Demographic features of Americans with mild intellectual disabilities were estimated in an analysis of the National Health Interview Survey. The cohort was compared to the population of the United States, persons with specific learning disabilities, and persons with mental retardation. Comparison on basic indices of adaptive functioning and SES suggest a large cohort of Americans who share many support needs and social and economic vulnerabilities with those labeled “mentally retarded.” The combined prevalence of intellectual disability/mental retardation was estimated to be 1.27%. Implications are discussed in terms of the “forgotten generation” report of the 1999 President's Committee on Mental Retardation and evolving conceptions of mental retardation.
The now classic report by the President's Committee on Mental Retardation (PCMR), “Six Hour Retarded Child,” crystallized emerging concerns over cultural bias and the arbitrariness of labels in the assessment of children with mild intellectual impairments (President's Committee, 1969). The title was a metaphorical reference to disability labels meaningful only in the school context. Often labeled as having “borderline” or, to a lesser extent, “mild” mental retardation, these children faired reasonably well, if not on par with their peers, outside the academic setting (Guskin & Spicker, 1968; Kirk, 1964). In the years since the report's publication, the field of mental retardation has largely turned its focus away from this cohort in reaction to critiques of the labeling process and emerging emphases on the unmet needs and inclusion of persons with very severe disability.
Thirty years later, the PCMR revisited this cohort in its report entitled the “Forgotten Generation” (President's Committee, 1999; Tymchuk, Lakin, & Luckasson, 2001). The central theme and guiding assumption of the 1999 report was that persons with mild intellectual impairments confront substantial challenge and economic risk in a society grown more complex in its demands of intellectual ability and technical skill. Vulnerability is exacerbated by their exclusion—by statute or personal choice—from services and supports available to persons with mental retardation. This is a cohort with few outspoken advocates.
Though not directly addressed in the “Forgotten Generation,” the concerns expressed in the report are deeply connected to a central thesis of the 1969 report—the arbitrariness and ambiguity of the mental retardation label, particularly among those with mild forms of intellectual impairments. That mental retardation is less a fixed entity than evolving concept should not surprise anyone involved in the field. The definition and underlying premise of mental retardation has undergone numerous revisions over the past century. Definitions emerge from two basic conceptions of disablement: person-based or context-based. In the former, conventionally referred to as the medical model, the key diagnostic criteria are found in the characteristics of the person. One possesses the array of characteristics or does not. Context-based definitions emphasize environmental context in the manifestation of mental retardation (e.g., Mercer, 1973). In such social models diagnosis is conceptualized as a function of the interface of persons with their environment. In the 9th and 10th revisions of the American Association on Mental Retardation's definition (Luckasson et al., 1992; Luckasson et al., 2002), environmental context is given greater weight than in previous classification schemes. In addition to the traditional criteria of IQ and adaptive skill, determination of mental retardation requires identification of the kinds and amounts of environmental supports required across multiple dimensions. The intent is to employ language that emphasizes integrated and inclusive supports.
The approach introduces significant additional ambiguity over who has or does not have mental retardation. The ambiguity is not unintentional. To the extent that responsibility for disablement is posited in the context of the person's environment, function will ebb and flow in conjunction with the capacity of the environment to provide necessary supports. Similar conceptualizations underlie the World Health Organization's newly revised International Classification of Functioning, Disability and Health (World Health, 2001) and represent a general zeitgeist in contemporary approaches to the assessment of disability (Fujiura & Rutkowski-Kmitta, 2001).
