Exploring chronic disease prevalence in people with intellectual disabilities in primary care settings: A scoping review

Abstract Background Primary care providers require accurate evidence on chronic disease prevalence in people with intellectual disabilities in order to apply this information into practice. This study aimed to map the broadness of literature on chronic disease prevalence in people with and without intellectual disabilities, and to explore main characteristics of these studies. Method A scoping review of peer‐reviewed literature was conducted, covering 2000 to February 2020, including literature that discussed chronic disease prevalence in people with and without intellectual disabilities, with similar data collection method for both groups. Results Nineteen studies were included. Chronic disease prevalence varied considerably between people with and without intellectual disabilities. Studies differed in their methodologies, country and age groups that were enrolled. Conclusions Primary care providers should interpret results on disease prevalence among people with intellectual disabilities in light of the study characteristics. Researchers should always interpret prevalence rates in the context of methodology.


| INTRODUCTION
Unambiguous information on chronic disease prevalence in people with intellectual disabilities is largely lacking (Macrae et al., 2015;Oeseburg et al., 2011). Varying and sometimes even conflicting prevalence rates are presented in the literature (Draheim, 2006;Macrae et al., 2015). Heterogeneity between studies can potentially be reflected in various factors such as sample size, type of data, or methods of identification of intellectual disabilities; making correct understanding and interpretation of chronic disease prevalence in people with intellectual disabilities more complex.
Primary care providers and actors in public health planning require accurate information on chronic disease prevalence to interpret results in terms of chronic diseases being more or less prevalent among people with intellectual disabilities as compared to people without intellectual disabilities (Cooper et al., 2018;Mccarron et al., 2017;Morin et al., 2012;Tyler et al., 2010). Such accurate evidence, that can be applied and translated into practice, is a first necessity in providing optimal healthcare (Lennox et al., 2015). A better insight into the aspects that relate to the inconsistencies in the literature is therefore necessary to help primary care providers and researchers to better understand and accurately interpret prevalence rates of chronic diseases in people with intellectual disabilities.
The impact of chronic diseases can be even higher for people with intellectual disabilities compared to the general population, as they experience limitations in adaptive behaviour and intellectual functioning (Schalock et al., 2010). As a result, it is more difficult for them to fully comprehend the implications of chronic diseases, and this complicates disease management and results in poorer health outcomes (Van Schrojenstein Lantman-De & Walsh, 2008).
As chronic diseases are mostly managed in primary care, this setting provides the most complete representation of everyone in the population with and without chronic diseases (Crooks et al., 2015;Harvey et al., 2002). Secondary care settings typically report higher prevalence estimates than primary care settings do, as patients in this setting are more likely to have a chronic illness but may be overrepresented by severe cases (Crooks et al., 2015). It is therefore most relevant to focus on prevalence studies on people with and without intellectual disabilities conducted in primary care settings. Information on the prevalence of diseases such as IHD, CVD, DM and COPD is also used to plan the size and the allocation of healthcare resources (Petrou & Wolstenholme, 2000). Accurate understanding of published prevalence rates is therefore essential. This scoping review therefore aims (1) to map the broadness of published literature on IHD, CVD, DM and COPD prevalence in people with intellectual disabilities compared to people without intellectual disabilities in primary care settings, and (2) to explore main characteristics of these studies.

| Study design
This study is a scoping review, a type of review commonly used to map existing literature that 'exhibits a large, complex or heterogeneous nature' (Peters et al., 2015, p. 141). They are particularly useful for describing research findings in more detail by taking different research designs into account (Arksey & O'malley, 2005;Munn et al., 2018;Peters et al., 2020). This way, study characteristics that may be deemed important can be mapped and discussed (Munn et al., 2018). This scoping review followed the PRISMA guidelines extension for scoping reviews (PRISMA-ScR) (Tricco et al., 2018).

| Search strategy
To identify eligible studies, the databases of Embase, Medline, PubMed, Web of Science and PsycInfo were electronically searched for publications issued between 1 January 2000 and 7 February 2020.
The search strategy was developed in collaboration with a medical research librarian and consisted of a combination of four concepts: intellectual disabilities, prevalence, chronic diseases, and comparison with the general population. Both broad (e.g., 'chronic diseases') and specific (e.g., 'diabetes mellitus') terms were used in order to ensure that all relevant studies were included in the search results. A complete overview of the search strategy is provided in the Supplement.

