https://orcid.org/0000-0002-9661-7472
Abstract
Research on caregiving in the United States has not clearly identified the scope of the gap between care needed and care received and the changes implied by ongoing and anticipated shifts in family structure. This article examines the magnitude of contemporary gaps in care among older adults in the United States and how they are likely to evolve through 2050. We use data from the Health and Retirement Study (1998–2014) to estimate care gaps, operationalized as having difficulties with activities of daily living (ADLs) or instrumental activities of daily living (IADLs) but not receiving care. We also estimate variation in care gaps by family structure. Then, we use data from demographic microsimulation to explore the implications of demographic and family changes for the evolution of care gaps. We establish that care gaps are common, with 13% and 5% of adults aged 50 or older reporting a care gap for ADLs and IADLs, respectively. Next, we find that adults with neither partners nor children have the highest care gap rates. Last, we project that the number of older adults with care gaps will increase by more than 30% between 2015 and 2050—twice the rate of population growth. These results provide a benchmark for understanding the scope of the potential problem and considering how care gaps can be filled.
Background
Nearly half of U.S. adults aged 65 or older have difficulty or receive help with self-care or mobility activities, such as bathing or getting in and out of bed, referred to as activities of daily living (ADLs; Freedman and Spillman 2014). Many also have difficulty with household tasks, referred to as instrumental activities of daily living (IADLs). When older adults need help with these tasks, partners and adult children (most often biological) are the main providers of unpaid caregiving (Freedman and Wolff 2020; Wolff et al. 2018). The idea that families will and should complete this work is embedded in the U.S. psyche (Patterson and Reyes 2022) and the country's long-term care policy framework (Gaugler 2021). Yet, older adults’ families are changing rapidly, with declines in marriage and fertility, potentially increasing the gap between the care needed and the care received. Understanding the magnitude of current and future care gaps in the United States is critical because they can threaten the ability to live independently (Freedman 2018), exacerbate health declines and mortality risks (He et al. 2015), and create strong financial and emotional pressures on individuals and families (Redfoot et al. 2013; Skufca and Rainville 2021).
Research on older adult caregiving in the United States has not clearly identified the scope of care gaps and the changes implied by ongoing and anticipated changes in family structure. Much has been made of anticipated declines in the caregiver support ratio (Feinberg and Spillman 2019; Redfoot et al. 2013), which is expected to decline from 7 to 3 caregivers between 2010 and 2050, potentially portending substantial challenges. Unfortunately, the caregiver support ratio is defined solely by the relative size of different cohorts alive at given periods: those aged 45–64 divided by those aged 80 or older in each of the focal years (Redfoot et al. 2013). This measure does not account for many demographic changes that affect older adults’ available family care networks, such as marriage and partnership, fertility and family size, stepfamilies, and kinlessness (Freedman et al. 2023). It also does not account for demographic variation in caregiving norms and patterns, the aging of the broader population, or the need for care among many individuals younger than 80. We can gain a clearer perspective of anticipated changes in care gaps by considering current patterns of health, family structure, and caregiving and then looking to the future to examine how increasing rates of gray divorce (Brown and Lin 2012), family complexity (Seltzer 2019; Strohm et al. 2009), kinlessness (Margolis and Verdery 2017; Verdery and Margolis 2017), and ethnic and racial diversity in middle-aged and older adults (Vespa et al. 2020), as well as general population aging and growth, will affect the magnitude of care gaps.
In this article, we examine contemporary gaps in care among middle-aged and older U.S. adults and how these gaps will likely evolve through 2050. In contrast to the cohort size–based definitions of the caregiving support ratio, our measure of the care gap captures respondents who reported difficulty with ADLs or IADLs but received no help with these tasks in the last three months. We first establish the frequency of experiencing care gaps with data from 1998 to 2014. Next, we explore how the care gap varies by family structure, with particular emphasis on associations between experiencing a care gap and the availability of partners and children and variation by race and ethnicity. Last, we estimate the potential for an increased care gap in the United States through 2050, shaped by structural population changes in population size, age structure, and family structure. Our results offer new insights into the current and impending magnitude of the care gap associated with the country's rapidly aging population.
Measuring Individual- and Population-Level Care Gaps
Despite the robust literature on disability and caregiving, there is no standard measure of care gaps (Ankuda et al. 2022; Freedman 2000; Freedman et al. 2004). The most-cited paper on the potential crisis in caregiving did not use individual-level data on care needs or care receipt; instead, it examined the ratio of potential caregivers (adults aged 45–64) to those assumed to require care (adults aged 80 or older) (Redfoot et al. 2013). This type of population-level analysis ignores fluctuating individual-level factors, such as the extent of family networks, and how gaps are filled across the population.