These emerging paradigms of disablement raise interesting questions regarding the appropriate boundaries for constituencies served by mental retardation policy. What are the implications of a highly elastic demarcation for disablement? In their critique of the 1992 AAMR definition, MacMillan, Gresham, and Siperstein (1993) argued that a definition heavily dependent on adaptation to environments was inherently unreliable and would dramatically increase the numbers of persons labeled as having mental retardation through misdiagnosis (among other critiques). In his counterpoint, Reiss (1994) argued that the fundamental intent of the new definition was neither to increase nor decrease those labeled but, rather, “to change how people think about mental retardation” (p. 1). Obscured in the debates over the relative merits of psychometric versus context-based definitions was the possibility that a significant population of persons with mild intellectual impairments, whether labeled or not, would benefit from representation in mental retardation policies and practice. Although intelligence and adaptability may be manifested on a continuum, the provision of services and supports is very much a dichotomy, with categorical labels serving a gate-keeping function.
Implications for policy and practice reside at the core of the forgotten generation thesis. The present study is a modest effort to begin addressing these issues by better understanding the demographic features of persons at the margins of current conceptions of mental retardation. The research question addressed in the study is, To what degree do Americans with mild intellectual impairments share many of the functional and socioeconomic characteristics of those who are formally identified as having mental retardation? Implicit in this question is the assumption that the social and economic impact of intellectual impairments manifest themselves along a continuum and that many persons with a mild intellectual disability cannot be readily partitioned from those formally diagnosed as having mental retardation. I sought corroborative evidence for the thesis by directly testing for the presence of these population dynamics.
Participants were a cohort of Americans with mild intellectual impairments (who in past years were labeled as having “borderline” or “mild” mental retardation) and who no longer were formally identified as having mental retardation but who shared common characteristics due to a generalized impairment in intellectual functioning. The term mild intellectual disability is employed to describe this group. Because national surveys are not designed to partition groups across levels of intellectual impairment, considerable effort was devoted to use of indirect measures as evidence for group membership. The mild intellectual disabilities group was compared to persons formally labeled as having mental retardation, persons with learning disabilities, and the general population. Comparisons served two purposes: (a) to provide evidence for the validity of the identification procedure and (b) to characterize their status on basic social and economic indicators.
Source of Data
Data from the 1994 and 1995 National Health Interview Surveys on Disability—NHIS-DS (National Center for Health Statistics, 1998) were employed in the analysis. The NHIS-DS is the first national-level household survey specifically targeting persons with disabilities across all ages (Verbrugge, 1995). Of special relevance to the present analysis was the large array of survey questions focused on the characteristics of impairment, function, and access to disability-related services. Relatively sophisticated case identification strategies could be developed based on combinations of survey items (e.g., Larson et al., 2000). A second important feature of the NHIS-DS was the fielding of the disability supplement across 2 years of samples (1994 and 1995). The capacity to combine these two survey years into a substantially larger sample enabled more reliable population estimates, an important consideration when working with low prevalence conditions such as mental retardation.
General structure of the NHIS-DS data
The NHIS-DS consists of three major survey components: a “core” data set and two phases of supplemental disability data collection. The core section includes a set of basic health and demographic questions largely unchanged over the years (the core has been conducted annually since 1957 by the National Center of Health Statistics). The disability supplement involved two phases of surveying. In Phase I, a set of questions was administered in a face-to-face interview and used to screen those who were likely to have a disability. Screening items included queries common to most disability definitions: form of impairment; activities of daily living; instrumental activities of daily living; functional limitations; mental health status; services and benefits received; childhood education, health, and service characteristics. There were a total of 202,560 interviewees (107,496 for 1994; 95,091 for 1995).
Those individuals selected for the follow-up data collection (Phase II Survey) were persons who (a) had specific diagnoses or impairments (e.g., mental retardation), (b) were a recipient or applicant of publicly funded income support programs, (c) had functional limitations, or (d) used assistive technology. Phase II interviewing occurred 10 months to 2 years following the completion of the screener questionnaire. In this second round of follow-up interviewing, disability specific information was collected on (a) service utilization and unfulfilled need for service, (b) functional limitations, (c) childhood disability issues, (d) employment, (e) use of service and benefits, (f) transportation, (g) personal assistance, (h) housing, (i) environmental barriers, and (j) participation in social activities.
