To begin to understand retirement, we examined age-related differences in (a) employment rates, employment hours, and rates of individual-plan employment goals; and (b) participation rates in unpaid formal day programs. We report weighted analyses of 2014-15 National Core Indicators data from 32 states. Unlike younger age groups, a similar proportion of workers with intellectual and developmental disabilities continued working beyond age 65 as for the general community. We found that most workers with intellectual and developmental disabilities retire in older age and that their retirement appears to be sudden, rather than a gradual reduction in work hours. Facility-based day programs were the only program with an increased participation rate in older age groups, revealing an even greater reliance on facility-based services for older participants.
More people with intellectual and developmental disabilities (IDD) are living into older age compared to past generations. The average life expectancy for a person with IDD is 66 years (Coppus, 2013). The number of people with IDD in the United States over the age of 60 is growing and is expected to double from 641,860 in 2000 to 1.2 million around 2030 (Heller, 2010), meaning that examination of aging and retirement are increasingly needed. Retirement was recently highlighted as one of four U.S. national goals for research, practice, and policy concerning aging of people with IDD (Hahn et al., 2016).
We define retirement as “withdrawal from paid work” (Stancliffe, Brotherton, O'Loughlin, & Wilson, 2017). Withdrawal from unpaid activities, such as attending a day program, is not within this definition. Retirement differs from unemployment, “seeking work but temporarily not being in paid employment” (Stancliffe et al., 2017), because retirement is usually permanent and the person is no longer seeking paid work. Thus, an individual with IDD who is currently not working but is seeking a job would be considered unemployed rather than retired. We recognize that many people with IDD never participate in paid employment, therefore, this definition of retirement does not apply to them. However, this definition accords with the common understanding of retirement, and enables comparison with employment and retirement in the general community. The emphasis on increasing competitive integrated employment in recent policies and initiatives, including the Workforce Innovation and Opportunity Act (2014) and Employment First state-level policies (Gunty, Dean, Nord, Hoff, & Nye-Lengerman, 2017), provides a further rationale to proceed in this way.
Retirement in the General Community
In the general population, retirement has received considerable research attention (Wang & Shultz, 2010). Retirement age in Western countries with developed systems of retirement income and social welfare is often related to the age of eligibility for financial support. In the United States, full Social Security retirement benefits can be accessed between 66 and 67, depending on when a person was born (Social Security Administration, 2017). Medicare eligibility begins at 65 (Social Security Administration, 2017). The average retirement age in the United States is around 62 (U.S. Census Bureau, 2014). Although retirement is commonly a phenomenon of older age, retirement can occur at any age (Bureau of Labor Statistics, 2015). Some retire by choice, or are pushed into early retirement by ill health, caregiving responsibilities, or redundancy. Others work well beyond the pensionable age by choice or necessity (Coile, Milligan, & Wise, 2016).
Planning for retirement is an important issue in the general community. Financial security in retirement is a key consideration. Planning encompasses retirement lifestyle, health, and adjustment to change, with poorer outcomes typically associated with little or no planning (Wang & Shultz, 2010). Voluntariness of retirement, planning for retirement, and engaging in a gradual transition to full retirement through bridge employment have all been associated with positive outcomes in retirement (Wang & Shultz, 2010). In contrast with the general community, retirement among people with IDD may be less financially driven, likely due to the complex interplay of accessing employment and benefits (Hahn, Fox, & Janicki, 2015). Research in Australia supports this assertion. Views of people with IDD about retirement are characterized by the near absence of comments about finances or retirement savings (Bigby, Wilson, Balandin, & Stancliffe, 2011; McDermott & Edwards, 2012).
Employment and People With IDD
Given our definition of retirement as withdrawal from paid work, retirement must be understood in the United States context of employment of people with IDD. Adults with IDD participate in community- and facility-based settings, both paid and unpaid. However, despite policy initiatives to support people with IDD to access community employment, rates of community employment for people with IDD have actually decreased in recent decades from 24.2% in 2001 to 19.1% in 2014 (Butterworth et al., 2016). Participation rates in community- and facility-based unpaid activities have grown, although it is important to note that some people participate in both paid and unpaid activities.
According to National Core Indicators Adult Consumer Survey (NCI-ACS) data collected from 12,720 adult service users in 26 states in 2012-13, many working-age people with IDD participated in an unpaid facility-based (44.3%) or unpaid community-based activity (22.6%; Butterworth, Hiersteiner, Engler, Bershadsky, & Bradley, 2015). Another 27.0% took part in a paid facility-based job (previously known as sheltered employment). Only 14.7% undertook paid community employment.
