Psychiatrists and specialist mental healthcare nurses affiliated with ID mental healthcare organizations.

Table 2 shows data which complicated the withdrawal process of antipsychotics for challenging behaviour.

Most participants stated that the antipsychotic drug had to be re-instated in some patients. Furthermore, behavioural worsening or other complications like withdrawal dyskinesia or re-emergence of sleep problems or irritability could occur, leading to delay in the discontinuation process. In addition, staff could be unable or need extra time to adapt their guiding style, or to change clients' environment or daily activities when clients became less sedated and more active because of the antipsychotic's withdrawal. These environmental factors could also lead to delay or sometimes failure in the discontinuation process. Nevertheless, complete withdrawal was also achieved in some of the patients, even those whose behaviour deteriorated upon dose reduction.

#### *3.3. Factors Related to Deprescribing Antipsychotics in the Working Field of Participants*

Twenty-three (70%) of the thirty-four participants indicated that the institutions they were affiliated with had encouraged their clinicians to reduce antipsychotic drugs because of the new Dutch Act on Involuntary care, which is aimed to reduce restrictive practice and the new Dutch guideline on problem behaviour in adults with ID. They also indicated that they were less likely to initiate antipsychotics for challenging behaviours.

Participants also mentioned some factors that facilitated successful discontinuation and barriers that hindered this process. For example, it was difficult to achieve complete withdrawal among those who received polypharmacy of antipsychotics. On the contrary, the concurrent use of other psychotropics (polypharmacy of psychotropics rather than antipsychotics) facilitated complete withdrawal. Other facilitators were the presence of side effects, first attempt to withdraw, and a positive attitude and involvement of clients' families and support staff. Barriers were resistance against discontinuation from support professionals or family, which was mentioned by 26 (76%) and by 25 (74%) participants, respectively, and unavailability of non-pharmaceutical treatments for clients' challenging behaviours, which was mentioned by 20 (59%) participants. In addition, lack of knowledge about the discontinuation process, e.g., time schedules and steps in dose reductions (20 participants, 59%), and lack of multidisciplinary team involvement (8 participants, 24%) were mentioned.

**Table 2.** Complications in withdrawing antipsychotics for challenging behaviour in patients with intellectual disabilities as experienced by participants in a survey study (N = 34).


#### **4. Discussion**

In this study, we explored the experiences and views of prescribers regarding withdrawal of long-term antipsychotic drug use for challenging behaviours in people with ID in The Netherlands. Although the long-term prescription of antipsychotics for challenging behaviours in the absence of a psychiatric illness is not recommended [16,46], we found that approximately half of the 21 ID physicians and 13 psychiatrists/specialist mental healthcare nurses who completed the survey questionnaire were still prescribing antipsychotics in the absence of a psychiatric diagnosis in 0–25% of their patients. However, all ID physicians and all but one of psychiatrists/mental healthcare nurses had attempted to deprescribe antipsychotics in the preceding five years. This percentage is higher than in the survey of Deb et al. [44], who found that about half of their 88 respondents had attempted to deprescribe antipsychotics for challenging behaviour in their patients with ID. We found a similar percentage to Deb et al. [44] (65% versus 60% of participants) who succeeded in complete withdrawal in 0–25% of their patients. However, none of the participants in our study was successful in completely discontinuing antipsychotic medications in over 50% of their patients within 12 months compared to 4.5% of participants in the study of Deb et al. [44]. Our participants stated that objections and concerns of family and professional caregivers, and unavailability of non-pharmaceutical treatments and multidisciplinary team input, were barriers in starting or continuing a withdrawal process.

Deb et al. [48] recently completed a qualitative analysis of free text data returned by the UK psychiatrists' during the questionnaire survey [44] and found that the psychiatrists in the UK like the participants in the current Dutch study found that the lack of resources such as multidisciplinary team input and lack of caregiver support for the withdrawal hampered their attempt to deprescribe antipsychotics. As far as we know, there are no studies from other countries reporting prescribers' experiences with antipsychotics discontinuation in patients with ID.

As described above and known from other studies, involvement of all stakeholders in decisions about the withdrawal of long-term used antipsychotics for challenging behaviours is a key factor for success [43,44]. Besides organizational barriers and timeconstraints in prescribers' attempts to apply the deprescribing policies [44], staff's attitudes and knowledge around psychotropic drug use are important factors in outcomes of discontinuation trajectories [40]. However, the most important factor may be the involvement of family-caregivers and patients with ID themselves. Studies have shown that familycaregivers, although they have knowledge on the challenging behaviour of their relative and would like to be heard, are often not involved in treatment decisions, e.g., psychotropic drug prescribing [41]. In addition, often individuals with ID themselves lack information and/or their voices are not heard in decisions in prescription of antipsychotics [41,49]. Previous studies of interviews of adults with ID have just touched on the issue of psychotropic medication use [50,51]. To address this knowledge gap, in another study, we also gathered information on the experience of adults with a mild ID who have gone through the experience of psychotropic withdrawal [52]. An important finding was their statement that their own coping style, next to a supportive environment and a good relationship with their doctor, was a key factor for successful discontinuation.

Legal issues may play a role. We found that 70% of the participants in our study indicated that the new Dutch Act on Involuntary care, aimed to reduce restrictive practice, and the new Dutch guideline on problem behaviour in adults with ID had encouraged their service providers to facilitate psychotropic drugs deprescription. However, clinicians are allowed to deviate from this Act and from the Dutch guideline on problem behaviour provided that they can substantiate this decision. Since the present study was the first in The Netherlands on the issue of experiences of prescribers in withdrawing antipsychotics for challenging behaviours of people with ID, it is difficult to know definitely about the possible facilitating effect of these government Act and guidelines, for which future studies are needed.

Other contextual factors also may play a role in our study, since we found differences in antipsychotics discontinuation policies between ID physicians and ID mental healthcare professionals, who have different working settings. ID physicians took significantly more time for the discontinuation process, possibly because their patients' stakeholders' networks are more complex, and it will take more time to coordinate all stakeholders' interests. Another explanation for the longer discontinuation trajectory may be the difference in confirmation of the presence of a psychiatric illness in patients who started the withdrawal process. This was significantly less often confirmed by ID physicians compared with ID mental healthcare professionals. The lack of confirmation will likely lead to uncertainty and fear of family, caregivers and clinicians regarding re-occurrence of maladaptive behaviour or psychiatric symptoms, and delay in the withdrawal process.

Our study has some limitations. First, the response rate of our survey was low. Approximately 10% of ID physicians had completed the survey questionnaire, and we estimated a similar percentage of the ID mental healthcare professionals (psychiatrists and specialist nurses working in ID mental healthcare and/or affiliated with ID institutions). We were not able to calculate the exact response rate of the latter group because there is no data on the number of those professionals. Therefore, the results of our survey may not be representative of all ID physicians, ID psychiatrists and ID specialist nurses in The Netherlands. However, the low response itself raises some interesting observations. A similar survey conducted two years ago in the UK with the same equivalent target audience i.e., psychiatrists who work with people with ID had a response rate of approximately 40%. This raises questions on the differences in attitudes of the prescribers towards this challenging topic. There could be complex confounders, such as time, resource, training, education, knowledge acquisition and cultural attitudes, which might have influenced these response rates. It could be that a different methodology such as face to face interviews might result in better response rates and better understanding of the barriers to the current survey response. However, such projects can be time consuming and resource intensive.

Furthermore, because this study was conducted in Dutch healthcare settings, the results cannot be generalized to other countries. However, it is known that, in many countries, prescribers encounter the same problems in deprescribing antipsychotics in their patients with ID [37]. Therefore, the results of this study may also be applicable to other Western countries.

#### **5. Conclusions**

In line with the available international guidelines and a new Dutch Act on Involuntary care, most of the thirty-four participants in our survey of ID physicians, psychiatrists and specialist mental healthcare nurses in The Netherlands had attempted to discontinue their long-term prescriptions of antipsychotics for challenging behaviours in patients with ID. However, complete discontinuation could be achieved in only a small proportion of cases. Concerns of relatives and care and support staff, lack of non-pharmaceutical treatments and lack of multidisciplinary teams' input were perceived as barriers to successful antipsychotic withdrawal attempts.

Besides improvement in the availability of multidisciplinary teams and non-pharmaceutical treatments, education of care-professionals in supporting clients in the management of medication use is recommended.

Facilitation and support from managers at the institutional level, and creating conditions at the political and professional level, seem to be prerequisites in the successful implementation of deprescribing antipsychotic medication for challenging behaviours in patients with ID.

**Supplementary Materials:** The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/ijerph192417095/s1, Supplementary File S1. Survey among prescribers of psychotic drugs to people with intellectual disabilities.

**Author Contributions:** The study was set up and conducted, and analysis of the results was done by J.d.H. under supervision of G.d.K. The conceptualization of the survey was done by S.D. and R.S. The writing of drafts was done by G.d.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** A written and signed participants' consent was not required because this was not a study involving patients, but volunteers who talked about their experiences.

**Data Availability Statement:** The data are available on request and with a substantiated explanation at GGZ Drenthe/department research https://ggzdrenthe.nl/research, (accessed on 14 December 2022).

**Acknowledgments:** We thank the intellectual disability physicians, psychiatrists and specialist nurses who completed the survey for their cooperation.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Withdrawing Antipsychotics for Challenging Behaviours in Adults with Intellectual Disabilities: Experiences and Views of Experts by Experience**

**Gerda de Kuijper 1,2,\*, Joke de Haan 1, Shoumitro Deb <sup>3</sup> and Rohit Shankar <sup>4</sup>**


**Abstract:** People with intellectual disabilities (PwID) are frequently prescribed long-term antipsychotics for behaviours that challenge (BtC) despite the lack of proven effectiveness and the increased risks for side effects of these medications in this population. National and international good clinical practice guidelines recommend deprescribing antipsychotics for BtC, which is often not successful due to environmental and other factors. The involvement of all stakeholders, including PwID, is crucial for deprescribing. However, studies showed that PwID and/or their families are often not involved in decision-making regarding the (de)prescribing of antipsychotics despite their desire to get involved. Moreover, studies on the views of PwID regarding their experiences of withdrawing from antipsychotics are lacking. The aim of this study was to gain insight into the views of PwID by investigating their experiences of discontinuation of long-term prescribed antipsychotics for BtC. A qualitative study was set up. Seven experts by experience with mild intellectual disabilities were interviewed. After six interviews, data saturation was achieved. Interviews were transcribed verbatim. Using phenomenological analysis, themes on lived experiences were extracted. Each consecutive interview was analysed. The four main themes extracted from the interviews were the quality of treatment, knowledge and information about psychotropics and the process of withdrawal, support from the participants' environment and the coping style of the interviewees themselves.

**Keywords:** intellectual disabilities; antipsychotics; challenging behaviour; discontinuation; experts by experience views; qualitative study; interviews

#### **1. Introduction**

People with intellectual disabilities (PwID) frequently engage in behaviours that challenge (BtC), especially those with higher care needs and/or who receive residential care [1]. BtC include aggressive behaviour (e.g., verbal abuse, threats and physical violence), destructive behaviour (e.g., breaking or destroying properties and other objects and setting fires), disruptive behaviour (e.g., repetitive screaming, smearing faeces, setting off fire alarms when there is no fire, calling the emergency services when there is no emergency), self-injurious behaviour (e.g., self-biting, head-banging) and sexually harmful behaviour (e.g., sexual assaults, rape and stalking), as described in the National Institute of Care Excellence (NICE, NG11) [2] guidelines.

There are many reasons for BtC, including biological (genetic disorders, physical problems including pain, medication side effects), psychological (psychiatric disorders including anxiety, depression, psychosis and post-traumatic stress disorders) and social factors (inappropriate accommodation and lack of support) [3]. The reported prevalence of

97

**Citation:** de Kuijper, G.; de Haan, J.; Deb, S.; Shankar, R. Withdrawing Antipsychotics for Challenging Behaviours in Adults with Intellectual Disabilities: Experiences and Views of Experts by Experience. *Int. J. Environ. Res. Public Health* **2022**, *19*, 15637. https://doi.org/10.3390/ ijerph192315637

Academic Editor: Paul B. Tchounwou

Received: 13 October 2022 Accepted: 22 November 2022 Published: 24 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

psychiatric disorders in PwID varies between 14.4 and 60% depending on the definition of psychiatric diagnosis used (e.g., the inclusion or exclusion of BtC, ADHD and ASD within the psychiatric disorder diagnosis), method of detection using different classification systems (e.g., ICD vs. DSM), use of assessment methods (screening instrument vs. structured interviews), population studied (institutionalised vs. clinic-based vs. communitybased), etc. [4–6]. Neurodevelopmental disorders, such as autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD), are common among PwID [7]. The rate of both psychiatric disorders and BtC increases in PwID with these comorbidities [7,8]. However, both false positive and false negative diagnoses of psychiatric disorders are common in PwID [6].

In the UK, NHS England (NHSE) embarked on a major campaign six years ago called 'STopping Over-Medication of People with learning disabilities, autism or both (STOMP)' [9] to which STAMP (supporting treatment and appropriate medication in paediatrics) was also added recently. The Royal College of Psychiatrists in the UK also published a position paper to support the STOMP STAMP initiative [10]. The National Institute for Health and Care Excellence (NICE) in the UK [2] and the World Psychiatric Association [11] developed guidelines for the use of psychotropics to address BtC among PwID, including recommendations for the initiation, monitoring and potential withdrawal of psychotropic medications.

BtC is best-treated non-pharmacologically, but in some situations, short-term prescriptions of psychotropics may be beneficial [2]. However, BtC is often treated with antipsychotics and prescribed long-term in the absence of a valid psychiatric diagnosis [12,13] despite the lack of evidence for its long-term effectiveness for BtC [14–17]. Moreover, PwID are vulnerable to the side effects of antipsychotics, especially neurological side effects, such as movement disorders; sedation/sleepiness; and metabolic side effects, such as obesity and diabetes [15,18–20]. Therefore, it is important to balance the pros and cons of antipsychotic use carefully and to discontinue them when there is no longer a valid indication, especially in the case of BtC [2].

Furthermore, it is important to involve all the stakeholders. This means that along with staff and healthcare professionals, the PwID themself, their family members and other representatives of PwID showing with BtC should get involved in the decision-making of the (de)prescribing of psychotropic drugs [21]. Indeed, studies showed that PwID and their family members would like to contribute to treatment decisions about BtC and medication use but they are often not involved in decision-making regarding the (ongoing) prescription of medication [22–24].

Although national and international guidelines recommend [2,11,25] that long-term antipsychotics for BtC should be deprescribed, discontinuation often fails [26–29] because of a variety of factors. It is important to involve all stakeholders to increase the proportion of successful discontinuation, as was shown in a study by Shankar and colleagues [28]. In particular, the role of the primary caregivers and other staff should be recognised as the key factors that influence the successful discontinuation of antipsychotics [30–32], as staff may lack knowledge of the effects and side effects of antipsychotics and may have unrealistic expectations of the beneficial effects of psychotropic medications on the psychological functioning of their clients [30,31]. In a recent study of interviews with support staff in the UK, some direct care staff felt that medication is necessary to control BtC in PwID, whereas others felt that these medications were 'chemical restraints' [32].

Furthermore, evaluation of the process of discontinuation is important to identify potential barriers and facilitators for the deprescribing of these agents. A survey among UK psychiatrists showed that family and staff concerns, lack of multidisciplinary input and unavailability of psychosocial interventions were barriers to attempting to discontinue antipsychotics [33]. A Dutch study among direct support professionals about their perceptions of the effect of antipsychotics on BtC and their willingness to cooperate in discontinuing these agents showed that they believed that antipsychotics were effective in the management of BtC [34]. However, they also observed the negative consequences

of side effects. Staff were willing to cooperate in reducing the antipsychotic drug use but were not confident about successful complete discontinuation [34].

The current study aimed to add knowledge about facilitators and barriers to the successful discontinuation of antipsychotic drug use for BtC in PwID. Although studies were done on the experiences and views of prescribers and support professionals regarding withdrawing antipsychotics for BtC, as far as we know, studies on users' perspectives are missing. Therefore, in this study, we investigated the views of people with mild ID on the discontinuation of their antipsychotic drugs and their experiences with the withdrawal process.

#### **2. Methods**

#### *2.1. Design and Participants*

This was a qualitative interview study. Potential participants were selected by their clinicians according to preset eligibility criteria. Eligible participants were adults aged between 18 and 65 years and had mild intellectual disabilities according to the DSM5 classification system (Diagnostic and Statistical Manual of Mental Disorders, 5th revised edition; American Psychiatric Association, 2013). They had to be currently undergoing withdrawal of antipsychotic medications that they had been receiving for more than six months, where either their antipsychotics had already been discontinued recently or they had gone through an attempted withdrawal in the recent past. Potential participants were required to have competency and language skills to take part in the interviews and be able to express their views in the Dutch language. Participants with dementia, bipolar disorder, schizophrenia and chronic psychosis were excluded.

#### *2.2. Procedures*

Potential participants were recruited from the responders of a survey that had previously been carried out among Dutch physicians specialising in intellectual disabilities, psychiatrists and mental healthcare specialist nurses [35]. Additionally, the first two authors recruited participants from their networks and by asking their colleagues at the outpatient clinic of the mental healthcare organisations they were affiliated with that provide services for people with intellectual disabilities.

Potential participants were contacted on the phone or visited by the second author (J.d.H.). During these calls or visits, they were informed about the goal and content of the interviews. In these meetings and the interview sessions, the interviewer (J.d.H.) used accessible language, such as short sentences and clear questions. Furthermore, the interviewees could ask a significant other, such as their personal support professional or family member, to accompany them during the interviews.

A topic guide was used for these interviews (see Table 1), the items of which were drawn from the relevant literature on the subject [21–25,36,37].

**Table 1.** Topic guide with subjects for interviews on users' experiences with the withdrawal of antipsychotics.


Reasons for antipsychotic drug use

Reason for discontinuation of antipsychotics

Information the participants received about their medication use and the discontinuation process Mental and physical health problems experienced during the discontinuation process Opinions of the participants about the decision and process of discontinuation Tips or advice from the participants for the professionals and other PwID about the

discontinuation process

A test interview was done by the second author (J.d.H.) to assess the accessibility of the content and comprehensibility of the interview questions. This test interview took place in the inpatient clinic of the mental healthcare organisation the interviewer was affiliated with. The interviewee was a client who was currently undergoing withdrawal from her antipsychotics. This interview was not recorded and no data from this interview were included in the results section. The test interviewee revealed that there was no need to adapt the topic guide, language use or other subjects.

#### *2.3. Ethics*

All participants consented to take part in the interviews and signed an informed consent form before the interview took place. They were informed that the encrypted data would be stored at a safeguarded place and analysed and published anonymously and that they could withdraw from the study at any moment without giving a reason.

The interview study was not a medical, scientific study according to the Dutch Act on Medical Studies with Human Beings, as no personal health-related data were collected, but only the participants' views were captured through these interviews. Therefore, no ethical approval was required for this study. The study was carried out under the responsibility of Mental HealthCare Drenthe (GGZ Drenthe) and was approved by their research committee (declaration is available on request). All data were collected, stored and safeguarded anonymously according to the European Act on Protection of Personal Information, which, in the Netherlands, was ratified in 2018.

#### *2.4. Analyses*

The interviews were video-recorded and transcribed verbatim. A phenomenological data analysis method was used to extract themes from the transcriptions by means of (a) thorough reading and re-reading of the transcripts to obtain a general sense of the whole content; (b) extraction of significant statements on the phenomenon under study and formulation and coding of the meaning of these statements; (c) sorting of categories, clusters of themes, and themes of these codes and meanings; and (d) integration of the findings and description of the fundamental structure of the phenomenon that ought to be validated by the interviewees [38].

The sample size was established according to the criteria of data saturation in qualitative studies, which means that qualitative data were collected to the point where a sense of closure was attained because new data yielded redundant information [39]. In the present study, each consecutive interview was analysed. When the last interview yielded no new codes and themes, data saturation was achieved and no new interviews/new participants were needed.

