to group together items in a logic and infor- conducted with one informant, 250 with two infor- mative manner to guide interviewers and informants mants, and 88 with three informants. Table 1B shows through the entire scale. This is especially relevant for demographic data for the informants per study group. informants who are generally not trained in behav- Informants were primarily caregivers working in ioral definitions and may misinterpret an item. Items day-care centers or residential facilities, and family addressed various potential dementia-related behav- members. ioral alterations within each section. Since BPSD may present rather heterogeneously over time [29], sec- Validity tions were not designed with the expectation that all items within a section would be relevant simultane- Face and content validity were already ensured for ously. the initial version of the BPSD-DS [30]. Results and clinical experiences from that first study guided opti- Population demographics mization of the scale to further strengthen its validity. In the context of discriminative ability, 23 irrelevant Using informant interviews, the BPSD-DS II was items were removed, shortening the scale from 83 to completed for 575 people with DS. According to 52 items (Fig. 1). These were subsequently compared exclusion criteria, 51 interviews were excluded for between groups (see below). various reasons (Fig. 2), resulting in 524 eligible interviews. For a total of 92 people with DS, the inter- Discriminative ability: item (ir)relevance viewer had been previously involved in the diagnostic procedure of the (questionable) dementia within the Here, we aimed to confirm the relevance of the care institution. Table 1A presents demographic data remaining and refined 52 items by comparing fre- for each of the three study groups: DS (N = 292; quency change and severity change across the three 55.7% of people with DS in this study), DS + Q study groups. Hereafter, items are described per clini- (N = 119; 22.7%), and DS + AD (N = 113; 21.6%). cally defined section with the corresponding p-values A total of 950 informants (17.5% men) participated for the comparison of frequency change scores (pfq ) in the 524 eligible interviews: 186 interviews were and severity change scores (psv ) between groups. Fig- 1514 A.D. Dekker et al. / BPSD in Down Syndrome Fig. 3. Significant frequency changes for items in section 1 (anxious behavior). Underlying frequency change scores are depicted in a simplified way: the proportion of individuals (%) per group showing any decreased, unaltered or any increased frequency comparing the last sixth months to the typical/characteristic behavior in the past. Proportions of individuals (%) with missing values is depicted in dark grey, but not in numbers. Item descriptions and p-values (Kruskal-Wallis) are provided in the text. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia. ures 3–9 visualize frequency changes in a simplified unusually long time with any (apparent) physical way per item by depicting the proportion of individu- reason (1.2; pfq < 0.001, psv < 0.001), being tense als in each group (DS, DS + Q, DS + AD) showing (1.3; pfq < 0.001, psv < 0.001), avoiding situations/ any decreased, unaltered, or any increased frequ- places that makes the person nervous (1.4; pfq < ency comparing the last sixth months to the typical/ 0.001, psv < 0.001), being scared to be left alone (1.5; characteristic behavior in the past. In general, fre- pfq < 0.001, psv < 0.001), and being easily panicked quency changes were more pronounced than severity (1.6; pfq < 0.001, psv < 0.001). Figure 3 visualizes changes. With the exception of a few items, the that increased anxiety was prominent for people with proportion of individuals within each group show- DS + AD, but also for DS + Q. For items 1.1–1.4, the ing severity changes (any decrease, no change, any difference between DS + Q and DS + AD was rather increase) largely followed the same pattern as fre- small, suggesting that increased anxiety may already quency changes, though less pronounced. Therefore, occur in an early stage of dementia. only frequency changes are visualized below. A sim- plified presentation of severity changes is available Section 2: Sleep problems in Supplementary Figures 1–6. This section evaluated finding it hard to fall asleep Section 1: Anxious behavior (2.1; pfq = 0.001, psv = 0.001), waking repeatedly during the night (2.2; pfq < 0.001, psv < 0.001), wan- The first section addressed worrying about up- dering around at night (2.3; pfq < 0.001, psv < 0.001), coming activities/events (item 1.1; pfq < 0.001, psv < waking long before it is time to get up/ the alarm goes 0.001), going to the toilet unusually often or for an (2.4; pfq < 0.001, psv < 0.001), finding it hard to get up A.D. Dekker et al. / BPSD in Down Syndrome 1515 Fig. 4. Significant frequency changes for items in section 2 (sleep problems). Underlying frequency change scores are depicted in a simplified way: the proportion of individuals (%) per group showing any decreased, unaltered or any increased frequency comparing the last sixth months to the typical/characteristic behavior in the past. Proportions of individuals (%) with missing values is depicted in dark grey, but not in numbers. Item descriptions and p-values (Kruskal-Wallis) are provided in the text. