Prevalence of intellectual disability among adults born in the 1980s and 1990s in the United States

Abstract

Background

Prevalence of intellectual disability (ID) is currently estimated through parent report on surveys of children. It is difficult to estimate the number of adults living with ID in the United States because no comprehensive survey or surveillance allows for identification. The purpose of this study was to estimate the prevalence and number of adults with ID born between 1980 and 1999 using multiple years of the National Health Interview Survey (NHIS) and Census data.

Methods

We concatenated the NHIS from 1997–2016 that evaluated parental response about whether a child aged 3–17 years had an ID. Using weighted survey analyses, we estimated the prevalence of ID among individuals across four birth cohorts—(1) 1980–1984, (2) 1985–1989, (3) 1990–1994, and (4) 1995–1999. The number of adults with ID was then extrapolated by applying these prevalence rates to Census population estimates (as of 1 July 2021) of respective birth cohorts.

Results

Weighted prevalence of ID varied by birth cohort, sex, race and ethnicity, and US Census Bureau regions. The overall prevalence rate was 1.066 [95% confidence interval (CI): 0.831–1.302] for adults born between 1980 and 1984, 0.772 (CI: 0.654–0.891) for adults born between 1985 and 1989, 0.774 (0.675–0.874) for adults born between 1990 and 1994, and 1.069 (CI: 0.898–1.240) for adults born between 1995 and 1999. Overall, we estimate that 818564 adults with ID who were approximately 21–41 years were living in the United States as of 2021.

Conclusions

This study provides researchers examining adult health outcomes with an estimated denominator of young and middle-aged adults living with ID in the United States. Policymakers can use this information to support justification for resource and service needs, and clinicians may benefit from understanding that ID is a lifelong developmental condition often with additional physical, emotional and developmental needs requiring tailored care.

Introduction

Individuals with intellectual disability (ID) are identified at birth and in early childhood, and evaluation of service access and health needs among people with ID in U.S. population-based surveys has been relegated to surveys assessing child health. Since 1987, the Centers for Disease Control and Prevention (CDC) via the National Health Interview Survey (NHIS) has ascertained whether a sample family member or individual has ‘mental retardation’ (1987–2003) or ‘intellectual disability also known as mental retardation’ (2004–2016). These questions quantify ID from parent-report and are weighted to produce population estimates. These results are used to quantify prevalence of children with this condition (e.g. Boyle et al. 2011; Zablotsky & Black 2020; Zablotsky et al. 2015, 2017, 2019), and as such, are an important metric for understanding potential system delivery needs.

Evaluating health service access and tracking changes in important health outcomes are often identified through population-based surveys and surveillance in the U.S. Although assessing health among children with ID is important, it is equally essential to quantify health outcomes and barriers to care among adults living with ID and other developmental conditions. Unfortunately, no U.S. population-based surveys ask questions about intellectual or developmental disability in adulthood or in self-report surveys of adult health. Further, while life expectancy for people with ID is similar to the general population (Bittles et al. 2002; Janicki et al. 1999), it is difficult to estimate the true number of adults with ID in the United States. This barrier affects needs assessments that characterise service delivery, analyses of emergency and inpatient hospitalisation records (e.g., via Health Care Utilization Project, H-CUP; AHRQ) and opportunities to evaluate state efforts at delivering care to people with disabilities through Medicaid and Medicare. Without accurate and available denominator estimates of the overall population of adults with ID in the United States, guesstimates from existing data, service delivery and future projections of resources are made from incomplete data. Anderson et al. (2019) provide a thorough systematic review of US prevalence among children and adults with ID and/or developmental disability and comment that one solution is to pool estimates from existing data sources.

The purposes of this brief report are to (1) describe the methodology and prevalence estimates of adults with ID living in the United States, who were born between 1980 and 1999 and who were approximately 21–41 years old in 2021 and (2) provide estimates of adults living with ID for researchers working with adult data sources. These estimates rely on publicly available data from the National Center for Health Statistics and US Census Bureau population estimates.

Methods

This study relied on publicly available data. The data were anonymised and no human subjects ethics review was needed for this analysis.