Broadly speaking, there were two distinct forms of data files associated with the Phase I and Phase II Surveys. Both were employed in the screening criterion for the current study. In the person file, individuals are the unit of sampling and impairment codes represent affirmative answers to multiple queries in the form of, “Do you have __________ or “What is the main condition that causes __________? ”
The other form of data file was the condition file. The impairment rather than person was the sampling unit. A respondent reporting multiple conditions (e.g., mental retardation and high blood pressure) would be represented by two records in the condition files. Up to 30 conditions may be reported per person. The significance of the condition file is that more detailed diagnostic information is available to the analyst. For example, an intellectual impairment may be present but not cited as a primary cause of a limitation in one of the 29 Disability Screens in Phase I.
Four groups were identified: (a) mental retardation, which included those persons labeled as having mental retardation; (b) mild intellectual disabilities (persons not labeled as having mental retardation but for whom evidence suggested a general learning impairment); (c) learning disabilities (persons having specific learning disabilities), and (d) U.S. nondisability, the population of Americans without any impairments or limitations. Respondents were identified as having mental retardation if (a) they replied that mental retardation was the primary cause of limitations in basic activities or for seeking various services and supports, (b) responded affirmatively to direct queries on whether the condition of mental retardation or the closely related conditions of Down syndrome, autism, or hydrocephaly were present, or (c) if mental retardation was cited as a condition present (via the condition file).
Those identified as having mild intellectual disabilities were persons reporting an activity limitation or need for formal programmatic supports due to (a) a generalized difficulty in learning or (b) a specific learning disability. The presence of a major life limitation or involvement in disability-related program supports was taken as evidence of more generalized deficits in learning rather than specific learning disabilities. The working assumption guiding the selection criterion was that mild and generalized intellectual impairments are often labeled by respondents as a “learning disability.” As Warren (2000) observed, the “learning disabilities” label is being increasingly adapted as an alternative descriptor or euphemism for mental retardation. Program supports cited by respondents varied by age group. For children and youth, participation in early intervention services, family support programs, development of Individual Education Plans (IEPs), or enrollment in special education classes was taken as additional evidence of generalized impairment beyond a specific learning disability. Among adults ages 18 years or older, program supports included enrollment in guardianship programs; supported, transitional, or sheltered employment programs; day services; or use of case management services. The intent was to look beyond the label by functionally imputing the impact of a learning impairment; however, the impairment was labeled (or not labeled) by the respondent.
In contrast, the learning disabilities group included only those persons identified as having learning disabilities but not reporting any significant impact on daily activities or need for program supports. The fourth group (nondisabled U.S. population) served as the reference group for comparisons and included all Americans who did not report any impairment or activity limitation. Thus, all other forms of disablement were excluded from the national reference group.
There were three major components to the analysis: (a) validation of the screening criteria, (b) derivation of population estimates for sociodemographic variables, and (c) estimation of service needs and service access.
Profiles of the groups were evaluated in terms of consistency with the attributions of mild intellectual disabilities and mental retardation in an effort to approximate a validation of the screening criteria. Support for the screening criteria would come in the form of evidence that the identified groups were similar and dissimilar in ways that would be expected consistent with the construct of intellectual performance. Comparisons were made on (a) activities of daily living or instrumental activities of daily living, (b) average number of activities of daily living or instrumental activities of daily living, and (c) types of community living supports needed. There were six activities of daily living (bathing, dressing, eating, mobility, toileting, and navigating the home environment) and six instrumental activities of daily living (preparing meals, shopping, money management, use of phone, heavy housework, and light housework) included in the comparisons. If a respondent was unable to do the activities of daily living or instrumental activities of daily living or required assistance in their completion, they were scored as having a limitation in that area. The average number of activities of daily living/instrumental activities of daily living could range from 0 to 6. In order to take into account developmental changes, I did not evaluate children under the age of 5 years on activities of daily living nor were those under the age of 18 years assessed on instrumental activities of daily living.