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia. in the morning (2.5; pfq < 0.001, psv = 0.002), being Section 4: Obstinate behavior tired or complaining of fatigue (2.6; pfq < 0.001, psv < 0.001), and sleeping in the daytime (2.7; Being self-willed (4.1; pfq = 0.436, psv = 0.362), pfq < 0.001, psv < 0.001). Figure 4 demonstrates that being argumentative, uncooperative, or obstructive the proportion of individuals showing an increase in (4.2; pfq = 0.089, psv = 0.007), not willing to accept sleep problems was consistently highest in DS + AD, necessary help (4.3; pfq = 0.420, psv = 0.415), and intermediate in DS + Q, and lowest in DS. Specifi- sighing/groaning (4.4; pfq < 0.001, psv < 0.001) were cally, for items 2.6 and 2.7, the proportion of ind- evaluated in this section. Only the latter item differed ividuals showing an increased frequency was already significantly between groups, with the proportion of substantial in more than one third of individuals with individuals showing an increased frequency being DS + Q. lowest in the DS group and higher in DS + Q and DS + AD groups (Fig. 5). Though not significant, items 4.1–4.3 showed an interesting bidirectional Section 3: Irritable behavior change: for a substantial proportion of individu- als a decreased frequency was reported, while for Three items evaluated were: being irritable, touchy another substantial proportion an increased frequency (3.1; pfq < 0.001, psv < 0.001), being impatient (3.2; was reported. Among the DS + AD group, item pfq < 0.001, psv < 0.001), and being short-spoken, 4.1 (14.2% decreased frequency; 15.9% increased responding grumpily (3.3; pfq < 0.001, psv < 0.001). frequency), 4.2 (13.3%; 24.8%), and 4.3 (15.0%; In general, the proportion of individuals showing an 23.9%) was found. increase was highest in both DS + Q and DS + AD, and lowest in DS (Fig. 5). The difference between Section 5: Restless & stereotypic behavior DS + Q and DS + AD was rather small, suggesting that increased irritable behavior may already occur This section included items on general restless- in an early stage of dementia. ness (5.1; pfq < 0.001, psv < 0.001), wandering 1516 A.D. Dekker et al. / BPSD in Down Syndrome Fig. 5. Significant frequency changes for items in section 3 (irritable behavior), section 4 (obstinate behavior), section 5 (restless & stereo- typic behavior), and section 6 (aggressive behavior). Underlying frequency change scores are depicted in a simplified way: the proportion of individuals (%) per group showing any decreased, unaltered or any increased frequency comparing the last sixth months to the typi- cal/characteristic behavior in the past. Item descriptions and p-values (Kruskal-Wallis) are provided in the text. Proportions of individuals (%) with missing values is depicted in dark grey, but not in numbers. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia. (5.2; pfq < 0.001, psv < 0.001), stereotypic behav- behavior (5.6; pfq = 0.018, psv = 0.005). Between ior (5.3; pfq < 0.001, psv < 0.001), repeatedly groups, the same pattern was observed for all items, dressing and undressing (more than necessary) with the proportion of individuals showing an (5.4; pfq = 0.001, psv < 0.001), verbal stereotypy increase being consistently highest in DS + AD and (5.5; pfq < 0.001, psv < 0.001), and compulsive lowest in DS (Fig. 5). A.D. Dekker et al. / BPSD in Down Syndrome 1517 Section 6: Aggressive behavior empathy (7.7; pfq < 0.001, psv = 0.109), and jaded emotional responses (7.8; pfq < 0.001, psv = 0.002). Verbally aggressive behavior (6.1; pfq < 0.001, Among all sections, apathetic behavior was increased psv < 0.001), destructive behavior (6.2; pfq = 0.004, most evidently in relation to the status of dementia, psv = 0.014), and physically aggressive behavior with proportions of individuals showing an increase towards others (6.3; pfq = 0.003, psv = 0.020) were up to 73.5 % (item 7.4). The proportion of individ- evaluated. Figure 5 demonstrates that an increased uals showing an increase was consistently highest in frequency of verbal aggression was more pronounced DS + AD, intermediate in DS + Q, and lowest in DS in DS + AD than in DS. Destructive and physically for all items (Fig. 6), with the exception of item 7.6 in aggressive behavior showed a rather similar pattern, which DS + Q and DS + AD were relatively similar. though less pronounced. A substantial proportion of individuals with DS + Q already demonstrated an increase frequency of most Section 7: Apathetic behavior apathetic items, thus suggesting that increased apa- thetic symptoms may already present in an early Different possible symptoms of apathetic behav- phase of dementia. ior were assessed: lack of initiative (7.1; pfq < 0.001, psv < 0.001), lack of interest in the direct living Section 8: Depressive behavior environment (7.2; pfq < 0.001, psv < 0.001), hard to motivate to get involved in familiar activities/tasks, Although many symptoms overlap, apathy and appearing lazy (7.3; pfq < 0.001, psv < 0.001), not depression are regarded as two distinct neuropsychi- independently completing activities/tasks, needs atric conditions [29]. Items in this section emphasized encouragement or help (7.4; pfq < 0.001, psv < 0.001), depressive over apathetic characteristics, including not participating much in conversation (7.5; rapid mood swings (8.1; pfq < 0.001, psv < 0.001), pfq < 0.001, psv < 0.001), social withdrawal (7.6; being sad and/or weeping a lot (8.2; pfq < 0.001, pfq < 0.001, psv < 0.001), lack of sympathy or psv < 0.001), being very downhearted and appearing Fig. 6. Significant frequency changes for items in section 7 (apathetic behavior). Underlying frequency change scores are depicted in a simplified way: the proportion of individuals (%) per group showing any decreased, unaltered or any increased frequency comparing the last sixth months to the typical/characteristic behavior in the past. Item descriptions and p-values (Kruskal-Wallis) are provided in the text. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia. 1518 A.D. Dekker et al. / BPSD in Down Syndrome Fig. 7. Significant frequency changes for items in section 8 (depressive behavior) and section 9 (psychotic behavior). Underlying frequency change scores are depicted in a simplified way: the proportion of individuals (%) per group showing any decreased, unaltered, or any increased frequency comparing the last sixth months to the typical/characteristic behavior in the past. Item descriptions and p-values (Kruskal-Wallis) are provided in the text. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia. to be in low spirits (8.3; pfq < 0.001, psv < 0.001), hav- commonly as compared to other sections. Figure 7 ing physical complaints without any apparent illness shows that the frequency of psychotic behavior or injury (8.4; pfq = 0.001, psv = 0.007), and moving increased in a larger proportion of individuals with and responding slowly (general slowness) dementia compared to those without dementia. (8.5; pfq < 0.001, psv < 0.001). For items 8.1, 8.2, and 8.5 the proportion of individuals showing Section 10: Disinhibited behavior an increase was evidently highest in DS + AD, intermediate in DS + Q, and lowest in DS (Fig. 7). Items addressed behaving impulsively For items 8.3 and 8.4, the proportion of individuals in (10.1; pfq = 0.001, psv = 0.111), making inappropriate DS + Q and DS + AD was rather similar as compared comments or jokes (10.2; pfq = 0.526, psv = 0.071), to DS. The proportion of individuals in the DS + Q and behaving in impolite or indecent ways (loss group showing an increased frequency was already of decorum) (10.3; pfq < 0.001, psv = 0.302). The substantial for items 8.2 and 8.5, suggesting that proportion of individuals demonstrating an increased these symptoms may present early in the course of frequency of disinhibited behavior was highest in the disease. DS + AD and lowest in DS (Fig. 8). Section 9: Psychotic behavior Section 11: Eating & drinking behavior This section concerned incorrect beliefs/thoughts This last section addressed drinking poorly, hav- (delusions) (9.1; pfq < 0.001, psv < 0.001) and abnor- ing to be encouraged to drink (11.1; pfq < 0.001, mal sensory experiences not experienced by others psv < 0.001), poor appetite, having to be encouraged (hallucinations) (9.2; pfq < 0.001, psv < 0.001). Psy- to eat (11.2; pfq < 0.001, psv < 0.001), eating slowly chotic symptoms were reported substantially less (11.3; pfq < 0.001, psv < 0.001), being picky about A.D. Dekker et al. / BPSD in Down Syndrome 1519 Fig. 8. Significant frequency changes for items in section 10 (disinhibited behavior) and section 11 (eating & drinking behavior). Underlying frequency change scores are depicted in a simplified way: the proportion of individuals (%) per group showing any decreased, unaltered or any increased frequency comparing the last sixth months to the typical/characteristic behavior in the past. Item descriptions and p-values (Kruskal-Wallis) are provided in the text. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia. food and drink (11.4; pfq = 0.078, psv = 0.116), and icantly between groups (p < 0.001): DS (median = 0; putting substances/objects in the mouth that are not 25% = –11; 75% = 4), DS + Q (median = 15; 25% = 3; intended for consumption (pica) (11.5; pfq = 0.002, 75% = 29), and DS + AD (median = 23; 25% = 7; psv = 0.162). For items 11.1–11.3, the proportion 75% = 54). In line with frequency change, also of individuals showing an increase was highest in total scale severity change scores differed signifi- DS + AD and lowest in DS (Fig. 8). cantly between groups (p < 0.001): DS (median = 0; Items with unaltered frequency change and sever- 25% = –11; 75% = 2), DS + Q (median = 5; 25% = –2; ity change scores (change = 0) for ≥ 85% of DS + Q 75% = 14), and DS + AD (median = 6; 25% = –5; and DS + AD were regarded irrelevant. Only 11.5 75% = 22). Evidently, total scale scores were signifi- (pica) fulfilled this criterion: frequency change was cantly higher in the group with dementia (DS + AD) unaltered for 98.3% of DS + Q and 98.1% for as compared to those without dementia (DS). DS + AD, and severity change was unaltered for 100% of DS + Q and 99.1% of DS + AD. None of the Discriminative ability: Sensitivity, specificity, other items were found to be irrelevant as substantial and predictive values changes were observed between groups. Although a diagnosis of dementia cannot exclu- Discriminative ability: Total scale scores sively be based on a purely behavioral assess- ment (without addressing cognitive and functional Kruskal-Wallis tests were used to compare sum decline), scores on a behavioral scale may aid scores for the total scale between the three groups. clinicians and ID (neuro)psychologists in the diag- Total scale frequency change scores differed signif- nostic process. Therefore, we aimed to identify 1520 A.D. Dekker et al. / BPSD in Down Syndrome cut-off scores for discrimination between groups, the total proportion of individuals in each group for i.e., DS versus DS + Q/DS + AD (cut-off 1) and whom informants experienced a decreased, unaltered DS/DS + Q versus DS + AD (cut-off 2). Between or increased care burden. Evidently, the proportion groups, frequency changes were more pronounced of individuals for whom an increase in care burden than severity changes. Therefore, Supplementary was reported, was the highest in the DS + AD group, Figure 7 shows the ROC curves for total scale fre- intermediate in the DS + Q group and lowest in the quency change scores for DS versus DS + Q/DS + AD DS group. In most sections, an increase was already and for DS/DS + Q versus DS + AD. Since sensitivity, reported in a substantial number of individuals in the specificity, positive predictive value and negative pre- DS + Q group. Care burden thus increased in rela- dictive value approached each other, Table 2 presents tion to the presence of dementia, most prominently a range of cut-off scores. for anxious, apathetic and depressive behavior: in these sections, care burden increased for over 65% Care burden of individuals in the DS + AD group. Care burden was scored per section considering Reliability those items within each section that were answered positively (frequency ≥ 1). Kruskal-Wallis tests IRR was determined for a subset of N = 82 ind- demonstrated that care burden change scores dif- ividuals (15.6%): DS (n = 52), DS + Q (n = 12), fered significantly between groups for all sections: DS + AD (n = 18). TRR was determined for a sub- 1) anxious behavior (p < 0.001), 2) sleeping prob- set of N = 50 individuals (9.5%): DS (n = 33), DS + Q lems (p < 0.001), 3) irritable behavior (p < 0.001), (n = 9), DS + AD (n = 8). Table 3 depicts the low- 4) obstinate behavior (p < 0.001), 5) restless & est and highest IRR and TRR (range) for individual stereotypic behavior (p < 0.001), 6) aggressive behav- items within each section for frequency change and ior (p < 0.001), 7) apathetic behavior (p < 0.001), severity change (percent agreement). Evidently, IRR 8) depressive behavior (p < 0.001), 9) psychotic was very high