Data

We used data on 128785 individuals aged 3–17 years from the 1997 to 2016 waves of the National Health Interview Survey (NHIS, 1989, 2004). Survey waves were pooled to create a single dataset following the NHIS guidelines of treating different sample design periods as statistically independent and different years within a sample design period as not statistically independent (NCHS, 2017). Our data spanned over three sample design periods as follows: (1) 1995–2005; (2) 2006–2015; and (3) 2016–2018. As such, adjustments for survey weights were made in the following manner: first, 9 years of data from 1997 to 2005 were pooled and sample weights were divided by nine. Second, 10 years of data from 2006 to 2015 were pooled, and sample weights were divided by 10. Third, pooled data from 1997 to 2005, pooled data from 2006 to 2015, and 2016 data were pooled, and final weights were obtained by dividing the sample weights by three.

Data for 1997 to 2003 were obtained from the Inter-university Consortium for Political and Social Research (ICPSR) website following a data use agreement request (1989). Data for 2004–2016 were obtained from the CDC website (2004). Population estimates by single-year age (as of 1 July 2021) and sex were obtained from the CDC WONDER Online Database.

Measures

The Sample Child file of the NHIS reports whether the sample child was ever told by a doctor or health professional to have intellectual disability or ‘mental retardation’ (Survey wording: ‘Has a doctor or health professional ever told you that [sample child name] had … mental retardation [1997–2010 surveys] … intellectual disability, also known as mental retardation [2011–2016 surveys]?’). A child was identified as having ID if ever told to have the condition. Responses ‘refused’, ‘not ascertained’, or ‘do not know’ were excluded (N = 95, 0.08% of the total sample).

The Person file of the NHIS reports all persons’ birth year. We categorised individuals by birth year ranging from 1980 to 1999. We also grouped individuals in four birth cohorts, each spanning over 5 years: (1) 1980–1984; (2) 1985–1989; (3) 1990–1994; and (4) 1995–1999.

Persons born between 1 January 1980 and 31 December 1999 were aged 21 years 6 months to 41 years 6 months on 1 July 2021. Because population estimates were reported in single-year age, at a point of time, we aligned those estimates with respective birth years via a simple assumption. We assumed that those who were of age t years on 1 July 2021, half of them were born in (2021-t) and half were born in [2021 − (t + 1)]. For example, those who were of age 21 years in 2021, 50% of them were assumed to be born in 2000 and 50% in 1999. Similarly, 50% of the individuals born in 1980 were assumed to be aged 41 years and 50% were aged 40 years.

Statistical analysis

For each birth year, we estimated the share of individuals having ID using adjusted survey weights of the pooled sample. Next, we estimated the prevalence of ID by birth cohort. For each birth cohort, we estimated the prevalence by sex (male and female), race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic and non-Hispanic other) and US Census Bureau regions (Northeast, Midwest, South and West) in accordance with recommendations (Anderson et al. 2019). All analyses were performed using Stata 18.0 software.

Results

Table 1 reports the unweighted sample size by birth year and survey waves for the 128785 individuals aged 3–17 years included in our estimates. Of these, 1177 were reported to have ID and 127608 had no reported ID.

Table 1.

Unweighted sample size by birth year and NHIS survey waves

		National Health Interview Survey Wave
		1997	1998	1999	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	Total
Birth Year	1980	838	348	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1186	12 821	1980–84	Birth cohort
	1981	802	778	311	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1891
	1982	788	724	724	360	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	2596
	1983	682	686	693	757	344	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	3162
	1984	734	672	710	791	751	328	0	0	0	0	0	0	0	0	0	0	0	0	0	0	3986
	1985	727	663	657	688	744	714	336	0	0	0	0	0	0	0	0	0	0	0	0	0	4529	28 647	1985–1989
	1986	688	673	622	721	714	697	663	344	0	0	0	0	0	0	0	0	0	0	0	0	5122
	1987	685	671	649	627	729	665	678	743	322	0	0	0	0	0	0	0	0	0	0	0	5769
	1988	669	634	642	647	695	602	629	705	767	242	0	0	0	0	0	0	0	0	0	0	6232
	1989	748	656	641	688	650	661	639	698	715	612	287	0	0	0	0	0	0	0	0	0	6995
	1990	737	666	679	641	657	652	608	688	696	609	580	271	0	0	0	0	0	0	0	0	7484	42 161	1990–1994
	1991	723	698	636	680	726	630	638	610	648	505	498	515	261	0	0	0	0	0	0	0	7768
	1992	790	671	667	691	680	621	603	575	615	512	488	519	690	317	0	0	0	0	0	0	8439
	1993	789	696	635	636	618	575	606	604	594	498	448	492	663	740	402	0	0	0	0	0	8996
	1994	458	725	682	688	684	570	565	558	623	432	477	477	650	684	806	395	0	0	0	0	9474
	1995	0	359	660	703	659	599	572	591	561	418	434	461	581	620	720	849	382	0	0	0	9169	45 156	1995–1999
	1996	0	0	381	675	712	574	521	545	592	445	423	383	559	584	694	737	771	411	0	0	9007
	1997	0	0	0	376	717	574	575	588	642	428	437	406	582	562	643	695	707	767	401	0	9100
	1998	0	0	0	0	426	612	598	602	575	467	410	427	494	589	649	703	697	847	746	330	9172
	1999	0	0	0	0	3	363	639	595	574	428	419	384	521	553	636	701	707	747	748	690	8708
	Total	10 858	10 320	9989	10 369	10 509	9437	8870	8446	7924	5596	4901	4335	5001	4649	4550	4080	3264	2772	1895	1020	128 785