Butterworth et al. (2015, p. 215) also reported that there were marked between-state differences in employment, with the rate of community-based employment ranging “from only 0.9% in Alabama to 38.1% in Connecticut.” Such differences have important implications for data analysis in the current study because we also use NCI-ACS data. Unweighted analyses will overestimate overall employment rates if high-employment states are overrepresented in the current 2014-15 NCI-ACS sample, and vice versa. Therefore, as is explained further in the Method section, we used state weights in all our analyses of NCI-ACS data to correct for unrepresentative state sample sizes.
The distinction between facility- and community-based employment is important for both policy and outcomes. At a policy level, full community participation, including community employment, is prioritized under U.S. disability policy (Centers for Medicare and Medicaid Services, 2011; 2014). In terms of outcomes, community employment results in higher wages and greater social inclusion (Butterworth et al., 2015). Given the interplay between employment, social inclusion, and financial preparation, differences between facility- and community-based employment may also be important when examining retirement.
Retirement and People With IDD
Research on retirement among people with IDD in the United States is scarce. We know of no robust research in the United States or elsewhere that documents the age of retirement by workers with IDD. Fesko, Hall, Quinlan, and Jockell (2012) identified three ways that people view retirement: (a) they do not want to retire, (b) they look forward to retiring, or (c) they want to gradually transition to retirement by reducing work hours over time. Consideration for the person's desire and choice is reflected in each of these approaches, although it is unclear the extent to which people with IDD choose when or how they retire. Australian research indicated that people in sheltered employment have no real choice about retirement (McDermott & Edwards, 2012). Retirement research in the general community shows that exercising control over manner and timing of retirement is associated with greater well-being in retirement (de Vaus, Wells, Kendig, & Quine, 2007). Information about individual plan goals related to employment or retirement could be a fruitful source of data about retirement. For example, a person who is currently working and has a work goal presumably does not plan to retire soon. Further, an individual who is not currently employed but has an employment goal should be considered unemployed rather than retired. Some research on individual plan employment goals is available (Butterworth et al., 2015), but we know of no analysis of them in the context of retirement.
Whether retirement happens suddenly or by gradually reducing work hours over time is another important consideration, given the ways that people with IDD regard their employment. Employment is a means whereby people access the community (Nord et al., 2015; Nye-Lengerman, Pettingell, Nord, & Hewitt, 2018), form social connections (Timmons, Hall, Bose, Wolf, & Windsor, 2011), and experience a sense of personal success and of playing a meaningful role in the community (Lysaght, Cobigo, & Hamilton, 2012). People with IDD express concern that they will lose access to meaningful activity (i.e., work) and social connections if they retire (Bigby et al., 2011; McDermott & Edwards, 2012). Facilitating retirement planning and authority in choosing when and how to retire accords with active aging initiatives for aging people with IDD to continue being physically, socially, and mentally active in their community (Hahn et al., 2016). A gradual transition can enable the person to develop a retirement lifestyle over time, well before ceasing work entirely (Stancliffe, Bigby, Balandin, Wilson, & Craig, 2015).
In the current study, we use weighted analyses of NCI-ACS 2014-15 data on employment, unpaid day activities, and individual plan employment goals to describe age-related trends and to answer the research questions. Consequently, our data are limited to questions contained in the NCI-ACS. The NCI-ACS data are cross-sectional and do not contain any items specifically about retirement, so we have no direct data on the age of retirement by people with IDD, or the circumstances under which they retire. Instead, we infer likely trends about retirement from age-related differences in rates of employment, or rates of individual plan goals about employment. The specific research questions addressed were:
How does the rate of paid employment among people with IDD compare to the general population by age group?
What is the participation rate of people with IDD by age group in different kinds of paid employment settings and unpaid activity settings?
Does the percentage of people with IDD with a community employment goal in their individual plan differ by age group or current employment status?
Are there differences by age group in the number of work hours and unpaid activity hours that people with IDD participate in that suggest whether retirement is sudden or gradual?
This study utilized data from the 2014-15 National Core Indicators (NCI) Project. NCI is a collaboration between the National Association of State Directors of Developmental Disabilities Services (NASDDDS), the Human Services Research Institute (HSRI), and state developmental disability agencies in participating states. NCI is intended to provide reliable and valid information about service users with IDD to improve the service system for people with IDD and their families (Bradley & Moseley, 2007). Data utilized in this study were collected in the NCI-ACS, which is administered in a face-to-face interview.