with lowest percent agreement of behavior (p < 0.001), 10) disinhibited behavior 92.7%, demonstrating that the defined scoring system (p = 0.019), and 11) eating & drinking behavior was not subject to evident interpretation differences (p < 0.001). Figure 9 shows a simplified represen- between raters. TRR was lower than IRR, but still tation of the results in stacked bar charts. It shows yields sufficiently high percent agreement (>70%) scores. Time interval between test and retest inter- views did not evidently affect percent agreement. IRR Table 2 and TRR results did not necessitate removal of items. Cut-off scores with corresponding sensitivity, specificity, positive predictive value, and negative predictive value IRR and TRR were also evaluated for total scale scores for frequency change and severity change. DS versus DS + Q/DS + AD Table 3 provides IRR and TRR for perfect agreement Cut-off score Sensitivity Specificity PPV NPV (identical scores) and with margins. Logically, IRR 4 76.7 72.6 69.0 79.7 5 75.0 75.7 71.0 79.2 and TRR were lower for the total scale scores as com- 6 74.6 78.4 73.3 79.5 pared to the individual items. Similar to individual 7 73.3 80.1 74.6 79.1 items, TRR for the total scale was lower than IRR. 8 72.0 82.2 76.3 78.7 In addition, IRR and TRR were calcu- 9 69.8 83.2 76.8 77.6 lated as percent agreement for section care DS/DS + Q versus DS + AD burden change scores: 1) anxious behav- Cut-off score Sensitivity Specificity PPV NPV ior (IRR = 95.1; TRR = 78.0), 2) sleeping 11 69.9 71.8 40.5 89.7 12 69.9 74.2 42.7 90.0 problems (96.3; 86.0), 3) irritable behavior 13 69.0 75.4 43.6 89.9 (97.6; 80.0), 4) obstinate behavior (95.1; 76.0), 14 66.4 76.4 43.6 89.2 5) restless & stereotypic behavior (95.1; 84.0), 15 65.5 78.4 45.5 89.2 6) aggressive behavior (100.0; 80.0), 7) apathetic 16 64.6 79.8 46.8 89.1 behavior (98.8; 86.0), 8) depressive behavior (97.6; The range of cut-off scores provided here starts with the first cut-off 66.0), 9) psychotic behavior (98.8; 96.0), 10) dis- score reaching a specificity ≥ 70%. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; inhibited behavior (98.8; 92.0), and 11) eating DS + AD, Down syndrome with diagnosed AD dementia; PPV, & drinking behavior (97.6; 92.0). IRR and TRR positive predictive value; NPV, negative predictive value. were also calculated for the total scale care burden A.D. Dekker et al. / BPSD in Down Syndrome 1521 Fig. 9. Care burden changes for each section. Underlying frequency change scores are depicted in a simplified way: the proportion of individuals (%) per group showing a decreased, unaltered or increased care burden comparing the last sixth months to the typical/characteristic behavior in the past. p-values (Kruskal-Wallis) are provided in the text. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia. change score for perfect agreement (IRR = 82.9; Internal consistency was evaluated by determining TRR = 40.0) and with margins –1 to + 1 point (93.9; Cronbach’s alphas for frequency change and sever- 64.0), –2 to + 2 points (98.8; 86.0), and –3 to + 3 ity change of individual items against the total scale points (100; 92.0). scores for frequency change and severity change, 1522 A.D. Dekker et al. / BPSD in Down Syndrome Table 3 Interrater reliability and test-retest reliability Section Number IRR (% agreement) TRR (% agreement) of items Frequency change Severity change Frequency change Severity change 1) Anxious behavior 6 93.9–97.6 95.1–97.6 76.0–92.0 76.0–92.0 2) Sleeping problems 7 92.7–98.8 93.9–98.8 72.0–92.0 78.0–98.0 3) Irritable behavior 3 96.3–98.8 96.3–98.8 70.0–90.0 82.0–88.0 4) Obstinate behavior 4 97.6–98.8 97.6–98.8 76.0–86.0 80.0–86.0 5) Restless & stereotypic behavior 6 98.8–100.0 98.8–100.0 84.0–94.0 84.0–96.0 6) Aggressive behavior 3 97.6–100.0 97.6–98.8 80.0–84.0 88.0–90.0 7) Apathetic behavior 8 96.3–100.0 95.1–100.0 78.0–96.0 80.0–98.0 8) Depressive behavior 5 97.6–100.0 97.6–100.0 76.0–94.0 78.0–96.0 9) Psychotic behavior 2 98.8–100.0 100.0-100.0 96.0–98.0 96.0–98.0 10) Disinhibited behavior 3 100.0-100.0 95.1–100.0 90.0–96.0 94.0–98.0 11) Eating & drinking behavior 5 98.8–100.0 98.8–100.0 88.0–98.0 94.0–100.0 Total scale (perfect agreement) 52 70.7 65.9 12.0 26.0 Total scale (margin –1 to + 1) 52 82.9 81.7 34.0 38.0 Total scale (margin –2 to + 2) 52 84.2 87.8 40.0 56.0 Total scale (margin –3 to + 3) 52 86.6 89.0 54.0 66.0 IRR and TRR for frequency change and severity change are provided as percent agreement. For each section, the range of agreement for individual items is given, i.e., the lowest and the highest percent agreement for items in each section. For each individual item, IRR, TRR, as well as internal consistency, are provided in Supplementary Table 1. For the total scale, both IRR and TRR were calculated for perfect agreement (identical scores) as well as with of plus or minus 1, 2, or 3 points. DS, Down syndrome without dementia; DS + Q, Down syndrome with questionable dementia; DS + AD, Down syndrome with diagnosed AD dementia; IRR, interrater reliability; TRR, test-retest reliability. respectively. Individual items had Cronbach’s alphas out evident effect on behavior according to clinical above 0.839 (frequency change) and above 0.788 judgement) did not significantly influence total scale (severity change) (Supplementary Material Table 1). scores for frequency change (p = 0.114) and sever- Among all items together, overall Cronbach’s alpha ity change (p = 0.128). Finally, groups with mild was 0.845 (frequency change) and 0.799 (severity ID (n = 61), moderate ID (n = 334), and severe ID change). In summary, reliability and consistency data (n = 129) were compared. ID level significantly in- for frequency change and severity change were good fluenced total scale score for severity change (F and confirm previous results [30]. (2,521) = 3.95, p = 0.020), but not for frequency change (F(2,521) = 0.81, p = 0.444). Effect of age, sex, depression, and ID level DISCUSSION Various potential confounding factors were already addressed in advance through clinical judgement (see Using the shortened and refined BPSD-DS II Methods section). In addition, effects of