Transferability was maintained (via a thorough reading and re-reading, as well as a rich description of the content of the interviews), along with credibility; confirmability; authenticity via peer-review of the coding of the meanings, categories and themes of the statements; and dependability via the interviewer's log-keeping [40].

Data were analysed by the second author (J.d.H.). A random sample of three interviews (out of seven in total) was independently analysed by the first author (G.d.K.) to assess the agreement with the analyses of J.d.H.

#### **3. Results**

Seven experts by experience with mild intellectual disabilities participated in the interview study. These were three men and four women, whose ages ranged from 22 to 62 years. Three used quetiapine; two used aripiprazole; one used olanzapine; and one used a combination of haloperidol, quetiapine and clozapine. All participants were white. The duration of the interviews was 25 min on average. After each interview, the extraction of significant statements and coding and categorising their meanings took place. After the sixth interview, no new categories and themes were extracted; therefore, data saturation was achieved. Four main themes could be extracted from the content of the interviews. These were (a) the quality of the treatment, (b) knowledge and information about the process of withdrawal, (c) the conditions that needed to be met before a discontinuation process could start and (d) the coping style of the person whose medication was discontinued.

(a) The quality of treatment.

All participants indicated that they had used antipsychotics for many years because of various reasons, but none was for a licensed indication. For example, one participant stated:

*'Finally, after many years, it was found that I have ADHD, and then quetiapine won't 'work very well, and Concerta will work better, you might say.'*

The reasons to discontinue antipsychotics were mainly because participants judged that they no longer had symptoms that needed treatment and/or because of side effects, such as weight gain, tremors, flattened affect and sleepiness. Participants stressed the importance of receiving good treatments from clinicians who take them seriously and provide a clear and understandable explanation about their treatment and the reason for the prescription of their medication, such as (i) What is it for? (ii) How does it work? (iii) What side effects could occur? (iv) When to consult my doctor? However, often they had not been provided explanation and information when the medicine was newly prescribed.

(b) Knowledge and information about the process of withdrawal.

Most participants had some fear of discontinuing because they were afraid that their behavioural problems may re-emerge.

*'I was afraid it turned out to go wrong.'*

All participants indicated that they and their caregivers were provided with oral information by their doctor or nurse about the discontinuation schedule and dose reductions and the possibility of withdrawal symptoms and changes in behaviour. Four participants experienced withdrawal symptoms, and most of them knew this could happen and what kind of symptoms could be expected.

*'I was like, this is just part of it, and then I went reading or doing something else and then it disappeared.'*

All participants had informed their caregivers, peers or colleagues that they were discontinuing their drugs and the possibility of withdrawal symptoms.

*'They have to know I may react differently.'*

(c) Conditions to be met before starting a discontinuation process according to participants' experiences.

Half of the participants found the withdrawal process difficult for various reasons. One participant had problems with providing care for her children, others had physical and emotional problems. In one case, a dose increase of the antipsychotic drug was necessary to relieve these symptoms.

All participants indicated the importance of a good, accessible and trusted relationship with their doctors. This was helpful in continuing their withdrawal process when they had questions or wanted to talk about negative feelings when they experienced difficulties with the withdrawal process.

*'All went well in agreement, I could always phone or come for consultation'.*

Moreover, others in the participants' environments were important for providing support.

*'Support from my parents, brother, sister-in-law, grandmother and grandfather.'*

*'Good to talk about it with others around you, anyway I think this is important because you won't make it on your own'.*

(d) Coping style of the participant.

Most participants indicated that an appropriate mindset and intrinsic motivation are important to start with and continue the discontinuation process.

*'Yes, you have to switch and stick to it.'*

*'Just say to yourself, I can make it.'*

*'It's also your own mindset, how to deal with yourself when you are afraid to discontinue, then it will be difficult. This was my experience, when you switch your mindset appropriately, then it will be all right.'*

Participants also provided tips and advice for their peers. 'Ask help'.

*'When you are angry, go to your support professional', 'stick to a structure in your daily activities.'*

#### **4. Discussion**

In this study, we explored the experiences and views of adults with mild ID regarding the withdrawal from their long-term-used antipsychotic drugs prescribed for BtC. This process may be difficult because PwID are often dependent on others and/or easily influenced by others and often need support in decision-making around deprescribing and maintaining the process of discontinuation [22,24,41].

Recent studies in the UK and the Netherlands have shown that even after longterm use, it is possible to discontinue antipsychotic medication in 25–61% of adults with intellectual disabilities and have a 50% dose reduction in another 11–19%, although, in up to 20% of cases, antipsychotics were re-instated within 3–4 years of discontinuation, primarily due to the resurgence of BtC [15,26,28]. However, several internal (e.g., comorbid ASD, baseline severity of BtC, an underlying psychiatric disorder) and external (e.g., baseline high dose of antipsychotics, lack of support and contingency plan during the withdrawal process) factors determine the success of any antipsychotic withdrawal [27,42].

The resurgence of BtC may not be related to the withdrawal of medication per se. The withdrawal may unmask a previously undiagnosed psychiatric disorder, such as psychosis, depression or bipolar disorder, or lead to a relapse of them. Discontinuation or dose reduction may lead to withdrawal syndromes, such as insomnia, anxiety and panic. A supersensitivity syndrome consisting of extrapyramidal symptoms, such as akathisia, Parkinsonism and dyskinesia, was associated with antipsychotic withdrawal [43,44]. All these can lead to BtC. In some cases, caregivers' anxiety may exacerbate the BtC or its perception, leading to the enhanced reporting of BtC. Therefore, the clinician will need to make a full assessment of the causes and effects of BtC using standard methodologies [45] (https://spectrom.wixsite.com/project/ (accessed on 1 November 2022).

It is known from other studies that the involvement of all stakeholders in decisions around the withdrawal of long-term-used antipsychotics for BtC is a key factor for success [22–25]. However, the most important factor may be the involvement of family caregivers and patients with ID themselves. Studies have shown that even though family caregivers have the appropriate knowledge of the BtC of their relatives and they would like to participate, they are often not involved in treatment decisions, e.g., psychotropic drug prescribing [21,41]. Furthermore, individuals with ID themselves often lack information and/or are not asked to contribute to decisions in the prescription of antipsychotics [22–25,41]. Although previous studies of interviews of adults with ID have touched on the issue of psychotropic medication use [41], as far as we know, the current study is the first one that specifically targeted the experience of PwID who have gone through the experience of psychotropic withdrawal.

In the current study, a phenomenological analysis of interviews with seven experts by experience extracted four themes that may be of importance in the successful discontinuation of long-term antipsychotic drugs. These themes were (a) the quality of treatment, (b) knowledge and information about the discontinuation process, (c) the conditions that

should be met in a withdrawal process according to participants' experiences of taking part in the withdrawal process and (d) the participants' own coping style.

The interviewed participants indicated that a good relationship with their doctor was important. This meant that they were taken seriously and had been provided information about the indications, effects and side-effects of the antipsychotic agents they received, the reason why this medication was no longer indicated/could be discontinued and information about the discontinuation process. Furthermore, their doctor was reliable, had a personal approach and was accessible. The prescribing doctor had to be available when there were problems during the discontinuation process, e.g., withdrawal symptoms or negative feelings. Furthermore, interviewees mentioned that support from their peers, family and support professionals was important. The finding of the fourth theme, namely, the interviewees' own coping style as an important factor in successful discontinuation, was new and surprising. This may suggest that promoting and stimulating patients' feelings of self-confidence and self-efficacy in the management of and decisions around their own medication use may be helpful in the successful discontinuation of inappropriate (long-term) medication. Perhaps, there is a role for support professionals and healthcare workers to support PwID to achieve this. A recently developed training programme for support staff to help with the deprescribing of psychotropics for people with ID, which is called SPECTROM (https://spectrom.wixsite.com/project/ (accessed on 1 November 2022), developed resources that could be used for this purpose [30].

Our study had some limitations. For the interview study, we had to recruit participants from our own networks, which may have influenced the representativeness of participants in the study. For example, all participants were white. Moreover, we did not interview representatives of people with moderate, severe and profound intellectual disabilities who discontinued antipsychotics; therefore, the results of this interview study are not applicable to these populations. Moreover, we only had seven participants in this qualitative interview study. However, we may assume that this sample size was sufficient because data saturation was achieved in the sixth interview. Finally, because this study was carried out among Dutch clients in Dutch healthcare settings, the results might not be generalisable to other countries. Yet, it may be assumed that the occurrence of withdrawal symptoms and the need for support during the process of medication withdrawal are universal among all people with mild intellectual disabilities. Therefore, the results of this study may also be applicable to other countries with comparable service- and care-providing systems for PwID.

#### **5. Conclusions**

The experiences and opinions of experts by experience with mild intellectual disabilities who had discontinued their antipsychotic drugs were that the reliability and accessibility of their doctors and being taken seriously; a clear explanation and information about the discontinuation trajectory; support from peers, families and care professionals; and their own coping styles were important factors in successful discontinuation.

In addition to the education of care professionals in supporting their clients with mild intellectual disabilities in the decision-making around their medication use, the clients' own coping style in decisions around the start of and management of difficulties during an antipsychotic discontinuation process may be an important factor for the successful completion of discontinuation.

**Author Contributions:** Conceptualisation, J.d.H.; data curation, J.d.H.; formal analysis, J.d.H.; investigation, J.d.H.; methodology, J.d.H.; supervision, G.d.K.; writing—original draft, G.d.K.; writing review and editing, G.d.K., J.d.H., S.D. and R.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Commissie Onderzoek en Zorginnovatie GGZ Drenthe at 28 June 2021.

**Informed Consent Statement:** Written and signed participants' consent forms were collected.

**Data Availability Statement:** The data are available on request and with a substantiated explanation at the GGZ Drenthe department research website (https://ggzdrenthe.nl/research, accessed on 1 November 2022).

**Acknowledgments:** We are very grateful for the cooperation of the experts by experience who consented to be interviewed on their experiences with the topic of this study.

**Conflicts of Interest:** R.S. received institutional and research support from LivaNova, UCB, Eisai, Veriton Pharma, Bial, Angelini, UnEEG and Jazz/GW pharma outside the submitted work. The other authors declare no conflict of interest.

#### **References**


## *Article* **Causes of and Alternatives to Medication for Behaviours That Challenge in People with Intellectual Disabilities: Direct Care Providers' Perspectives**

**Shoumitro (Shoumi) Deb 1,\*, Bharati Limbu 1, Gemma L. Unwin <sup>2</sup> and Tim Weaver <sup>3</sup>**


**Abstract:** Behaviours that challenge (BtC), such as aggression and self-injury, are manifested by many people with intellectual disabilities (ID). National and international guidelines recommend nonpharmacological psychosocial intervention before considering medication to address BtC. Support staff play a pivotal role in the prescription process. Using coproduction, we developed a training programme for support staff, called SPECTROM, to give them knowledge and empower them to question inappropriate prescriptions and ask for the discontinuation of medication if appropriate and instead look for ways to help people with ID when they are distressed without relying on medication. We have presented data from two focus groups that we conducted during the development of SPEC-TROM: one that included support staff, and another that had service managers and trainers. In these focus groups, we explored participants' views on the use of medication to address BtC with a particular emphasis on the causes of and alternatives to medication for BtC. Along with the participants' views, we have also presented how we have addressed these issues in the SPECTROM resources.

**Keywords:** people with intellectual disabilities; the causes of behaviours that challenge; alternatives to medication for behaviours that challenge; social care services; support staff; service/home managers; trainers

#### **1. Introduction**

Behaviours that challenge (BtC), or challenging behaviour, can be defined as "culturally abnormal behaviour of such an intensity, frequency or duration that the physical safety of the person or others is likely to be placed in serious jeopardy, or behaviour which is likely to seriously limit the use of, or result in the person being denied access to, ordinary community facilities" [1]. BtC is common in people with intellectual (learning) disabilities (ID), with up to 60.4% of adults with ID showing at least one form of BtC [2–4]. BtC includes aggression, destructive behaviour, and self-injurious behaviour [5]. BtC can be difficult to manage and may lead to exclusion from community facilities, community placement breakdown and hospitalisation, and the use of restrictive practices such as physical restraint and inappropriate medication use [2,5].

To help a person with BtC, it is crucial to understand the reason behind the BtC rather than to inappropriately use medication [2,6]. BtC can be considered a form of communication whereby a person with ID conveys their distress [7]. For example, if someone is in pain or frustrated because of the demand put on them, they may shout and scream if they cannot communicate their distress to others. According to Matson and colleagues [8], the function of BtC could be categorised under six headings: attention

**Citation:** Deb, S.(.; Limbu, B.; Unwin, G.L.; Weaver, T. Causes of and Alternatives to Medication for Behaviours That Challenge in People with Intellectual Disabilities: Direct Care Providers' Perspectives. *Int. J. Environ. Res. Public Health* **2022**, *19*, 9988. https://doi.org/10.3390/ ijerph19169988

Academic Editors: Paul B. Tchounwou and Marco Scarselli

Received: 18 June 2022 Accepted: 11 August 2022 Published: 13 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

107

(receive attention), escape (avoid something), non-social (factors internal to the person), physical (physical problems such as relief from pain), and tangible (achieve something). The function of the behaviour is assessed using functional behavioural analysis through Antecedent (situation before the BtC), Behaviour (the description of the actual behaviour), and Consequences (the consequences of the behaviour) ABC charts [9].

Both pharmacological and non-pharmacological psychosocial interventions such as positive behaviour support (PBS) [10] have been used to manage BtC [11,12]. A recent metaanalysis found a significantly long-lasting moderate overall effect of non-pharmacological interventions on BtC (effect size = 0.573) [13]. Interventions combining mindfulness and behavioural techniques, such as a PBS approach, showed greater impact than other interventions. Other PBS-based non-pharmacological interventions have also been shown to be effective in reducing BtC when compared with treatment as usual [14]. Another recent study has shown that a significantly higher number of adults with intellectual disabilities managed to come off their psychotropic medications when PBS was implemented in comparison to the group where no PBS was implemented [15].

Despite the poor evidence for the effectiveness of medications in managing BtC [6], psychotropic medications are used widely among people with ID (49–63%) and often off-license [16], which is a major public health concern [17].

Support (care) staff play a pivotal role in influencing the prescription process, such as by asking doctors to prescribe medication for BtC to begin with, and, given the lack of evidence of the effectiveness of the medications, are overly optimistic about the medication's potential effect [18,19]. Support staff also are most anxious and obstructive to psychiatrists' attempts to withdraw antipsychotic medication when appropriate [20]. Previous surveys of support staff in Australia [21] and the Netherlands [22] showed that most staff felt the use of psychotropic medications for BtC is justified. In our recent focus groups of support staff in the UK, some staff felt the use of medication is appropriate, whereas others felt that it is a 'chemical restraint' [23]. Among other factors, poor staff training and organisational policies are crucial factors in successfully withdrawing psychotropic medications [19–21]. Proper training and support for support staff are thus particularly important for successful programmes concerning the rationalisation of psychotropic medication use in adults with ID.

Training programmes are shown to be useful in many psychiatric disorders, including schizophrenia (30 RCTs) [24] and bipolar disorder [25], in improving patients and their caregivers' quality of life (QoL). They have also been found useful in different neurodevelopmental disorders. For example, our recent meta-analysis has shown a moderate effect size of parental training on improving autism symptoms in children with autism spectrum disorder (ASD) [26]. A review of training programmes directed at parents and teachers included four RCTs and found improvement in attention deficit hyperactivity disorder (ADHD) symptoms with effect sizes of 0.05 to 0.77 [27]. A Cochrane review showed that training support staff helped to reduce antipsychotic prescriptions in people with dementia by 40–50% [28]. This was reflected in a nationwide trend in the reduction of antipsychotic prescriptions by 11% over 10 years (2005–2015) in the UK [29].

We have addressed the aforementioned issues by developing online training resources implemented through face-to-face interactive workshops for support staff caring for adults with ID in community settings. The training programme, SPECTROM (Short-term Psycho-Education for Carers To Reduce Over Medication of people with intellectual disabilities) (https://spectrom.wixsite.com/project accessed on 12 July 2022), was developed using a coproduction method [30,31]. This was achieved by putting stakeholders' experiences at the centre of the study and ensuring close and equal collaboration among them from the outset. The ultimate aim of SPECTROM is to empower, inform, and equip support staff with skills to understand the person they support, manage their own psychological responses to their behaviour, negotiate the care pathway, advocate on behalf of the person they care for, and take the views of adults with ID fully into account. The intended goal is to reduce requests from staff for medication and encourage staff to ask the prescribers questions about the

necessity of the continued use of psychotropic medication, which should lead to a significant reduction in the use of medication and an increase in psychosocial interventions instead.

As part of the development of SPECTROM training, focus groups were conducted to explore staff perceptions on the use of psychotropic medication to manage BtC in people with ID and gather suggestions regarding the contents and format of SPECTROM. The aim of this paper is to present the findings of the first set of focus groups where support staff, service/home managers, and PBS trainers discussed their perceptions of and views on the use of psychotropic medications to address BtC in adults with ID. In this paper, we have presented two themes related to BtC: participants' views on 'the causes of BtC' and 'the alternatives to medication for BtC'.

#### **2. Materials and Methods**

We conducted two focus groups: one involving support staff and another with house/service managers and PBS trainers. These focus groups explored participants' views on the use of psychotropic medication for BtC in adults with ID with a particular emphasis on the potential 'causes of BtC', and 'alternatives to medication for BtC'.

#### *2.1. Participants*

We invited nine support staff, seven of whom agreed to attend, but eight ultimately took part. Five service/house managers and three trainers were invited, and all took part in the focus groups. We have not collected any demographic data on the participants.

#### *2.2. Conduct of the Focus Groups*

A topic guide (see Supplementary Material S1) was developed based on the literature review findings and the project's aims and objectives. After discussion with the core team (BL, SD, TW, and GU) and other relevant stakeholders, the topic guide was finalised. This was employed flexibly and was open to emergent themes but framed using the Theory of Planned Behaviour model [32], thus examining beliefs and attitudes (e.g., about psychotropic medication and alternative approaches such as PBS) and how these might influence behaviour (e.g., in terms of requesting support from professionals to prescribe medication or provide help with alternative approaches). The interviews started with an exploration of participants' experiences dealing with BtC in adults with ID and the use of medication for BtC. Then, the topic guide moved into issues related to potential causes of BtC, including physical, psychiatric, and environmental causes. From there, the topic guide entered into the issue of participants' views on alternatives to medication, particularly psychosocial interventions such as PBS, for BtC. The sample size was a pragmatic decision. No formal sample size calculation was required for this study. This was a small study with limited resources and included primarily qualitative data collection. The minimum sample size we aimed for was 6–8 participants in each focus group. Participants were purposively sampled for each group to include a range of support staff in terms of their experience and from different organisations.

A researcher (BL) with previous experience in conducting qualitative research ran the focus groups with the help of the chief investigator (SD) under the supervision of an expert in qualitative research (TW). We used the approach utilised in our previous studies of interviewing the carers of people with ID as well as head injury [33–35]. Any paid carers, service/home managers, and trainers working with people with ID who showed BtC were eligible to participate in the study. A research advertisement was sent through the UK Voluntary Organisations Disability Group (VODG), an umbrella organisation of more than 35 social care service providers in the UK (social service, voluntary, and independent sectors). Nine service provider organisations agreed to take part, but, ultimately, only eight got involved. The organisations are Mencap, Challenging Behaviour Foundation, Achieve together, AT-Autism, Avenues Group, Dimensions UK, Milestones Trust, and National Autistic Society. Each organisation identified one available manager and trainer

for the focus groups, and each manager identified two support staff. The focus groups were held face-to-face at a venue in London, England in March 2019. Eligible participants were then sent a SPECTROM study summary and information on the study. Once participants agreed, written informed consent was taken. Two separate focus groups were conducted, one with support staff and the other with service managers and trainers. Each lasted for approximately 90 min. The same topic guide was used for both groups, and a semistructured interview was carried out. The focus groups were recorded using pseudonyms and professionally transcribed. Two authors (BL and SD) independently analysed the data to achieve a consensus. In the focus groups, care staff, service managers, and trainers explored issues around their perception and views on medication use for and causes and assessment of BtC.

