Little is known about the health status of and health risks faced by adults with intellectual disability who do not use intellectual disability services. Self-report data collected from 1,022 people with mild intellectual disability in England indicated that people who do not use intellectual disability services are more likely to smoke tobacco and less likely to access some health services and promotion activities than those who do use these services. In addition, they are more likely to be exposed to some known social determinants of poorer health (greater material hardship, greater neighborhood deprivation, reduced community, and social participation).
People with intellectual disability have significantly poorer health than their peers who do not have intellectual disability (Krahn, Hammond, & Turner, 2006; Ouellette-Kuntz, 2005; Sutherland, Couch, & Iacono, 2002; U.S. Department of Health and Human Services, 2002). This disparity in health status is, in part, attributable to potentially avoidable factors, including increased rates of exposure of people with intellectual disability to key social determinants of health (e.g., poverty, social exclusion) and institutional discrimination in the operation of health care systems (Disability Rights Commission, 2006; Emerson & Hatton, 2007a, 2007b, 2007c; Krahn et al., 2006; Mencap, 2007; Michael, 2008; Ouellette-Kuntz, 2005). As such, the poorer health status of people with intellectual disability can be appropriately described as an example of health inequity (Graham, 2005; Ouellette-Kuntz, 2005; Whitehead, 1992).
Concern about the heath inequalities faced by people with intellectual disability has generated a number of practice and policy initiatives (Department of Health, 2007, 2009; Lennox et al., 2002; Meijer, Carpenter, & Scholte, 2004; U.S. Department of Health and Human Services., 2002). However, accurately monitoring and understanding the health of people with intellectual disability is beset by a number of rather intractable methodological problems (National Center on Birth Defects and Developmental Disabilities, 2010). Perhaps foremost among these are the practical difficulties associated with constructing representative population-based samples of adults with intellectual disability (Campbell & Fedeyko, 2001).
Virtually all research and health surveillance activity undertaken in the field of intellectual disability utilizes convenience samples drawn from administratively defined populations (i.e., people who are known to agencies providing specialized services for people with intellectual disability). This is problematic because it appears likely that in most jurisdictions, a significant proportion of adults with intellectual disability do not use specialized intellectual disability services and are not known to the agencies providing such services (Edgerton, 2001). In England, for example, the administrative prevalence of intellectual disability drops precipitously from 4.2% among children in the education system (Emerson, in press) to 0.6% among adults aged 20 to 29 (Emerson & Hatton, 2004). Similar reductions in administrative prevalence have been reported in the United States (Kiely, 1987; National Center on Birth Defects and Developmental Disabilities, 2010).
It is highly implausible that such reductions in prevalence can be accounted for by either excess mortality or sudden improvements in intellectual functioning. Rather, it is likely that they reflect the impact of a combination of factors that will vary in importance across jurisdictions and cultures. These include (a) a general decrease in health/disability surveillance in posteducation health and welfare agencies; (b) the rationing of specialized health and welfare supports to adults with disabilities; (c) the stigma associated with intellectual disability, which can lead to unwillingness to use specialized services or self-identify as having intellectual disability; and (perhaps) (d) the less disabling impact of the intellectual impairments associated with intellectual disability in noneducational settings.
The last point is qualified because the available evidence suggests that people with mild or borderline intellectual disability are at risk of significant levels of disadvantage, social exclusion, and poor health. First, in a small number of population-based longitudinal studies, researchers have followed-up children with mild (Maughan, Collishaw, & Pickles, 1999; Richardson & Koller, 1996; Wells, Sandefur, & Hogan, 2003) or borderline (Firkowska-Mankiewicz, 2002; Seltzer et al., 2005; Vaillant & Davis, 2000) intellectual disability into adulthood. Although the results of these studies are somewhat mixed, they do indicate that when compared with their peers, people with mild or borderline intellectual disability tend to have lower levels of psychological well-being, poorer self-rated health, reduced access to social capital, more problems in personal and social relationships, lower occupational prestige, and lower income (Firkowska-Mankiewicz, 2002; Maughan et al., 1999; Richardson & Koller, 1996; Seltzer et al., 2005; Vaillant & Davis, 2000; Wells et al., 2003). Although these investigators did not discriminate between participants who were and were not receiving support from intellectual disability agencies, it is likely that the vast majority were not receiving such support (Edgerton, 2001). Second, in their pioneering ethnographic studies, Edgerton and his colleagues documented in great detail the difficulties faced by deinstitutionalized adults with mild intellectual disability in the absence of formalized support systems (Edgerton, 1967, 1984, 1996; Edgerton & Bercovici, 1976). Third, a small number of cross-sectional studies have highlighted the extent of disadvantage, social exclusion, and poor health of people with borderline or mild intellectual disability (Ferrari, 2009; Tymchuk, Lakin, & Luckasson, 2001). The results of this body of research strongly suggest that it would be unwise to assume that the health and well-being of adults with mild intellectual disability (but who are not identified as such in administrative agency records) is not problematic or equivalent to that of their peers with no disabilities.