age, sex, scale, dementia-related behavioral changes in 524 depression, and ID level were evaluated. Since age is individuals with DS were evaluated in relation to the major risk factor for dementia [38], the DS + AD their status of dementia to further study reliability group was expectedly older. A linear regression anal- and discriminative ability, such as confirmation of ysis in the group without dementia (DS) showed previous results about item relevance [30]. Com- that age did neither significantly influence total scale paring item change scores between groups revealed scores for frequency change (coefficient b = 0.07, prominent changes in frequency and severity for anx- p = 0.526) and severity change (b = 0.010, p = 0.926). ious, sleep-related, irritable, restless & stereotypic, Next, regardless of dementia status, effects of sex, apathetic, depressive, and eating & drinking behav- depression and ID level were studied. Comparing ior. For most individual items, an overall pattern men (n = 278) and women (n = 246) showed that emerges in which the proportion of individuals dis- sex did not significantly affect total scale scores for playing a frequency/severity increase was highest frequency change (p = 0.41) and severity change in DS + AD, intermediate in DS + Q, and lowest in (p = 0.57). Comparing groups with (n = 21) and wit- DS. Interestingly, for various items within sections hout depression (n = 503) demonstrated that the about anxious, sleep-related, irritable, apathetic and presence of (effectively) treated depression (with- depressive behaviors, already more than 1/3 of the A.D. Dekker et al. / BPSD in Down Syndrome 1523 DS + Q group showed an increased frequency (and the risk of erroneously attributing changes to demen- sometimes even higher than in the DS + AD group), tia, each individual was carefully evaluated for other which suggests that changes in these items might be causes of decline/change, such as major life events early ‘alarm signals’ for AD in DS. These results are or comorbidities that may present with dementia-like largely in accordance with the initial study publica- symptoms. tion [30], in which a more extensive contextualization Results from the initial study [30] guided opti- of results with literature is found. Care burden was mization of the scale. The BPSD-DS scale is not found highest for DS + AD and lowest for DS. As intended as purely an academic research tool but reported in the general population [8, 12], care burden was developed together with daily practice experts. evidently increased with the presence of dementia and To enhance usability and reduce assessment time, the presence of BPSD. Finally, reliability measures the scale was shortened and digitized, also enhanc- were promising: IRR and TRR data show high per- ing data completeness and quality. Another strength cent agreement and internal consistency with alpha concerns the large representative study population around 0.80 was found for individual items and the with, among others, different baseline ID levels and total scale. living situations (e.g., different care institutions and regions/countries). To the best of our knowledge, this Study strengths study is the largest behavioral study in DS so far, enhancing generalizability of results. Virtually all individuals with dementia present with one or more behavioral changes [9, 10]. Whereas Study limitations these symptoms are well studied in the general population [8, 10, 39], BPSD have not been com- Among the 524 eligible interviews, the interviewer prehensively studied in DS [29]. To that end, we was previously involved in the diagnostic procedure developed the BPSD-DS scale exclusively focus on of (questionable) dementia of 92 individuals with DS. behavioral alterations in DS [30]. Such a modular Although one could argue that this is a strength (well approach corresponds to the way in which dementia aware of the diagnostic process and its thoroughness), is assessed in the general population using specific this might also be regarded as a risk of bias. Therefore, behavioral questionnaires like the Neuropsychiatric it is important to note that the scores were provided by Inventory [27, 28] in addition to cognitive and func- informants (not the interviewer). Moreover, diagnosis tional assessments. of dementia, and thus division into study groups, was Diagnosing dementia in DS is rather complex established without considering the outcomes of the due to the baseline level of functioning, presence of BPSD-DS II. (life-long) characteristic behavior and the frequent In retrospective interviews, recall bias and a de- presence of comorbidities which may contribute to gree of subjectivity—especially regarding (variable) dementia-like symptoms [29, 40–42]. Considering behavior—may influence results. Although the selec- this diagnostic complexity and specific circumstances tion and quality of informants was considered in ad- relating to individuals with DS, an important strength vance by the interviewer, differences in informants’ of this scale is its specific adaptation to the target personal attentiveness to signal changes could be group. To account for life-long characteristic behav- a potential limitation. In the context of assessing ior, the central aim of the scale is to identify ‘change’. dementia in individuals with ID, however, Jamieson- The scoring system has been developed as such that Craig et al. showed that ‘retrospective carer report the individual with DS is compared to oneself. As of change in everyday function was as good as, if people with ID often find it difficult to verbally not better than, prospective ratings to identify demen- express their feelings and emotions, or may not under- tia’ [42]. Nevertheless, certain informants may have stand the items, identification of changes is based on exaggerated or trivialized behavioral alterations. That interviews with key informants [43]. is why an interview design was chosen rather than In this study, the diagnosis of (questionable) self-completion by informants. Experienced inter- dementia was based on existing clinical multidisci- viewers may, in part, recognize and