Research on unmet needs is one subset of scholarship on care gaps, focusing on the adverse consequences of unmet care needs (Beach and Schulz 2017; Freedman and Spillman 2014). However, this type of measure may be overly conservative because it measures only instances with dire consequences, leading to a narrow picture of care dynamics. Many older adults might need help with ADL and IADL tasks. Such individuals might experience daily stress but not perceive that their lack of help has consequences significant enough to prompt them to report distress in a standard survey. Given an extensive literature highlighting the impacts of daily stressors on health (Almeida et al. 2005), the number of such individuals is also a consequential consideration.
Few studies have addressed the amount of care needed and not received. Many studies have not differentiated between care needs and care access, focusing instead on the contrast between those who receive care and those who do not (Suanet et al. 2012). This focus conflates individuals reporting no difficulty with tasks and receiving no care with individuals reporting difficulty and receiving no care. Some important exceptions focused on care for older adults with dementia and examined those reporting difficulty with and without help (Edwards et al. 2020; Yang et al. 2022). Another exception estimated the projected increase in adults who will need help with their health or disability in England in 2032 relative to 2007, projecting a likely gap of 160,000 caregivers in 2032 (Pickard 2015). To our knowledge, Pickard (2015) is the only analysis examining how shifting family structure will affect the macro-level demand for caregiving. Our study adds to the knowledge base by examining the care gap and how it will evolve in the United States.
Family Structure and Caregiving in the Contemporary United States
Family members play an essential role in providing help with ADLs and IADLs for older adults. Family caregiving yields important benefits, including reducing health care costs and reducing or delaying nursing home admission (Friedman et al. 2019). However, a lack of family caregivers can lead to reliance on paid services and higher risks of nursing home admission or mortality (He et al. 2015; Potter 2019). Spouses and adult children are the most common care providers, and they provide the most care of all caregivers, including paid caregivers (Freedman and Wolff 2020; Wolff et al. 2016). Many older adults receive mixed care types, including help from unpaid and paid caregivers (Jacobs et al. 2018). However, most helpers are from the family and unpaid caregiver networks, and only 10% are paid (Freedman and Spillman 2014). Although family members provide a great deal of care for older adults requiring help, care gaps might be common, especially among those with thin family networks.
The components of care gaps—having difficulty with ADLs and IADLs and receiving caregiving—display well-established differences by race, ethnicity, and gender. The oldest-old, women, and Black adults have the highest care need levels (Freedman et al. 2002), with married adults from all racial and ethnic groups reporting lower rates of disability than those who are unmarried (Liu and Zhang 2013). Over and above demographic differences in care needs, patterns of caregiving and unmet needs also vary demographically. For example, unmarried older adults are at an increased risk of having unmet needs (negative consequences from caregiving gaps) (Beach and Schulz 2017; Lima and Allen 2001), and women report more unmet care needs than men (Lima and Allen 2001). Moreover, caregiving patterns also vary by race and ethnicity. Black and Hispanic older adults with disabilities report higher rates of inadequate help than White older adults with disabilities (Berridge and Mor 2018; Lima and Allen 2001). However, despite the vast evidence of variation in caregiving gaps, there is no clear picture of variation in care gaps across the contemporary U.S. population or how such gaps are evolving.
Impending Structural Increases in the Care Gap
Population-level changes in marriage, fertility, and aging will likely produce a structural increase in care needs among older adults. Marriage declines and recent increases in gray divorce are increasing the proportion of older adults who are unmarried. The percentage of adults aged 45–64 who were unmarried rose from 22% to 34% between 1980 and 2009 (Lin and Brown 2012). In 2015, 37% of baby boomers were unmarried (Brown and Wright 2017). Many older adults who repartner cohabit rather than marry (Carr and Utz 2020) or choose a living-apart-together relationship (Connidis et al. 2017), and adults in such relationships are less likely than married older adults to receive care from their partners (Broese van Groenou et al. 2019; Noël-Miller 2011). Fertility decline, increasing childlessness, and changing patterns of proximity and residential status of children might also influence the future availability of family caregivers. Help is more forthcoming for older adults with more children (Wolf et al. 1997), and children are the primary source of informal care for single older adults (Kwak et al. 2021).
Population aging is another force that will also likely increase the size of the population needing care. The population of middle-aged and older adults in the United States is growing rapidly. The Census Bureau projects a population of 157 million individuals aged 50 or older in 2050, up from 113 million in 2016, and staggering growth among those aged 65 or older, from 49 million in 2016 to 86 million in 2050 (Vespa et al. 2020). Demographers anticipate that expected years of life remaining at age 65 will increase from nearly 18.8 years for men and 21.4 years for women in 2020 to 21.0 years for men and 23.7 years for women in 2050 (Medina et al. 2020). In 2050, there will almost surely be many more older adults than today, and these older adults will be much older, on average. However, the force of population aging varies across demographic groups, and we do not have a clear picture of whether the increase in the number of older adults will occur among groups with care gaps.