| Study selection
Studies were included if they: • were written in the English language; • reported original data; • were published in peer-reviewed journals; • discussed the prevalence of at least IHD, CVD, DM or COPD; • addressed the prevalence within (a subgroup of) people with intellectual disabilities compared to people without intellectual disabilities; • used a data collection method that was identical for people with and without intellectual disabilities.
Studies were excluded if they focused solely on conditions where intellectual disabilities cannot be assumed (i.e., cerebral palsy, autism spectrum disorder); assessed the prevalence of chronic conditions after certain interventions; focused on children only (aged 18 or below); and took place in secondary care settings only (such as hospitals or specialist care).
The initial search was conducted by the first author (MvdB), with the second author (MC) screening a random sample of 10% of all titles and abstracts. Next, the remaining articles were screened full-text by the first and the second author to assess eligibility. Disagreements were solved by discussion.

| Methodological quality assessment
To better judge the results of included studies, all studies deemed eligible for inclusion were evaluated on methodological quality to assess risk of bias. The appraisal tool used -Joanna Briggs Institute Prevalence Critical Appraisal Toolwas created specifically to evaluate studies reporting prevalence data (Munn et al., 2015). The checklist consisted of nine questions and addressed the following issues: sampling, sample size, (non)response rates, description of study participants and country, appropriate statistical analysis, and valid and reliable methods to identify the condition of intellectual disabilities.
The first and the second author assessed the studies separately and later reached agreement by discussion.
The results of the quality appraisal checklist were combined into four main topics in order to provide a more structured overview. First, the findings regarding the sample were summarised; this concerned issues such as representativeness, sampling methods and sample sizes.
Second, attention was paid to the method of identification of people with intellectual disabilities. Possible influencing factors such as the use of proxy respondents, identification of intellectual disabilities based on formal diagnosis or otherwise, and method of recruiting respondents with intellectual disabilities were taken into account.
Third, the manner of identification of chronic diseases was summarised, such as diagnoses in medical records or self-reported diseases. Last, the type and detail of statistical analyses performed in each study were summarised. For each topic, studies were assessed on a three-point scale ranging from sufficient (+) to insufficient (À).
The assessments are presented in the Supplement.

| Data extraction and calculations
All data on relevant chronic diseases were extracted from the included articles. Some studies reported chronic disease prevalence for men and women separately (Mcdermott et al., 2007b) or for age groups separately (Flygare Wallen et al., 2018). In order to achieve comparability, new prevalence rates were calculated by determining the mean of the rates for men and women (not weighted due to unavailability of population size rates) and weighted mean of the rates for the age groups. Thus, one mean prevalence rate for the total study population was computed.
Characteristics of the included studies were described. First, different types of data can be used to report on chronic disease prevalence, such as register or (primary care) medical data. Next, the definition of intellectual disabilities is the way in which intellectual disabilities were operationalised in the included study. Methods for identifying someone as having intellectual disabilities consisted of a medical record of a diagnosis of intellectual disabilities, various screening methods, or information on received services or supports specifically for people with intellectual disabilities (e.g., income support programmes, social services and residential care). Country was defined by the country in which the studies were performed, along with their dominant lifestyle and health policies and their organisation of healthcare. Next, age groups were the ages of the included study groups that were taken into account. Lastly, sample size was the size of the group of people with intellectual disabilities and the comparison group.

| RESULTS
The initial search resulted in 4311 papers, excluding duplicates. After title and abstract screening, 98 articles were assessed full-text. There was disagreement on 14% of the articles (n = 14), on which consensus was reached by discussion. This resulted in 19 studies meeting the inclusion criteria ( Figure 1). A complete overview of study characteristics and prevalence rates is shown in Table 1     †, no significant difference; ‡, significant difference, p-level not reported; *p < 0.05; **p < 0.01; ***p < 0.001; §, significance not reported.

| Characteristics of the included studies
The results of the quality appraisal are depicted in Appendix S1. Eight studies received a high appraisal (++ or +), eight a medium appraisal (±), three a low appraisal (À).
The characteristics of the included studies are described in

| IHD prevalence
Studies (n = 10/19) reported IHD prevalence rates between 0.0% and 5.7% for people with intellectual disabilities, and 0.0% to 7.7% for people without intellectual disabilities ( Figure 2). In most studies, IHD prevalence was lower for people with intellectual disabilities compared to people without intellectual disabilities. One study that stratified by severity levels of intellectual disabilities reported higher IHD prevalence in more severe levels (Jansen et al., 2013). The highest IHD prevalence rates among people with and without intellectual disabilities were found among the studies with a high-quality appraisal (Jansen et al., 2013;Tyler et al., 2010) (Table 2). The range in IHD prevalence was higher in studies where the population of people with intellectual disabilities was identified through relevant diagnoses in medical records rather than through other methods (Figure 3). The two studies identifying intellectual disabilities through support or services both focused on adults aged 50 years or older (Jansen et al., 2013;Mccarron et al., 2017), of which one shows highest IHD prevalence among people with intellectual disabilities (Jansen et al., 2013). In studies performed in the Unites States, IHD prevalence had the highest range for people without intellectual disabilities compared to other countries (Figure 4). Studies performed in Great