address this plinary assessment, the current gold standard [16]. during the interview. Participants were not subjected to new dementia Another possible limitation is the fact that this assessments. To minimize the effect of other poten- study did not consider individuals with profound ID tial causes of decline than dementia and to reduce who often face other disabilities as well, such as 1524 A.D. Dekker et al. / BPSD in Down Syndrome motor or sensory disabilities. They require a spe- changes, though less pronounced, a short version of cific approach to identify symptoms of dementia [44] the BPSD-DS II only including frequency changes as (many) specific skills have never been developed might be considered. Moreover, results point at irrel- or care professionals have taken over. Consequently, evance of item 11.5 (pica), which could potentially be items addressing such skills are not relevant as they removed in a new version. Furthermore, those items cannot demonstrate decline, i.e., they cannot be that did not differ significantly between groups might indicative of dementia [16, 45]. Currently, no adapted be further scrutinized. Finally, consideration of fur- scales are available for dementia in this subpopula- ther practical experiences obtained in this study may tion. Potentially, a future selection of BPSD-DS II refine the scale and procedure for future use. items might serve useful in this population. In rela- In addition, it would be valuable to apply the tion to individuals’ capacities, changes in psychotic BPSD-DS II in qualitative as well as longitudinal behavior were reported less frequently as compared to studies. Qualitative studies may further contextualize other sections. This most likely relates to the fact that the added value of the BPSD-DS II in daily practice. observing such symptoms largely relies on the (often Whereas the current cross-sectional study retrospec- limited) self-report abilities by the person with DS. tively assessed changes, longitudinal studies would Lastly, we faced practical difficulties to schedule enable evaluation of intraindividual changes over more retest interviews. Although we extended the time. Such a longitudinal approach is essential to time interval from maximum 4 weeks to 7 weeks, it address the temporal relationship between behavioral was still not feasible on many occasions due to, e.g., alterations and dementia staging and confirm early holidays, illness, far travel distances to the interview behavioral ‘alarm signals’ of dementia in DS. location, illness, full agendas. CONCLUSION Future implications The optimized BPSD-DS II has been studied in Primary goal of this study was to optimize and a large representative DS study population in rela- further study an adapted assessment tool for BPSD tion to the status of dementia. Results are largely in adults with DS. We have shown that the BPSD- in accordance with previous findings and further DS II, capturing a wide range of dementia-related confirm item relevance and reliability. Comparing behavioral changes, appears to be reliable and may study groups, the BPSD-DS II revealed an overall aid identification of those at risk to develop dementia pattern in which the proportion of individuals dis- (‘alarm signals’). The scale serves useful to sys- playing a frequency/severity increase was highest tematically evaluate a person’s behavioral changes, in DS + AD, intermediate in DS + Q, and lowest in important for person-centered care (planning) [46, DS. Interestingly, changes in various items concern- 47]. Indeed, in the context of improving or main- ing anxious, sleep-related, irritable, apathetic, and taining quality of life, it is of essence to determine depressive behaviors might be early ‘alarm signals’ the needs of the person to adapt daily care and for AD in DS. Care burden increased in relation to (behavioral) treatment. Furthermore, identifying dementia. Systematic evaluation of behavioral cha- changes would also help increasing understanding nges in DS using the BPSD-DS II may contribute and insight into the progress of dementia among to early identification of those at risk to develop caregivers. Additionally, disease progression and dementia, adaption of daily care/treatment and under- effects of therapeutic interventions could possibly standing among caregivers/family members. The be assessed in daily clinical practice using the scale may be a useful addition to already existing BPSD-DS II. cognitive questionnaires for dementia in DS. The BPSD-DS II provides rich, detailed informa- tion about the person’s behavioral changes over time. Depending on the status of dementia, such interviews ACKNOWLEDGMENTS take an hour, on average. It is conceivable that in spe- cific settings, a more rapid screening is preferred (at This study was financially supported by the J. the cost of less detail), similar to, e.g., a brief clinical Th. Guepin Stichting Onderzoek Down Syndroom version of the Neuropsychiatric Inventory in the gen- and further supported by in kind contributions from eral population [48]. Since severity changes in this the participating care institutions and European study largely followed the same pattern as frequency expertise centers. Founding work of this project A.D. Dekker et al. / BPSD in Down Syndrome 1525 [30] was financially supported by the Research Authors’ disclosures available online (https:// School Behavioral and Cognitive Neurosciences www.j-alz.com/manuscript-disclosures/20-1427r3). (RUG/UMCG) and the Gratama Stichting/Stichting Groninger Universiteitsfonds (2015-04). In Belgium, Institute Born-Bunge/University of Antwerp was SUPPLEMENTARY MATERIAL granted a subsidy from Research Foundation Flan- ders (G053218N). In Spain, this study was supported The supplementary material is available in the by Fondo de Investigaciones Sanitario, Instituto de electronic version of this article: https://dx.doi.org/ SaludCarlos III (PI14/01126, PI17/01019 to JF) and 10.3233/JAD-201427. the CIBERNED program, partly jointly funded by Fondo Europeo de Desarrollo Regional, Unión Euro- pea, Una manera de hacer Europa. This work was REFERENCES also supported by the National