The changes in family structure, alongside the increases in population aging, mean that in 2050, the projected size of the population aged 50 or older without living kin of different types relevant to caregiving, such as children and partners, might increase (Verdery and Margolis 2017). It is unclear how the changing U.S. demographic composition by race and ethnicity will influence care gaps, but there is room to expect potential countervailing effects. White and Asian adults are more likely to be married than Hispanic and African American adults (Smock and Schwartz 2020), and the United States will be more racially and ethnically diverse in 2050 than in 2015. Even though Black adults are more likely to lack close kin than White adults (Verdery and Margolis 2017), they also may have care networks extending beyond spouses or children (Taylor et al. 2022) and different notions of family obligation (Dilworth-Anderson et al. 2005). How will structural changes in families and increases in population aging shape the population needing help and not receiving it?
The Present Study
This study has three aims. First, we investigate how common it is for U.S. older adults to have care gaps, which we operationalize as reporting difficulty with a task but not receiving help with it in the last three months. We examine difficulties with ADLs and IADLs separately because care with ADLs (personal care and mobility) is more intimate than help with IADLs (household tasks), and patterns might differ across the two task types. We also present three levels of care gaps: at least one, at least two, and at least three tasks with care gaps for each type. Second, we test how individual-level care gaps vary by family structure, focusing on differences by partnership and having children. We also test whether those with neither a partner nor children (i.e., who are kinless) are more likely to have care gaps than those with close kin. Finally, we examine population-level care gaps to gauge the potential for increased care gaps in the United States through 2050 as a result of structural population changes (population size, age structure, and family structure) using the results of demographic microsimulations. This method allows for the calculation of changes in family structure and population composition by age, race, ethnicity, and sex. These results provide a benchmark for understanding the scope of the potential problem and considering how care gaps can be filled.
Data
We use the Health and Retirement Study (HRS) to estimate the magnitude of care gaps in the contemporary United States and how it varies by family structure. The HRS is a panel study of U.S. individuals aged 50 or older, including those living in nursing homes, who are followed until death. The study began in 1992 and has been continuously refreshed to be nationally representative. The HRS is ideal for our study because it provides information about respondents’ difficulties with ADLs and IADLs, whether the respondent receives help with each task for which they report difficulty, data on family structure, and information about other key covariates (e.g., education and wealth). Further, unlike studies that examine only adults 65 or older (e.g., the National Health and Aging Trends Study), the HRS contains data on individuals as they age into the years when they need greater assistance, providing preliminary indicators of care needs in younger cohorts. Recent trends indicate that successive cohorts of older adults are receiving help from family at younger ages (Wolff et al. 2018), another reason why the lower age limit of HRS is important for this study.
Our analysis focuses on 1998–2014. Question wording regarding ADLs and IADLs in 1992–1996 was not consistent with that in later waves. The analysis extends through 2014, the most recent year for which we have data from both the RAND HRS Longitudinal file and the RAND Family file for information about key family members. The sample for these years contains 170,631 person-waves for respondents aged 50 or older. We exclude the 0.3% of respondent-waves with missing information on all ADLs and IADLs, questions about help received, or demographic and socioeconomic characteristics (432 person-waves). Our final analytic sample contains 170,199 person-waves for respondents aged 50 or older in 1998–2014.
To analyze what demographic changes imply for the evolution of care gaps in the coming decades, we use data from a previously reported and validated demographic microsimulation (Margolis and Verdery 2019; U.S. Census Bureau 2014; Verdery and Margolis 2017). The simulation used historical, contemporary, and projected rates of demographic events (e.g., fertility, mortality, marriage, and divorce) to examine changes in family structure over the next several decades. The simulated data were created using the Berkeley microsimulation program, Socsim (Hammel et al. 1976; Hammel et al. 1990; U.C. Berkley Demography Lab n.d.). The simulated population reflects population and kinship dynamics for a large subset of the population: single-race, native-born, non-Hispanic Black and White individuals. This subset represents the largest racial groups in the United States, especially among older adults, accounting for 78% of the population aged 50 or older in 2016 and projected to compose 61.4% of those aged 50 or older in 2050 (Vespa et al. 2020). We estimated separate models for Black and White individuals because demographic rates vary across these groups and can bias the distribution of kin (Ruggles 1993). The simulation does not model intermarriage between racial groups, immigration, or emigration because the methods are not well-suited to modeling these dynamics. Despite having limitations, the simulated data provide novel insights into potential shifts in changes in family structure, population size, and population structure by 2050, allowing us to forecast the percentage and number of people with care gaps in 2050.