#### *2.3. Thematic Analysis*

The transcriptions were analysed using thematic analysis. Thematic analysis is the process of identifying patterns and themes within qualitative data and texts. The data are interpreted to examine underlying meaning and ideas. Thematic analysis can be topdown deductive, driven by specific research questions, or bottom-up inductive, driven by data [36]. We used a top-down deductive approach to analyse the gathered data based on the topic guide and research questions we developed. NVivo 12 plus for windows software [37] was used to manage and analyse the data. The authors first familiarised themselves with the data. The transcripts were then read and interpreted for meaning and significance. Then, a code was given to each interpretation, and an initial coding frame was developed to organise the identified codes. This initial frame included main themes under 'physical cause of BtC', 'psychiatric cause of BtC', 'environmental cause for BtC', and 'psychosocial interventions for BtC'. The coding frame developed was continuously reviewed and refined as new codes emerged or as it was searched for patterns. The data and quotes were indexed on identified codes or new emerging codes. The coding framework was then searched for patterns for emerging themes. Similar codes were categorised together to produce patterns and themes. The identified categories were also reviewed to ensure the emerging category was discrete and modified as necessary. The identified themes were then reviewed and revised if needed. Once no new themes emerged, the themes were finalised and defined. The themes and analysis process were also overseen and verified by two authors (GU and TW) who were experienced qualitative researchers.

#### **3. Results**

We have presented two themes and six subthemes from the thematic analysis of focus group discussions: (a) causes of BtC, including psychiatric disorders, and (b) alternatives to medication including PBS, person centred approaches, understanding the person, developing relationships, and collaborative working. In the results section, we present the main themes and subthemes followed by relevant quotes directly taken from the transcripts. We have used codes in the parenthesis such as 'SS' for support staff, 'SM' for service/house manager, and 'TR' for trainer to identify the quotes from the different groups of participants.

#### **4. Causes of BtC and How Understanding Them Can Help the Person with ID**

Participants discussed the causes of BtC and how understanding them would help address BtC using alternatives to medication. The central emerging theme was 'behaviour is a means of communication'. Triggers for BtC were discussed, including the environmental factors, particularly if the people are not placed in the right environment, and the help they need to accommodate any changes in their life. Participants felt it would be good to have a checklist to recognise the factors that lead to BtC. We have addressed this issue by developing a Checklist for the Assessment of Triggers for the behaviour of concern Scale (CATS) [38]. Staff can use this list to identify the triggers for BtC, which will help complete the Antecedent-Behaviour-Consequence (ABC) chart and functional behaviour analysis.

" ... *when she was aggressive, and all she wanted was the curtains tied back. So, what someone else would think of that, you know it's just, it's just knowing.*" (TR)

"*I think it was literally a checklist of going through the whole thing with her. What could it be?*" (SM)

"*To be considered, the environment, how they feel in the environment, do they feel comfortable there?*" (SS)

" ... *placement staff comes into it a lot on whether someone who is in the right placement.*" (SM)

Some participants felt it is not helpful when people come with a label such as being very aggressive. They thought they should thoroughly reassess the person and their behaviour to draw a new person-centred support plan.

"*Looking at the support plan. I mean had to rewrite her support plan with in the space of two weeks because she came labelled as being a very difficult person, very challenging. And naturally she was scared so her way of reacting was to fight. And the treatment assessment unit, she was in where I visited her, I was scared, so she must have been absolutely petrified.*" (SM)

Participants discussed how physical problems, such as pain in the body, can manifest as BtC and be wrongly treated with antipsychotics.

"*I know for a fact people have been given anti-psychotic medication and actually it's something physical that's wrong. They've got toothache or they've got tummy ache or they don't like the colour of their room. Something as simple as that, you know.*" (SM)

"*Um, but recently I had someone who was exhibiting behaviour but it was because he had chest infection and he was in pain but he couldn't vocalise or tell someone I'm in pain. So, then staff log it as challenging behaviour, we need to talk to a psychiatrist.*" (SM)

The issue of communication came up in the discussion. SPECTROM has a module on effective communication and another on effective engagement with people with ID. Both modules encourage staff to learn the best ways to communicate with the person they support so that they can concentrate on their skill-building rather than BtC.

"*I think often people are perceived as showing behaviours because of the condition, but actually its external factors and things in the environment or how they're being supported. Um, the communication strategies that they've got.*" (SM)

"*Sometimes when you have challenging behaviours it's not, it's because someone wants something. It's like you said. Yeah, they can't communicate. They're telling you they want something done or to do something. Communication. It's always communication.*" (TR)

#### *Causes of BtC: Psychiatric Disorder*

While discussing the causes of BtC, the role of psychiatric disorders was raised. All participants unanimously agreed that no meaningful training currently exists on the issue of psychiatric disorders in ID, although there is an increasing awareness of Autism Spectrum Disorder (ASD).

" ... *we've not really, we've never really given any sort of any specific training on psychosis or um, mine was just basically what I've read from you know carer plans and things like that. Um, but nothing, we've never ever attended any training on it.*" (SS)

Staff found it difficult to distinguish between psychiatric disorders and BtC. They felt that when the person is trying to communicate their needs through BtC, it is often misinterpreted as a psychiatric illness.

"*When is it challenging behaviour and when is it psychosis, like you say? When is it that he's hearing voices that's telling him to do something and when is it when he's doing it because of his own accord that he's trying to communicate something or maybe an expression?*" (SS)

" ... *how much of what they're displaying is because of their mental illness or how much is that just because they're trying to tell you something and you're not able to understand it? As support staff, something I struggle with personally is knowing the difference.*" (SS)

Participants felt that when there is a crisis and the person with ID displays BtC, it is often considered part of a psychiatric disorder. As a result, they do not explore other factors, including the environment, to check what is causing the behaviour.

" ... *when he attacked staff but he didn't hear voices, was that because of the psychosis or what that because he was having a, he was going through, he was going through that crisis.*" (SS)

There was significant confusion about the overlap between what may be ASD symptoms (e.g., living within an inner world) and psychiatric symptoms. SPECTROM has comprehensive modules on ASD and ADHD. The ASD module discusses the potential overlap between the ASD phenotype and psychiatric symptoms in detail, but acknowledges that many psychiatric disorders, including psychosis, are common in people with ASD [39].

" ... *there was too much stimulation going on and that frustrated him so that he just lashed out. Because he's also diagnosed with autism, there's also that to consider as well. So that was due to the environment why he acted that way rather than it being psychosis because something told him to lash out.*" (SS)

Participants discussed the difficulty of distinguishing between trauma-related symptoms and psychiatric symptoms. Some participants thought that psychotic symptoms might, in fact, be the memories of past traumatic events spoken aloud rather than genuine psychosis. There is also the possibility of past traumatic experiences inducing real psychotic symptoms [40,41]. This issue has been discussed in the psychiatric disorder module of SPECTROM.

"*He believes that he hears people, and he will say that they made me do it and things.* ... *.is that because he's saying that but actually is that just a past traumatic event that's happened that then he's remembering it?*" (SS)

One support staff expressed frustration on the lack of awareness among some general practitioners who take a medical approach to BtC and say that these are often due to a psychiatric disorder.

" ... *one of the GP I discussed it. But what do you expect? He's got mental health problems. I'm talking about uh, educating people.*" (SS)

#### **5. Alternatives to Medication**

#### *5.1. Positive Behaviour Support (PBS)*

After discussing the possible causes of BtC, the discussion progressed to how to best address BtC without relying on medication. This started with a discussion around PBS and how this framework can be used to reduce inappropriate uses of medication. Some larger service provider organisations seem to have their own PBS support team, which staff found very helpful. However, even in big organisations, the PBS support team resource is not always adequate, and most smaller organisations will not have this kind of support available at all. Most staff in large organisations also only have basic training on PBS.

"*She got into crisis, was in crisis for months and regardless of whatever we were doing as a carer, as a care staff, and you know we brought in um, um positive behaviour support team. We were working really closely with them. We've got really, really robust PBS plans.*" (SS)

"*Part of our PBS was to give him space, so we gave him space.*" (SS)

"*So now that he's come* ... *..to our company, we said we were going to support him for positive behaviour support plan.* ... *..we've seen a different person. He's living a* *better life and that's because of the service that he's in that's allowing him to live a better life.*" (SS)

"*You know, staff have PBS training, um, we work with behavioural support analysts that come in and collect data.*" (SM)

The issue of the cost of providing the service was discussed. One trainer highlighted the fact that using the PBS approach can reduce the frequency of BtC and the use of medication, which will save money in the long run. Therefore, a case could be made for social and health care commissioners to invest the funds needed to implement PBS now in order to save money in the future.

"*So that's got big budget implications as well, because certainly similar examples where somebody has come out of long stay and they were on, you know, X amount of medication and all the support plans are saying three to one support in our community.* ... *.but two years down the line that three to one support becomes one to one support because somebody's taken a holistic approach to look at what do we need to do to support this person?*" (TR)

#### *5.2. Person-Centred Care Approach*

Other alternatives to medication based on the PBS approach were discussed, including a person-centred care approach. There seems to be more awareness within the larger social care provider organisations in the UK and their staff about using alternatives to medication to address BtC through a person-centred care approach.

"*So our staff are very aware of the, um, other ways to help behaviours rather than prescribing medication.*" (SM)

"*Right now, there has been a lot of awareness that's been created that people are using all those alternatives rather than medication.*" (SM)

" ... *He's more engaged. He's more willing to communicate, willing to do things. Willing to participate because we are allowing him to do.*" (SS)

#### *5.3. Understanding the Person*

Another person-centred care approach the participants discussed was understanding the person behind the behaviour. Participants felt that, once they got to know the person well, it became easier to help them without using medication when they were distressed. Sometimes people come from a previous placement with a label of being aggressive, but staff felt that this should not deter them from trying to get to know the person and help them with their skill development instead of concentrating on their BtC. One service manager mentioned that, if you do not understand the person and the reason for their BtC, you tend to blame the person for their behaviour which does not help.

"*Um she's not written up for anything, PRN at all so her behaviour are managed through um, staff really knowing her well.*" (SS)

"*So, if you don't get to know what they are on about, then they will start to display challenging behaviour.*" (SS)

"*And you find out what they like. Nobody asked them. And show positive interest in what they like and you know to talk to them like they're a real human being.*" (SS)

"*Instead of taking the baggage of their history with them. And it's about stripping all of that out, and almost starting again and saying let's look at this person as a whole now and get the right people involved.*" (TR)

"*And I believe that does have a lot to do with staff attitudes and behaviours. There are still some places where they blame the person and they are behaving because that's the way they are, rather than then behaving in this way because someone has taken something from that person and they don't know how to control it.*" (SM)

#### *5.4. Developing Relationships*

Another vital person-centred care approach that involved developing a positive relationship with the person they support was discussed. One participant said that, even after being scratched by the person with ID on one occasion, she continued to build a relationship with the person, which helped to improve the BtC.

"*So, I stayed longer and now we've got a bond that when I'm on annual leave it's a problem because we build that relationship and it got to a point that everything, she wants to do has to be with me.*" (SS)

" ... *where he became heightened and within that he was able to scratch me which left a scar. From that incident, I didn't change my approach towards him but obviously I was cautious because I'm not trying to get injured. But through that, because I hadn't changed the way I was, and no more the next day and even after that, after that incident he saw me at breakfast, I'll still engage with him, still carried on as normal. The relationship that we have now is that you know he, he trusts.*" (SS)

"*But now we got to a point where I can take her to any appointment, we can sit down. So, I, we build a bond.*" (SS)

#### *5.5. Collaborative Work*

Participants felt that multi-disciplinary work was in the best interests of the person with ID. This includes other relevant professionals like doctors and nurses involved in the care of the person and the person with ID themselves and their families.

"*With the prescribing, it's also working with families and psychiatrists, because sometimes there's a parent who is going to tell you I know my child and I think they need this, they have been using this.*" (SM)

"*If the families are involved or any other person involved in their circle of support, they will also attend the meeting. And it's a best interest meeting as well.*" (SM)

" ... *um, how you work as a team with those different professionals. We have a pharmacy check.*" (SM)

One service manager mentioned that they asked key support staff to accompany the person with ID to the clinic for the medication review. She also emphasised the importance of gathering all relevant information before attending the clinic and involving the person with ID and their family, if possible. SPECTROM provides a checklist for staff to go through in preparation for a formal medication review and a set of questions that staff could or should ask the prescribers in the clinic.

" ... *before we go to that meeting, let's have a discussion first so there's no surprises. What are we going to say, what are we presenting to the clinical psychiatrist that doesn't know this person as well as what we do. What does the family members got to say about it? So, we go into that meeting with that individual, if they're open to input and they have the capacity to do that, to feed into that meeting so that it's productive for that person.*" (SM)

#### **6. Discussion**

In this paper, we have presented a wealth of data on wide-ranging issues from focus groups, primarily on participants' perception of the causes of BtC and methods to help people with ID without using medication (alternative to medication).

#### *6.1. Causes of BtC Including Physical and Psychiatric Disorders*

As for the causes of BtC, the discussion primarily revolved around physical disorders, psychiatric disorders, and environmental factors. There was a consensus among the participants that BtC is often a means of communication and that the causes and effects

of BtC need a thorough assessment to draw up an effective person-centred behaviour support plan to reduce overreliance on medication to address BtC. In SPECTROM, we have provided a module on the assessment of behaviour and examples of two assessment schemas: **B.M.P.P.S.** (assessment of the **B**ehaviour itself, **M**edical issues, the **P**erson with ID, **P**sychiatric/psychological issues, and **S**ocial issues) [5] and **H.E.L.P.** (assessment of **H**ealth and medical conditions, **E**nvironment and support, **L**ived experience and emotional well-being, and **P**sychiatric illness) [42].

Physical disorders are more prevalent in people with ID than in the general population [43]. Although it is not uncommon for physical problems that are common in people with ID, such as pain in the body (headache, toothache), constipation, and acid reflux, to lead to BtC, particularly in people who cannot communicate their feelings, it is not always easy to identify these causes. Therefore, a high degree of suspicion and thinking about a physical cause for BtC should help ensure the correct diagnosis. If necessary, a therapeutic trial with a painkiller where pain is suspected may help. We have developed a 'Physical disorder' module in SPECTROM and provided examples of proformas that could be used to detect and rate the severity of pain. Another problem mentioned was the difficulty of carrying out required investigations such as blood tests, X-rays, etc., for many people with ID. SPECTROM provides practical guidelines for addressing these issues, such as using accessible information including pictures of the tests, teaching the person relaxation techniques, etc. SPECTROM also provides videos on 'what happens during a health check', 'what happens during a blood test', 'what happens during an MRI', etc., to familiarise people with ID before they go for any of these investigations.

The issue of the relationship between psychiatric disorders and BtC is a complex one [39]. Most staff did not receive any training or information on this. Participants raised the question of how to distinguish between psychosis and BtC due to environmental factors or when caused by traumatic life events. Therefore, SPECTROM has developed a psychiatric disorder module to provide information to support staff.

The participants discussed the issue of overlap between ASD and BtC. ASD and ID commonly co-occur; 38% of people with ASD have ID, and a similar proportion of people with ID are also known to have ASD. ADHD is equally common in both ASD and ID, although this issue was not discussed in the focus groups. The extensive overlap of symptoms between ID, ASD, and ADHD often makes it difficult to tease apart these diagnoses in people with ID [39,44,45]. As BtC is common in ID (18–60%) [2], ASD (10– 15%) [46], and ADHD [44], respectively, the rate of BtC increases when all three conditions co-exist [39]. Therefore, SPECTROM has developed comprehensive modules on ASD and ADHD, with information on strategies to help people with ID and ASD and/or ADHD to reduce BtC. Currently, there is more awareness of the trauma-induced symptoms in people with ID [40], some of which may manifest as psychiatric disorders or BtC [40,41]. Many support staff in large provider organisations seem to be more conscious of this possibility when they draw up a person-centred care plan for the people they support.

#### *6.2. Alternatives to Medicine through the Use of Person-Centred PBS Care Planning*

After discussing the causes of BtC, participants progressed to discuss how to address BtC, particularly by using a psychosocial approach and avoiding medication use. The most well-known psychosocial approach is PBS [10], which provides a framework to apply in practice [47]. Most staff from large provider organisations seem familiar with PBS principles, which they use to draw person-centred care plans [48] for people they support. This approach seems to reduce the need to use medication to address BtC. One study found that community team-led implementation of PBS framework did not reduce BtC in community settings [49]. However, others found that PBS and other psychosocial approaches were useful in reducing aggression among people with ID [14].

Participants acknowledged that understanding the person they support is key to understanding BtC and improving them without relying on medication. Service managers and trainers mentioned not focusing on the 'labels' people with ID bring with them but concentrating on the whole person in order to understand their likes and dislikes, strengths and weaknesses, and develop a positive relationship. All participants agreed on a holistic approach to care planning and addressing BtC.

#### *6.3. Shared Decision Making*

All participants in the focus group agreed on the need for multidisciplinary work and acknowledged that there is scope to improve shared decision-making by involving the person with ID and their families. In a previous study, we found that adults with ID did not feel sufficiently involved in making the decision about their medication [50]. Some adults with ID who we interviewed were dissatisfied with their medication, mainly due to lack of involvement in the treatment decision, adverse effects, lack of efficacy, and a 'desire to lead a normal life'. In another study, most of the six adults with ID who were interviewed complained about not having enough information on their medication, particularly in an accessible format [51].

A similar sentiment has been echoed by family members, many of whom felt that there is an over-reliance on medication instead of taking a holistic approach to address BtC [52]. In our study, many of the 20 family caregivers interviewed complained about not having enough information about the care of their loved ones and not having enough involvement in the decision-making process [53]. Family caregivers play an essential part in the care of people with ID, even when they do not live in the family home; they are the only constant presence in the life of the person with ID, whereas professionals and care staff come and go. Therefore, the knowledge of their loved ones is of paramount importance in care provisions for people with ID. To address these issues, we have devoted an entire module to 'effective liaison with families' in SPECTROM. This module teaches staff to respect family caregivers' views, acknowledge family caregivers' expertise, communicate with them without jargon, and include family caregivers and the person with ID in care planning, including prescriptions of medication, from the outset in order to promote more shared decision-making.

Although several studies explored care staff views on medication use, they rarely reported on staff perceptions of the causes of BtC, which was at the heart of the discussions within our focus groups. In our previous study, few care staff explicitly reflected that their own behaviour might influence aggressive behaviour in adults with ID [54]. Furthermore, only 16% of staff interviewed in this study mentioned issues around communication despite much of the aggressive behaviour being considered to be communicative [7]. Staff felt that they would benefit from training and information about potential triggers to help them think more about environmental conditions and their own role in precipitating BtC. Many other researchers have highlighted support staff frustration for not having the right training and their desire to gather more knowledge and training on (a) mental health issues, (b) medication prescribing (when to use them and why, and when not to use them and why), (c) medication side effects, (d) and when and how medication could be safely withdrawn [55]. All of these issues are addressed in the SPECTROM programme.

#### *6.4. Strengths of the Study*

Previous studies of staff surveys primarily concentrated on staff knowledge of psychotropic medication. They rarely explored their views on the causes of BtC, which our study has addressed. One strength of this study is that it examined not only support staff's opinions, but also service managers' and trainers' opinions. Previous studies have primarily focused on the experiences and perceptions of support staff and family caregivers. Another strength of this study is that support staff, service managers, and trainers were interviewed separately to avoid the influence of service managers on support staff responses. The support staff, managers, and trainers were also from different service provider organisations, allowing us to capture different aspects of the participants' experiences. In addition, the anonymised data analysis allowed support staff to express their opinion freely, thus increasing the face validity of our findings. Another strength was that the data were analysed by two authors.