The community living support areas were based on the domains specified under the dimension of Intellectual and Adaptive Skills in the AAMR classification system: Communication, Self-Care, Home Living, Social Skills, Community Use, and Work (Luckasson et al., 2002). Survey questions on impairments, activities of daily living, instrumental activities of daily living, and other skill areas were used to assess the need for assistance in each domain: Communication (5 items on communication impairments), Self-Care (5 activities of daily living items), Community Use (5 instrumental activities of daily living on home living and community use and 2 items on communication impairments), Social Skills (2 items on friendship and socialization and 2 communication items), and Work (1 item on employment limitations and 2 items on communication impairments). For example, a person was deemed in need of supports in the area of Community Use if they (a) had difficulty communicating with persons outside of the family (a communication impairment), (b) had difficulty shopping or could only do so with supervision (instrumental activities of daily living), or (c) had difficulty getting around by self (instrumental activities of daily living) or through use of public transportation due to “cognitive problems” (instrumental activities of daily living items from Phase II follow-back).
Economic status and profile of services
A second set of analyses assessed whether there were differences across the cohorts on basic indices of social and economic status (SES) and in the profile of needed services and access to services. The forgotten generation thesis suggests that many persons, regardless of label, share similar needs due to their intellectual impairment. Thus, the analysis compared the degree of similarity between the mild intellectual disabilities cohort and persons with mental retardation on three global indicators of social status and economic functioning: (a) employment, (b) poverty, and (c) unmet service needs.
Respondents who were from 22 to 64 years of age were classified as employed if they were engaged in compensated work during the 2 weeks prior to the interview. Household poverty was based on the aggregate income of all members of the household. Aggregate rather than individual income is likely a better measure of economic status in the community because most persons with mental retardation live in family and other multi-person households (Fujiura, 1998).
The gap between service needs and access among respondents was imputed through a series of questions about support services that were part of the Phase II follow-up survey. The questions were generally in the form of “Did you receive __________ services in the past 12 months?” and “Did you need __________ service in the past 12 months?” Fourteen services were included in the areas of allied health therapies (audiological, counseling, occupational, physical, recreational, and speech), health services (nursing), and other supports (day care, independent living, personal care, residential, sign interpreter, social work, and transportation). Respondents were coded as having an unmet need if they cited a service that was needed but not received during the past 12 months. In addition, an unmet need was identified for any adult respondent who was unemployed and reported no participation in a listing of 25 different day-activity, vocationally related services, or employment options.
National estimates were derived using population weights associated with each household and individuals within the household in the annual samples. Weights, developed by the National Center for Health Statistics (NCHS), were adjusted on the basis of probability of selection, stratification across demographic cohorts, and nonresponse rates within cohorts (National Center, 1998). Variance estimation in standard statistical software typically underestimates sample variances for data based on complex sample designs such as the NHIS. Therefore, I constructed variance estimates using Sudaan 7.5 software (Shah, Barnwell, & Bieler, 1997), which includes adjustments for cluster-correlated data. These adjusted variance estimates were employed in developing confidence intervals for the core indices.
A 1.27% prevalence rate (standard error [SE] = .03) for the combined populations of mild intellectual disabilities and mental retardation was found. As expected under an expanded selection criterion, the 1.27% figure is considerably higher than the categorically defined mental retardation rates found in other analyses of national data sets: (a) 0.78% in Larson et al.'s (2000) analysis of the 1994–1995 NHIS-DS; (b) 0.60% in the U.S. Bureau of the Census 1992–1992 Survey of Income and Program Participation (McNeil, 1993); and (c) 0.54% in the 1992 NHIS Core data set (LaPlante & Carlson, 1996). The derived 1.27% prevalence rate is comparable to large-sample studies with investigators employing diagnostic assessments (versus self-report method employed in the large national surveys) to estimate the prevalence of both mild and severe mental retardation: 1.0% in Mercer (1973) and .90% in Boyle et al. (1996). McLaren and Bryson (1987) found estimated rates converging on 1.2% in total population screenings of both mild and severe mental retardation. The juxtaposition of rates from the present study, national surveys, and large-scale population screenings suggests potentially significant underreporting of intellectual disability in the major national data sets (NHIS and Survey of Income and Program Participation) through the exclusion of mild intellectual disabilities.