My aim in the present paper is to compare the health and exposure to risk factors associated with poorer health among samples of adults with mild intellectual disability who were and were not receiving support from specialized intellectual disability health and welfare agencies. Quantifying the similarities and differences between these two groups could aid in estimating the health needs and risks associated with the “hidden majority” of adults with mild intellectual disability who are not identified in administrative records.
In this study I conducted secondary analysis of data extracted from the survey Adults With Learning Difficulties in England 2003/4. (The term learning difficulties that is used in this survey is synonymous to the term intellectual disability.) Full details of the study design are available at a dedicated website (http://www.ic.nhs.uk/statistics-and-data-collections/social-care/adult-social-care-information/adults-with-learning-difficulties-in-england-2003-2004), in the project report, and several associated papers (Emerson & Hatton, 2008a, 2008b; Emerson, Malam, Davies, & Spencer, 2005). The main aspects of the design are summarized below.
The original survey involved interviews with 2,898 adults with intellectual disability living in either private households (independently or with relatives) or in some form of long-term supported accommodation. Adults with intellectual disability who were homeless, incarcerated, or living temporarily in a health-related treatment facility were not included in the survey. I restricted the present analyses to a subsample of participants with intellectual disability who were the primary informants during the interview.
The survey sample was drawn from five sampling frames that in combination provided comprehensive coverage of adults with intellectual disability in England.
General Household Omnibus Survey
Screening questions (see appendix) to identify adults with intellectual disability were placed in weekly General Household Omnibus Surveys (based on random locale sampling) operated by a market research organization for 56 weeks between July 2003 and August 2004. This led to the identification of 2,214 potential participants, 74% of whom gave permission to be re-contacted and for whom there was sufficient information for them to be included. Of these, 35% were deemed ineligible. Information was collected on 750 people (70% of the issued eligible sample). Of these, 58 (8%) were subsequently excluded because their academic attainments made it highly unlikely that they had an intellectual disability.
Administrative records of adults with intellectual disability living in private households
Twenty Councils With Social Services Responsibilities were selected to ensure coverage by type of council, percentage of population from minority ethnic communities, affluence, and geographical region. Administrators at each Social Services Council randomly selected 50 adults with intellectual disability who were living in private households and receiving services from the social services councils. Information was collected on 480 people (71% of the eligible issued sample).
Administrative records of adults with intellectual disability living in differing forms of supported accommodation
Three nonoverlapping sampling frames were employed to draw random samples of adults with intellectual disability living in three administratively defined systems of supported accommodation: (a) Registered Residential Care Homes, (b) supported accommodation funded through the national Supporting People program, and (c) medium- to long-term accommodation operated by the United Kingdom's National Health Service. From each database, a random sample of providers was selected with probability proportionate to size (number of residents or households). Each provider was requested to randomly sample 10 respondents from an alphabetically sorted list of all residents. This procedure resulted in information on 919 people living in Registered Residential Care Homes (70% of the eligible issued sample), 562 people living in Supporting People settings (68% of the eligible issued sample), and 263 people living in National Health Service accommodation (83% of the eligible issued sample).
Information was collected by face-to-face computer-assisted personal interviews undertaken at the participants' home or service setting. A number of strategies were adopted to maximize the active participation of the participant with intellectual disability. These included providing specific training for all interviewers (part of which was undertaken by trainers who had intellectual disability), simplifying the wording of questions, employing visual aids, and encouraging interviewers to rephrase questions (Emerson et al., 2005).