#### *6.5. Limitations of the Study*

All participants were recruited from large service provider organisations. Therefore, the views and experiences of staff working in smaller organisations were not captured, which may have differed significantly. Another weakness is that service managers identified the support staff for the focus groups, thus managers may have chosen support staff that were particularly eager to get their voices heard on this topic.

#### **7. Conclusions**

Our exploration of the views of support staff, service managers, and trainers on the causes and management of behaviours that challenge in people with intellectual disabilities used focus groups to reveal that most participants have some knowledge of the physical, psychological, and environmental causes of BtC. Most were also familiar with the concept of positive behaviour support that could be used to support people with intellectual disabilities who display behaviours that challenge without relying on medication. However, it is worth remembering that all participants in our study were employed by large social care service provider organisations in the UK. Most of these organisations are likely to have their own positive behaviour support teams; smaller provider organisations are unlikely to have this in-house support. Participants unanimously expressed concern about their lack of knowledge of psychiatric disorders and their relationship with behaviours that challenge in people with intellectual disabilities. There were no major discrepancies in the views of the support staff and those of the service managers and PBS trainers. All parties have agreed on the need for more information on medication, their indications, and side effects. We have addressed these issues and others raised by the participants in SPECTROM modules.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph19169988/s1. S1: Topic Guide for SPECTROM focus groups 1 & 2.

**Author Contributions:** S.D. is the grant holder. All authors were involved in the conception and design of the study. S.D. conducted the focus groups. B.L. analysed focus group data, and S.D. acted as the second-rater. All authors contributed substantially to the preparation of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by the UK's National Institute of Health Research (NIHR) Research for Patient Benefit (RfPB) Programme (grant PBPG-0817-20010). The Imperial Biomedical Research Centre Facility, which is funded by the NIHR, has provided support for the study. The views expressed in this article are those of the authors and not necessarily those of the National Health Service, the NIHR, or the Department of Health, UK.

**Institutional Review Board Statement:** Not applicable as no new or patient-related data were collected. Also, the participant's views were analysed anonymously, and no personal data were collected from the participants.

**Informed Consent Statement:** Written informed consent was collected from the participants before the start of the focus groups.

**Data Availability Statement:** Transcripts of the focus group discussions are available from the authors subject to approval from the study sponsors and the funders.

**Acknowledgments:** We thank adults with intellectual disabilities, Cornwall Learning Disability Advisory Group, family caregivers and advocates, other stakeholders including service provider organisations, Achieve Together, Dimensions-UK, National Autistic Society, MENCAP, Milestones Trust, Avenues Group, VODG, Challenging Behaviour Foundation, AT-Autism, Mandy Donley and Jeffrey Chan from the Australian National Disability Insurance Scheme Quality and Safeguards Commission, community learning disability team members, Saadia Arshad, Bini Thomas, Sujit Jaydeokar for taking part in the project. We also acknowledge the help from the co-applicants and project group members, Umesh Chauhan, John Rose, Jean O'Hara, David Branford, Mike Crawford, Rohit Shankar, Caroline Finlayson, Georgina Samuels, Mike Wilcock for helping with the development of SPECTROM.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Predictors of Change in Stepping Stones Triple Interventions: The Relationship between Parental Adjustment, Parenting Behaviors and Child Outcomes**

**Matthew Sanders 1, Nam-Phuong T. Hoang 1,\*, Julie Hodges 1, Kate Sofronoff 1, Stewart Einfeld 2, Bruce Tonge 3, Kylie Gray 3,4 and The MHYPEDD Team**


**\*** Correspondence: a.phuong@uq.edu.au

**Abstract:** The current study explored the process of change in Stepping Stones Triple P (SSTP) using a community-based sample of 891 families of children with developmental disabilities (DD) who participated in an SSTP intervention at a community level. A preliminary analysis of outcome data indicated that SSTP intervention was effective in reducing parental adjustment difficulties, coercive parenting, and children's behavioral and emotional difficulties immediately after the intervention. The effects were maintained at 12-month follow-up. The results also indicated that change in parental adjustment over the course of intervention was significantly associated with a change in parenting behaviors. However, change in parenting behaviors but not change in parental adjustment, predicted children's behavioral and emotional problems following the intervention. The results suggest that positive parenting skills are the most salient ingredient driving the change in child behaviors in SSTP interventions.

**Keywords:** mechanism of change; developmental disability; evidence-based parenting; Triple P

#### **1. Introduction**

*1.1. Relations between Parental Adjustment, Parenting Behaviors and Children's Outcomes in Families of Children with DD*

Research into children with developmental disabilities (DD) has consistently pointed to the elevated risk of the development of behavioral and emotional problems among children of this group [1]. As reported in a recent meta-analysis, the rate of behavioural problems in children with DD is two to three times higher than in typically developing children [2–4]. Different factors can contribute to the development and exacerbation of behavioral and emotional issues within this population. Parental stress has consistently been identified as one of the most prominent contributors. Neece et al. [5] followed two groups of children (144 were typically developing children and 93 were diagnosed with a type of DD) from age three to nine years. Their cross-lagged panel analyses indicated that behavioral problems and parental stress covaried across time, but parental stress consistently arose as a predictor of child behavior problems while the effect of early child behavior problem on parental stress was much less consistent. This finding was supported by Lin et al. [6] study which examined the transactional relations between parenting stress and both internalizing and externalizing behavioral problems in 75 young children with ASD over 1.5 years. The findings also indicated that early parenting stress was significantly associated with later children externalizing problems.

**Citation:** Sanders, M.; Hoang, N.-P.T.; Hodges, J.; Sofronoff, K.; Einfeld, S.; Tonge, B.; Gray, K.; The MHYPEDD Team. Predictors of Change in Stepping Stones Triple Interventions: The Relationship between Parental Adjustment, Parenting Behaviors and Child Outcomes. *Int. J. Environ. Res. Public Health* **2022**, *19*, 13200. https:// doi.org/10.3390/ijerph192013200

Academic Editors: Paul B. Tchounwou, Pentti Nieminen and Fernando García

Received: 19 July 2022 Accepted: 11 October 2022 Published: 13 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Coventry CV4 7AL, UK

While parental stress directly affects children's emotions and behavior [7], it also influences parenting which may in turn further exacerbate children's behavioral and emotional difficulties. When parents are distressed, they are less likely to be sensitive to their children's behaviors, they might also be more likely to engage in irritable transactions and poor disciplinary practices (inconsistency, coercion) that reinforce undesirable behaviors in children, thus making them more likely to occur again [8,9]. Totsika et al. [10] analyzed data from 555 families of children with DD and found that early parental distress (at nine months) significantly predicted child behavioral and emotional problems at both age seven and age 11 years. This relationship was mediated by adversarial parenting practices between ages 3 and 5 years. Day et al. [11] also surveyed 1392 families of children with a disability aged between 2 and 12 years and found that parental adjustment difficulties (depression, anxiety and stress) were among the strongest predictors of coercive parenting.

Due to the reciprocal transaction between parental stress, parenting behaviors and children's behaviors, it is not uncommon for parents of children with DD who exhibit elevated behavioral problems to also experience a high level of stress and use more coercive parenting. Interventions to reduce behavioral problems in children with DD thus commonly aim to address both parental stress and dysfunctional parenting behaviors [12].

#### *1.2. Stepping Stones Triple P*

A number of different parenting programs have been shown to be effective in managing behavioral problems in children with DD. Among those, Stepping Stones Triple P (SSTP) [13]—(a variant of the Triple P Positive Parenting Program) is one of the most extensively studied and widely used. Built on social learning theory, SSTP recognizes the reciprocal nature of parent–child interactions surrounding dysfunctional behaviors [13]. Therefore, the occurrence of child behavioral and emotional problems in children with DD is viewed as both a consequence and an antecedent of dysfunctional parenting and parental adjustment difficulties. In SSTP, the goals are to help parents learn to manage their children's behavior problems without using coercion escalation or harsh discipline and to adopt better strategies to regulate their emotion [13].

Throughout the SSPT program, parents are encouraged to choose their own goals, develop plans and execute their plans which includes the capacity to plan and anticipate, regulate their emotions, solve problems, and collaborate with significant others to provide care for their children. This self-regulatory approach is expected to help reduce parents' use of coercive and punitive disciplining and promote parents' capacity to regulate their emotions and behavior throughout intervention [13]. In addition to the core Triple P strategies, the program incorporates a number of additional disability-related components to reflect the additional challenges faced by parents of children with disabilities as well as a focus on community living and family support movements (such as Being part of the community). These include: (1) Identifying additional factors that are more likely to contribute to the development of behavior problems in persons with disabilities (e.g., the accidental reward for stopping disliked activities). (2) Incorporating other behavior change strategies from the disability literature (such as setting up an activity schedule). (3) Developing additional protocols to deal with self-injurious behavior, repetitive behaviors, and pica that are more prevalent among children with disabilities. (4) Modifying wording and examples in parenting materials to make them more acceptable and sensitive to parents of children with disabilities [13].

Meta-analyses of randomized controlled trials of SSTP have consistently shown that SSTP effectively reduces harsh parenting, child behavioral and emotional problems, and parental distress. For example, Tellegen and Sanders [14] analyzed both controlled and uncontrolled design studies. They found that SSTP had moderate to large effects in reducing coercive parenting behaviors, moderate effects in reducing child behavioral and emotional problems, and moderate effects in reducing parental adjustment problems. In a recent systematic review and meta-analysis of all SSTP levels, Ruane and Carr [15] found that SSTP has small to medium effects on parental adjustment and co-parenting. For parenting behaviors and child behaviors, the effect sizes were medium to large. The growing literature supports the efficacy of SSTP in reducing harsh parenting, child behavioral and emotional problems, and parental adjustment difficulties. Yet, little is known about the process of change explaining the effects of SSTP for families of children with DD. It is not yet understood if a change in parental adjustment over the course of the intervention assists change in parenting behaviors and vice versa. It is also yet to know if changes in parental adjustment or changes in parenting behaviors contribute to the change in children's behaviors and emotional outcomes as proposed by the literature.

#### *1.3. Current Study*

The present study examined the process of change in SSTP interventions, explicitly emphasizing the bidirectional association between change in parental adjustment and change in coercive parenting over the course of the SSTP intervention and their association with children's subsequent outcomes. Specifically, we examined (1) how parents' experience of emotional difficulties (parental adjustment) and use of coercive parenting affect one another before, during, and after the intervention and (2) how the change in parental adjustment and use of coercive parenting during the intervention predict subsequently reported a decrease in child behavioral and emotional problems. We hypothesized that: (1) the decrease in parental adjustment difficulties over the course of intervention will be associated with the simultaneous decrease in coercive parenting behaviors and (2) the decrease in parental adjustment difficulties and coercive parenting from pre- to postintervention will predict a subsequent decrease in child behavioral and emotional problems at follow-up.

#### **2. Materials & Methods**

#### *Participants*

Participants in this study were 891 parents and caregivers living in the Australian states of Victoria and Queensland and enrolled in the Mental Health of Young People with Developmental Disabilities (MHYPeDD) research study. These were caregivers of a child aged between 2 and 12 years who were recruited via a variety of pathways, e.g., posters, brochures, and newsletters prepared by the project team and disseminated by their current service provider, a project-specific Facebook page, direct contact from the project team or via their child's school. Interested parents then provided evidence of a of diagnosis of DD provided by a suitable professional such as a psychiatrist, psychologist, speech pathologist, neurologist or pediatrician. Although most parents responded to questionnaires online, there was also the option of telephone interviews or hard copies if needed. Most participants (87.65%) were mothers (biological, stepmother, adoptive mother) of at least one child with DD. The majority of target children were male (77.10%) aged between 2 and 12 years old (*M =* 4.98, *SD =* 2.65). At the time of enrolment, 76.77% parents reported their child also had a diagnosis of ASD (Table 1).

**Table 1.** Demographic characteristics.




Note: The accumulate percentage might not equal to 100% due to missing data.

#### **3. Procedure**

The study received ethical clearance from the Behavioral and Social Sciences Ethical Review Committee at the University of Queensland. Professionals involved in the MHYPeDD project received training on at least one SSTP program, and they participated in a two-year implementation period to deliver interventions to families of children with DD. Parents were referred by their current service providers, directly via the SSTP project team (including the Facebook page) or their child's school to attend either Primary Care SSTP (3–4 brief individual sessions), SSTP seminars (120-min large-group presentations), Group SSTP (5 group sessions and follow-up telephone calls), Standard SSTP (10 individually delivered sessions); Self-directed Triple P or Triple P Online (self-administered). Parents completed a short package of measures before the intervention, after the intervention and then at six months following their attendance.

To minimize site differences, all practitioners received identical competency- and accreditation-based training and all interventions were delivered with the same practitioners and parent resources. This is a widely used method of Triple P dissemination, ensuring fidelity to the program. Studies of Triple P regular service delivery have shown that there are few differences between training outcomes for practitioners from different disciplines, countries, and levels of programs [16,17].

#### **4. Measures**

#### *4.1. Demographics*

Demographic variables used for analysis in the present study included the child's age and gender, type of DD (with or without ASD), caregivers' relationship to the child, financial hardship and level of intervention. Responses were mainly based on the primary carer. Financial hardship was assessed using the question: "*Suppose you only had one week to raise \$2000 for an emergency. Which of the following best describes how hard it would be for you to get that money*?", with responses ranging from 1 (*I could easily raise the money*) to 4 (*I don't think I could raise the money*). This item has been demonstrated to be a good index of financial hardship [18].

#### *4.2. Parental Adjustment*

The parent adjustment subscale of the Parent and Family Adjustment Scale developmental disability version (PAFAS-DD) was used to measure parental adjustment difficulties. The PAFAS-DD has 30 items measuring parenting and family adjustment [19] on a scale from 0 (None at all) to 3 (Very much/most of the time) with higher scores indicating a higher level of dysfunction within families. The Parental Adjustment subscale has five items that assess parents' emotional adjustment at the time of the survey. Examples of questions include: "*I feel satisfied with my life*" or "*I cope with the demands of being a parent*". PAFAS-DD Parental Adjustment subscale has been shown to be a reliable measure to assess parental adjustment difficulties with internal consistency found in previous studies ranging from *α* = 0.81; [11] to *α* = 0.82 [19]. PAFAS-DD has also been shown to have satisfactory construct and convergent validity [19].

#### *4.3. Coercive Parenting*

The Coercive Parenting subscale of the PAFAS-DD was used to measure participants' levels of coercive parenting. The Coercive Parenting subscale is comprised of five items that describe parenting behaviors such as: "*I shout or get angry at my child when they misbehave*" or "*I get annoyed with my child*". Parents indicated how true the statement is to their parenting practice on a scale from 0 (None at all) to 3 (Very much/most of the time). A higher score indicates more use of coercive parenting. This scale has been demonstrated to be a valid and reliable measure of coercive parenting in families of children with DD. Composite internal consistency found for this subscale was 0.75, [19] and internal consistency was *α* = 0.73 [11].

#### *4.4. Child Behavioral and Emotional Problems*

Parents reported child behavioral and emotional problems using the Child Adjustment and Parenting Efficacy Scale—Developmental Disability (CAPES-DD) [20]. CAPES-DD has 30 items that describe different behavioral and emotional problems in children. Examples of items are: '*breaks or destroys things*' (Behavioral problems) and '*seems fearful and scared*' (Emotional problems). Parents indicate how accurately the problems describe their children by rating on a scale from 0 ('*Not true of my child at all*') to 3 ('*True of my child very much, or most of the time*'). The CAPES-DD yields three scores. The Behavioral Problems score, Emotional Problems Score and Total Problems score. CAPES-DD has consistently been found to have good internal consistency with Cronbach's alpha ranging between (*α* = [0.80–0.90]) for both the subscales and the total score [11,21]. CAPES-DD total problems scale also correlates strongly with the total behavior problems scale of the Developmental Behavior Checklist for both Primary Carer and Under-4 versions [21].

#### *4.5. Analytic Strategies*

Repeated measures ANOVA was adopted to calculate the change of scores across three time points. The relationship between parental adjustment and coercive parenting was explored using latent growth modelling (LGM) in which intercepts represent baseline score and slopes represent latent change over time. LGM allows researchers to explore the growth of individual constructs while simultaneously examining the relationship between several constructs. Bi-directional relationships were estimated with the error covariances, and the unidirectional relationships were estimated with path coefficients. A comparative fit index (*CFI*) value ≥ 0.95 and the root means square error approximation (*RMSEA)* value ≤ 0.08 indicates a good fit.

To estimate the contribution of change in parental adjustment and change in coercive parenting to subsequent child behavioral and emotional problems, hierarchical multiple regression was conducted. Demographic variables of families and child behavioral/emotional problems at baseline were controlled at Step 1 and Step 2 before changes in parental adjustment and coercive parenting (Time 1–Time 2) were entered at Step 3 to predict Time 3 child's emotional and behavioral problems. The analyses were undertaken using AMOS and R software for statistic computing.

#### **5. Results**

#### *5.1. Missing Data Analysis*

The analysis of missing data indicated there was 18.7% missingness in total. Little's MCAR test was not significant (*X<sup>2</sup> =* 175,818.70, *df* = 190,183, *p* > 0.05). The maximum likelihood estimation method was used to handle missing data.

#### *5.2. Change in Parental Adjustment, Parenting Behaviors, and Child Behaviors*

Table 2 shows the mean score of PAFAS- Adjustment, PAFAS—Coercive, CAPES-DD-Behaviors and CAPES-DD Emotion at Time 1, Time 2, and Time 3. The PAFAS- Adjustment score of 5.75 (*SD =* 3.00) pre-intervention, decreased significantly (*F*(1,890) = 80.37, *p* < 0.05) to 5.11 (*SD* = 2.61) post-intervention, and was maintained at the 12-month follow-up period ending at 5.02 (*SD =* 2.56). A significant reduction in the Coercive parenting score was also observed from Time 1 to Time 2. At Time 1, PAFAS- Coercive score was *M =* 9.66 (*SD* = 2.61) which reduced significantly to *M =* 8.73 (*SD =* 2.27) at Time 2 (*F*(1,890) = 230.31, *p* < 0.05). This effect was maintained at Time 3 at *M* = 8.85, *SD* = 2.09 (*F*(1,890) = 157.20, *p* < 0.05).


**Table 2.** *Mean*, *SD* of Variables.

To examine whether the change scores are different across different levels of intervention, analysis was conducted controlling for the level of intervention. Results showed no interaction effect between time and level of intervention for any of the variables either short-term or long-term.

Child behavior problems estimated with the CAPES-DD Behavior started at Time 1 at *M* = 23.04 (*SD* = 6.65) then significantly reduced to *M* = 21.61 (*SD* = 6.15) at Time 2 (*F*(1,890) = 116.27, *p* < 0.05). The comparison between Time 3 and Time 1 was also significant (*F*(1,890) = 153.74, *p* < 0.05) indicating that the effect was maintained at Time 3.

Child emotional problems measured by CAPES-DD Emotion was 5.04 (*SD* = 1.80) at Time 1 and reduced significantly to *M* = 4.73 (*SD* = 1.57) (*F*(1,890) = 42.95, *p* < 0.05). This change was maintained at Time 3 (*F*(1,890) = 14.75, *p* < 0.05).

#### *5.3. The Association between Change in Parental Adjustment and Change in Coercive Parenting*

The LGM model to test the correlation of changes between parental adjustment and coercive parenting is presented in Figure 1 and Table 3. The model fits the data well (*X2* = 29.508, *df* = 7, *p* < 0.05; *CFI* = 0.993, *RSMEA* = 0.060). At Time 1, parental adjustment was significantly and positively associated with coercive parenting (*covariance coefficient* = 0.40, *p* < 0.05), indicating that those who experienced more adjustment difficulties at Time 1 were more likely to use coercive parenting strategies and vice versa. The examination of the change scores showed that changes in parental adjustment throughout intervention were significantly and positively associated with the change in coercive parenting (*covariance coefficient* = 0.16, *p* < 0.05). This finding suggests that a reduction in parental adjustment difficulties was associated with a reduction in coercive parenting.