Abbreviation: NHIS, National Health Interview Survey.

Source: Author’s analysis of National Health Interview Survey, 1997–2016.

The prevalence of ID in the full sample for individuals born between 1980 and 1999 was 0.95% (95% CI: 0.84–1.05). While it was 0.86% (95% CI: 0.74–0.97) among individuals born in the 1980s (i.e., 1980 to 1989), it increased to 0.94% (95% CI: 0.83–1.04) among individuals born in the 1990s (i.e., 1990 to 1999). Figure 1 presents the ID prevalence by birth year. The prevalence was the lowest (0.54%, 95% CI: 0.37–0.72) for individuals born in 1992 and was the highest (1.50%, 95% CI: 0.97–2.03) for individuals born in 1999.

Figure 1.

Intellectual disability prevalence by birth year. Note: Estimates were obtained using complex survey weights of the NHIS. Vertical lines across the markers refer to 95% confidence intervals. The dashed line represents intellectual disability (ID) prevalence for all respondents born between 1980 and 1999 and shaded green represents the 95% confidence interval for that estimate. Source: Author’s analysis of National

Table 2 presents the ID prevalence rates by birth cohort, sex, race, ethnicity and US Census region. The prevalence ranged from 0.77% to 1.07% across birth cohorts. The prevalence rates were generally higher among males (0.83% to 1.25%) than those among females (0.62% to 0.95%). Of note, the difference in prevalence rates between males and females was not statistically significant for the 1980–1984 and 1985–1989 birth cohorts. Among White individuals, the prevalence ranged from 0.67% to 1.04% across birth cohorts. Among Black and Hispanic individuals, it ranged from 0.98% to 1.79% and 0.70% to 1.10%, respectively. No particular patterns in the prevalence rates were observed across regions.

Table 2.

Intellectual disability prevalence by birth cohort, sex, race, ethnicity and US Census region

	ID Prevalence
	All	Male	Female	Non-Hispanic White	Non-Hispanic Black	Hispanic	Non-Hispanic other	Northeast	Midwest	South	West
Birth cohort
1980–84	1.066 (0.831, 1.302)	1.173 (0.853, 1.494)	0.953 (0.616, 1.290)	0.948 (0.655, 1.241)	1.791 (1.066, 2.517)	1.096 (0.636, 1.555)	0.480 (−0.047, 1.007)	1.106 (0.660, 1.552)	1.069 (0.523, 1.616)	1.399 (0.947, 1.852)	0.446 (0.157, 0.734)
1985–89	0.772 (0.654, 0.891)	0.826 (0.657, 0.994)	0.717 (0.547, 0.887)	0.668 (0.522, 0.814)	1.269 (0.892, 1.645)	0.821 (0.540, 1.101)	0.541 (0.142, 0.940)	0.754 (0.458, 1.050)	0.555 (0.327, 0.783)	0.933 (0.717, 1.150)	0.654 (0.386, 0.923)
1990–94	0.774 (0.675, 0.874)	0.928 (0.776, 1.080)	0.615 (0.487, 0.743)	0.728 (0.589, 0.867)	0.977 (0.733, 1.222)	0.696 (0.513, 0.879)	0.965 (0.595, 1.335)	0.734 (0.508, 0.961)	0.955 (0.683, 1.227)	0.670 (0.525, 0.815)	0.600 (0.402, 0.798)
1995–99	1.069 (0.898, 1.240)	1.245 (0.970, 1.520)	0.883 (0.676, 1.091)	1.037 (0.796, 1.279)	1.620 (1.012, 2.227)	0.799 (0.597, 1.000)	1.039 (0.565, 1.514)	1.149 (0.730, 1.568)	0.821 (0.563, 1.080)	1.159 (0.838, 1.480)	1.155 (0.732, 1.579)

Note: Estimates were obtained using complex survey weights of the NHIS. 95% confidence intervals are in parentheses.