Institutes of Health (NIA grants 1R01AG056850–01A1, R21AG056974, [1] de Graaf G, Engelen JJM, Gijsbers ACJ, Hochstenbach R, R01AG061566 to JF), Fundació La Marató de TV3 Hoffer MJV, Kooper AJA, Sikkema-Raddatz B, Srebniak MI, van der Kevie-Kersemaekers AMF, van Zutven LJCM, (20141210 to JF), Fundació Catalana Sı́ndrome de Voorhoeve E (2017) Estimates of live birth prevalence of Down, Fundació Vı́ctor Grı́fols i Lucas and General- children with Down syndrome in the period 1991–2015 in itat de Catalunya (SLT006/17/00119 to JF). the Netherlands. J Intellect Disabil Res 61, 461-470. The authors are grateful for the participation of [2] Loane M, Morris JK, Addor M-CC, Arriola L, Budd J, Doray B, Garne E, Gatt M, Haeusler M, Khoshnood B, all caregivers/familiy members for being infor- Klungsøyr Melve K, Latos-Bielenska A, McDonnell B, mants as well as support staff of the various Mullaney C, O’Mahony M, Queißer-Wahrendorf A, Rankin care institutions for the local organization and J, Rissmann A, Rounding C, Salvador J, Tucker D, Wellesley administration. We wish to thank all interview- D, Yevtushok L, Dolk H, Klungsoyr Melve K, Latos- Bielenska A, McDonnell B, Mullaney C, O’Mahony M, ers who took part in the project (per institute): Queißer-Wahrendorf A, Rankin J, Rissmann A, Rounding Mylou Pool, Marije Ravesteijn, Nadi de Vos C, Salvador J, Tucker D, Wellesley D, Yevtushok L, Dolk H (Amerpoort), Lizan Exterkate, Rachel Kemna, Nar- (2013) Twenty-year trends in the prevalence of Down syn- dine Lukassen, Danielle Oosterling, Carla Wensink drome and other trisomies in Europe: Impact of maternal age and prenatal screening. Eur J Hum Genet 21, 27-33. (Aveleijn), José Eleveld, Desiree van Leth, Sharina [3] Ballard C, Mobley WC, Hardy J, Williams G, Corbett A Grefelman, Cobien Wever (Cosis), Sanne Apper- (2016) Dementia in Down’s syndrome. Lancet Neurol 15, loo, Eline de Jong, Sandra Kleijer (De Twentse 622-636. [4] Mann DMA (1988) Alzheimer’s disease and Down’s syn- Zorgcentra), Jade Oostrom (Dichterbij), Marscha drome. Histopathology 13, 125-137. Brunia, Rosanne Derksen, Esther Scholten, Marije [5] McCarron M, McCallion P, Reilly E, Dunne P, Carroll van Dijk-van der Werff (Elver), Lyanne Hassefras, R, Mulryan N (2017) A prospective 20-year longitudinal Marloes Lansbergen, Simone Wilson-Koudenburg follow-up of dementia in persons with Down syndrome. J Intellect Disabil Res 61, 834-852. (Ipse de Bruggen), Janette Drolinga, Tineke Arts- [6] Fortea J, Vilaplana E, Carmona-Iragui M, Benejam B, Wiegmink (Nieuw Woelwijck), Hannelore Broers, Videla L, Barroeta I, Fernández S, Altuna M, Pegueroles Regine van Duijvenboden, Maarten Faber, Annebeth J, Montal V, Valldeneu S, Giménez S, González-Ortiz S, Maarsman, Lilian van der Meer, Romy Schegget- Muñoz L, Estellés T, Illán-Gala I, Belbin O, Camacho V, Wilson LR, Annus T, Osorio RS, Videla S, Lehmann S, Hol- man, Karen van de Weijer (Philadelphia), Merel land AJ, Alcolea D, Clarimón J, Zaman SH, Blesa R, Lleó Duim, Kirsten Hendrickx, Britta te Nijenhuis, Loes A (2020) Clinical and biomarker changes of Alzheimer’s Velner (Reinaerde), Eden van den Akker, Lobke disease in adults with Down syndrome: A cross-sectional Berendsen (Severinus), Fenna Blaas, Herco Elbert- study. Lancet 395, 1988-1997. [7] Janicki MP (2011) Quality outcomes in group home demen- sen, Marjon van der Poel (Sherpa), Marieke Groen, tia care for adults with intellectual disabilities. J Intellect Corien Rikkers-van Nes, Corine van Essen (’s Heeren Disabil Res 55, 763-776. Loo), Annemarieke Bronswijk, Ybelina de Jong- [8] Finkel SI (2001) Behavioral and psychological symptoms van der Meulen (Sprank), Natascha Albers, Eva of dementia: A current focus for clinicians, researchers, and caregivers. J Clin Psychiatry 62 Suppl 2, 3-6. Smit, Nienke Stap (Talant/Alliade Care Group), [9] Gauthier S, Cummings J, Ballard C, Brodaty H, Grossberg Aart-Jan Lenstra, Tijs van der Linden, Marian Roffel- G, Robert P, Lyketsos C, Brodaty H, Grossberg G, Robert P de Jong (Vanboeijen), Monique Bomer-Veenboer, (2010) Management of behavioral problems in Alzheimer’s disease. Int Psychogeriatrics 22, 346-372. Lisette Meinster, Deliah Ormskerk (Zuidwester), [10] Cerejeira J, Lagarto L, Mukaetova-Ladinska EB (2012) Gianluca Radice (Institut Lejeune), and Laura Videla Behavioral and psychological symptoms of dementia. Front (Sant Pau). Neurol 3, 73. 1526 A.D. Dekker et al. / BPSD in Down Syndrome [11] Kales HC, Gitlin LN, Lyketsos CG (2015) Assessment and [26] Janowsky DS, Barnhill LJ, Davis JM (2003) Olanzapine management of behavioral and psychological symptoms of for self-injurious, aggressive, and disruptive behaviors in dementia. BMJ 350, h369. intellectually disabled adults: A retrospective, open-label, [12] Finkel SI (2000) Introduction to behavioural and psycholog- naturalistic trial. J Clin Psychiatry 64, 1258-1265. ical symptoms of dementia (BPSD). Int J Geriatr Psychiatry [27] Cummings JL, Mega MS, Gray K, Rosenberg-Thompson 15(Suppl 1), S2-4. S, Carusi DA, Gornbein J (1994) The Neuropsychiatric [13] Adams D, Oliver C, Kalsy S, Peters S, Broquard M, Basra T, Inventory: Comprehensive assessment of psychopathology Konstandinidi E, McQuillan S (2008) Behavioural charac- in dementia. Neurology 44, 2308-2314. teristics associated with dementia assessment referrals in [28] Cummings JL (1997) The Neuropsychiatric Inventory: adults with Down syndrome. J Intellect Disabil Res 52, Assessing psychopathology in dementia patients. Neurology 358-368. 48, S10-16. [14] Iacono T, Bigby C, Carling-Jenkins R, Torr J (2014) Taking [29] Dekker AD, Strydom A, Coppus AMW, Nizetic D, Ver- each day as it comes: Staff experiences of supporting peo- meiren Y, Naudé PJW, Van Dam D, Potier M-C, Fortea ple with Down syndrome and Alzheimer’s disease in group J, De Deyn PP (2015) Behavioural and psychologi- homes. J Intellect Disabil Res 58, 521-533. cal symptoms of dementia in Down syndrome: Early [15] Cleary J, Doody O (2017) Nurses’ experience of caring for indicators of clinical Alzheimer’s disease? Cortex 73, people with intellectual disability and dementia. J Clin Nurs 36-61. 