**Table 3.** Estimate, Standard error of Coefficient and *p*-values of slopes and intercepts association.


**Figure 1.** Latent growth Model of Intervention Outcomes. Note: \*\*\* *p* < 0.001.

*5.4. The Association between Changes in Parental Adjustment and Coercive Parenting to Change in Child Behavioral and Emotional Problems*

The association between variables are presented in Table 4.

**Table 4.** Cross-correlation between variables.


Note: \*\* *p* < 0.01.

Prior to conducting the regression analysis, all assumptions for multiple regression including linearity, multivariate normality, multicollinearity, and homoscedasticity were conducted. Data met the assumptions required for multiple regression.

After accounting for demographics and child behavioral/emotional problems at Time 1, the change in coercive parenting throughout Intervention (Time 1–Time 2) significantly predicts the behavioral problems (*β* = 0.10, *p* < 0.01) and child's emotional problems (*β* = 0.11, *p* < 0.01) at Time 3. However, changes in parental adjustment did not significantly predict either child emotional or behavioral problems at Time 3 (Table 5).


**Table 5.** Hierarchical multiple regression to predict child Behavior and Emotion problems at Time 3.

\* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001. ASD: Autism Spectrum Disorder.

#### **6. Discussion**

This study sought to extend the literature by examining how changes in parental adjustment and parenting skills at key timepoints throughout SSTP intervention affect each other and subsequently influence children's outcomes. To our knowledge, this is the first mechanism of change analysis of the SSTP intervention. Results of this study suggested two main findings: First, the changes in parental adjustment over the course of intervention were associated with the changes in coercive parenting such that a decrease in parental emotional adjustment was correlated with a decrease in coercive strategies used. These change processes appear to co-occur such that when there is a reduction in parental emotional adjustment, there is also a reduction in coercive parenting. Second, the decrease in coercive parenting but not parental emotional adjustment achieved via intervention significantly contributed to children's behavioral and emotional performance at follow-up.

When examining the relationship between baseline performance and the trajectory of changes, we found that parents who used more coercive parenting and reported more emotional difficulties at baseline demonstrated greater changes over the course of intervention. Such findings were consistent with and supported previous studies of a behavioral parenting intervention, suggesting that families with more problems at baseline respond better to intervention [22,23]. Families with more problems at baseline may have greater scope for growth thus demonstrating better progress than families with fewer initial concerns of which little improvement was needed.

Research conducted on families of children with DD in the past two decades has highlighted that an effective intervention for children with DD needs to address parental distress in order to bring about change in children. This argument is based on evidence to the link between parental adjustment difficulties and children's outcomes [8,12]. In this study, we hypothesized that as parent's emotional difficulties decrease, children's behavioral and emotional problems will mutually deescalate. In contrast to our hypothesis, we found that change in parental adjustment throughout intervention was significantly associated with the change in coercive parenting, but its increase or decrease over the course of intervention did not contribute directly to the increase or decrease of subsequent child behavioral and emotional problems.

To explain this finding, we need to understand the SSTP program's model of change. SSTPs and Triple P programs are built on the foundation of self-regulation principles. According to Sanders [24], self-regulation refers to the ability to change one's own behavior and become an independent problem-solver by gaining the skills necessary to achieve one's personal goals. In social cognitive theory, self-regulation is viewed as an essential process through which individuals can guide their behavior through changing circumstances over time. It involves modulating thought, affect, behavior, or attention with specific mechanisms and supportive meta-skills. As a result, SSTP training encourages parents to identify their goals, plan and self-select the most appropriate strategies to manage their emotions, and conduct self-evaluations addressing changes if needed [25]. The parents are thus able to attribute changes to their own behavior and effort rather than the child's difficulties. Furthermore, self-evaluation might help parents become more aware of their own behavior and better able to assess situations before responding to them. By selfregulating, parents may have been able to distinguish their own emotions from their children's difficulties and avoid passing on emotional disturbances to their children [26].

The findings of this study have implications for the development of parenting programs that support families of children with developmental disabilities; suggesting that parenting skills are key to influencing children's behavior. Meanwhile, it is also important to highlight that although parental adjustment was not directly correlated with change in children's behaviors, change in parental adjustment throughout the intervention, actually fostering healthy functioning parenting. As parents experience less emotional distress, they might be able to focus on building a positive, nurturing relationship with their children, resulting in a decrease in child behavioral and emotional problems

The finding of a non-significant relationship between change in parental adjustment and subsequent child outcomes nevertheless needs to be interpreted with care. It is possible too that some aspects of parental adjustment, such as the presence of parental mental health problems (anxiety and depression) might take longer to recover and change in response to reductions in problematic child behavior. There might also be a floor effect for problems, as most parents in our sample did not experience elevated adjustment difficulties hence a relatively small effect size was observed. Future studies with multiple data collection points and with clinically elevated samples might be useful to understand the cumulative and interactionally dependent between parental adjustment and child behavior change over time.

Finally, the mechanism for testing relationships between variables was restricted to three-time point data and was based on single parent reports which might require further validation. As Rutter [27] has suggested, to answer causal questions about development, there is a need for integrating longitudinal data in experimental intervention. More data points would allow a more accurate trajectory of change across individuals using LGM. Additional assessment time points would also allow a more definitive conclusion of the causal relationships which would enhance our understanding of the mechanisms by which changes in parent-related variables produce changes in child outcomes.

This study suggests that positive parenting skills are the most salient intervention ingredient driving the change in child behaviors, and the continued focus on building parenting capacities and parenting skills is justified. Although conducting a moderator analysis was not the primary focus of this study, our findings of baseline effect on family changes over the course of intervention are valuable in pointing to subgroups that might benefit the most from SSTP intervention. Our regression model findings also indicate several potential moderators (baseline status of families, DD with or without ASD, parental education, and family financial hardship) that can impact families' capacity to change through intervention. Future studies with adequate sample sizes and more advanced analytic techniques could conduct moderator-mediator analysis to explore how mechanisms of change might vary by moderator groups.

#### **7. Conclusions**

Using a community-based sample of SSTP roll-out, this study emphasizes the significant role of parenting behaviors in improving children's outcomes and suggests that

developing warm, positive relationships between parents and their children should continue to be a priority in evidence-based parenting programmes for parents of children with DD. Additionally, this study also highlights the importance of promoting parental emotional well-being throughout intervention in order to mitigate the tendency to engage in coercive or negative parenting practices that are detrimental to children's development.

**Author Contributions:** Conceptualization, M.S. and N.-P.T.H.; Methodology, N.-P.T.H.; Software, N.- P.T.H.; Validation, N.-P.T.H. Formal Analysis N.-P.T.H.; Investigation, M.S. and N.-P.T.H.; Resources M.S. and J.H.; Data Curation M.S. and N.-P.T.H.; Writing—Original Draft Preparation, N.-P.T.H. and M.S.; Writing—Review and Editing M.S., J.H., K.S., S.E., B.T. and K.G.; Visualization, N.-P.T.H.; Supervision, M.S.; Project Administration, J.H., K.S. and The MHYPEDD team; Funding Acquisition, M.S., J.H., K.S., S.E., B.T. and K.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by a grant from the Australian Government's National Health and Medical Research Council (NHMRC programme grant APP1016919) for the Mental Health of Young People with Developmental Disabilities (MHYPeDD) project.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of The University of Queensland (protocol code 2012001065, Date of approval: 20 September 2012).

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

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethics commitment of data security.

**Conflicts of Interest:** The Parenting and Family Support Centre is partly funded by royalties stemming from published resources of the Triple P—Positive Parenting Program, which is developed and owned by The University of Queensland (UQ). Royalties are also distributed to the Faculty of Health and Behavioral Sciences at UQ and contributory authors of published Triple P resources. Triple P International (TPI) Pty Ltd. is a private company licensed by Uniquest Pty Ltd. on behalf of UQ, to publish and disseminate Triple P worldwide. The authors of this report have no share or ownership of TPI. Authors Sanders, Hoang and Hodges receive or may in future receive royalties and/or consultancy fees from TPI. TPI had no involvement in the study design, collection, analysis or interpretation of data, or writing of this report. Authors Sanders, Hoang and Hodges are employed by UQ; author Sofronoff holds an honourary appointment with UQ. Other authors have no conflict to declare.

#### **References**


## *Article* **The Global Deterioration Scale for Down Syndrome Population (GDS-DS): A Rating Scale to Assess the Progression of Alzheimer's Disease**

**Emili Rodríguez-Hidalgo 1,2, Javier García-Alba 3, Ramon Novell 1,2 and Susanna Esteba-Castillo 1,2,\***


**Abstract:** The aim of this study is to adapt and validate the global deterioration scale (GDS) for the systematic tracking of Alzheimer's disease (AD) progression in a population with Down syndrome (DS). A retrospective dual-center cohort study was conducted with 83 participants with DS (46.65 ± 5.08 years) who formed the primary diagnosis (PD) group: cognitive stability (*n* = 48), mild cognitive impairment (*n* = 24), and Alzheimer's disease (*n* = 11). The proposed scale for adults with DS (GDS-DS) comprises six stages, from cognitive and/or behavioral stability to advanced AD. Two neuropsychologists placed the participants of the PD group in each stage of the GDS-DS according to cognitive, behavioral and daily living skills data. Inter-rater reliability in staging with the GDS-DS was excellent (ICC = 0.86; CI: 0.80–0.93), and the agreement with the diagnosis categories of the PD group ranged from substantial to excellent with κ values of 0.82 (95% CI: 0.73–0.92) and 0.85 (95% CI: 0.72, 0.99). Performance with regard to the CAMCOG-DS total score and orientation subtest of the Barcelona test for intellectual disability showed a slight progressive decline across all the GDS-DS stages. The GDS-DS scale is a sensitive tool for staging the progression of AD in the DS population, with special relevance in daily clinical practice.

**Keywords:** global deterioration scale; rating scale; cognitive decline; cognitive testing; Alzheimer's disease; dementia; down syndrome; intellectual disability

#### **1. Introduction**

Life expectancy in people with Down syndrome (DS) has increased considerably, to up to sixty years on average [1], evidencing an accelerated aging phenotype [2]. The risk of developing Alzheimer's disease in people with DS (AD-DS) is greatly elevated due to the gene overexpression produced by the chromosome 21 trisomy, in which amyloid precursor protein and other gene modulations involve an accumulation of extracellular amyloid beta and neurofibrillary tangles [3,4]. AD neuropathology is present in more than 60% of people with DS by the age of 40 to 50 years [5]. Therefore, people with DS are considered at absolute risk of developing AD [6] and become a genetic paradigm for translational research with regard to the general population [7]. Despite these conceptual advances, AD-DS is usually diagnosed around the age of 50 years [8,9], later than expected [10], and these individuals are usually excluded from the clinical trials of typically developing individuals with AD [11].

Some major stages of the natural course of AD-DS have been suggested. Diagnostic categories in some studies are as follows: asymptomatic, cognitively stable, preclinical, prodromal AD, AD dementia, and uncertain [12–14]. Recently, under the biological definition and diagnosis relying on biomarkers, five stages have been proposed for AD-DS [15]:

**Citation:** Rodríguez-Hidalgo, E.; García-Alba, J.; Novell, R.; Esteba-Castillo, S. The Global Deterioration Scale for Down Syndrome Population (GDS-DS): A Rating Scale to Assess the Progression of Alzheimer's Disease. *Int. J. Environ. Res. Public Health* **2023**, *20*, 5096. https://doi.org/10.3390/ ijerph20065096

Academic Editor: Shoumitro Deb

Received: 29 December 2022 Revised: 9 March 2023 Accepted: 13 March 2023 Published: 14 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

cognitively stable (CS), preclinical AD, mild cognitive impairment (MCI), prodromal AD and dementia. Bearing these data in mind, it seems self-evident that there is a lack of consensus on the proposed stages to capture the entire clinical continuum of AD-DS which, on the contrary, are based primarily on biological criteria. Furthermore, recent studies in the general population suggest different initial stages on the dementia continuum. In this sense, a subjective cognitive decline (SCD) stage has been proposed as a risk condition announcing possible dementia [16,17], but it must be supported by objective cognitive impairment to be considered part of the AD continuum [18]. Additionally, mild behavioral impairment (MBI) [19,20] has been proposed as a prodromal neuropsychiatric stage of emergent dementia. Due to the high risk of developing AD in subjects with DS, and considering that AD-DS pharmacological interventions are not efficient enough [21,22], it is necessary to arrange a grading scale that captures the whole clinical spectrum of the AD-DS with the inclusion of these recently proposed diagnostic categories to detect early therapeutic windows.

Although the biological perspective research framework has meant a forward leap in the prediction of MCI and AD, cognitive markers continue to be essential. There are several cases in daily clinical practice in which neuroimaging and biomarker findings suggest a certain diagnosis but without being supported by a neuropsychological examination, as in cases of biomarker positivity for AD without associated or progressive cognitive impairment [18]. Because biomarkers may fail in the diagnosis of AD, a neuropsychological examination can provide crucial cognitive information in the diagnostic and follow-up processes of the ID population [23]. In this sense, the aspects that should be explored in any longitudinal assessment for the entire continuum of AD-DS are dysexecutive cognitive or behavioral features, episodic memory, orientation, and general global cognitive decline [24–27].

The high prevalence of MCI and AD in individuals with DS has resulted in much research concerning the detection of both diagnoses. In a recent paper, the diagnostic criteria to delimit MCI in people with DS (MCI-DS) have been proposed [28], a stage that can be considered as prodromal of AD [29]. These authors showed that scores in the behavior rating inventory of the executive function parent form (BRIEF-P) [30] combined with scores on abstract thinking and verbal memory are useful values for detecting MCI-DS. Furthermore, the memory, language, and communication sections of the National Task Group-Early Detection Screen for Dementia (NTG-EDSD) seem to be sensitive to MCI-DS [31]. The Cambridge Cognitive Examination for Older Adults with Down's syndrome—Spanish version (CAMCOG-DS) [32], with different cut-offs points, has been provided to detect prodromal AD or MCI and AD in people with DS and mild or moderate ID [13,32]. Furthermore, visuospatial-paired associate memory, hand–eye coordination, and semantic verbal fluency may be relatively sensitive events in the prodromal stage of AD-DS [10]. Finally, the performance of episodic memory using a modified cued recall test discriminates between individuals with preclinical, prodromal, and clinically manifest AD, albeit of mild degree [33]. Compared with the aforementioned major phases, the later stages of the AD-DS spectrum have not received special attention nor an international official definition.

The progression of the disease in the clinical setting is assessed through cognitive, behavioral, psychiatric, and clinical tests. Additionally, rating scales provide a common language to diagnose, monitor, and evaluate therapeutic interventions. A good rating scale would be practical and properly validated in the target population, embracing different domains beyond cognition, and be sensitive to change in all stages to measure therapeutic effects [34], both pharmacological and non-pharmacological. In the general population, there are two reference scales for grading dementia. One of them is the clinical dementia rating scale (CDR) [35], a semi-structured interview with a patient and a close caregiver, covering six domains (memory, orientation, judgment and problem solving, work in the community, performance at home and in hobbies, and personal care). The results are distributed on a five-point scale from cognitive normality to severe dementia. Interestingly, an adaptation of the CDR, the CDR for frontotemporal lobar degeneration (CDR-FTDL) [36], was developed with the inclusion of language and behavior domains not contemplated in the original CDR, and has been shown to be sensitive in distinguishing disease progression between FTLD and AD in the general population [37]. Additionally, a modified CDR for adults with DS based on questionnaires and patient/caregiver interviews is available [38], capturing the progressive deterioration of AD in this population.

The second grading dementia scale is the global deterioration scale (GDS [39], which determines the degree of functional loss based on the severity of the cognitive impairment. It consists of descriptions of the clinically differentiated stages of the AD continuum, from stage 1 (normal) to 7 (severe AD). Subsequently, the functional assessment staging (FAST) [40] provides the GDS with a division of phases 6 and 7, involving the progressive inability to maintain basic activities of daily living. The GDS is easier to apply than the CDR as it correlates with neuropathological, functional, global, and cognitive changes in the progression of MCI to AD [41] and with the hippocampal volume [42]. Furthermore, in our context, the GDS is the instrument of choice that determines the introduction and withdrawal of pharmacological and non-pharmacological treatment in patients with AD [43,44]. Therefore, it would be desirable to arrange a global deterioration rating scale which contemplates the amnesic [45,46] and the behavioral [47,48] forms of onset classically described in AD-DS. Consequently, such a scale would improve both the response to health needs according to the time point of disease progression and provide researchers in this field with a common language in our context.

The main objective of the present study is to examine the feasibility of an adapted GDS for its use in people with DS on the continuum of AD. We expect to find that the stages proposed for GDS-DS are anchored to the performance of cognitive and behavioral instruments that are well established in people with DS and a mild or moderate level of ID.

#### **2. Materials and Methods**

#### *2.1. Study Design and Description of the Sample*

This is a retrospective, dual-center cohort study of Caucasian adults with DS. A total of 87 participants were identified from the Servicio Especializado en Salud Mental y Discapacidad Intelectual (Specialized Mental Health ID Unit, Institute of Health Assistance, Girona) and the Unidad de Adultos con Síndrome de Down (Adult Down Syndrome Unit, La Princesa University Hospital, Madrid, Spain).

A neurological, psychiatric, and laboratory examination was applied to all the participants. The neurological examination consisted of taking the participant's history (e.g., previous central nervous system alterations, relevant drug treatment, substance abuse, sleep disorders) and a physical examination. Psychiatric data were collected using the psychiatric assessment schedule for adults with a developmental disability (PAS-ADD) [49]. Blood samples were obtained in order to detect hypothyroidism, vitamin B12 deficiency, and anemia.

The inclusion and exclusion criteria of the study are displayed in Table 1. Going into detail, the age cut-off was chosen based on previous studies which reported that being above 39 years of age represents a high risk of developing cognitive decline from a previous level of efficiency in people with DS [50]. The level of ID was determined according to the DSM-5 criteria [51]. All participants were required to have a reliable informant available to report on the present and past adaptive skills and behavior of the participants.

Those with sensory impairments that prevented the completion of the research protocol, those with a history of alterations of the central nervous system (e.g., brain tumors, head injury, stroke), those with uncontrolled sleep disorders (e.g., obstructive sleep apnoea), and individuals suffering from substance abuse were excluded. Patients were also excluded if they had untreated anemia, vitamin B12 deficiency, or uncontrolled hypothyroidism because of their potential risks of influencing behavior.


**Table 1.** Inclusion and exclusion criteria.

It should be noted that cases of conduct disorder, depression, and anxiety were not automatically ruled out. Those cases caused by a stressful event in the past six months, were excluded but those (un)treated cases related to normal daily functioning were not, according to clinical judgment.

Finally, participants exposed to a high anticholinergic burden through polypharmacy from psychotropics, gastrointestinal and cardiovascular medications were excluded if the treatment was considered ineffective, according to clinical judgment.

Our study was conducted in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki). The study protocol was approved by the Clinical Research Ethical Committee of the Parc Hospitalari Santa Caterina (Girona). Written informed consent was obtained from parents and written and pictorial assent were additionally obtained.

#### *2.2. Clinical Assessment*

The protocol consisted of a cognitive test and behavioral and daily living skills questionnaires that have been shown to be most sensitive to cognitive impairment in Spanishspeaking subjects with DS.

#### 2.2.1. Cognitive Assessment

The protocol with the following test was applied to all the participants:


2.2.2. Informants' Questionnaires


The participants of the PD group were classified according to three diagnostic categories: cognitive stability, MCI and AD. The diagnosis of cognitive stability was made when participants had no cognitive impairment or decline in adaptive skills, according to the CAMDEX-DS informant section. As there is currently no internationally accepted official definition for MCI-DS [57], the diagnosis of MCI or AD was based on expert multidisciplinary clinical judgment according to recent publications, as is recommended in standard practice for DS [26,28,58–61]. In detail, a participant fulfilled the diagnosis of MCI when presented with a single or multiple cognitive decline and/or loss of functionality according to the CAMDEX-DS informant section [26]. The diagnosis of AD was obtained if participants had memory decline or another cognitive impairment, such as aphasia, apraxia, agnosia, or dysexecutive syndrome, or loss of functionality [62]. In both conditions, the information about changes from previous levels of performance must be supported by a close caregiver [26,58,60]. It should be noted that the CAMDEX-DS was used only for diagnostic purposes, whereas performance on the neuropsychological tests confirmed the diagnosis through a longitudinal clinical study [26].