Due to the severity of their intellectual impairments, however, a significant proportion of the sample had difficulty answering interview questions. In such cases, interviewers asked proxy responders to provide information. In addition, participants were given the opportunity to be supported during the interview by a person of their choice. When supported, the interviewer coded at 34 separate points during the interview the identity of the person providing the majority of answers to the questions asked in the preceding section: the adult with intellectual disability, a combination of the adult with intellectual disability and the support person, or the support person.
In order to avoid problems associated with combining self-report and proxy data on subjective states (Schneider & Schimmack, 2009), only participants who were either interviewed alone (no support person being present at any point) or who themselves answered the majority of questions related to health and well-being as well as the majority of answers to more than 50% of questions posed in other sections of the interview were chosen. Twenty-two people who either showed evidence of response bias or acquiescence or for whom the data used to determine response bias or acquiescence were missing were excluded (Emerson, 2010; Emerson & Hatton, 2008a). The final sample contained 1,022 adults with intellectual disability.
This sample was further divided into people who were considered to be (or not to be) in contact with specialized health or social care services for people with intellectual disability. Participants were identified as being in contact if either: (a) they reported, in response to specific questions contained in the interview, receiving assistance from social services departments, National Health Services staff, or paid support persons in any of the following six areas (housing, looking after their children, managing money, receiving welfare benefits, looking for employment, and/or receiving personal support; (b) they reported attending a day center; (c) they were recruited from samples based on administrative records. The resulting subsamples contained 227 participants (22%) considered not to be in contact with services and 795 (78%) participants considered to be in contact with services. Of the participants in contact with services, 213 (27%) were living in general households (either independently or with relatives); the remainder (582, 73%) were living in staffed or semi-staffed supported accommodation.
Consent and Ethical Approval
Ethical approval was obtained from the relevant agencies. The ability of each potential participant to give informed consent was assessed by the interviewer providing them with a verbal and written overview of the project and then assessing ability to recall relevant information. Of the participants in the selected subsamples, 92% were judged able to give informed consent (by giving positive responses in each of the four areas). For the remaining 8%, agreement to participate was sought and gained from a relative.
The Adults With Learning Difficulties In England 2003/4 Survey resulted in wide-ranging information about the life experiences of participants. Key measures I used in the present analyses are described below.
Participants were asked, “In the last year would you say your health was very good, fairly good, or not good?”
Five interview items were related to aspects of psychological well-being. First, interviewees were asked, “How do you feel about your life at the moment? Very happy, quite happy, sometimes happy/sometimes unhappy, mostly unhappy. Please point to the face.” They were then asked to select one of these four options from a visual cue card. Second, they were asked four questions of similar format from the Millennium Poverty and Social Exclusion Survey, which were slightly amended for use in the present survey (Pantazis, Gordon, & Levitas, 2006). “All of us feel a bit unhappy or worried at times. Do you ever feel sad or worried?” If the respondent answered yes, they were then asked “Is that a lot or just sometimes?” This question format was repeated for left out of things, helpless, and confident. Participants were shown a visual cue card to help identify the psychological states in question.
A number of variables related to the personal characteristics of participants and their living situation were extracted from the data set: age, gender, level of support needs (Emerson et al., 2005). The Support Needs Scale was specifically designed for the project after co-researchers who had intellectual disability rejected the use of existing scales of adaptive behavior on the grounds that they were too demeaning to people with intellectual disability. The scale commenced with a general statement (All of us need help at times to do things that we find difficult. I now want to ask you some questions about how much help you usually need to do different things) and then a question to determine how much support the person needed (I can do it on my own, I need a bit of help, I need a lot of help, I need someone to do it for me) in undertaking 11 activities that varied in complexity from drinking a cup of tea to paying money into a bank. Visual cues were used to illustrate each item and the level of support required. The scale showed acceptable levels of internal consistency for the full sample, α = .91, and for the subsample of participants included in our analyses, α = .79.
Two indicators of socioeconomic disadvantage were extracted from the data set: neighborhood deprivation quintile from the English Indices of Deprivation (Noble et al., 2004) and a measure of material hardship. The latter contained nine items derived from the Millennium Poverty and Social Exclusion Survey (Pantazis et al., 2006). Participants were asked, “Sometimes, when money is tight, people have to go without things. In the last year, have you always had enough money for _____ [item] when you wanted it/them?” The specific items included were new clothes, new shoes, food, heating, telephoning friends or family, going out, visits to the pub or a club, a hobby or sport, a holiday. Visual cues were used to illustrate each item. The Hardship Scale showed acceptable levels of internal consistency for participants in our analyses, α = .89.