Validity of Screening Criteria
The summary of activities of daily living and instrumental activities of daily living limitations shown in Table 1 is congruent with expectations of increased functional limitations as one moves along the continuum of intellectual disability. Shown in the table are the percentages of activities of daily living/instrumental activities of daily living limitations and the average number of limitations per person across four groups: (a) those in the U.S. general population without a disability (defined as the absence of any limitation in activity), (b) those with learning disabilities, (c) persons identified in the screen as having a mild intellectual disability, and (d) persons with mental retardation. Because many activities are age-related, indices were compared across four different age cohorts: children and youths (21 years and younger), young adults (22 to 44 years), older adults (45 to 64 years), and seniors (65 years or older).
Within the group if individuals ages 21 years and younger, the maximum number of limitations was less than 6.0 because activities of daily living are not assessed for children less than 5 years, and children under 18 years are excluded from instrumental activities of daily living evaluations. Thus, the proper base of comparison is within age group.
The profile of community living supports across the four groups is graphically shown in Figure 1. The vertical lines represent 95% confidence intervals around the estimated percentage of each group requiring supports.
Shown in the figure are the proportions of each of the four groups requiring some form of support in each of the six domains. Data were consistent with the working hypothesis of a larger intellectual disabilities cohort bound by need rather than diagnostic classification: The mild intellectual disabilities and mental retardation groups were very similar in degree of need and higher than the other groups. With the exception of community use, the differences between mild intellectual disabilities and mental retardation were not statistically significant.
Economic Status and Profile of Services
The analysis of unmet needs among the cohorts suggests a lack of access to services most pronounced among those with an intellectual disability. Proportions of respondents with an unmet need in the clinical and support services reviewed were nearly identical among persons with mental retardation and mild intellectual disabilities (46.8% and 46.4%, respectively), but lower among persons with learning disabilities (34.3%). Data should be interpreted cautiously; however; rates of nonresponse on the Phase II service questions were extraordinarily high, ranging from 54% among persons with learning disabilities to 82.7% for persons with mental retardation. The high rate of nonresponding introduces substantial biases of indeterminate type. Relatedly, unmet needs for specific services could not be evaluated due to small sample sizes.
Shown in Table 2 are the two status variables used as proxies for economic vulnerability: unemployment and household poverty. Again, findings were consistent with the forgotten generation thesis. The table reflects the substantial gulf in economic well-being between Americans with mild intellectual disabilities and mental retardation and the U.S. population as a whole.
If the general proposition of the forgotten generation is correct, then there should be a sizable population of Americans with the characteristics of mild intellectual disabilities that is clearly distinguishable from the general population of the United States as well as those with specific learning disabilities. Study findings were consistent with the proposition. The profile of skills and service needs among persons identified as having mild intellectual disabilities and mental retardation was generally consistent with a continuum of functioning ranging from mild to severe mental retardation. There were substantial numbers of Americans at the margins of current diagnostic conceptions of mental retardation who have substantial commonality with those formally labeled with mental retardation in terms of their social and economic vulnerability.