| CVD prevalence
CVD prevalence in the included studies (n = 10/19) varied from 0.3% to 5.7% among people with intellectual disabilities, and from 0.0% to 4.4% among people without intellectual disabilities (Figure 2). One study reported prevalence by severity levels: the higher the severity level of intellectual disabilities, the higher the CVD prevalence (Jansen et al., 2013). The range in prevalence among people with intellectual disabilities was higher when diagnoses of intellectual disabilities in medical records were used as the indicator (Figure 3). The United States had the highest range in CVD prevalence among people with intellectual disabilities. In the UK, the range in CVD prevalence was higher among people without intellectual disabilities (Figure 4).
The highest CVD prevalence among people both with and without intellectual disabilities was reported by a study including adults aged 50 years and older (Jansen et al., 2013); the lowest prevalence rates were reported by  who included ages 40 years or less ( Table 2). The highest difference in prevalence rates between people with and without intellectual disabilities could be found among the study using the smallest samples . Studies

| DM prevalence
The prevalence of DM varied in studies (n = 18/19) from 0.7% to 11.5% among people with intellectual disabilities, and from 0.4% to 19.3% among people without intellectual disabilities (Figure 2). DM prevalence was mostly higher for people with intellectual disabilities than for people without intellectual disabilities, except in studies that found high prevalence rates among people without intellectual disabilities (>10%). Only two studies distinguished between Type 1 and 2 diabetes (Mcdermott et al., 2006(Mcdermott et al., , 2007a. Both the highest and the lowest DM prevalence for people with and without intellectual disabilities were found in studies using diagnoses related to intellectual disabilities in medical records (Table 2). DM prevalence among people with intellectual disabilities was generally lower in the Unites States compared to those without intellectual disabilities, whereas the opposite was true for Western-Europe ( Figure 4). The studies with highest appraisal were performed in Western-Europe (Cooper et al., 2015;Hedgeman et al., 2017). The two studies focusing on all ages reported the highest prevalence among people with intellectual disabilities and the lowest DM prevalence among people without intellectual disabilities (Hedgeman et al., 2017;Perera et al., 2019). Lastly, the smallest sample size corresponds with the highest DM prevalence in people both with and without intellectual disabilities , whereas the lowest prevalence rates can be found in the largest sample size (Perera et al., 2019). The highest DM prevalence among people with and without intellectual disabilities was reported in a study from the Unites States with smallest sample, which focused on the oldest age groups (40-79 years) compared to the other studies .

| COPD prevalence
Studies on COPD (n = 8/19) reported prevalence rates from 1.1% to 6.4% among people with intellectual disabilities, and from 1.4% to 9.5% among people without intellectual disabilities (Figure 2). In all but one study (Durbin et al. (2019), the prevalence of COPD was lower in people with intellectual disabilities compared to people without intellectual disabilities. The highest COPD prevalence was reported by two studies with a medium appraisal (Mcdermott et al., 2006(Mcdermott et al., , 2007a. COPD prevalence was highest in the USA compared to studies performed in other countries, and showed the largest differences between people with and without intellectual disabilities ( Figure 4). Prevalence rates in the UK were more comparable between people with and without intellectual disabilities, and overall lowest across the included studies. The only study considering all ages reported the lowest COPD prevalence (Perera et al., 2019) (Table 2).
A larger sample size was accompanied by a lower COPD prevalence (Perera et al., 2019), a smaller sample size by a higher prevalence (Mcdermott et al., 2006(Mcdermott et al., , 2007a.