We use the simulation's results by family structure in addition to its projected compositional changes by age, gender, and race and ethnicity. Demographic microsimulation is the best available method for estimating such family structure changes. Standard demographic projections do not include estimates of family structure change. Alternative methods, such as newer tools in the formal demography of kinship, can provide projections of changes in expected average counts of blood relations (e.g., Alburez-Gutierrez et al. 2023), but their implementations lack tools for modeling the partnership dynamics that are particularly relevant to caregiving and do not offer estimates of distributional changes (e.g., they can identify declines in average numbers of children but not changes in the percentage of the population with no children).
Measures
Dependent Variables
We measure care gaps among middle-aged and older adults who reported difficulty with ADLs or IADLs. Respondents were asked about whether a health or memory problem led to difficulty with any of six ADL tasks (dressing, walking, bathing, eating, getting in and out of bed, and toileting) and five IADL tasks (preparing a hot meal, shopping for groceries, making phone calls, taking medications, or managing money) in the last three months. For each task with which respondents reported having difficulty, they reported whether anyone ever helped them with it. We define a care gap as a reported difficulty with a task for which the respondent received no help in the last three months. Care provided by professional or paid caregivers is included, even if respondents received no unpaid help from family members or others. We include paid and unpaid care to capture the broad range of care sources used, especially among those without close kin. We gauge whether respondents have any care gaps (one or more), two or more gaps, and three or more gaps to gauge the intensity of pressures some respondents might face. We analyze ADL and IADL care gaps separately because ADLs are more severe. Further, the two task types might have different associations with family structure, given that help with ADLs is more personal and more often requires additional time together in the household, whereas IADLs could be more easily provided by someone outside the family or household.
Independent Variables
The key independent variable captures respondents’ family structure. We measure which of the respondent's close kin are alive by categorizing each respondent as (1) partnered and has biological children, (2) partnered and has no biological children, (3) unpartnered and has biological children, or (4) unpartnered and has no biological children. We include both married and cohabiting partners. We focus on biological children rather than including stepchildren because parents in stepfamilies are less likely than those in biological families to receive help from adult children (Wiemers et al. 2019) and receive necessary care (Patterson et al. 2022).
We control for demographic characteristics associated with both family structure and the likelihood of receiving care: gender, age, race and ethnicity, and residing in a nursing home. Race and ethnicity are categorized as non-Hispanic White, non-Hispanic Black, and Hispanic/other. We also control for socioeconomic characteristics: educational attainment, employment status, and total wealth. Educational attainment is coded as the highest attainment, categorized as less than high school, high school diploma, some college, and college degree. Employment status is measured as not working, working part-time, and working full-time. Total wealth is categorized as less than $0, $0–$49,999, $50,000–$149,999, $150,000–$499,999, and $500,000 or more.
Methods
First, we present key descriptive statistics for care gaps among adults aged 50 or older in the contemporary United States. We calculate weighted estimates of care gaps at three different intensities: difficulty with at least one, at least two, and at least three tasks. We also examine the number of care gaps by the number of tasks with which respondents reported difficulty, allowing us to measure differences in the intensity of the care gap by the number of reported task difficulties. We explore the specific ADL and IADL tasks, the prevalence of difficulty with each task, and the prevalence of care gaps for each task. Descriptive statistics reveal distinct patterns between ADL and IADL care gaps, showing that it is crucial to test them separately. For example, ADL tasks often have larger care gaps than IADL tasks.
Second, we gauge variation in care gaps by respondents’ family structure. We use six multivariate logistic regression models to predict three ADL and three IADL outcomes: at least one, two or more, and three or more care gaps. These models adjust for demographic and socioeconomic characteristics in addition to family structure. All results use person-level weights that include the nursing home sample and robust standard errors to account for the survey design and the clustering induced by some respondents contributing multiple observations in our pooled sample.
Finally, we move from the individual level to the population level to explore changes in U.S. care gaps over the next several decades. We use microsimulation data for the size of population groups by family structure in addition to age, gender, and race for non-Hispanic White and non-Hispanic Black older adults in 2015 and 2050. Then, we use the HRS to analyze predicted probabilities by family structure, age group, race, and gender. We estimate the predicted probability of having one or more, two or more, or three or more ADL difficulties; we estimate the same respective numbers for IADLs. Second, conditional on reporting ADLs or IADLs, we test the predicted probabilities of our different measures of care gaps (one or more, two or more, and three or more) for ADLs and IADLs separately by family structure, age group, gender, and race and ethnicity. To examine the expected future U.S. care gap, we apply the predicted probabilities of care gaps estimated with the HRS to the population size and structure (in terms of age, gender, race and ethnicity, and family structure) in 2050 as estimated in the microsimulation.