Beyond the general contours of function, support needs, and SES presented here, it is difficult to specify with any degree of certainty who is represented in this cohort. Some may be representative of those Edgerton first identified as “passing” in his longitudinal studies of former institutional residents (e.g., Edgerton & Bercovici, 1976). Others may represent the very vulnerable of our nation who live their lives with only sporadic assistance or largely disconnected from the formal mental retardation/developmental disabilities services infrastructure. A third possibility are those persons who knowingly denied a formal diagnosis of mental retardation. In her analysis of welfare reform policy, for example, Boggs (1994) suspected that substantial numbers of persons with mild or even moderate mental retardation (or their proxy respondents) attributed limitations to other conditions in order to conceal the fact of an intellectual disability (pp. 83–85). Reliably distinguishing these and other possible subgroups within the mild intellectual disabilities cohort is well beyond the capabilities of national level survey data. Whether through concealment, erroneous attribution, or some combination, self- or proxy-based reporting very likely results in the routine underreporting of recognized diagnoses of mental retardation in survey data sets. This is an important limitation because data were based on self- or respondent report, and the degree of validity in diagnostic identification is unknown. Thus, it would be wrong to assume all members of the mild intellectual disabilities cohort represented a homogeneous grouping whose impairments were “hidden” or forgotten.
Nonetheless, prevalence estimates for mild intellectual disabilities and mental retardation in this analysis paralleled those derived from large population screenings in which some form of diagnostic testing was employed for case identification (Boyle et al., 1996; McLaren & Bryson, 1987; Mercer, 1973). This juxtaposition is strongly suggestive of a conservative bias in the estimates of the population of persons affected by an intellectual impairment when only mental retardation is considered.
Is the more expansive criterion the better representation of the scope of intellectual disability in the population? A disability studies perspective is assumed here; the term mental retardation is treated as a dynamic construct with multiple possible operationalizations rather than as a static population with fixed boundaries and features. As argued elsewhere in the literature on statistical treatments of disability definition (Fujiura & Rutkowski-Kmitta, 2001; Zola, 1993), rather than pursuing an illusory best value, disability statistics are better represented as different aggregations with greater or lesser relevance under different conditions. From this perspective one would argue that the search for a single “true” number of persons within our statistical systems using ever more refined criteria is a largely futile effort. An intellectual disability is whatever one decides it should be for the particular issue at hand. Definition should be guided by context. Such an approach is consistent with recent trends in the use of multidimensional or multiperspective approaches in identifying who has mental retardation and disability generally (Altman, 2001).
Study results raise questions regarding what constituencies and which issues should be targeted by a “mental retardation” policy agenda. The lack of attention to persons with mild intellectual disabilities is problematic for a society imposing increasingly more complex demands upon its members. Do our notions of intellectual disability, however defined, serve to exclude large numbers of Americans from potentially beneficial advocacy and programmatic support? The issue of access is further complicated by one of the central findings of the present analysis—the extraordinarily high rates of poverty among the mild intellectual disabilities constituency. There are few if any domains of care or support for which access is not further compromised by the status of being poor: welfare and employment, housing, and health care, among others. Over 3 decades ago, Hurley (1969) wrote of mental retardation and poverty being, “largely relegated to regions beyond the American conscious” (p. 3). The attribution remains relevant today. Poverty as a topic is something of a “white noise” for our field—acknowledged, but largely in the background of our deliberations. The present study is presented not as an argument on behalf of a re-labeling movement but rather as an effort to elevate awareness of individuals not currently supported by professionals in the field who, although qualitatively different than our contemporary conceptions of people with mental retardation, may share considerably in need for supports, vulnerability, and exposure to poverty. Differences exist across a continuum and the demarcation between the status of disability and nondisability is externally imposed. As our classification schemes continue to evolve in the ongoing reconsideration of the meaning of mental retardation, we must remain aware of those constituencies at the margins of our conceptions of disablement.
Preparation of this article was supported, in part, by Grant No. H133B980046 (Rehabilitation Research and Training Center) on Aging and Mental Retardation) and from H133A990017 (Center on Emergent Disability) from the National Institute on Disability and Rehabilitation Research.
Author: Glenn T. Fujiura, PhD, Associate Professor of Human Development, Department of Disability and Human Development (M/C 626), University of Illinois at Chicago, 1640 W. Roosevelt Rd., Chicago, IL 60608. firstname.lastname@example.org