The NCI-ACS surveys adult users of publicly funded IDD services who receive case management and at least one other IDD service. The full NCI-ACS 2014-15 sample consisted of 25,820 adults with age data. Of these, 416 participants aged 18-19 years were omitted to enable accurate comparisons with Bureau of Labor Statistics (BLS) data for the general community in research question one. The 1,018 participants enrolled in public school were also omitted because they were not available for full-time work. Due to the use of state weighting, we excluded all data (N = 319) from the sole substate entity, the Mid-East Ohio Regional Council (MEORC), because we had no means of calculating an appropriate weight for a subsection of a state such as MEORC. Finally, participants were omitted with missing data on employment and other daily activities. The final sample was N = 21,481 from 32 states (AR, CA, CO, CT, DC, DE, FL, GA, HI, IL, IN, KS, KY, LA, ME, MI, MN, MO, NC, NH, NJ, NY, OH, OK, PA, SC, SD, TN, TX, UT, VA, VT). Participants' age averaged 42.73 years (SD = 14.49, range 20-97). Research question one utilized Bureau of Labor Statistics data for 2015 for the civilian noninstitutional population of the United States. (Bureau of Labor Statistics, 2015).
State sample sizes ranged from 195 (DC, 0.91% of the NCI-ACS sample) to 7,425 (CA, 34.57% of the NCI-ACS sample). The CA sample was overrepresented because, in 2014, adult service users from CA made up only 13.94% of the total number of adult service users across the 32 states (Larson et al., 2017). Given the disproportionately high numbers from some states (e.g., CA, DE, HI, NH, VT) and unrepresentatively low numbers from other states based on their size (e.g., IL, MI, NY), we used state weights in our analyses so our overall findings more accurately reflected the situation for the population of service users across the 32 states. The effect of using state weights was to compensate mathematically for the over- or under-representation of the NCI-ACS sample from individual states, relative to each state's proportional representation among all adult service users from these 32 states. Use of state weights has no effect on the overall number of participants; it simply corrects the bias caused by unrepresentative state sample sizes in some states in the NCI-ACS.
Weights were calculated using data for the same 32 states from the 2014 Residential Information Systems Project (RISP; Larson et al., 2017). We used RISP data that showed the total number of adults (aged 22+) by state who used the major federal funding programs—home and community-based services (HCBS; Table 2.2) and intermediate care facilities (ICF; Table 2.6). First, each state's proportion of the respective 32-state total sample was calculated separately for the NCI-ACS and RISP samples. Next, weights were calculated by dividing the RISP proportion by the NCI-ACS proportion for each state. For example, the weight applied to participants from CA was 0.403 (i.e., 0.1394/0.3457) to compensate for their over-representation in the NCI-ACS sample. Weights ranged from 0.233 (DE) to 5.575 (NY). The greater than twentyfold range in weights demonstrated clearly the need for state weighting. Analyses using these weights are labeled as weighted analyses.
The NCI Adult Consumer Survey 2014-15 version was used for data collection. The current study uses data only from the NCI-ACS Background Information section, which asks about the service user's personal characteristics, functional capacities, health issues, diagnoses, residential status, service usage, employment, and daily activities. These data are typically provided by case managers based on individual client records, although sometimes direct support providers, family members, or individual service recipients provide these data. Mandatory training for interviewers incorporates manuals and videos, scripts for scheduling interviews, presentation slides, a review of the survey instrument, picture response formats, and a list of frequently asked questions. Information on the psychometric properties of the NCI-ACS can be found in National Association of State Directors of Developmental Disabilities Services and Human Services Research Institute (NASDDDS and HSRI; 2012).
NCI data on employment and day activities
The following analyses examine the participation rates and hours of participation in paid work and unpaid day activities by age group. The NCI-ACS gathers data on four types of day activities: (a) paid community work, (b) unpaid community activities, (c) paid facility-based work, and (d) unpaid facility-based activities. A community setting is defined as “a place where most people do not have disabilities.” A facility-based setting is “a place where most people do have disabilities” (NASDDDS & HSRI, 2014, p. 19). The NCI-ACS data provide a snapshot for the most recent 2-week period of participation and hours in the four types of activities listed. Data from the NCI-ACS item, “Is community employment a goal in this person's service plan?” was also analyzed. The NCI-ACS does not specifically ask questions on retirement.
Participants were grouped by age category to mirror the Bureau of Labor Statistics (BLS) dataset. In most analyses, these were 10-year age groupings (e.g., 35-44), but in some cases 5-year age groupings were used (e.g., 35-39, 40-44). Analyses consisted of simple descriptive data of the percentage of participants within each age group that took part in various forms of work or day activities, their hours of participation, and the percentage with a community employment goal. Chi-square analyses with Bonferroni pairwise comparisons were used to identify significant differences in participation rate by age group. ANOVA was used to compare work hours across age groups.
Comparing Employment With the General Population
Research question one
This analysis compared the percentage participation rate in paid work by age for people with IDD versus the general population. The general population data were collected using a representative sample (Bureau of Labor Statistics, 2015), so no state weighting was needed for these data.