Approach to Analysis
In order to simplify interpretation and address the nonnormality of variable distributions, nonbinary indicators of social participation and hardship were reduced to binary variables by splitting variables at the median scale point.
Selected personal characteristics of participants are presented in Table 1. Given the significant between-group differences in age, support needs, ethnicity, marital status, and the trend toward significant between-group differences in gender, I made all subsequent between-group comparisons controlling for any potential confounding effects these differences in personal characteristics may have had on health status. Specifically, I used logistic regression to compare the reference group (people not in contact with intellectual disability services) with the two comparison groups (people in contact with intellectual disability services living in general households or supported accommodation) after controlling for between-group differences in age, support needs, ethnicity, marital status, and gender.
Hardship and Social Capital
Indicators of hardship and social capital are presented in Table 2. When compared to both groups of people receiving services, participants not receiving services were significantly more likely to experience material hardship and to live in a materially and socially deprived neighborhood, significantly less likely to have regular contact with friends who have intellectual disability, and significantly less likely to have participated in an above median number of community activities in the previous month. When compared to people receiving services living in general households, they were more likely to be in paid employment. When compared to people receiving services living in supported accommodation, they were more likely to see their family but less likely to feel safe in the area in which they were living.
Health and Access to Health Care
Indicators of health and access to health care are presented in Table 3. When compared to both groups of people receiving services, participants not receiving services were significantly less likely to have visited a dentist in the previous year and significantly more likely to smoke. When compared to people receiving services living in general households, they were less likely to have poor self-rated health. When compared to people receiving services living in supported accommodation, they were less likely to feel confident and less likely to have had their vision and hearing tested in the previous year.
Adults with intellectual disability who do not use intellectual disability services were more likely to smoke tobacco and less likely to access some health services and promotion activities than were adults with intellectual disability who do use such services. In addition, they were more likely to be exposed to some known social determinants of poorer health (e.g., greater material hardship, living in more deprived neighborhoods, reduced community and social participation). These differences were generally more marked when comparisons were made between adults with intellectual disability who do not use intellectual disability services and adults with intellectual disability in supported accommodation services. When compared with adults with intellectual disability known to intellectual disability services but who lived in general households (either independently or with relatives), a much more mixed pattern of health risks is evident.
Strengths and Limitations of the Study
The main strengths of the study are the inclusion of adults with intellectual disability who do not receive intellectual disability services, the use of a relatively large nationally representative sample of adults with less severe intellectual disability, and the use of multiple indicators of health status and risk. The main limitations are reliance on self-report measures of health status and use of health services, failure to verify the respondents' level of intellectual disability, likely failure to identify nonusers of intellectual disability services, and the inability within the data to independently verify that people did not use intellectual disability services.
With regard to the first point, although there is extensive evidence that self-rated health is a robust predictor of mortality in general populations (Idler & Benyamini, 1997), it is clearly not equivalent to health status (Sen, 2002). Indeed, the relationship between general self-reported health and morbidity is complex and likely to reflect such factors as individual and group differences in (a) interpreting the question (e.g., the time span over which health is to be evaluated, whether health includes mental health as well as physical health), (b) expectations regarding what would constitute good or poor health, and (c) the extent to which ill health impacts on meeting the demands of everyday life. Thus, for example, evidence suggests that low socioeconomic position may be associated with an underreporting of ill health, an association that would lead to measures of self-reported health underestimating social gradients in health status (Blane, Power, & Bartley, 1996).
With regard to the second point, I have assumed that selection on the basis of performance in interview (answering most questions independently, showing no evidence of response bias) will have led to the selection of people with mild intellectual disability. This was not, however, possible to independently verify.
As noted in the introduction, there is good evidence to suggest that the majority of adults with mild intellectual disability do not use intellectual disability services and, thereby, constitute a ‘hidden majority.’ In the present study, however, only a minority of participants (22%) were nonusers of intellectual disability services. It is likely that the apparent contradiction between my estimates and sample reflect the undersampling of people with mild intellectual disability who do not use intellectual disability services resulting from, among other things, an unwillingness for people to self-identify as having intellectual disability. The extent to which the health status of study participants who do not use services corresponds to the assumed wider population of nonusers is not known.