This analysis makes a key assumption to limit the scope of factors to explore. We assume that the levels of ADLs, IADLs, and care gaps by gender, age, race and ethnicity, and family structure will be constant between the estimated period (1998–2014) and 2050. Changes in U.S. care gaps between 2015 and 2050 in this analysis are due to changes in population size, age structure, and family structure. Any population-level increase in disability over the next several decades will change the size of care gaps. This analysis provides insights into the structural increase in care gaps, given changes in family structure, population aging, and population growth.
Results
Care Gaps in the Contemporary United States
Table 1 presents the contemporary care gap prevalence. Overall, 17.1% of adults aged 50 or older reported difficulty with at least one of six ADLs. Among this group, 75.2% had at least one care gap with at least one difficult task in the last three months. Among the 9.4% of respondents reporting difficulty with two or more ADLs, 50.3% had two or more care gaps, representing 4.7% of the age 50+ population. Among respondents reporting difficulty with three or more ADLs (6.0% of the age 50+ population), one third reported three or more care gaps.
A similar pattern is evident for IADLs. Among middle-aged and older adults, 14.5% reported difficulty with at least one IADL, and among this group, more than one third (35.8%) had at least one difficult task for which they received no help in the last three months. Among respondents reporting difficulty with two or more IADLs, 11.1% received no help with those tasks. Among respondents reporting three or more ADL difficulties, most reported some help; only 4.2 % reported three or more care gaps.
Table 2 shows variation in the prevalence of care gaps by the number of tasks reported as difficult. As the number of respondents reporting difficulty with ADLs increases, a greater percentage of respondents received at least some help (i.e., no care gaps). Among respondents reporting difficulty with only one ADL task, 78.1% had a care gap. Among respondents reporting difficulty with all six ADLs, many received some help (59.7%), but 40.3% had care gaps. Those with care gaps might include respondents needing help with all six ADL tasks and having at least some help with only some of those tasks over the last three months. The pattern is different for respondents reporting difficulty with IADLs: the prevalence of receiving help with all difficult IADL tasks was similar regardless of the number of difficulties (63% to 65%).
Figure 1 shows each of the individual ADL and IADL tasks, with the total column showing the percentage of adults aged 50 or older who reported difficulty with that task. Within each column, we indicate whether help was received with that task in the last three months or whether there was a care gap. Among ADLs (left panel), the most respondents reported difficulty dressing (9.8%), whereas the fewest respondents reported difficulty with eating (3.5%). The ADL tasks with the highest care gap levels were dressing (4.7%), walking (4.0%), using the toilet (3.9%), and getting in/out of bed (3.7%); bathing (2.9%) and eating (1.5%) had lower care gaps.
The overall prevalence of reporting difficulty with IADLs (Figure 1, right panel) among adults aged 50 or older does not differ much from that for ADLs. The key difference is that IADL care gaps are much less common. Grocery shopping was the task that most respondents reported difficulty with (9.6%), but only 1.4% of older adults received no help with this task in the last three months. Respondents less frequently reported difficulty with other IADL tasks. Managing money and preparing a hot meal were difficult for 6.8% and 6.7% of older adults, respectively, but less than 1.5% of older adults received no help with these tasks. Only 1% to 1.5% of respondents reporting difficulty with IADLs had a care gap.
Family Structure and Care Gaps
The second part of our analysis examines variation in care gaps by family structure. Table 3 shows descriptive results. Family structure is associated with the prevalence of reporting difficulty with at least one ADL and care gaps. Respondents with a partner and children had the lowest probability of reporting difficulty with any ADL (12.5%) and the lowest likelihood of having at least one care gap (73.9% of those with any ADL difficulty). Respondents with no partner or children (i.e., kinless) had a much higher prevalence of reporting difficulty with one or more ADLs (22.3%) and the highest care gaps: 80.1% of those with any ADL difficulty reported at least one care gap. Patterns for IADLs are similar. Overall, both ADL and IADL care gaps were significantly higher for kinless respondents. These patterns by family structure likely reflect health selection into partnership and the influence of kin support and other sociodemographic differences between those with versus without kin.
Table 4 assesses family structure differences in care gaps in a multivariate framework. It presents results from logistic regression models to predict ADL and IADL care gaps, net of all controls for demographic and socioeconomic characteristics. We examine our three measures of care gap magnitude for both ADLs and IADLs (one or more, two or more, or three or more care gaps).
Two key results emerge across the six outcomes. First, whereas unpartnered respondents with or without children were significantly more likely to have care gaps than those with a partner and children, partnered respondents with or without children did not differ significantly. Second, kinless older adults had higher odds of having care gaps of all magnitudes than respondents with any other family structure group.