Figure 1 shows dramatically higher levels of participation in paid employment among the general U.S. community compared to adults with IDD in all age groups except 65 and over. Both groups display a clear drop in the employment participation rate for older workers, but the decline in participation in employment is far steeper in the general community group. A notable percentage of both groups aged 65 or more were in paid employment. Some people in both groups worked beyond age 75. Unlike any other age group examined, the participation rate in employment is similar for both groups over age 65 (18.9% for the general community vs. 18.8% for people with IDD).
Overall, weighted analysis showed that 7,373 (34.3%) of the NCI sample were in paid work (community- or facility-based employment, or both). Figure 1 shows the weighted NCI participation rate by 5-year age groups. In any age group for the NCI sample, participation rates did not exceed 41%. Chi-square analysis showed a significant difference in participation rate by age group within the NCI sample, χ2 (11, N = 21,481) = 339.87, p < .001. The oldest age groups had a significantly lower employment participation rate than younger age groups.
Participation Rate of People With IDD by Age in Kinds of Paid and Unpaid Settings
The remaining weighted analyses focus solely on adults with IDD from the NCI sample of adults aged 20 or older (N = 21,481).
Research question two
Table 1 shows a weighted analysis of the number of participants who accessed the different types of employment and day activities in the most recently available 2-week period. Most individuals (53.0%) took part in one type of day activity, but 4,405 people (20.5% of the sample) participated in more than one type. Another 26.5% was involved in none.
Age-related patterns of participation in community- and facility-based employment were examined separately, as well as unpaid day activities in each type of setting. The rationale for including unpaid activities in an article on retirement was to identify the activities people undertake in older age. Figure 2 shows weighted participation rates by age group in four forms of employment and day activities.
These analyses show a marked decline in the percentage of people participating in paid work in older age groups. The percentage in employment differed significantly by age group, for both community-based employment, χ2 (5, N = 21,481) = 338.99, p < .001, and facility-based employment, χ2 (5, N = 21,481) = 157.17, p < .001. The percentage in paid community employment was significantly lower for the 55-64 age group than for all the younger groups, and significantly lower again for the 65+ age group. For facility-based employment, the age-related decline in participation rate is not as marked, with the 55-64 age group not differing significantly from most younger age groups. Participation for the 65+ group was significantly lower than for several younger age groups (35-44, 45-54, 55-64), but did not differ significantly from others (20-24, 25-34).
In relation to unpaid activities, there are contrasting age-related trends for community- versus facility-based activities. The weighted percentage in unpaid community activities declines with age, whereas there is a strong increase with age in participation in unpaid facility activities. For community activities, a significantly smaller percentage of the 65+ group participated than for all younger age groups. By contrast, for facility-based activities, the groups aged 45 or older (45-54, 55-64, 65+) all had a significantly larger percentage participating than for any of the younger groups (20-24, 25-34, 35-44). The percentage involved in no formal out-of-home activities was also significantly higher for the 65+ age group than all younger age groups except 20-24.
Community Employment Goals and Individual Plans
Research question three
We examined the percentage of people who had a community employment goal by current employment status (community-based employment, facility-based employment, no employment). Of the 20,662 participants with valid data about having such a goal, 402 worked in both community- and facility-based employment. To avoid double counting, these 402 individuals were classified into the community group. Figure 3 shows the percentage of participants with a goal by age group and employment status.
Overall, 5,122 (24.8%) participants had a community employment goal. There were marked differences by employment status in presence such a goal, ranging from 74.2% of those employed in community-based work, to 27.0% of individuals in facility-based work, and 14.0% of those with no paid work.
All three employment groups had significant differences by age group in the percentage with a community employment goal. In each group, the percentage fell with increasing age.
Weighted chi square analyses for people in community work showed a significant decline in the percentage with a community employment goal with increasing age, χ2 (11, N = 2,805) = 81.36, p < .001. However, the magnitude of the decline from the 20-24 age group (87.5%) to the 75+ group (40.0%) was only moderate. Even in the oldest age groups, a substantial percentage of participants had a goal, likely indicating that they would continue to work and were not planning to retire.
Facility-based work and no work
The age-related trend was similar for individuals in facility-based work, χ2 (11, N = 4,196) = 179.24, p < .001, and no work, χ2 (11, N = 13,620) = 531.78, p < .001. The widespread absence of community employment goals for participants in these groups does not necessarily indicate a plan not to work in the future. They may plan for facility-based employment, but the NCI-ACS does not gather data on individual plan goals for such employment.