| DISCUSSION
This scoping review is the first to map the broadness of published literature on chronic disease prevalence in people with intellectual disabilities compared to people without intellectual disabilities. Chronic disease prevalence varied considerably between studies and differed when study characteristics were taken into account. This study builds upon existing chronic disease prevalence reviews by exploring their observations that methodological differences in the included studies could possibly be important in explaining variances in prevalence rates. The reviews mention methodological differences such as operational definition and method of identification of intellectual disabilities, differences in study groups in terms of sex and aetiology of intellectual disabilities, method of data collection, sample size and method of diagnosis of chronic diseases (Jansen et al., 2004;Macrae et al., 2015;Mcvilly et al., 2014;Oeseburg et al., 2011). Other similar reviews either did not take the role of methodological choices into account or focused on different health problems (Fortin et al., 2012;Jansen et al., 2004). This study is therefore the first to offer guidance to primary care providers and researchers in interpreting chronic disease prevalence in people with intellectual disabilities.
This review described characteristics of included studies and identified five valuable aspects that are important when interpreting chronic disease prevalence in people with intellectual disabilities; being type of data, identifying of intellectual disabilities, country, age of the study groups and sample size. These aspects are discussed one by one: First, when interpreting results, one should always be aware of the consequences of different types of data. Studies relying on self-reported values are at risk of potential bias, which may result in an over-or underestimation of a person's ill-health. In people with intellectual disabilities, self-reporting can be accompanied by extra challenges (Fujiura & Rrtc Expert Panel on Health Measurement, 2012), and therefore studies often resort to using proxy respondents. However, proxy reporting decreases the validity of the results (Cummins, 2002;Emerson et al., 2013) and complicates comparison between people with and without intellectual disabilities.
Second, this study emphasises the value of recognising the way in which intellectual disabilities are identified and defined across studies.
Although most included studies used similar methods for identifying intellectual disabilities (via medical records or records of specific services), chronic disease prevalence was still diverse in these studies.
This finding suggests that studies using the same methods for identifying people with intellectual disabilities do not necessarily include the same populations, as people with intellectual disabilities are identifiable via multiple sources. Earlier research supports the finding that using different identification methods as well as different definitions of intellectual disabilities may complicate estimating prevalence rates (Lin et al., 2013).
Only a few countries have national registers from which intellectual disabilities can be identified in a relatively reliable manner; other methods are often less conclusive (Mcconkey et al., 2019). Frequently, many different conditions related to intellectual disabilities were examined simultaneously, but in conditions such as autism or cerebral palsy intellectual disabilities cannot always be assumed (Bryson et al., 2008;Reid et al., 2018).
Third, the country in which studies were performed was relevant for interpreting chronic disease prevalence. Interestingly, in the United States, the prevalence of cardiovascular diseases (IHD and CVD) was consistently higher among people with intellectual disabilities compared to people without intellectual disabilities, whereas COPD and DM in the USA were more prevalent among people without intellectual disabilities. Prevalence of IHD, CVD, DM and COPD was high in the United States among people both with and without intellectual disabilities compared to other countries. A possible explanation is the higher prevalence of unhealthy lifestyles, and consequently obesity levels, in the Unites States (Wang & Beydoun, 2007), given that these diseases are all related to unhealthy lifestyles (Forey et al., 2011;Willett et al., 2006). In addition, some argue that American health promotion policies can be prone to reinforce health inequalities (Goldberg, 2012), whereas European policies seem more inclusive (Fosse, 2011). Furthermore, the differences in primary care systems in the Unites States and European countries can result in different timings in diagnosis and management of chronic diseases (Erler et al., 2011;Mcglynn et al., 2003). When interpreting and comparing health statuses of people with intellectual disabilities residing in the Unites States and Western-Europe these differences should therefore always be kept in mind.
Fourth, the role of age should always be noted in studies on chronic disease prevalence. Although the life expectancy of people with intellectual disabilities has increased, they often show earlier signs of aging compared with people without intellectual disabilities (Evenhuis et al., 2012), resulting in higher mortality rates . Results and comparability between people with and without intellectual disabilities can be affected by this earlier aging effect, as the occurrence of chronic diseases is generally higher with increasing age (Buist et al., 2007;Thomsen & Nordestgaard, 2014), and as several chronic diseases are more common among aging people with intellectual disabilities than among aging people without intellectual disabilities (Krahn et al., 2006). In line with these previous findings, this review found that studies only taking older age groups into account were more likely to report higher prevalence of chronic diseases in people with intellectual disabilities.
Fifth, sample sizes should be critically evaluated when one is interpreting differences in prevalence rates of chronic diseases. In the case of COPD and DM, it could be seen that a higher sample size was accompanied by a lower prevalence, and vice versa. This can be explained by the fact that larger sample sizes are generally better suited to make more precise claims and are more likely to have included a representative sample (Charter, 2003).