Finally, it was not possible to independently verify that people did or did not use intellectual disability services. As a result, some degree of misclassification may have occurred, thereby increasing the extent of error in the results. One could argue that it is more likely that people would have failed to mention that they used a service than falsely claimed that they were using a service. As such, the nonuser group could contain a proportion of service users, thereby reducing the probability of finding between-group differences.
As noted in the introduction, there is extensive evidence that people with intellectual disability who use intellectual disability services have significantly poorer health than do their peers (Disability Rights Commission, 2006; Graham, 2005; Krahn et al., 2006; Michael, 2008; Ouellette-Kuntz, 2005; Sutherland et al., 2002; U.S. Department of Health and Human Services, 2002). The results of the present study failed to identify any consistent difference in health status between adults with mild intellectual disability who were and those who were not using intellectual disability services. The results did, however, suggest that some health risks (e.g., exposure to material hardship, neighborhood deprivation, smoking) were higher among adults who did not use intellectual disability services. In general, the results suggest that the health inequalities faced by users of intellectual disability services may be mirrored among the much larger numbers of people with intellectual disability who constitute the hidden majority of people with intellectual disability who do not use and are probably unknown to intellectual disability services. The results are broadly consistent with the small number of studies in which researchers have investigated the health and social status of people with mild or borderline intellectual disability (Edgerton, 1967, 1984, 1996; Edgerton & Bercovici, 1976; Ferrari, 2009; Firkowska-Mankiewicz, 2002; Maughan et al., 1999; Richardson & Koller, 1996; Seltzer et al., 2005 ,Tymchuk et al., 2001; Vaillant & Davis, 2000; Wells et al., 2003).
Results of the present study are also consistent with the growing body of research on the association between intelligence and indicators of health, including mortality (Batty, Deary, & Gottfredson, 2007; Batty, Gale, Tynelius, Deary, & Rasmussen, 2009), long-term illness and self-rated health (Batty, Der, Macintyre, & Deary, 2006), and mental health (Batty et al., 2006; Emerson & Einfeld, 2010; Urfer-Parnas, Mortensen, Sæbye, & Parnas, 2010). In these studies researchers have tended to report clear monotonic associations between lower intelligence and poorer health status either across the full ability spectrum or, more commonly, across the bottom median or quartile of ability. As such, it appears that the health inequalities faced by people with intellectual disability represent the extreme end of a broader continuum rather than a discrete or unique association by intellectual ability and health.
The implications for improving the health status of (and reducing the health inequalities faced by) people with intellectual disability are clear; focusing solely on the visible minority of people who use and are known to intellectual disability services will prove an inadequate strategy. Rather, initiatives that address the determinants of the poorer health of the wider population of people with intellectual disability (and borderline intellectual disability) will be necessary. At a general level, this will require the mainstreaming of concern with intellectual disability and low intelligence within more generic policies aimed at reducing health inequalities or improving population health (Emerson et al., 2009; The Marmot Review, 2010).
Unanswered Questions and Future Research
Accurately documenting (and understanding) the health and well-being of the population of adults with intellectual disability who do not use services is problematic (Campbell & Fedeyko, 2001; National Center on Birth Defects and Developmental Disabilities, 2010). Further research is required to develop efficient and valid means of identifying people with probable intellectual or borderline intellectual disability through large-scale health and social surveys and administrative databases operating in health care systems. This is a complex and difficult task given the reluctance of many people with intellectual or borderline intellectual disability to self-identify as such and the lack of knowledge about intellectual disability in many mainstream health services. In addition, future researchers should seek to include independent verification of health status through, for example, the use of medical examination or administrative data from health records.
Editor-in-Charge: Susan Parish
Eric Emerson, PhD (e-mail: email@example.com), Professor of Disability and Health Research, Centre for Disability Research, Lancaster University, Lancaster, LA1 4TY, United Kingdom, and Visiting Professor, Australian Family and Disability Studies Research Collaboration, Cumberland Campus, University of Sydney, PO Box 170, Lidcombe, NSW 1825, Australia.