Differences in Number of Work Hours and Kind of Activity
Research question four
This examination of age-related differences in work hours was intended to explore whether retirement is sudden or gradual. A pattern of sudden retirement should see an age-related decline in the percentage of people in employment, without a marked age-related change in the hours of work for those who remain employed. Gradual retirement may be characterized by people remaining employed but reducing their work hours over time, as shown by shorter average work hours in older groups who are still working.
Work and activity hours per 2 weeks—All participants
Data on work and activity hours were missing for 1,703 participants, reducing the sample to 19,778 for these analyses. Figure 4 shows stacked bars with weighted average hours of paid work and unpaid activities per 2 weeks by age group, distinguishing between community- and facility-based work and activities. Figure 4 shows that the total number of hours of work and activities peaks in the 45-54 age group and declines slightly from there. However, average hours are affected by participation rates. Activities with lower participation rates (e.g., community employment) have higher numbers of people with zero hours, which reduces the overall average. As Figure 2 showed, unpaid facility-based activities had the highest participation rate and paid community employment had the lowest in each age group, so it is not surprising that the mean hours of participation in these activities differed in a similar way.
We compared participation hours in each type of work and activity, using ANOVA and post-hoc pairwise Bonferroni comparisons across age groups.
Total hours of activity
The combined total hours of paid and unpaid activities (shown by the top of the stacked bars) differed significantly by age group, F(5,19772) = 23.60, p < .001, with the 65+ age group participating for significantly fewer hours than the 35-44, 45-54 and 55-64 age groups.
Paid community work
There was a significant difference in hours of work by age group, F(5, 21156) = 43.89, p < .001. The 65+ age group participated in significantly fewer hours of community work than all other age groups (p < .001).
Paid facility work
Significant differences were also evident among age groups in hours of facility work, F(5, 20945) = 19.30, p < .001. The 65+ age group participated in significantly fewer hours than the 35-44 and 45-54 age groups (p = .009, p < .001 respectively).
Unpaid community activity
There was a significant difference between age groups, F(5, 20917) = 6.02, p < .001. The 65+ age group participated in significantly fewer hours of unpaid community activity than the 20-24 and 25-34 age groups (p < .001).
Unpaid facility activity
Contrary to other activities, there was a modest trend toward increased average hours of unpaid facility activity in older age groups, F(5, 20681) = 27.88, p < .001. The 65+ age group participated in significantly more hours of facility activity than the 20-24, 25-34, and 35-44 age groups (p < .001).
Average hours of participation in a given type of activity by age group were strongly affected by the percentage participation rate in that activity by age group. The decline in average total hours of activity in older age appeared to be driven by drops in the participation rate in paid work (especially community-based) and the consequent fall in average hours. These effects were strongest for the 65+ age group. By contrast, an increase in the participation rate in unpaid facility-based activities in older age groups was related to the age-related increase in average hours of participation in unpaid facility activities reported above.
Work and activity hours per 2 weeks—Current active participants only
Participation hours in the four types of activity by age group were re-examined, but each analysis was restricted to individuals who were currently taking part in that type of activity. This approach enabled us to compare mean participation hours by age group independent of the participation rate to determine whether there was a true drop in total participation hours in older age groups, or whether the decline shown in Figure 4 was due to the increased percentage of nonparticipants in older age groups. These results allowed us to infer whether people gradually reduced work hours as they aged, or if they suddenly withdrew completely from work. The total hours by age group in these revised analyses are shown in Figure 5 with hours for each type of employment and activity program.
Analyses of the combined total hours of paid and unpaid activities by age group indicated that the 65+ group did not differ significantly from any other group, other than being higher than the 20-24 group (p = .051). The significant decline in mean total hours of total activities per week shown in Figure 4 was due to a higher percentage of people in older age groups with no activities at all (i.e., a lower participation rate), rather than a reduction in hours among those who continued to take part in one or more types of work or day activity.
ANOVA comparison of work hours by age group participating in community employment (n = 2,627) showed that, overall, there was no significant difference between groups, F(5, 2620) = 1.46, p = .20. Similarly, for facility-based employment (n = 4,317), there were no significant age-group differences, F(5, 4312) = 1.47, p = .20, as was also the case for unpaid community activities (n = 3,402), F(5, 3397) = 1.31, p = .26. Age-group differences were significant for unpaid facility-based activities (n = 8,385), F(5, 8380) = 4.24, p = .001, but this was due solely to the 20-24 year age group having fewer total hours of unpaid facility-based activities than several older age groups. The 65+ group did not differ significantly from any other group. These findings suggest that participation hours were much the same across age groups for active participants, a pattern more consistent with sudden rather than gradual retirement from either community-based or facility-based employment.