Finally, demographic and socioeconomic covariates associated with care gaps are noteworthy. Men had higher odds of ADL care gaps than women, but fewer significant differences emerged for IADLs. Respondents in their 50s or 60s were more likely to have care gaps than respondents aged 70 or older. Non-Hispanic White respondents were more likely to have care gaps for ADLs and IADLs than other racial and ethnic groups. Regarding sociodemographic variables, respondents in the labor force were more likely to have care gaps than those not working. Moreover, despite the few differences in ADL care gaps by educational attainment, respondents with more education were more likely to have IADL care gaps. Last, respondents living in nursing homes were much less likely to have care gaps than others.
Projecting the Future Care Gap
Finally, we estimate the size of the population with care gaps in 2050. First, Figure 2 presents the current care gap level, showing predicted probabilities of one or more ADL care gaps by family structure, age group, gender, and race and ethnicity estimated from the HRS data. The predicted probability of having one or more ADL care gaps is highest among respondents in their 50s and decreases as with age among all demographic groups. Looking across the six panels by gender, race, and ethnicity, respondents without close kin have a higher likelihood of having any care gap with ADLs than respondents with a partner and biological children, especially among White women and men. The corresponding numbers and cell sizes for Figure 2 are shown in Table A2 (all tables indicated with an “A” appear in the online appendix); comparable numbers for IADLs are shown in Table A3.
Next, to examine the potential for increased care gaps in the United States through 2050 resulting from structural population changes (population size, age structure, and family structure), we apply the estimated predicted probability of care gaps to the projected population size stratified by these dimensions from the microsimulation data, which focuses on non-Hispanic Black and White older adults. Table 5 presents the simulated care gap sizes based on changes in family structure and population aging and growth.
As shown in the top section of Table 5, the total size of the population aged 50 or older will increase from 80 million to 92 million from 2015 to 2050—a 15.3% increase. However, the magnitude of change in all measures of the care gap shown will be larger than this, some by more than half. The number of adults with any care gap among the population with at least one ADL is projected to grow by roughly 31.6% in 2050, and the number with two or more and three or more care gaps will rise by more than 25%. The number of individuals with any IADL care gap is projected to increase by 31.3% between 2015 and 2050, with slightly lower growth in the population with care gaps for two or more and three or more IADLs. The results in Table 5 by gender, race, and ethnicity indicate a much larger projected growth in care gaps for Black older adults than White older adults, which is driven partly by the notable expected increase in the number of never-married Black adults without children (Verdery and Margolis 2017).
Sensitivity Analysis
In addition to the main results examining family structure and care gaps, we measured respondents’ coresidence and geographic proximity to adult children. The results from these models are shown in Tables A4 and A5. We find no differences in care gaps for ADLs (Table A4) and IADLs (Table A5) between partnered respondents, regardless of whether their adult children were coresident, lived within 10 miles, or lived farther than 10 miles away. Next, we turn to unpartnered respondents, who are more likely to rely on children for help. We find that for ADL care gaps (Table A4), respondents with coresident children had a higher likelihood of care gaps than partnered adults but a lower likelihood than unpartnered adults whose children live nearby (within 10 miles) or farther away. For IADL care gaps, unpartnered respondents with coresident children were comparable to partnered respondents, and unpartnered respondents with noncoresident children had a higher likelihood of an IADL care gap. For both ADLs and IADLs, kinless respondents had a higher likelihood of care gaps than unpartnered adults with noncoresident children, indicating an increased risk of care gaps among kinless adults.
A second set of sensitivity analyses tests whether the results differ for measures of family structure that include children of all types (stepchildren, adopted children) versus only biological children. These results, comparable to those shown in the main analysis (Tables 3 and 4), appear in Tables A6 and A7. We find the same patterns as those reported in the main analysis.
Discussion
Care Gaps in the United States
As the U.S. population ages and families shift, many people may have trouble getting the care they need with personal activities, mobility, and household tasks. Our analysis highlights the contemporary care gap among U.S. adults 50 or above and finds that care gaps are very common. Among U.S. adults aged 50 or older, 13% had at least one ADL care gap (i.e., they reported difficulty with a task but received no help with it in the last three months), 4.7% had two or more, and 1.9% had three or more. The ADLs with the highest prevalence of a care gap were dressing (4.7%), walking (4.0%), toileting (3.9%), and getting in or out of bed (3.7%). IADL care gaps were slightly less common: among adults aged 50 or older, 5.2% reported at least one IADL care gap, and 0.8% reported two or more. Between 1.0% and 1.5% of adults aged 50 or older reported difficulty with any IADL task but received no help in the last three months. These descriptive statistics are stark, highlighting that by 2015, 9.2 million and 3.7 million U.S. adults will have an ADL and IADL care gap, respectively.