26, 620-631. [30] Dekker AD, Sacco S, Carfi A, Benejam B, Vermeiren Y, [16] Sheehan R, Sinai A, Bass N, Blatchford P, Bohnen I, Beugelsdijk G, Schippers M, Hassefras L, Eleveld J, Grefel- Bonell S, Courtenay K, Hassiotis A, Markar T, McCarthy man S, Fopma R, Bomer-Veenboer M, Boti M, Oosterling J, Mukherji K, Naeem A, Paschos D, Perez-Achiaga N, GDE, Scholten E, Tollenaere M, Checkley L, Strydom A, Sharma V, Thomas D, Walker Z, Strydom A (2015) Demen- Van Goethem G, Onder G, Blesa R, zu Eulenburg C, Cop- tia diagnostic criteria in Down syndrome. Int J Geriatr pus AMW, Rebillat A-S, Fortea J, De Deyn PP (2018) The Psychiatry 30, 857-863. Behavioral and Psychological Symptoms of Dementia in [17] Keller SM, Janicki MP, Esralew L (2016) Dementia: Screen- Down Syndrome (BPSD-DS) scale: Comprehensive assess- ing, evaluation, diagnosis and management. In Health Care ment of psychopathology in Down syndrome. J Alzheimers for People with Intellectual and Developmental Disabili- Dis 63, 797-820. ties across the Lifespan, Rubin IL, Merrick J, Greydanus [31] Hithersay R, Hamburg S, Knight B, Strydom A (2017) Cog- DE, Patel DR, eds. Springer International Publishing, pp. nitive decline and dementia in Down syndrome. Curr Opin 1449-1463. Psychiatry 30, 102-107. [18] Chapman M, Lacey H, Jervis N (2018) Improving services [32] McKhann GM, Knopman DS, Chertkow H, Hyman BT, for people with learning disabilities and dementia: Find- Jack CR, Kawas CH, Klunk WE, Koroshetz WJ, Manly ings from a service evaluation exploring the perspectives of JJ, Mayeux R, Mohs RC, Morris JC, Rossor MN, Schel- health and social care professionals. Br J Learn Disabil 46, tens P, Carrillo MC, Thies B, Weintraub S, Phelps CH 33-44. (2011) The diagnosis of dementia due to Alzheimer’s dis- [19] Lautarescu BA, Holland AJ, Zaman SH (2017) The early ease: Recommendations from the National Institute on presentation of dementia in people with Down syndrome: Aging-Alzheimer’s Association workgroups on diagnostic A systematic review of longitudinal studies. Neuropsychol guidelines for Alzheimer’s disease. Alzheimers Dement 7, Rev 27, 31-45. 263-269. [20] Bekkema N, De Veer AJE, Hertogh CMPM, Francke AL [33] World Health Organization. International Statistical Clas- (2015) “From activating towards caring”: Shifts in care sification of diseases and Related Health Problems, 10th approaches at the end of life of people with intellectual dis- Revision. abilities; A qualitative study of the perspectives of relatives, [34] Trevethan R (2017) Sensitivity, specificity, and predictive care-staff and physicians. BMC Palliat Care 14, 33. values: Foundations, pliabilities, and pitfalls in research and [21] Ogg-Groenendaal M, Hermans H, Claessens B (2014) A practice. Front Public Health 5, 307. systematic review on the effect of exercise interventions on [35] McHugh ML (2012) Interrater reliability: The kappa statis- challenging behavior for people with intellectual disabili- tic. Biochem Med 22, 276-282. ties. Res Dev Disabil 35, 1507-1517. [36] Feinstein AR, Cicchetti DV (1990) High agreement but low [22] Heyvaert M, Saenen L, Maes B, Onghena P (2014) Sys- kappa: I. The problems of two paradoxes. J Clin Epidemiol tematic review of restraint interventions for challenging 43, 543-549. behaviour among persons with intellectual disabilities: [37] Benjamini Y, Hochberg Y (1995) Controlling the false dis- Focus on effectiveness in single-case experiments. J Appl covery rate: A practical and powerful approach to multiple Res Intellect Disabil 27, 493-510. testing. J R Stat Soc Series B Methodol 57, 289-300. [23] Nicoll M, Beail N, Saxon D (2013) Cognitive behavioural [38] Alzheimer’s Association (2020) 2020 Alzheimer’s disease treatment for anger in adults with intellectual disabilities: facts and figures. Alzheimers Dement 16, 391-460. A systematic review and meta-analysis. J Appl Res Intellect [39] Gottesman RT, Stern Y (2019) Behavioral and psychiatric Disabil 26, 47-62. symptoms of dementia and rate of decline in Alzheimer’s [24] Chowdhury M, Benson BA (2011) Use of differential disease. Front Pharmacol 10, 1062. reinforcement to reduce behavior problems in adults with [40] Sabbagh M, Edgin J (2016) Clinical assessment of cognitive intellectual disabilities: A methodological review. Res Dev decline in adults with Down syndrome. Curr Alzheimer Res Disabil 32, 383-394. 13, 30-34. [25] Hässler F, Reis O (2010) Pharmacotherapy of disrup- [41] Zigman WB, Devenny DA, Krinsky-McHale SJ, Jenkins tive behavior in mentally retarded subjects: A review EC, Urv TK, Wegiel J, Schupf N, Silverman WP (2008) of the current literature. Dev Disabil Res Rev 16, Alzheimer’s disease in adults with Down syndrome. Int Rev 265-272. Res Ment Retard 36, 103-145. A.D. Dekker et al. / BPSD in Down Syndrome 1527 [42] Jamieson-Craig R, Scior K, Chan T, Fenton C, Strydom [45] Llewellyn P (2011) The needs of people with learning A (2010) Reliance on carer reports of early symptoms of disabilities who develop dementia: A literature review. dementia among adults with intellectual disabilities. J Policy Dementia 10, 235-247. Pract Intellect Disabil 7, 34-41. [46] Dodd KD (2015) Care considerations for dementia in peo- [43] Sturmey P, Reed J, Corbett J (1991) Psychometric assess- ple with Down’s syndrome: A management perspective. ment of psychiatric disorders in people with learning Neurodegener Dis Manag 5, 293-304. difficulties (mental handicap): A review of measures. Psy- [47] Molony SL, Kolanowski A, Van Haitsma K, Rooney chol Med 21, 143-155. KE (2018) Person-centered assessment and care planning. [44] Dekker AD, Wissing MBG, Ulgiati AM, Bijl B, van Gool Gerontologist 58, S32-S47. G, Groen MR, Grootendorst ES, van der Wal IA, Hobbe- [48] Kaufer DI, Cummings JL, Ketchel P, Smith V, MacMillan len JSM, De Deyn PP, Waninge A. Dementia in people A, Shelley T, Lopez OL, DeKosky ST (2000) Validation with severe or profound intellectual (and multiple) disabil- of the NPI-Q, a brief clinical form of the Neuropsychiatric ities: Focus group research into relevance, symptoms and Inventory. J Neuropsychiatry Clin Neurosci2 12, 233-239. training needs. Journal of Applied Research in Intellectual Disabilities. In press.