To improve the understanding of retirement for older adults with IDD, this study provided initial information about age-related trends in participation in formal paid work and unpaid activity programs, as well as plans for community employment. Retirement was defined as withdrawal from paid work. Because the NCI-ACS does not provide direct data on the age or manner of retirement, our approach was to compare participation in employment and formal day activities by age group to understand retirement in the context of employment across the adult lifespan. All analyses of NCI-ACS data for adults with IDD were weighted to compensate for unrepresentative NCI state sample sizes.
Before discussing our results, we remind readers that (a) more than one-fourth (26.5%) of NCI participants took part in no formal work or day programs, and (b) one-fifth (20.5%) participated in more than one type of work or day program (Table 1). In the discussion that follows, for clarity, we will generally focus on the different types of work or day programs as if they were mutually exclusive. We recognize, however, that (a) some individuals both work and attend a day program, which means that, when they retire from paid work, they likely will continue to attend a familiar day program; and (b) retirement and transition to new activities in retirement simply do not apply to people who participated in no formal work.
Comparison With the General Community
Far fewer adults with IDD worked, compared to the general community (Figure 1). Both groups showed a steady and continuing decline in employment rate from about middle age, consistent with a normative expectation that many people will retire in older age. The magnitude of change was greater in the general community because of that group's higher percentage participation rate. Some individuals in both groups continued to work well beyond age 65. In the oldest age groups (70-74, 75+), the difference in percentage participation rate disappeared between workers with IDD and those in the general community (Figure 1).
Retirement policy for people with IDD should provide the opportunity to retire or to continue to work for as long as the person wishes to and is able. We presume that those over 70 who currently worked also worked when younger. NCI-ACS data are cross-sectional, so to begin to understand the proportion of people who had ever worked that continued in paid work in old age, we could only compare across age cohorts. Expressed as a proportion of the highest participation rate (40-44 in the general community, 45-49 for adults with IDD), a much higher proportion of people with IDD (almost one-fifth) were still working beyond age 74 than for the general community (one-tenth). Some people with IDD in the oldest age groups may make different decisions from their general-community peers about continuing to work. Future research should examine factors associated with the decision to keep working among older people with IDD. One important issue is whether this decision is self-determined or made by others (see McDermott & Edwards, 2012).
Trends Across Age Groups for People With IDD
Because there is no equivalent to facility-based employment among the general community, comparisons with the general community involved combining community- and facility-based employment for people with IDD (Figure 1). However, for comparisons by age-group among people with IDD, it was important for policy and practice to analyze these forms of employment separately. Similarly, separate age-group comparisons of participation in community- and facility-based day programs enabled identification of what unpaid day programs people participate in at different ages. It was also of interest to know whether people who retire participate in day programs or have no formal day activity.
Community inclusion for older adults and disability policy
The majority of older adults in this study received services in facility-based day and employment settings, not community-based settings (see Butterworth et al., 2015; 2016; Nye-Lengerman et al., 2018). This highlights the IDD system's dependence on facility-based services. This robust facility-based infrastructure likely affects the types of settings and activities available to older adults with IDD. Just as efforts to improve full inclusion in the workforce gain momentum, so too should those that ensure that retirement services offer choice and inclusion.
Recent policy directives from the Centers for Medicare and Medicaid have focused on transforming HCBS to ensure that IDD services and supports, including those for older adults, are provided in integrated, community-based settings. Although states have until 2022 to come into compliance with the final rule on HCBS, it provides the system an opportunity to further transform and maximize inclusion (Centers for Medicare & Medicaid, 2011; 2014).
Retirement: Sudden or gradual
Fesko et al. (2012) noted that some people with IDD want to transition to retirement gradually by reducing their work hours over time. If this approach were common, we should have seen lower average work hours in older age groups, relative to their younger peers. Figures 1 and 2 showed that the participation rate falls substantially in the older age groups as more people retire. Average work hours will, therefore, be lower in the older age groups simply because more people are not working (Figure 4). Valid comparisons across age groups require analyses of work hours only among individuals who are still currently employed. Using that approach we found that, for both community- and facility-based employment, workers in older age groups worked a similar number of hours to their younger peers (Figure 5).
Taken together, our findings of lower employment participation rates in older age groups, but similar work hours for those who are still employed, suggest that retirement for people with IDD mostly occurs suddenly. If some people gradually reduced their work hours before retirement, there were too few to affect the overall result. Our cross-sectional data do not permit us to estimate how many participants retired gradually. Longitudinal studies of work hours and retirement are needed to identify what proportion of people experience a gradual retirement. This is a markedly different experience from older Americans without disabilities, where retirement is often gradual (Cahill, Giandrea, & Quinn, 2006).
We found that part-time work predominates for workers with IDD, with average work hours less than 35 hours per 2 weeks (Figure 5; Butterworth et al., 2015). Therefore, older workers have time available for nonwork activities and could be supported to begin to develop a retirement lifestyle (e.g., volunteering, leisure, and social activities) while still working, without disrupting working hours. This approach provides another pathway toward gradual retirement.