Our measure of the care gap captures respondents who reported difficulty with a task but received no help with that task in the last three months. This measure adequately identifies the gaps that could be filled with help, but it is imperfect. It does not capture those receiving insufficient help with a task. For example, this measure would consider respondents who received only occasional or sporadic help dressing as receiving some care and not having a care gap for this particular task. Further, our measure does not perfectly account for respondents’ personal preferences for help. A respondent might report a task difficulty but prefer to undertake the activity without help or with adaptive equipment.
Family Structure and Care Gaps
Our analysis highlights the importance of family structure for shaping care gaps among older adults in the contemporary United States. Kinless older adults had the highest ADL and IADL care gap rates for all three care gap levels (at least one, at least two, and at least three care gaps). Partnered respondents with or without children had the lowest likelihood of care gaps, and unpartnered respondents with children had a care gap likelihood between those with both types of kin and those with none. The sensitivity analysis also indicates that having a partner might be more important for care needs than having children, regardless of children's type or proximity. Why is the presence of adult children not filling care gaps? Some studies indicated that adult children might misidentify their older parents’ needs (Walz and Mitchell 2007) or lack the skills or availability to be effective caregivers (Brank and Wylie 2016), leading to care gaps. Thus, even though unpartnered adults with children might receive some help from their children, care gap levels could be quite similar to those of unpartnered adults without children.
Kinless older adults might still have access to other family members in their broader network (Reyes et al. 2021) and friends who could provide ADL and IADL help (Mair 2019). Just as perceptions of access to social support from family increase in older age, so do perceptions of access to social support from friends, albeit more weakly (Verdery and Campbell 2019). Among those receiving substantial help with healthcare activities, roughly 17% report help from other relatives, and 3% report help from nonrelatives (Wolff et al. 2016). Among nonfamily caregivers, friends are more likely to provide help with personal care than neighbors (LaPierre and Keating 2013), and some nonkin caregivers report being the primary or sole caregiver for a dependent older adult (Barker 2002). However, despite research findings that broader network members do help out, kinless older adults in the United States remain distinctly disadvantaged: they are much more likely to report difficulty with key tasks but receive no help with them. Future work can explore the reasons for such care gaps, determining whether older adults report difficulty but do not want help, are too embarrassed to ask for help, or have no one to turn to.
Covariates and Care Gaps
Some demographic and socioeconomic indicators are strongly associated with whether older adults receive support or care when needed or whether they have care gaps. First, men were more likely to report care gaps than women, perhaps because women have more extensive care networks (Lima and Allen 2001). Second, non-Hispanic Black, Hispanic, and other adults of other races and ethnicities are less likely than non-Hispanic White older adults to have care gaps. The cultural values of family obligations to provide caregiving might encourage non-Hispanic Black and Hispanic adults to help other family members, especially Black adults, who are likely to help friends or neighbors (Dilworth-Anderson et al. 2005; Dilworth-Anderson et al. 2002). Thus, even unpartnered adults without children might still receive some help. Third, despite some evidence that economically disadvantaged older adults might not have enough care resources (Wiltshire et al. 2009), our findings did not demonstrate a clear pattern of care gaps by socioeconomic status. We found that respondents with little wealth were not the most likely to have ADL care gaps, perhaps because Medicare provides some support (Dunlop et al. 2002). Higher educated adults with IADL difficulty are more likely to have care gaps, possibly because those with more education tend to have less social support from family than those with less education (Verdery and Campbell 2019). This association between care gaps and education may be related to higher levels of individualism, emphasizing independence, autonomy, and looser kinship ties among the White middle class (Pyke and Bengtson 1996). Finally, we found a lower likelihood of care gaps among adults living in a nursing home than among other adults. Future research can further examine these important covariates of care gaps.
Looking to the Future
We project that the number of adults aged 50 or older with care gaps will increase by more than 30% from 2015 to 2050—twice the rate of population growth in this age group—because of shifts in family structure, population aging, and population growth. First, the shifting family structures of older adults in the United States means that in the future, fewer older adults will be partnered, more will be childless, and those with children will have fewer children, on average. As a consequence of having fewer children, more people will not have sons (Allendorf 2015; Pandian and Allendorf 2022) or, perhaps more importantly for caregiving, daughters, who tend to provide more care to parents (Kasper et al. 2015). Because partners and children are the family members who provide most unpaid family caregiving, lacking close kin may leave a gap to fill in caregiving. If other relatives, friends, neighbors, or paid workers do not step in to assist older adults, increasing numbers of adults will be left without assistance, potentially accelerating health declines, reducing independent living, and increasing healthcare costs. Population aging and population growth also contribute to care gaps as the average age of our population increases and the number of older adults increases. If disability is delayed in the life course, caregivers for older adults may themselves become older with each successive cohort, introducing new potential gaps in caregiver availability.