#### 2.2.3. Global Deterioration Scale for Down Syndrome (GDS-DS)

The GDS-DS began at stage 1 (cognitive and behavioral stability), in which subjects were placed if they had neither subjective complaints nor cognitive or behavioral impairments identified by the instruments of the study protocol. Stage 2 (subjective cognitive and/or behavioral impairment) included those subjects with changes in cognition, behavior or adaptive skills, as self-reported or reported by caregivers but not supported by objective data. Stage 3 (mild cognitive and/or behavioral impairment) corresponded to those cases

with reports of cognitive or behavioral impairment by the patient or confirmed by a reliable informant, supported by objective data, with no or very mild loss of adaptive skills. Stages 4 (mild Alzheimer's disease), 5 (moderate Alzheimer's disease) and 6 (advanced Alzheimer's disease) were reserved for those subjects presenting mild to severe deterioration in cognitive and behavioral domains, as reported by an informant, supported by objective data and demonstrating affected adaptive skills (Table 2).


**Table 2.** Stages and diagnostic criteria for each stage of the GDS-DS.

GDS-DS, global deterioration scale for people with Down's syndrome; BRIEF-P BRI, behavior rating inventory of executive function, parents' form behavioral regulation index; CAMCOG-DS, Cambridge cognitive examination for older adults with Down's syndrome and other intellectual disabilities; ID, intellectual disability; CAMDEX-DS, Cambridge examination for older adults with Down's syndrome and other intellectual disabilities.

#### • GDS-DS adaptation

The GDS-DS was constructed on the basis of the global deterioration scale (GDS) [39]. Some modifications were introduced to adapt this scale to the study population. First, the original GDS scale consists of seven stages (1–7), whereas the GDS-DS consists of six stages (1–6). Secondly, because growing evidence suggests that subjective cognitive decline [16,17] and mild behavioral impairment [19,20] could predict AD, subjective cognitive and behavioral aspects were considered. Therefore, stage 2 was labeled as subjective cognitive and/or behavioral impairment and stage 3 as mild cognitive and/or behavioral impairment.

In general, for the delimitation of the GDS-DS stages, the performance of the participants in the cognitive instruments and the information of the behavioral questionnaire and caregivers gathered from the study protocol were classified as mandatory or supporting criteria, according to the recent literature about AD and DS (Table 2).

• Mandatory criteria

The mandatory criteria were established based on those cognitive and behavioral instruments that had obtained sufficient quantitative data to differentiate MCI and AD, in addition to adaptive skills. The purpose was to establish criteria that were as inclusive as possible; thus, the following instruments and scores were considered:


It should be noted that the BRIEF-P (BRI index) and the CAMCOG-DS, in addition to providing a quantitative basis for each stage of the GDS-DS, also cover the symptoms of the two most frequent forms of AD onset in the DS population, either the amnestic [45,46] or the behavioral variants [47,48].

• Supporting criteria

The supporting criteria were established based on those cognitive and behavioral instruments of the protocol study that demonstrated a decrease associated with an increase in the clinical intensity of AD-DS from the MCI-DS phase. These criteria might not be mandatory, but their presence helped determine the stage of the GDS-DS that would be assigned to each participant. The instruments considered for this purpose were:

1. BT-ID: the orientation, semantic fluency (eating and drinking), formal fluency, delay verbal memory (stories) and visual discrimination subtests.

#### 2. CAMCOG-DS: abstract thinking subtest.

Recent studies claim that significantly lower scores on the above subtests can be indicative of MCI-DS and the early phase of AD [10,13,26,28]. Therefore, these were added as supporting criteria for AD-DS from stage 3 (mild cognitive and/or behavioral impairment) upward (stages 4, 5 and 6).

It should be emphasized that in order to identify the stages of AD (4, 5 and 6), in addition to the mandatory and supporting criteria established, the degree of deterioration (mild, 4; moderate, 5; and advanced, 6) would have to be determined by the judgment and clinical experience of specialists, who play a decisive role in this regard.

For stage 6 (advanced Alzheimer's disease), an additional criterion was established. Based on our clinical experience and other studies of populations with DS, it has been suggested that some tests cannot be administered to those with advanced stages of AD or in populations with severe intellectual disabilities [57] because they do not provide enough information due to the "floor effect" [22]. Therefore, the criterion of not being able to administer all or any cognitive tests from the screening protocol to participants would, on its own, be an indicator of advanced AD, supported by affected adaptive skills.

#### *2.3. Procedures*

Cognitive, behavioral and adaptive skills data from the PD group were used retrospectively to place each participant in a stage of the GDS-DS and thus form the GDS-DS group. Two neuropsychologist specialists in DS blinded to the diagnosis of the PD group (SEC, JGA) placed each subject into one of the six levels of the proposed GDS-DS rating scale, according to the level of ID of each participant and the mandatory and supporting criteria established (Table 1). It should be highlighted that in order to avoid circularity, the two specialists were different to those who made the diagnoses for the PD group. The degree of agreement between the examiners in placing the subjects in the stages of the GDS-DS was checked. The inter-rater reliability test was applied to the two raters' first GDS-DS classifications. Then, the classifications of the two raters was transformed into single values as follows: if they matched the GDS-DS assignment, the same value was maintained, and when the two raters did not agree on the classification using the GDS-DS, the case was discussed until a consensus was reached. Additionally, the demographic and the diagnostic categories of the PD and the GDS-DS groups were analyzed in order to observe possible similarities.

The agreement between the two raters classifying the participants using the GDS-DS compared with the PD group was analyzed. For this purpose, the stages 1 (cognitive and behavioral stability) and 2 (subjective cognitive and/or behavioral impairment) of the GDS-DS were associated with the cognitive stability diagnosis of the PD group; stage 3 (mild cognitive and/or behavioral impairment) of the GDS-DS was associated with the diagnosis of mild cognitive impairment of the PD group; and stages 4, 5 and 6 (Mild, Moderate and Advanced Alzheimer's disease) of the GDS-DS were associated with the Alzheimer's disease of the PD group. Codes associated with each category of the PD group are displayed in Table 3.

**Table 3.** Recode numeric values.


GDS-DS, global deterioration scale for people with Down's syndrome.

Finally, as the progressive deterioration of adaptive skills was an immovable criterion and an analysis with this variable could lead to a circularity problem, the authors checked for a possible association of the GDS-DS stages with a selection of the cognitive and behavioral instruments included as mandatory and supporting criteria for each stage.

#### *2.4. Statistical Analysis*

We performed all statistical analyses using the software program SPSS (version 27.0; SPSS Inc., Chicago, IL, USA). Bilateral significance levels were set at a *p*-value of less than 0.05. The normality of data was assessed using the Kolmogorov–Smirnov test, and subsequently, non-parametric analyses were carried out. For the GDS-DS group, demographic characteristics, level of ID, and performance on selected neuropsychological tests were analyzed by a non-parametric Kruskal–Wallis test followed by the Bonferroni correction for post hoc pairwise comparisons or Pearson's chi-square test for category data. Spearman's correlation analysis was used to determine the associations between the stages of the GDS-DS with the cognitive and behavioral data selected. A Mann–Whitney U test was performed with pairs of the GDS-DS groups because the effect size between the cognitive and behavioral performance across the GDS-DS stages was computed using the non-parametric probability of superiority estimation (PSest), interpreted as small (≥56), medium (≥0.64), and large (≥0.71) [63]. Inter-rater agreement and concordance between the stages issued by rater 1 and rater 2 with the primary diagnosis of the study were analyzed by using Cohen's weighted kappa values.

#### **3. Results**

#### *3.1. Sample and Demographics*

Of the subjects selected, four ultimately did not agree to participate in the study. The final sample formed the primary diagnosis (PD) group and consisted of 83 subjects (46.65 ± 5.08 years; male = 46 (55.4%), female = 37 (44.6%)) and was divided into three groups: (1) cognitively stable group, with 48 subjects (45.10 ± 3.83 years; male = 24 (50%), female = 24 (50%)); (2) mild cognitive impairment group, with 24 subjects (47.46 ± 5.35 years; male = 16 (66.7%), female = 8 (33.3%)); and (3) Alzheimer's disease group, with 11 subjects (51.64 ± 6.05 years; male = 6 (54.5%), female = 5 (45.6%)). Seven of the initial candidates had to be excluded because they could not undergo the neuropsychological assessment.

The demographic details of the PD group are reported in Table 4. Of the total 83 participants, about half were diagnosed as being cognitively stable, one third as having mild cognitive impairment and a smaller number as having Alzheimer's disease. The mean age of the subjects with AD was significantly higher than the stable subjects but not than the MCI group. No statistical differences were observed in the gender and level of ID across the groups.


**Table 4.** Demographics of the primary diagnosis sample.

Age values are shown as means and range. For the sex and ID level, percentages regarding the group are shown. ID, intellectual disability. \* *p* < 0.05, according to Kruskal–Wallis test with Bonferroni correction for age, or χ2-test for gender and ID level. a, between cognitive stability and Alzheimer's disease.

Within the GDS-DS group (Table 5), 27 participants were diagnosed as having cognitive stability (stage 1) and with subjective cognitive and/or behavioral impairment (stage 2), 25 with mild cognitive and/or behavioral impairment (stage 3) and 31 participants with any degree of AD (stages 4, 5 or 6). Interestingly, comparing these two groups, a similar number of participants were classified as MCI with both methods and 20 more subjects as AD. There were more participants diagnosed as more cognitively stable in the PD group than those placed in stage 1 of GDS-DS (48 and 9, respectively). As expected, subjects diagnosed with AD in the PD group and those placed on the stage 6 of the GDS-DS (advanced Alzheimer's disease) were the oldest.


**Table 5.** Demographics for each stage of the global deterioration scale for people with Down's syndrome.

Age values are shown as means and range. For the sex and ID level, percentages regarding the group are shown. ID, intellectual disability. \* *p* < 0.05, according to Kruskal–Wallis test with Bonferroni correction for age, or χ2-test for gender and intellectual disability level. a, between subjective cognitive/behavior impairment and advanced Alzheimer's disease; b, between mild cognitive/behavior impairment and advanced Alzheimer's disease.

#### *3.2. Cognitive and Behavioral Data across the GDS-DS Stages*

These analyses were conducted without the data from the stage 6 subjects, who were excluded because they could not complete the study protocol.

The analysis of the correlations revealed negative significant correlations between performance on the tests and the GDS-DS, except for the BRIEF-P and the visual discrimination BT-ID subtest. The correlations were moderate between the GDS-DS and the CAMCOG-DS Total score, orientation subtest and formal fluency, but weak with abstract thinking subtest of the CAMCOG-DS, free memory delay (stories) and semantic fluency of the BT-ID (Table 6).

The performance on the cognitive and behavioral tools across the GDS-DS stages is displayed in Table 7. Overall, the performance on the CAMCOG-DS total score and the orientation subtest (BT-ID) decreased significantly across stage 1 (cognitive and/or behavioral stability) to 5 (moderate Alzheimer's disease).


**Table 6.** Correlations between GDS-DS stages and cognitive/behavioral assessment.

GDS-DS, global deterioration scale for people with Down's syndrome; BRIEF-P, behavior rating inventory of executive function parents' form; BRI, behavioral regulation index; CAMCOG-DS, Cambridge cognitive examination for older adults with Down's syndrome and other intellectual disabilities; BT-ID, Barcelona test for intellectual disability. \*\* *p* < 0.01, (two-tailed test), using Spearman's correlation coefficient.



Values are given as median and range: GDS-DS, global deterioration scale for people with Down's syndrome; BRIEF-P, behavior rating inventory of executive function parents' form; BRI, behavioral regulation index; CAMCOG-DS, Cambridge cognitive examination for older adults with Down's syndrome and other intellectual disabilities; BT-ID, Barcelona test for intellectual disability. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, according to Kruskal–Wallis test with Bonferroni correction. a, between cognitive/behavioral stability and mild Alzheimer's disease; b, between cognitive/behavioral stability and moderate Alzheimer's disease; c, between subjective cognitive/behavioral impairment and mild Alzheimer's disease; d, between subjective cognitive/behavioral impairment and moderate Alzheimer's disease; e, between mild cognitive/behavioral impairment and moderate Alzheimer's disease; f, between mild cognitive/behavioral impairment and mild Alzheimer's disease. Significant differences in bold.

In terms of sectors, performance significantly decreased regarding:


• Semantic fluency (BT-ID) between stages 2 (subjective cognitive and/or behavioral impairment) and 5 (moderate Alzheimer's disease).

Note that (i) the performance on abstract thinking, orientation, delay stories, and formal fluency subtest (BT-ID) did not decrease between stage 1 (cognitive and behavioral stability) and 2 (subjective cognitive and/or behavioral impairment); (ii) the performance on the BRI index (BRIEF-P) decreased between 1 (cognitive and behavioral stability) and 3 (mild cognitive impairment), but was heterogeneous from stage 3 (mild cognitive and/or behavioral impairment) to stage 5 (moderate Alzheimer's disease); (iii) the performance on delay stories subtest (BT-ID) did not show significant oscillations between stage 2 (subjective cognitive and/or behavioral impairment) and 3 (mild cognitive and/or behavioral impairment); and the performance on the visual discrimination subtest (BT-ID) was similar across all the stages. These data had an impact on the posterior analysis of the effect sizes values.

The raw differences that showed a progressive decrease between performances on the selected tests across the GDS-DS stages are displayed in Table 8 and Table S1 of the online Supplementary Materials.


**Table 8.** Effect sizes between the GDS-DS stages.

GDS-DS, global deterioration scale for people with Down's syndrome; BRIEF-P, behavior rating inventory of executive function parents' form; BRI, behavioral regulation index; CAMCOG-DS, Cambridge cognitive examination for older adults with Down's syndrome and other intellectual disabilities; BT-ID, Barcelona test for intellectual disability. P, *p* value; PSest, probability of superiority effect size; \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *<* 0.001; <sup>a</sup> PSest ≥ 56, <sup>b</sup> PSest ≥ 0.64, <sup>c</sup> PSest ≥ 0.71. Significant effect sizes and *<sup>p</sup>* values with progressive decrease on the performance across the stages (Table 7) in bold.

Overall, small to large effect sizes were observed between stage 2 (subjective cognitive and/or behavioral impairment) and 5 (moderate Alzheimer's disease) for the CAMCOG-DS total score and orientation subtest (BT-ID).

The effect sizes ranged from small to medium between stages 1 (cognitive and behavioral stability) and 3 (mild cognitive and/or behavioral impairment) for the BRI index (BRIEF-P). Medium to large effect sizes were observed for the formal fluency subtest (BT-ID) between stage 2 (subjective cognitive and/or behavioral impairment) and stage 4 (mild Alzheimer's disease).

The effect sizes ranged from small to medium for free delay memory and semantic fluency subtest (BT-ID) between stage 4 (mild Alzheimer's disease) and 5 (moderate Alzheimer's disease).

A small effect size was observed between stage 3 (mild cognitive and/or behavioral impairment) and 4 (mild Alzheimer's disease) on the *visual discrimination* subtest (BT-ID).

#### *3.3. Reliability*

The inter-rater reliability for staging using GDS-DS was excellent, with a mean Cohen weighted κ value of 0.86 (CI: 0.80–0.93). The agreement between the two raters classifying the PD group using the GDS-DS was excellent, as determined by specialists, with κ values of 0.82 (CI: 0.73–0.92) and 0.85 (CI: 0.77–0.94).

#### **4. Discussion**

The main goal of the present study was to devise a global rating scale for AD in people with DS based on the GDS scale [39]. The resulting GDS-DS amplifies the classical phases of the AD-DS (stability, prodromal AD and AD). Our purpose was to provide a scale that captures progressive decline in the entire AD-DS continuum. The GDS-DS meets this requirement from the cognitive stability to AD stages, especially in relation to performance on the CAMCOG-DS *total score* and *orientation* subtests of the BT-ID and dementia.

The GDS-DS ranges through six stages from cognitive and behavioral stability (stage 1) to advanced Alzheimer's disease (stage 6), contrasting with the seven stages of the global deterioration scale (GDS) [39] for the general population. In view of the results obtained in the present work, in which an overall decrease in deterioration is detected by the CAMCOG-DS total score, six stages capture the entire continuum of AD in people with DS and, in addition, they are consistent with the stages of AD proposed in research of the National Institute on Aging and the Alzheimer's Association (NIA-AA) [6].

The six stages of the GDS-DS differ from the five stages of the clinical dementia rating scale (CDR) [35] and the modified clinical dementia rating scale for people with Down syndrome questionnaire (CDR-QDS) and interview (CDR-IDS) forms [38]. These two rating scales (CDR and CDR-IDS/CQR-IDS), as opposed to our GDS-DS, do not include stage 2 (subjective cognitive and/or behavioral impairment). This can lead to misdiagnosis because subjective cognitive impairment is increasingly seen as an early symptom of dementia risk in the general population [16,17] and although it is not a sensitive aspect by itself [18], it must be considered with complementary clinical data as a different entity from MCI. In addition, behavioral aspects are also not considered in these scales, although they are important factors for detecting possible changes associated with prodromal AD in subjects with DS [47]. Furthermore, in our clinical dementia contexts, the global deterioration scale (GDS) [39] is used to monitor, design and modify appropriate pharmacological treatment in patients with AD in the general population [43,44]. Thus, the authors consider the GDS-DS a closer instrument for daily clinical practice and recommend its use for clinical trials in subjects with AD-DS.

Seven participants were placed in the stage 6 (advanced Alzheimer's disease) category of the GDS-DS in our study. The criterion to be placed in this stage was that the subject did not endure the neuropsychological examination proposed in the protocol and this was, in turn, an exclusion criterion for the PD group on which the present work is based. Some DS adults in the symptomatic stages of AD are not able to complete a neuropsychological examination [22]. The authors of the present study agree, as routine follow-up visits have encountered this setback. As a result, despite progress in terms of instruments adapted for people with ID, the neuropsychological tests used in the aforementioned studies and in daily clinical practice are not sufficiently valid to capture the deterioration in advanced stages of AD-DS nor for severe/profound intellectual disability [57,64,65]. Therefore, in terms of daily clinical practice, for people placed in stage 6 in our study, the authors recommend the use of adapted tools, such as the cognitive exploration scale for people with intellectual disability and extended support (ECDI-SE) or the modified ordinal scales of psychological development for people with intellectual disability (M-OSPD-ID). However, at the same time, this also means that using these instruments in the early stages of AD can lead to a ceiling effect on the scores. Future studies should focus on valid instruments for all phases of AD, avoiding ceiling and floor effects, regardless of the severity of symptoms.

Performance in the orientation subtest of BT-ID decreased significantly throughout the GDS-DS stages. Although not pathognomonic of AD, temporal disorientation is frequently observed in daily clinical practice in subjects with cognitive impairment. This requires semantic and episodic information activation [66] and has been linked to atrophy in the posterior hippocampus [67] and the disconnection between the posterior part of the right medial temporal gyrus and the posterior cingulate cortex [68]. The progressive loss of orientation has already been described in the transition from MCI to AD in people with DS [26]. In view of these facts, this subtest should be present in longitudinal follow-ups in individuals with ID.