Individual plan community employment goals
The employment goal data utilized here may not provide a comprehensive picture of a person's employment goals. NCI-ACS data were available for community employment goals, but not facility-based employment goals. Secondly, individual plan goals reflect the collective decision of the planning team, not only the aspirations of the person with IDD. Nevertheless, the available data provided useful information about future community employment, and the employment data reflected the current situation. Thus, an individual who currently works in community employment but has no community employment goal may well be planning to retire from community employment.
All groups showed a significant age-related drop in the percentage with a community employment goal (Figure 3). For the community employment group, this finding is consistent with more people retiring from such employment in older age groups. One interpretation of this consistent trend is that sources of satisfaction and role expectations change as people enter middle and older age. Ra and Kim (2016) found that the quality of life (QOL) of individuals with disabilities was more stable for those in their 50s, 60s, or 70s, and that their QOL was not necessarily tied to their employment status. Although employment is often an indicator of QOL and satisfaction, it may not play as significant a role later in life. Wik and Tøssebro (2014) found that older people with disabilities in Norway were substantially less likely than their younger counterparts to consider that paid work was possible. Further, if thought possible, older participants were significantly less likely to want to work. Wik and Tøssebro argued that employment is a normative role expectation for young and middle-aged adults and an important part of their identity, whereas, for older people, the social role of retiree is acceptable.
Despite the significant age-related decline in the percentage with a community employment goal among the community employment group, 40% or more had such a goal in every age group, including the oldest groups (Figure 3). There was a clear trend toward planning to work well beyond the nominal retirement age of 65. Future research should investigate whether there is something about community employment that motivates people with IDD to continue to work into old age.
A striking feature of the community employment goal data was the vastly higher percentage of those currently working in community employment with such a goal, compared to other groups (Figure 3). The former group likely had mainly positive experiences of community employment and, thus, most planned to continue. Those with only facility-based work or no work may not have had such experiences. In addition, the community employment group likely had milder levels of ID and fewer co-occurring impairments (but we did not directly assess this issue), and may have been seen as more capable of undertaking community employment.
The strong association between community employment and having a community employment goal underlines the likely importance of such goals. Having a goal and receiving support consistent with it may well be a useful means of increasing community employment among people with IDD.
What do people with IDD retire to?
Alongside an age-related fall in the participation rate in employment, we found that the participation rate in unpaid community activities also fell in older age-groups, whereas there was an increase with age in participation in unpaid facility-based activities (Figure 2) to just over 50% in the 65+ age group. Relative to the peak employment years of middle age, there was a significant increase in older age groups of those not participating in any formal employment or day activity program. Overall, these findings suggest that, in older age, most people who have previously worked stop working. In retirement, they may take up an unpaid formal day activity (typically a facility-based day program) or may have no formal day activity. Given that participation rates in community-based day activities also decrease in older age groups, it seems unlikely that large numbers of U.S. retirees with IDD take up socially inclusive community activities in retirement, despite policy mandates for such outcomes (Centers for Medicare and Medicaid, 2011; 2014).
These findings present a picture of predominantly segregated activities in retirement. It is important to identify which features of the disability service system may contribute to such outcomes. In the United States, the primary providers of day and employment services for people with IDD are funded through Medicaid waiver programs (Butterworth et al., 2016), and nationally, most agencies (69%) provide both work and day programs (Domin & Butterworth, 2013). This arrangement may partially explain why retirement appears suddenly for people with IDD. That is, we speculate that a person may finish working but be transferred immediately to a day program offered by the same provider. Some may see this arrangement as: (a) meeting the person's need for out-of-home activity, (b) providing similar “covered hours” (i.e., when the person's residential setting does not need to provide support), and (c) resulting in no loss of funding to the provider agency. These notions need to be tested directly through future research.
Retirement-related services and supports do not have to be facility based, and retirement need not be sudden. For example, gradual transition to retirement was facilitated through participation in mainstream community groups and/or volunteering opportunities by Stancliffe et al. (2015; Stancliffe, Wilson, Gambin, Bigby, & Balandin, 2013). There is a great deal of policy and practice attention to inclusive workforce participation for people with IDD, such as Employment First policies (Gunty et al., 2017) and the passage of the Workforce Innovation and Opportunity Act (2014). These policies will likely have an impact on the retirement of people with IDD in coming years, and could result in increased demands for socially inclusive retirement activities from both policy makers and people with IDD (Brotherton, Stancliffe, Wilson, & O'Loughlin, 2016).