| Strengths and limitations
This review has some limitations. First, we restricted our scope of chronic disease to IHD, CVD, DM and COPD. Diseases that are more prevalent among people with intellectual disabilities, for instance epilepsy (Mcdermott et al., 2005) (Jansen et al., 2004), they focused solely on the prevalence of several health problems that were not included in this review, such as epilepsy and sensory loss. The current review is in line with another review that explored how methodological choices may influence multimorbidity prevalence rates (Fortin et al., 2012). Comparable to the current review, the authors concluded that type of data, country and age groups are important in assessing multimorbidity in the general population. However, intellectual disabilities were not taken into account (Fortin et al., 2012). This review therefore offers direction in interpreting studies on chronic disease prevalence in people with intellectual disabilities. Second, it offers a first insight into the comparative health regarding chronic diseases of people with intellectual disabilities compared to the general population. Third, a large variety of studies have been taken into account. Although study characteristics such as age or sex are better known influences on prevalence rates (Flygare Wallen et al., 2018;Perera et al., 2019), this review highlights the significance of other, less often examined characteristics, such as type of data. In traditional reviews, the great heterogeneity in study designs, populations and countries is associated with challenges in summarising evidence, but by performing a scoping review it was possible to explore such characteristics in greater depth. Fourth, the fact that we were able to perform a quality assessment increases the legitimacy of the claims made.

| Recommendations for future research
This review provides a fruitful basis upon which to build future research on chronic diseases in people with intellectual disabilities.
First, as the current review is the first to explore the role of study designs, populations and countries in chronic disease prevalence, this study can be used as a valuable basis for conducting further research, such as a meta-analysis. In addition, no studies conducted in non-Western countries were identified. Research demonstrates that chronic diseases represent a high burden in non-Western, low-or middle-income or less developed countries (Boutayeb & Boutayeb, 2005;Wagner & Brath, 2012). The situation of people with intellectual disabilities is also very different in such countries, but this global difference is not often studied (Emerson et al., 2008). The prevalence rates of IHD, CVD, DM and COPD as presented in this review are therefore a representation of Western countries.
Next to the use of different methods or countries, this review has also identified several important aspects that future research should take into account when both studying and interpreting chronic disease prevalence in people with intellectual disabilities. First, future research should disclose as much as possible the study and population characteristics. Existing guidelines for prevalence studies, such as STROBE or RECORD (Benchimol et al., 2015;Von Elm et al., 2014), are useful tools and should be utilised widely. This way, the need for valid and reliable information on the health of people with intellectual disabilities (Ruddick, 2005) can be better met. Second, in order to make useful claims future studies on chronic disease prevalence should take into account multiple interacting factors, such as age Jansen et al., 2013;Perera et al., 2019) or sex (Flygare Wallen et al., 2018;Mcdermott et al., 2007a), but also factors such as type of data or identification of intellectual disabilities. Additionally, future research should report chronic disease prevalence by severity levels of intellectual disabilities if possible. The few studies that do so report possibly important patterns in chronic diseases (Cooper et al., 2018;Heslop et al., 2019;Jansen et al., 2013). Third, large population studies should be conducted in order to obtain reliable and valid prevalence estimates. In this type of study, entire populations can be taken into account, resulting in thoroughly defined and representative study populations (Lieb, 2013). Because it currently still is difficult to identify people with intellectual disabilities in population datasets , future studies should be transparent in the methods used to identify people with intellectual disabilities.
Lastly, comparisons between incidence and prevalence rates can prove interesting research subjects. While prevalence rates are useful for indicating disease burden, incidence rates give insight in the occurrence rate of chronic diseases in populations (Keiding, 1991).

| CONCLUSION
This review adds to the literature by providing a first exploration of the broadness of published literature on chronic disease prevalence in people with intellectual disabilities and by describing main characteristics of these studies. Chronic disease prevalence varies greatly between people with and without intellectual disabilities across studies. Although study characteristics such as country and age group are more apparent influencers in chronic disease prevalence, this review also highlights the importance of other factors that are less often examined, such as type of data and definition of intellectual disabilities. Researchers should therefore acknowledge the influence of study characteristics and methodologies when studying chronic disease prevalence in people with intellectual disabilities. This review underscores the need for transparent and comparable prevalence studies. The great heterogeneity in study characteristics and methodologies complicate generalisation of study results. Rather, this review argues that prevalence rates should always be interpreted in the context of methodology. Only then, primary care providers and public health planners are able to utilise prevalence rates of chronic diseases and apply them into practice.

ACKNOWLEDGEMENT
We would like to acknowledge the medical librarians of Radboudumc for their help with the search strategy.