Performance on executive functions (abstract thinking, free delay memory (stories), semantic and formal fluency) shows a slight decline with modest sustained effect sizes in some stages of the AD continuum. The anatomical substrate of these functions is linked with the temporoparietal, precuneus-posterior cingulate and occipital areas [24], which show a decreased volume [41,60] and loss of integrity in white matter tracts [58,69] in individuals with AD-DS. In addition, an increase in alpha band synchronization using magnetoencephalography has been found in the functional connectivity in the AD continuum in the general and in the DS population and is associated with cognitive decline in executive function, language and working memory [62]. Additionally, MCI-DS individuals with confirmed amyloid positivity who progress to AD have shown a pattern of increased delta activity in frontal regions [70]. The clinical relevance of these findings suggests that all these cognitive functions also should be examined in longitudinal clinical follow-ups

Declines in the CAMCOG-DS total score have been shown to be related to changes in all of the stages of the GDS-DS. A decline in performance on CAMCOG-DS has been found in the entire AD-DS continuum [13,26,28,62], regardless of whether the level of ID is mild or moderate [22]. Interestingly, performance on CAMCOG-DS is linked to amyloid deposition [71], and it is recommended that longitudinal studies assess cognitive changes related to ID and dementia [72]. Therefore, CAMCOG-DS could be a suitable instrument for anchoring the GDS-DS stages of our context, in a similar way that the mini examen cognoscitivo (MEC) [73] is anchored to the global deterioration scale [39] in the general population.

The *BRI* index of BRIEF-P [30] does not change significantly across the GDS-DS stages, but it drops slightly from the cognitive/behavioral stability to mild cognitive and/or behavioral impairment, stages 1 to 3, respectively. The authors selected the *BRI* index because worse scores on this index could differentiate between healthy and MCI-DS subjects [28], and our results partially replicate this. In light of these results, the inclusion of behavioral impairment in the GDS-DS seems appropriate. In addition, in the general population, mild behavioral impairment [19] improves the specificity of MCI as an atrisk state for incident dementia [74] and has been associated with higher AD polygenic risk scores [75]. Considering that behavioral changes are able to herald AD in the DS population [47], the authors recommend the use of the *BRI* index scores to detect behavioral impairment in the DS population in the earlier stages.

The main strength of the present study is that the GDS-DS, with the inclusion of behavioral aspects, allows the capture of subjects in the initial stages of AD-DS either due to memory or behavioral difficulties, in addition to being associated with performance in cognitive and behavioral tests that have been shown to be reliable in the DS population.

Additionally, the GDS-DS could be the first step in unifying the criteria for classifying the continuum of Alzheimer's disease in people with DS. It allows a common language for communication between clinicians and researchers and could be a basic instrument for the selection of samples in clinical trials in people with DS.

Furthermore, the GDS-DS, can also be used to inform families about the stages of Alzheimer's disease, when sufficient data from future studies have been collected to construct specific clinical profiles in combination with functional assessment staging. Knowing the continuum can complement the neurological diagnosis and facilitate family members' understanding of the expected prognosis.

However, some limitations have to be acknowledged. In this study, we used crosssectional information that relied on the longitudinal data of a three-year follow-up study. We first wanted to verify that the GDS-DS was applicable to people with DS. As this has been proven to be feasible, future studies could investigate the application of the scale in longitudinal follow-ups to minimize possible cohort effects.

Additionally, our study was based on data collected from neuropsychological tests, behavioral questionnaires and informant interviews. Today, neuroimaging or neurophysiological techniques provide rich complementary data that correlate brain changes with cognitive features in aging and AD. Thus, future studies must include the linking of neuroimaging and neurophysiological data with the stages of GDS-DS.

#### **5. Conclusions**

In summary, GDS-DS is not designed as a diagnostic tool but as a quantitative measure of disability. The insights delivered in this paper show that the proposed GDS-DS rating scale represents an important attempt at staging people with DS throughout the continuum of AD, providing a unique method of classification that can be useful in clinical trials and in daily clinical practice. As noted in the development of the original GDS [39], the limits of each of the GDS-DS stages are not axiomatic, but they do allow graduation as a guideline that facilitates the monitoring of the AD-DS continuum.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph20065096/s1, Table S1: Effect sizes between groups of the GDS-DS stages.

**Author Contributions:** Conceptualization, E.R.-H. and S.E.-C.; methodology, E.R.-H. and S.E.-C.; software, E.R.-H.; validation, S.E.-C., R.N. and J.G.-A.; formal analysis, E.R.-H.; investigation, E.R.-H. and S.E.-C.; data curation, E.R.-H.; writing—original draft preparation, E.R.-H.; writing—review and editing, S.E.-C., R.N. and J.G.-A.; supervision, S.E.-C. and R.N.; project administration, S.E.-C., J.G.-A. and E.R.-H.; funding acquisition, S.E.-C., R.N. and J.G.-A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Spanish Government, grant number PI12/02019, PSI-2014- 53524-P. The APC was funded by S.E.-C,'s SESMDI research start-up funds.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board and Research Ethics Committee of the Parc Hospitalari Martí i Julià (Approval Code: S041-775; Approval Date: 4 July 2012).

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

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank the participants and their families enrolled in this study.

**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.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

## *Article* **The Pictorial Screening Memory Test (P-MIS) for Adults with Moderate Intellectual Disability and Alzheimer's Disease**

**Emili Rodríguez-Hidalgo 1, Javier García-Alba 2, Maria Buxó 3, Ramon Novell 1,3 and Susana Esteba-Castillo 1,3,\***


Assistance (IAS), Parc Hospitalari Martí i Julià, Catalonia, 17190 Girona, Spain

**\*** Correspondence: susanna.esteba@ias.cat; Tel.: +34-972-18-26-00

**Abstract:** In this study, we examined normative data and diagnostic accuracy of a pictorial screening test to detect memory impairment for mild cognitive impairment (MCI) and Alzheimer's disease (AD) in Spanish-speaking adults with intellectual disability (ID). A total of 94 volunteers with ID (60 controls, 17 MCI, and 17 AD), were evaluated by neuropsychological tests including the PMIS-ID in a cross-sectional validation study. Discriminative validity between the MCI, AD, and control group was analyzed by the area under the ROC curve. A cut-off score of 4.5 on the immediate recall trial had a sensitivity of 69% and a specificity of 80% to detect memory impairment (AUC = 0.685; 95% CI = 0.506–0.863) in the AD group. The PMIS-ID is a useful screening test to rule out a diagnosis of memory decline in people with moderate level of ID and AD, and it shows good psychometric properties.

**Keywords:** screening; memory; mild cognitive impairment; Alzheimer's disease; dementia intellectual disability

#### **1. Introduction**

Intellectual disability (ID) is a neurodevelopment disorder affecting intellectual functioning and adaptive behavior [1]. Globally, the prevalence of ID varies between 0.05 and 1.55% [2]. The underlying health conditions and the increased life expectancy of this population makes them more vulnerable to developing mild cognitive impairment (MCI) and dementia [3,4]. Although the epidemiological data of MCI and dementia in people with ID without Down's syndrome (DS) has not been accurately set, recently, in Japanese people, the prevalence data for MCI hovers around 3% from the age of 45 onwards, and from 0.8% to 13.9% in those aged between 45 and 74 years old for dementia [5].

There is increasing evidence of a link between neurodevelopment disorders as ID and dementia [6]. Specifically, Alzheimer disease (AD) is the most common syndrome of dementia in DS [7], with incidence rates from 75% to 100% in those aged 60 and older [8]. Beyond DS, people with autism spectrum disorder (ASD) with ID are also at high risk for developing early onset AD (EOAD) [9], and ASD behaviors may be present in geriatric people with MCI and AD without ID [10]. Cerebral palsy (CP) by itself increases the risk of EOAD and related dementias [11], having an accelerated aging that predisposes patients to MCI and AD [12]. Based on these facts, certain ID conditions increase the risk of suffering AD. Therefore, early detection of MCI and AD in people with ID is required to implement appropriate interventions on the optimal therapeutic window.

It is well-known that identifying MCI or AD in people with ID poses a challenge to clinicians. Strengths and weaknesses of the cognitive and behavioral phenotype of

153

**Citation:** Rodríguez-Hidalgo, E.; García-Alba, J.; Buxó, M.; Novell, R.; Esteba-Castillo, S. The Pictorial Screening Memory Test (P-MIS) for Adults with Moderate Intellectual Disability and Alzheimer's Disease. *Int. J. Environ. Res. Public Health* **2022**, *19*, 10780. https://doi.org/10.3390/ ijerph191710780

Academic Editor: Paul B. Tchounwou

Received: 13 July 2022 Accepted: 20 August 2022 Published: 30 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

each etiology must be taken into account when considering changes that may herald MCI or dementia in aging [13]. Therefore, some studies try to cognitively characterize the stages of MCI and AD in ID population. In people with mild or moderate levels of ID and unspecified etiology, decline in orientation questions and depressive symptoms [14] or decline in memory measured by a paired-learning task [15] have been found. By the Dementia Questionnaire for Mentally Retarded People [16], memory and orientation were altered [17]. Furthermore, cognitive decline is similar between people with ID and the general population with AD [18], exhibiting difficulties in learning and visuoverbal memory, semantic verbal fluency, and attention/executive functions, measured with the Fuld Object-Memory Evaluation [19], the Controlled Oral Word Association Test [20], and the Color Trail Test [21], respectively. In people with moderate and severe levels of ID, significantly lower scores were found in autobiographical memory and orientation in people with AD compared to healthy people [22]. In this same work, performance on a modified Objective Memory Test was lower in the group with AD, but only in the immediate recognition trial and in both immediate and delayed memory subtests of a Picture Recognition Test, a visuoverbal learning and memory test developed for the purpose of their work, that has proven able to detect memory troubles in people with moderate and severe levels of ID and DS [23]. Despite the different etiologies studied, memory decline arises as a common factor associated with MCI or AD.

As AD is the most common form of dementia and it is especially linked with DS, the majority of studies focused on this population have used cognitive measures. In this sense, a recent study of functional brain connectivity in people with DS [24] showed that low scores on the Cambridge Cognitive Examination for Older Adults with Down's Syndrome (CAMCOG-DS) Spanish version [25] and decreased performance in verbal and visual memory appear to be key indicators of MCI and AD. Also, a slight impairment in delayed verbal and visual memory could be considered as a potential cognitive marker of MCI, with an increase in memory deficit in AD stage [26,27]. However, more important is the proposal model for MCI diagnosis in which decline or change in the Behaviour Regulaltion Index (BRI) from the Behaviour Rating Inventory of Executive Function-Informant's Report (BRIEF) [28], and in the delayed verbal and visual memory domains from the Barcelona Test for Intellectual Disability (TB-DI) [29] are good predictors for MCI diagnosis [30]. Upon collecting these data, it is evident that memory declines across the AD continuum. Hence, there is a need to develop sensitive and screening test for daily clinical practice to detect memory changes associated with MCI or AD in this population.

Memory tests including a delayed recall trial are useful to detect longitudinal changes [31], as are recognition trials that improve the sensibility to detect retrieval alterations [32]. The word-based Free and Cued Selective Reminding Test (FCSRT) [33] for the general population and the pictorial Cued Recall Test (CRT) [34] for people with DS are tests with controlled learning techniques to optimize the coding processes [35] that are specially altered in the early stage of AD. It is well-known that scores of word and pictorial versions are not equivalent [36,37], although it is suggested that poor results in the free recall essay are associated with a reduction of the hippocampal volume in both tests [38]. As of today, a few screening tests are available that yield robust data with high accuracy in detecting memory deficits in the general population. For instance, the Memory Impairment Screen (MIS) [39] is a brief tool of four items that controls learning and cued recall, providing excellent properties to detect individuals at higher risk to develop AD. Different MIS-based versions are currently available, such as the Picture-Based Memory Impairment Screen (PMIS) [40], with four pictorial stimuli for English people with low levels of education. A Spanish pictorial memory test is also available that reports preliminary results for the general population with amnesic MCI and AD [41], as well as for people with DS [42]. When compared to the general population, there is no picture memory screening test available for Spanish people with ID. Furthermore, the administration of the available tests is hard to acquire for people with ID and time is short for routine appointments. Clearly, there is a need to develop valid memory screening tests suitable for primary care and specialized services in this population.

In light of these facts, the main aims of the current study were (a) to adapt the PMIS [40] for people with ID, (b) to provide normative data, and (c) to assess its feasibility as a screening tool for memory decline to discriminate healthy ID subjects from those with amnestic impairment typical of MCI or AD.

#### **2. Materials and Methods**

#### *2.1. Participants*

We undertook a prospective single-center cross-validation study with a convenience sample from December 2017 to December 2018. A total of 116 subjects recruited from the Specialized Service in Mental Health Unit for Adults with ID (SESM-DI, Parc Hospitalari Martí i Julià, Girona, Spain) were identified. All the participants were 18 years old or older, with a mild or moderate level of ID according to DSM-5 criteria (5th ed.; DSM-5) and with no drug treatment that could had significant effects on cognition. Those showing psychiatric or neurological conditions that could cause a dementia-like presentation or cognitive decline (depression, clinical hypo/hyperthyroidism, uncontrolled B9/B12 vitamin deficiency, seizures, delirium) and uncorrected auditory or visual sensory impairment that would make a neuropsychological assessment impossible were excluded. Of the subjects selected, 12 did not agree to participate in the study, 4 were excluded due to an absence of expressive language, and 6 more were excluded due to non-stable medical conditions at the moment of the assessment. The final sample consisted of 94 adults with ID (mean age = 47.33 ± 5.18 years; males = 57.4%, females = 42.6%) due to different etiology (54 Down's syndrome, 4 tuberous sclerosis, 4 fragile X syndrome, 2 cerebral palsy, and 30 unknown). The baseline level of at least one year of all participants was available through annual follow-up in our service, in which most of the cognitive and functional tests that are part of the present study are routinely administered. The sample was divided into three groups: CN group (subjects without symptoms of MCI or AD), MCI group (subjects fulfilling criteria for MCI diagnosis), and AD group (subjects with AD diagnosis). The diagnosis of MCI or AD was based on expert multidisciplinary clinical judgment according to recent publications [27,30,43–46]. The diagnosis of MCI was made when participants presented a single or multiple cognitive decline(s) without significant functional loss. On the other hand, the diagnosis of AD was made in participants with memory decline and another cognitive impairment as aphasia, apraxia, agnosia, or disexecutive syndrome, and loss of functionality. In both conditions, changes from previous level of performance had to be supported by information obtained from a close caregiver [27,43,45].

The CN group included 60 subjects (47.47 ± 5.78 years; males = 65%, females = 25%) and was used to obtain normative data of the PMIS-ID. Both the MCI group, composed of 17 subjects (45.76 ± 4.22 years; male = 47.05%, females = 52.95%), and the AD group, composed of 17 subjects (48.41 ± 3.34 years; males = 41.18%, females = 58.52%), were used to gain the diagnostic accuracy properties of the PMIS-ID.

#### *2.2. Instuments*

Each participant underwent a comprehensive clinical and neuropsychological assessment. A neuropsychologist administered a large cognitive evaluation and informed-based measures during three different sessions.

#### 2.2.1. To Detect ID Level


2.2.2. Neuropsychological Assessment with Cognitive Tools Adapted and Validated for ID Spanish-Speaking Population


#### 2.2.3. Parents Interview

• Cambridge Examination for Mental Disorders of Older People with Down's syndrome and Others with Intellectual Disabilities (CAMDEX-DS) Spanish version [25]. It consists of a structured informant-based interview, cognitive evaluation, diagnostic criteria guide, and recommendations for the interventions. The Spanish version presents an internal consistency of α = 0.93, considering performance on the memory subtest of the cognitive form and the memory section of the informant interview.

2.2.4. Picture Memory Impairment Screen for People with Intellectual Disability (PMIS-ID)

The PMIS-ID consists of four-color photographs semantically unrelated in each quadrant of a DIN-A4 sheet. It includes four distinct parts: Identification (I), Learning (L), Immediate Recall (IR), and Delayed Recall (DR).

In the I and L parts, a sheet with four different categories of photographs (*horse*, *ludo*, *sofa*, *cherry*) is presented to the subject who has to name each one. Afterwards the subject has to identify them according to a cue (category) provided by the examiner (*animal*, *board game*, *fruit*, *furniture*). If the subject does not recognize or identify a photograph, the administration is ruled out. If items are correctly identified, the sheet is removed and the subject is told that they will be asked to repeat the words in a short time. Exploration must go on with another non visuoverbal task.

After three minutes, in the IR part, the participant is asked to recall the name of the four photographs (Immediate Free Recall, IFR). If any of the four items is missed, the examiner provides a category cue [Immediate Cued Recall (ICR)]. In case of failure, the target stimulus and two more distracters of an equal semantic category are orally provided (*cherry*, *pear*, *kiwi*; *goose game*, *cards*, *ludo*; *table*, *sofa*, *chair*; *horse*, *cow*, *tiger*) and the subject has to detect the right stimulus [Immediate Recognition (IRC)]. Correct stimulus has to be provided again if failure persisted. Finally, after twenty minutes, the DR part is administered, with the same tasks as in the IR part: Delayed Free Recall (DFR), Delayed Cued Recall (DCR) and a Delayed Recognition Recall (DRC).

The scores are calculated as two points for each correct response in FR, one point for each one in CR task, and 0.5 point in the RC. Immediate Total Recall (ITR) and Delayed Total Recall (DTR) are calculated separately (FR scores + CR scores + RC scores) of the FR, the CR and the RC parts. Total PMIS-ID (TPMIS-ID) score (0–16) is the sum of ITR and DTR, both ranging from 0 to 8.

#### • PMIS-ID adaptation

The PMIS test [40] was adapted by: (1) introducing different items and categories suitable for people with mild and moderate ID; (2) translating the instructions with easiest vocabulary; (3) introducing Delayed Recall (DR) and Recognition (RC) tasks both for the Immediate Recall (IR) and Delayed Recall (DR) trials; and (4) by implementing a new scoring system. The numbers of items were consistent with the original version. An iterative procedure in line with practices recommended by Muñiz, Elosua, and Hambleton (2013) was followed, considering the particularity of this memory visuoverbal test. A pool of 12 color photographs belonging to four different semantic categories according to Spanish typicality norms were extracted [49,50]. To provide the sufficient complexity and avoid the ceiling effect, six stimuli corresponded to the first third while the other six to the second third of the total responses by category. The photographs were shown to 30 volunteers with mild and moderate ID (men age 43.8 ± 3.55; males = 55%; females = 45%) who were not enrolled in the validation study. Then, stimuli were reduced to the four most recognized and the remaining ones were introduced as distracters for the recognition task. Two native-English specialized psychologists in ID translated the test instructions. The two versions were discussed by the research team. An independent English linguist completed the back-translation of the document. Finally, a neurologist and a speech therapist reviewed the process and agreed to a pre-final version. The pre-final version was rounded off to implement further modifications during a pilot test in the same sample for the stimuli selection phase. During test administration, the examiner controlled the execution time and also checked the comprehension of the instructions, asking the volunteer to repeat and to explain them with their own words. The PMIS-ID was applied again in a convenience subsample of 20 participants within four weeks to assess test–retest reliability and in another 20 participants by two different examiners (SEC, ERH) to assess inter-rater reliability.

#### *2.3. Data Analysis*

The working database includes entries from February to December 2018. A descriptive analysis was applied to the entire group sample to describe the demographic variables (age and sex), the ID level, and performance on the cognitive protocol (TB-DI memory and orientation, CAMCOG-DS memory subtest). Due to the small sample size in some subgroups, a nonparametric statistical analysis was conducted. Means comparisons were made by independent samples t-test or ANOVA for qualitative data, and χ2-test for category data. These data were presented as median and interquartile range (IQR) and were compared by Kruskal–Wallis test with Bonferroni adjustment. Multiple linear regression analysis was used to verify the possible influence of sociodemographic variables (age, sex) and the level of ID (mild, moderate) on the PMIS-ID immediate total score, delayed total score, and total score. Reliability was estimated by the test–retest and inter-rater methods and by calculating Pearson and intraclass correlation coefficients, respectively. For the MCI and AD groups, a descriptive analysis was applied for the sociodemographic variables (age, sex, and ID level). Construct validity of the PMIS-ID was verified using the coefficient corrected kappa statistic between the PMIS-ID total score and MCI and AD groups. Spearman's rho was run to evaluate convergent validity between the PMIS-ID total score and the memory subtest performance of the TB-DI and the CAMCOG-DS. Normative data was presented in line of the assumption of the MCI or AD prevalence (%) for the different PMIS-ID cut-scores (PPV and NPV). Receiver operating characteristic curve (ROC) and the Youden index were used to determine the optimal cut-off point of the PMIS-ID (immediate, delayed, and total scores) as a screening memory test for MCI or AD. The areas under the curve (AUC) were compared between the different trials [51].