The lack of detailed, publicly available data related to retirement provides a challenge to understanding the experiences of older adults with IDD. This study used NCI-ACS data with a weighted multistate sample large enough to support relatively fine-grained age-group analyses for age spans as small as 5 years. Even so, sample size in the oldest age groups was smaller, so the precision of estimates was not as good as for younger age groups.
Cohort effects may have influenced employment outcomes, because participants in different age groups likely experienced different employment opportunities throughout life. For example, individuals currently aged 65 or more left school (if they attended school) over 45 years earlier, a period when community employment was rarely available. Such experiences may partly explain the lower rates of community employment in older age groups. That is, some older individuals may never have experienced community employment, meaning that their lower current participation rate may not be solely due to retirement. Longitudinal studies can help clarify cohort effects. Cohort effects may have influenced employment participation rates, but should have had no effect on our analyses of work hours that involved currently employed participants only (Figure 5), because such analyses were not affected by participation rate.
The NCI-ACS contains no specific questions about retirement, so we were unable to examine fundamental issues such as individual age of retirement or reasons for retiring. NCI-ACS data are cross-sectional, so do not show what happens with individuals over time. Only data on participation in formal work or day programs was analyzed, but such data may not provide a comprehensive picture of all activities engaged in by people in retirement.
A final important limitation concerns the representativeness of the NCI-ACS sample. By design, the NCI-ACS only surveys IDD service users. Nonusers—such as people with IDD who may be in competitive employment and do not receive publicly funded IDD supports—are not represented in NCI-ACS data. In 2014, Larson et al. (2017) reported that 1.17 million people (25%) received IDD services nationally out of an estimated 4.68 million people with ID in the United States. This situation likely means that considerably larger numbers of older people with IDD are retiring than was shown using NCI-ACS data for IDD service users. The support needs in retirement of these unserved individuals are currently unknown. Further, the absence of this group from the comparisons with the general community (Figure 1) may have affected our findings.
Having identified that older people with IDD do retire, a number of important but unanswered questions arise regarding the circumstances and timing of retirement, and their postretirement lifestyle.
Future research should explore the factors associated with older people with IDD continuing to work, especially given our intriguing finding of a relatively larger proportion of workers with IDD continuing to work past age 65 than in the general population. Based on the prevalence of individual plan community employment goals, we provided initial evidence that this trend was strongest among those currently working in community employment, but the reasons for this outcome are unknown.
Second, choice and self-determination have been emphasized in recent policy initiatives. Evaluating the extent of choice in employment and activities across the age span will help ensure that older people with IDD are afforded choice about what they do in their retirement years.
Third, policy requirements for provision of services and supports in inclusive community settings (Centers for Medicare and Medicaid, 2011; 2014) will likely mean that IDD provider agencies will benefit from evidence-based guidance on how to do this effectively with people who have retired. Well-controlled research demonstrating the feasibility and effectiveness of inclusive retirement programs can provide such guidance (e.g., Stancliffe et al., 2015).
Fourth, to obtain a more representative picture, it will be important to find ways to include people with IDD who do not currently use IDD services in retirement research. Analysis of general population surveys that report retirement data could be an effective approach, if people with IDD can be identified accurately.
Finally, this cross-sectional study drew no firm conclusions about the process of retirement over time. Longitudinal research examining participants' retirement planning and their work hours and day activities over time, with concurrent information about personal and environmental characteristics, would strengthen the literature base about when and how people with IDD retire. One cost-effective approach would be to undertake secondary analysis of available large-scale longitudinal studies involving aging and retirement in the general community, provided that people with IDD can be identified within the larger sample. Research questions of interest include (a) the relative importance of financial factors in retirement decisions and postretirement lifestyle of people with and without IDD, and (b) the use and effects of retirement planning.
This study found that older adults with IDD do decrease their participation in paid employment as they age. However, most older adults with IDD are not in paid employment, and understanding retirement (withdrawal from paid work) is not relevant to these individuals. Our findings suggest that retirement mostly happens suddenly and may be followed by participation in a (facility-based) day program. Regardless of disability, retirement is a typical stage of the life course for older adults. Age makes community inclusion no less important. The changing needs of older adults with IDD will require collaborative efforts in research, policy, and practice to ensure that inclusive services and supports are accessible. The current lack of research in retirement for people with IDD in the United States provides limited guidance to practitioners and policy makers. Consequently, research into the retirement activities, services, planning, and supports for older workers with IDD should be expanded.
This publication is supported by Cooperative Agreement #H133B080005 from the National Institute on Disability Independent Living and Rehabilitation Research (NIDILRR) within the Administration for Community Living (ACL), Department of Health and Human Services (HHS), which was awarded to the Research and Training Center on Community Living (RTC-CL) at the University of Minnesota.