We note several important caveats to this study. First, we rely on survey questions about whether respondents received any help with each task in the last three months. In a sense, this measure is conservative because respondents could be getting some, albeit insufficient, help. Other data sources could be used to investigate whether care needs are being met sometimes, most of the time, or all the time and whether these frequencies vary by task type. Daily diary methods, for instance, could illuminate day-to-day care gaps and the possible daily stresses they might induce (Almeida et al. 2005). Moreover, future research could investigate the influence of supplemental help for care gaps, including the changing role of technology and other assistive devices, nursing home use, and other professional care services.
A second issue to note is that we do not model cohort or period changes in ADLs and IADLs. Our analysis of the HRS data tested for interactions between age and cohort in the probability of having ADL difficulties and found no significant interactions for any gender and racial or ethnic group except for Black women. Because most of the interactions we measured were insignificant, we did not explicitly model changes over time in ADL or IADL difficulties among future cohorts of older adults. However, recent research highlighted poorer age-specific health in the cohorts now aging into older adulthood, ages not yet included in the HRS. For example, increases in ADL and IADL difficulties are expected in the future because of large increases in obesity among adults aged 45–64 (Zajacova and Montez 2018). Future analyses can examine how unmet caregiving needs will be affected by a potential increase in the proportion needing ADL help, perhaps at younger ages.
A third caveat that can be addressed in future research is that these demographic changes are accompanied by social changes beyond family structure that affect caregiver availability. Shifts in families’ geographic proximity, women's labor force participation at all ages, and possibly caregiving norms will shape whether those with family members will have someone to care for them. As gender norms continue to shift in the labor force and at home (e.g., more acceptance of women in work, the normalization of men providing care), the associations and traditional gendered patterns of caregiving could change. In addition, family members’ health status (e.g., a partner's self-rated health or medical conditions) might impact their ability to help older adults. In an additional sensitivity analysis (not shown), we found that respondents whose spouse/partner self-reported poor health were more likely to have care gaps than those whose partners self-reported good health. Marriage is certainly not a panacea for addressing all care needs, especially when a partner has health problems.
Furthermore, studies differ in how they inquire about ADL and IADL caregiving. First, the HRS uses a three-month window of caregiving, potentially excluding people who rarely need help. Second, we measure care gaps by examining whether respondents reported receiving no help with a given task, including help from various sources (family and professional, paid caregivers). Some surveys (e.g., the National Health and Aging Trends Study, the National Study of Caregiving) also interview caregivers, potentially allowing researchers to address differences in care reporting between children and parents.
Finally, selection might influence our results. The sample composition changes as respondents age, potentially shifting the socioeconomic composition of those analyzed. Our use of sample weights partially adjusts for this phenomenon. However, research should delve deeper into the educational differences in IADL care gaps we find and consider socioeconomic differences in the dynamic changes in health status and care needs across the life course.
Conclusions
Who will provide help to older adults requiring help with basic life tasks is a critical question when aiming to improve older adults’ lives today and in the future. Who will unpartnered and childless older adults turn to as their health declines? Family is just one part of the support network available to older adults, but it is often a crucial one. Many older adults, especially those who are unpartnered and childless, have close friends and relatives in their network and have fostered these relationships over time. On the other hand, nonfamily ties might be weaker, especially for individuals unexpectedly without close kin in older age (Roberts et al. 2018). The U.S. long-term care system relies heavily on family for providing care, contributing to severe financial, physical, and mental burdens on individuals. For these reasons, it is important to track caregiving trends (Gaugler 2021). A lack of available family and unpaid caregivers will likely increase demands for paid care and institutionalization (Pickard et al. 2012), as well as lead to more older adults forgoing care and living with care gaps. Our analysis highlights the need for a thoughtful discussion about how we might fill the already-large care gaps that will increase substantially through 2050.
Acknowledgments
We thank the participants at the Michigan Center on the Demography of Aging (MiCDA) and National Study of Caregiving (NSOC) Demography of Family Caregiving Network Annual Meeting (May 2022) for feedback. We acknowledge support from the National Institute on Aging (1R01AG060949, T32AG000221, K99AG073473); the Penn State Population Research Institute, which is supported by an infrastructure grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2C-HD041025); the Government of Canada–Canadian Institutes of Health Research (MYB-150262) and Social Sciences and Humanities Research Council (435-2017-0618 and 890-2016-9000); and the Population Studies Center at the University of Michigan, funded by a NICHD Center Grant (P2CHD041028). The content is solely the authors’ responsibility and does not necessarily represent the official views of the National Institutes of Health or other funding sources. The Health and Retirement Study is sponsored by the National Institute on Aging (U01AG009740) and conducted by the University of Michigan.