All statistical analyses were conducted using the software program G-Stat (version 2.0) and the statistical software program SPSS (version 27.0; SPSS Inc., Chicago, IL, USA). Bilateral significance levels were set at a *p*-value of less than 0.05.

#### **3. Results**

#### *3.1. Demographics*

Results in Table 1 shows that the CN group was similar to the MCI and AD groups in mean age (*p* = 0.315), level of ID (*p* = 0.159), and sex distribution (*p* = 0.136).


**Table 1.** Demographic characteristics and cognitive performance by group.

Values are given by means and range; sex in percentages for each group. ID, intellectual disability; TB-DI, Barcelona Test for People with Intellectual Disability; CAMCOG-DS, Cambridge Examination for Mental Disorders of Older People with Down's Syndrome and Others with Intellectual Disability (brief neuropsychological battery); MCI, mild cognitive impairment; AD, Alzheimer's disease. \* *p* < 0.05, \*\*\* *p* < 0.001, using Kruskal–Wallis with Bonferroni correction or χ2-test for category data between MCI and AD groups compared with the control group.

#### *3.2. Between-Group Comparison of Cognitive Performance*

Compared to the CN group, the scores were significantly lower for the AD group on verbal learning, delayed free recall subtest, and false positive scores of the TB-DI and on new learning, remote, recent, and memory total score subtest of the CAMCOG-DS. Performance on the subscales of both tests was similar between the MCI and CN groups (Table 1).

In people with mild ID, the performance of the CN and MCI groups was similar on the entire subtest. Between the CN group and AD group, significant differences were shown in performance on verbal leaning, delayed free recall, and false positives of the TB-DI. No significant differences were observed on the subtests of the CAMCOG-DS (Table 2).


**Table 2.** Cognitive scores by group for mild intellectual disability sample.

Values are given by means and range. TB-DI, Barcelona Test for People with Intellectual Disability; CAMCOG-DS, Cambridge Examination for Mental Disorders of Older People with Down's Syndrome and Others with Intellectual Disability (brief neuropsychological battery); CN, control group; MCI, mild cognitive impairment; AD, Alzheimer's disease. \* *p* < 0.05, \*\* *p* < 0.01, using Kruskal–Wallis with Bonferroni correction between MCI and AD groups compared with the control group.

In people with moderate ID, no significant differences were observed between the CN group and the MCI group on all the subtests. Significant differences were observed on the false positives score, and no significant differences but moderate decline were observed on verbal learning and delayed free recall subtests of the TB-DI. Significant differences were observed on new learning, remote, recent, and memory total score of the CAMCOG-DS (Table 3).


**Table 3.** Cognitive scores by group for moderate intellectual disability sample.

Values are given by means and range. TB-DI, Barcelona Test for People with Intellectual Disability; CAMCOG-DS, Cambridge Examination for Mental Disorders of Older People with Down's Syndrome and Others with Intellectual Disability (brief neuropsychological battery); CN, control group; MCI, mild cognitive impairment; AD, Alzheimer's disease. \*\* *p* < 0.01, \*\*\* *p* < 0.001, using Kruskal–Wallis with Bonferroni correction between MCI and AD groups compared with the control group.

#### *3.3. Between-Group Comparison of PMIS-ID Performance*

The AD group performance differs from the CN group on free recall and total scores from both immediate and delayed part, as well as on the total PMIS-ID score, whereas the performance of the MCI and CN groups was similar on all the trials (Table 4). Analyses according the ID level show that these differences were valid for adults with a moderate level of ID, but not for those with mild ID (Table 5).

**Table 4.** PMIS-ID descriptive scores by group.


Median (first quartile–third quartile) for each variable is summarized. PMIS-ID, Picture Memory Impairment Screen for People with Intellectual Disability; CN, control group; MCI, mild cognitive impairment; AD, Alzheimer's disease; <sup>a</sup> Between CN and AD group. <sup>b</sup> Between MCI and AD group. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 using Kruskal–Wallis with Bonferroni correction between MCI and AD groups compared with the control group.


**Table 5.** PMIS-ID descriptive scores by group and level of intellectual disability.

Median (first quartile–third quartile) for each variable is summarized. PMIS-ID, Picture Memory Impairment Screen for People with Intellectual Disability; ID, intellectual disability; CN, control group; MCI, mild cognitive impairment; AD, Alzheimer's disease. <sup>a</sup> Between CN and AD group. <sup>b</sup> Between MCI and AD group. \* *p* < 0.05, \*\* *p* < 0.01, using Kruskal–Wallis with Bonferroni correction between MCI and AD groups compared with the control group.

#### *3.4. Analysis of Control Group*

To study the demographics and level of ID impact on the PMIS-ID scores, a multiple linear regression analysis for the PMIS-ID immediate total score (*r2adj* = 0.032; *F* = 1.006; *p* = 0.397), delayed total score (*r2adj* = 0.515; *F* = 1.102; *p* = 0.356), and total score (*r2adj* = 1.357; *F* = 1.271; *p* = 0.293) was performed for the CN group. Neither the level of ID (immediate: *t* = −1.523, *p* = 0.133, *B* = −0.615; delayed: *t* = −1.419, *p* = 0.161, *B* = −0.895; total: *t* = −1.509, *p* = 0.069, *B* = −1.509) nor the age (immediate: *t* = −1.078, *p* = 0.286, *B* = 0.0.38; delayed: *t* = −0.689, *p* = 0.494, *B* = −0.037; total: *t* = −0.001, *p* = 0.999, *B* = −0.001) nor sex (immediate: *t* = −0.602, *p* = 0.549, *B* = −0.615; delayed: *t* = −0.350, *p* = 0.7274, *B* = −0.222; total: *t* = −0.568, *p* = 0.572, *B* = −0.466) were significantly related to the PMIS-ID total scores. Hence, no adjustment of the PMIS-ID total scores was required.

#### *3.5. Reliability*

The Pearson's correlation coefficient for the total PMIS-ID score was 0.90 (*p* < 0.0001), a very strong test–retest association. The inter-examiner agreement was found to be good, with a mean intra-class correlation coefficient (ICC) of 0.96 (*p* < 0.0001).

#### *3.6. Validity*

#### 3.6.1. Convergent Validity

The correlations patterns between the PMIS-ID total score for the whole sample and by level of ID and the TB-DI and CAMCOG-DS subtest are displayed in Table 6.

Considering the total sample, the overall correlations were significant. For the mild ID sample, the total score of the PMIS-ID was low but positive in correlation with the verbal learning and delayed free recall subtest of the TB-DI and with the new learning ant total memory score of the CAMCOG-DS. In the moderate sample, satisfactory positive correlations were found between verbal learning and delayed free recall subtest of the TB-DI, and new learning, recent, and total memory score subtest of the CAMCOG-DS. Low negative correlation was found between false positives score of the TB-DI and total PMIS-ID total score. Weak positive and negative correlations were found for the other subtest of both sample and the PMIS-ID total score.


**Table 6.** Convergent validity values by groups.

TB-DI, Barcelona Test for People with Intellectual Disability; CAMCOG-DS, Cambridge Cognitive Examination Adapted for Individuals with Down Syndrome; PMIS-ID, Picture Memory Impairment Screen for People with Intellectual Disability; ID, intellectual disability. \* *p* < 0.05, \*\* *p* < 0.01 (two-tailed test), using Spearman's correlation coefficient.

#### 3.6.2. Discriminative Validity

Three ROC curves for each total PMIS-ID trial with enough discriminatory power between CN and AD groups with moderate ID were plotted (Figure 1) and comparisons between AUCs were calculated (Table S1 of the Online Supplement). The AUCs to discriminate between CN and AD groups had good diagnostic utility. For immediate total recall (ITR), the AUCs were 0.685 (CI 95%: 0.506–0.863) and 0.757 (CI 95%: 0.612–0.903) for the total delayed recall (TDR) and 0.749 (CI 95%: 0.597–0.901) for the TPMIS-ID. Paired comparison of the AUCs of the three totals scores were not statistically significant (*z* = −0.842, *z* = −1.021, *z* = 0.229). Hence, immediate total recall (ITR) was the best and most timeefficient trial for our purpose.

**Figure 1.** Receiver operating characteristics (ROC) of the PMIS-ID as a screening memory tool for AD in people with a moderate level of ID.

#### *3.7. Normative Data*

The values of sensitivity, specificity, positive likelihood ratio (+ LR), negative likelihood ratio (−LR), and the Youden indexes for various cuts-off scores of the PMIS-ID for the AD group with mild and moderate level of ID were calculated. For the group with moderate level of ID, the delayed and total recall trials normative data are presented in Tables S2 and S3 of the Online Supplement, and for the immediate recall (IR) in Table 7. Data for the sample with mild ID and AD is presented in Tables S4–S6 of the Online Supplement. Considering that the base rate in the moderate ID sample was 25.5%, for the immediate total recall (ITR), a cut-off score of ≤4.5 provided a moderate sensitivity (69%) and a high level of specificity (80%).

**Table 7.** Normative data of the PMIS-ID immediate total score for different cut-off points for AD in moderate level of ID sample.


MIS-ID, Picture Memory Impairment Screen for People with Intellectual Disability; S, sensitivity; Sp, specificity; *J*, Youden's J statistic; PPV, positive predictive values; NPV, negative predictive values. <sup>a</sup> 25.5% is the prevalence of AD in the sample.

#### **4. Discussion**

This is the first validation study of a pictorial screening memory test for people with ID, providing normative data and assessing its diagnostic utility to detect memory decline. The PMIS-ID demonstrates good discriminant validity for distinguishing between people with moderate ID and AD from healthy population with ID, and exhibits good convergence validity and reliability.

Our results show that it is possible to detect memory impairment with the PMIS-ID in people with moderate ID in a quick and simple way. It is easy to administer, brief (no more than five minutes), and cost-effective. Also, it shows good convergent validity with memory subtest of the TB-DI and CAMCOG-DS. Considering that the verbal memory subtest of the TB-DI presents good internal consistency [29] and the CAMCOG-DS is recommended for follow-up studies [52], this proves that the PMIS-ID measures memory processes. Furthermore, AUC is acceptable (above 0.7) and the discriminant validity for the proposed cut-off score of the PMIS-ID (4.5) shows good specificity (86%) and appropriate sensitivity (69%). Values of specificity are in line with the majority of the MIS-based screening tests in the general population that yield specificities values higher than 80%: the MIS [39], MIS-S [53], MIS-E [54], MIS-D [55], and PMIS [40]. The PMIS-ID also identifies healthy subjects correctly as the aforementioned screening tests. Consequently, the PMIS-ID is an excellent screening memory test for use in daily clinical specialized services for people with moderate ID.

Data gained from the immediate total recall (ITR) part of the PMIS-ID in isolation well enable detection of memory impairment in adults with moderate ID and AD. The inclusion of a delayed recall is tautological, opposite to the improvement described in the general population with MCI [53], but closer to the use of pictorial memory test based in learning trials without delayed recall in people with low educational level [41] and in people with ID [42]. Our results reveal that when performance is scarce in the immediate recall, it is also in the delayed recall. Moreover, in our sample, people with lower performance in free recall do not gain with the inclusion of a cued recall in the immediate or in the delayed recall. These results are aligned with those described for the MIS-S [54] and cued recall test [42]. Also, learning of relational material depends on the functionality of the hippocampus [55] and free recall trials of the FCSRT picture versions could be considered as an indicator of hippocampus structural integrity [38]. Furthermore, longitudinal memory score variation is specifically associated with volume change in the hippocampus [45,56]. For these reasons, it can be concluded that the PMIS-ID is a noninvasive tool to detect memory impairment due to hippocampus dysfunction in people with moderate ID and AD.

In our study, significant memory decline is the predominant symptom in people with moderate level of ID and AD. Memory decline has been described in MCI and AD in the general population and those with ID [57]. Especially, in people with DS, a slight memory impairment is usually found in MCI, with a significant decline in AD [24,26,27,30,58]. Possibly, memory processes in people with moderate level of ID are not semantic-dependent, and other strategies should be considered because deep processing is beyond their capability [59]. Other promising alternatives could involve developing a test under the associative learning paradigm (binding) that depends on the integrity of the medial temporal lobe structures, as in the general population with hopeful results for AD [60], MCI [41], and in adults with mild and moderate ID [61].

Although stimuli selection in our work was accurate, the scores on the PMIS-ID have a low ceiling effect. Despite the apparent easiness of the four proposed pictorial stimuli, the PMIS-ID measures the same function as a standard memory test. High concordance was obtained between the PMIS-ID total score and memory subtests of the CAMCOG-DS [25] and the TB-DI [29], especially for people with a moderate level of ID. For instance, the satisfactory concordance in our study for the verbal learning subtest of the TB-DI (*r* = 0.62, *p <* 0.001) has also been described between the MIS-E [62] and the analogue subtest of the original version of the TB-DI, the Barcelona Test [63] (*r* = 0.78; *p <* 0.001) in the general population. In this sense, it is important to consider that assessing memory in people with ID is a complex task for clinicians. In clinical practice, pictorial tests seem to be better accepted due to an increased feeling ability to solve the task. In return, the ceiling effect in pictorial tests has already been considered in the general population because the selected drawings are excessively simple and with little ecological validity [64]. This is consistent with the enhancement of performance with pictorial memory test in older adults [37] and in a Spanish population with amnesic MCI [41] and with AD [36] compared with word memory test. Our results confirm the classical dual-coding information theory [65] that postulates superiority in picture processing against words. In this context, increasing the number of stimuli could reduce the ceiling effect, improving its diagnostic capacity for adults with mild ID.

Our results show a slight decrease of the performance on the PMIS-ID for MCI compared with the CN group, both for mild and moderate level of ID groups. Medians are similar in both groups for each trial, but values of the first and third quartile are wider again in each trial, especially for the moderate ID group. This fits with the discrete accuracy to detect MCI compared with the satisfactory data for AD in people with ID. Until recently, few studies had shed light on how to detect preclinical or prodromal stages of AD. Thus, in the recent proposed diagnostic criteria for MCI in DS [30], three variables from a comprehensive neuropsychological examination have proven sensitive enough for this fact: the BRI of the BRIEF and the abstraction and delay memory subtest from the TB-DI. Also, the PAL first-trial memory is one of the most sensitive variables to detecting changes between the preclinical and prodromal phases of AD in people with DS [7]. Furthermore, it is feasible to diagnose AD with neuropsychological tools such as the CAMCOG-DS or the modified Cued Recall Test (mCRT) [34]. In this sense, a decline of performance on both tests

was evident in the continuum of AD, but performance on the mCRT was not discriminative for people with DS in the prodromal stage of AD. Analyzing these results, one can deduce that the most tangible changes occur when AD is established, but not in the prodromal phases. Possibly, the course of AD is different in comparison to the general population, in which MCI can be considered a slowly progressive transitional phase in cases of conversion to AD; sensitive neuropsychological instruments are available to detect these changes. Based on results in subjects with ID, changes between the preclinical and prodromal phase would be minimal and undetectable with current neuropsychological instruments, and are only sensitive when performance declines abruptly. From this data, it can be hypothesized that memory deficits with more or less intensity are part of the phenotype of almost all people with ID, and this implies that falls in memory are more difficult to detect than in the general population, in which the margins of scores are greater. This also might suggest that the course of the disease may not be slowly progressive and early detection is essential to initiate proper intervention. Therefore, neuropsychological tools with normative data for all ID ranges are needed to reduce misdiagnosis and to interpret cognitive profiles better in normal and pathological ageing.

Overall, our clinical experience reveals that in the general population, the diagnostic of MCI or DA poses a challenge in those with high educational level, just as learning disorders can be masked in exceptional children. In this sense, intelligence, education, and occupational level influence the onset and course of deterioration due to the reserve cognitive assumption [66]. This is also evident in people with ID in which the presentation and natural history of AD varies according to the level of ID [46]. Furthermore, declines in CAMCOG-DS scores are more evident in people with moderate ID [56], and adults with Klinefelter syndrome with higher values in intelligence tests performed better in working memory and executive functions [67]. Thus, the impact of the level of intelligence on neuropsychological tool performance seems contrasted. That is why sufficiently reliable neuropsychological tools should be available for different ranges of intellectual capacity and the level of cognitive reserve must be regarded.

There are some limitations that need to be considered when interpreting our findings; some caution is required.

First, even if the sample is acceptable for a preliminary study, participants were classified by level of ID, decreasing the size of the MCI and AD groups and limiting statistical and discriminatory power. Therefore, future research should be carried out with a representative sample, calculating power estimation before the onset of the study.

Second, alternatives forms are desirable to use in clinical practice and research in neuropsychology. In our study, the PMIS-ID stimuli were chosen according to the data in the general population, which could contribute to the ceiling effect of the scores. To avoid this bias and to expand its use for people with mild ID, we recommend carrying out studies to provide norms for word prototipicity and picture familiarity, according to appropriate cultural context, for people with ID.

Third, the predictive capacity of the PMIS-ID cannot be evaluated reliably because it is a cross-sectional study. Also, we have not considered the stage or severity degree of AD. Our objective was to develop a rapid measure of memory decline associated with MCI or AD to be applied in daily primary care, but it is well-known that follow-up of cognitive decline is required in people with ID to confirm a diagnosis of MCI or AD. Therefore, we recommend carrying out longitudinal studies with various time points and adapting current staging scales in the general population for people with ID.

#### **5. Conclusions**

Limitations notwithstanding, the PMIS-ID is a valid memory screening test for people with a moderate level of ID for use in primary health care centers and in clinical specialized services. The findings in the current pilot study suggest that the PMIS-ID does not provide a comprehensive memory assessment but may be useful as a first step in the diagnostic process to help clinicians in healthcare settings to determine the need to carry out a broader diagnostic evaluation.

**Supplementary Materials:** The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ijerph191710780/s1, Table S1: Paired comparisons of the areas under the curve of the PMIS-ID trials diagnostic accuracy; Table S2: Normative data of the PMIS-ID delayed total score for different cut-off points for AD in moderate level of ID sample; Table S3: Normative data of the PMIS-ID total score for different cut-off points for AD in moderate level of ID sample; Table S4: Normative data of the PMIS-ID immediate total score for different cut-off points for AD in mild level of ID sample; Table S5: Normative data of the PMIS-ID delayed total score for different cut-off points for AD in mild level of ID sample; Table S6: Normative data of the PMIS-ID total score for different cut-off points for AD in mild level of ID sample.

**Author Contributions:** Conceptualization, E.R.-H. and S.E.-C.; methodology, E.R.-H. and S.E.-C.; software, E.R.-H. and M.B.; validation, S.E.-C. and J.G.-A.; formal analysis, M.B.; investigation, E.R.-H. and S.E.-C.; data curation, M.B.; writing—original draft preparation, E.R.-H.; writing—review and editing, S.E.-C. and J.G.-A.; supervision, S.E.-C.; project administration, S.E.-C., J.G.-A. and E.R.-H.; funding acquisition, S.E.-C., J.G.-A. and R.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Spanish Government, grant number PI12/02019, PSI-2014- 53524-P. The APC was funded by R.N.'s SESMDI research start-up funds.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of Research Ethics Committee of the Parc Hospitalari Martí i Julià (Approval Code: S041-775; Approval Date: 4 July 2012).

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

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank the Down Syndrome Foundation of Girona (ASTRID-21), Els Garrofers Foundation, and participants and their families enrolled in this study.

**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.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*International Journal of Environmental Research and Public Health* Editorial Office E-mail: ijerph@mdpi.com www.mdpi.com/journal/ijerph

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel: +41 61 683 77 34

www.mdpi.com ISBN 978-3-0365-7415-8