Source: Author’s analysis of National Health Interview Survey, 1997–2016.

We estimated that 818564 (95% CI: 623480–1013323) adults born between 1980–1999 (approximately aged 21–41 years) with ID are living in the United States as of 2021. Of them, 348070 (95% CI: 255934–439916) are female and 470494 (95% CI: 367547–573408) are male.

Conclusions

To our knowledge, this study is the first published report estimating the number of adults living with ID in the United States based on historical birth cohort data from parental survey report. We estimate that approximately 818564 individuals with ID between 21 and 41 years of age were living in the United States as of 2021. This corresponds to an overall prevalence of 0.95% (CI: 0.84–1.05), or 9.5 per 1000 adults. Our prevalence estimates are consistent with older published survey research on this topic but are lower than estimates using administrative data (e.g., Boyle et al. 2011; Zablotsky et al. 2015, 2017; Anderson et al. 2019; McGuire et al. 2019; Zablotsky et al. 2019; Zablotsky & Black 2020; McBride et al. 2021). A recent systematic review reports that US survey estimates range from 0.52% to 1.21% (5 to 12 per 1000 individuals); however, estimates are higher (13.7% or 137 per 1000 individuals) when using Social Security Administration data (Anderson et al. 2019). Prevalence from international sources such as Ireland are consistent with our estimates and range from 0.3% to 1.5% (3 to 15 per 1000 individuals) for people aged 16 years and older, with other international estimates ranging from 0.05% to 0.8% or 1 to 8 per 1000 individuals (McBride et al. 2021).

ID prevalence variability may be influenced by many things, including changes in medical care, diagnostic practices, changes in language about ID, and/or changes in awareness (Committee to Evaluate the Supplemental Security Income Disability Program for Children with Mental Disorders 2015; McKenzie et al. 2016). Our work extends the work of others describing prevalence of ID in the United States in two key takeaways: first, our analyses generate estimates of ID prevalence by birth year and birth cohorts by pooling 20 years of nationally representative US survey data. Prevalence estimates reported in Boyle et al. (2011) were obtained by pooling 3 years of data with overlaps across ages. Further, Boyle et al. (2011) included NHIS data until 2008, and our data extend this work by including data until 2016. Our findings have implications for researchers aiming to include denominators of the number of adults with ID from which to base analyses of Medicaid waiver uptake, or among utilisation studies investigating emergency department visits or inpatient hospitalisation without sample denominators of the full population. Our findings also have implications for the training of healthcare providers. Typically, healthcare providers learn content about people with ID in paediatric courses and research settings; however, providers themselves identify a lack of training, and understanding affects ability to work with people with disabilities, including people with ID (Iezzoni et al. 2021). Our study provides evidence that training in ID should be included in adult rotations due to the significant size of this population of adults with ID.

There are several known limitations to this study. Our estimates are solely about individuals with ID, not developmental disability broadly, and are based on parent reports on national surveys, therefore parent recall bias is a concern. Sampling from these national surveys excluded respondents who were institutionalised respondents or who were US military, potentially contributing to an underestimate. Further, these parent reports are not confirmed with evaluation of actual diagnosis by a clinician, and therefore, these estimates should be validated with further surveillance of adults with ID living in the United States. Changes in the survey wording in 2014 are noted in our methods and are a potential source of variation in prevalence; however, Zablotsky et al. (2015) found that the wording did not affect prevalence estimates. Our estimates do not apply to middle-aged and older adults with ID who were born prior to 1980, for whom we have limited information on which to base prevalence. Lastly, we do not adjust for mortality or migration in aging up the childhood prevalence when applied to adult populations. However, lacking surveillance among adults, this study is the first to provide supportive evidence of estimates of young and middle-aged adults living with ID in the United States for future research, practice, and policy evaluation.

Data availability statement

Data used in this analysis are publicly available from the Centers for Disease Control and Prevention (CDC) (https://www.cdc.gov/nchs/nhis/data-questionnaires-documentation.htm) and University of Michigan’s Inter-university Consortium for Political and Social Research (ICPSR) (https://www.icpsr.umich.edu) websites.