Abstract
Objectives
We provide national estimates of caregiving networks for older adults with and without dementia and examine how these networks develop over time. Most prior research has focused on primary caregivers and rarely on change over time.
Method
We identify a cohort of older adults continuously followed in the National Health and Aging Trends Study between 2011 and 2015 and receiving help from family members or unpaid caregivers in 2015 (n = 1,288). We examine differences by dementia status in network size, types of assistance and task sharing, and composition—differentiating between “specialist” and “generalist” caregivers helping in one versus multiple activity domains. Multinomial regression is used to estimate change over time in network task sharing and composition.
Results
In 2015, older adults with dementia had larger caregiving networks involving more task sharing than those without dementia and more often relied on generalist caregivers, especially the subset assisting with medical, household, and mobility or self-care activities. Uniformly greater reliance over time on these more intensely engaged generalist caregivers chiefly accounts for larger dementia networks.
Discussion
Findings lend support to the need for caregiver training on managing multiple task domains and—for dementia caregivers in particular—task-sharing skills. More generally, the design of new approaches to better support older adults and their caregivers should consider the complexity, heterogeneity, and change over time in caregiving networks.
Keywords: Caregivers, Community-based care, National Health and Aging Trends Study, Task sharing
About 90% of older community residents with disabilities receive care from family or unpaid caregivers, and about two thirds receive assistance only from these informal sources (Freedman & Spillman, 2014; Kaye, Harrington, & LaPlante, 2010; Spillman & Black, 2005). Dementia exacerbates caregiving demands: although about 10% of older adults have dementia, their family and unpaid caregivers provide 41% of aggregate care hours for older adults (Kasper, Freedman, Spillman, & Wolff, 2015). Dementia caregivers provide more hours of assistance on average than non-dementia caregivers, even though dementia care networks generally are larger. Dementia caregivers also more often report substantial negative consequences of caregiving, such as exhaustion and lacking time for themselves (Spillman, Wolff, Freedman, & Kasper, 2014).
In addition to assisting with self-care, mobility, and household activities, family and unpaid caregivers also play an important role in health care for older adults. Medical-related assistance, including medication management and interaction with providers, is more common among older adults with dementia and is associated with significantly higher caregiver time demands, emotional difficulty, restriction of participation in valued activities, and work productivity loss for family and unpaid caregivers (Wolff & Spillman, 2014; Wolff, Spillman, Freedman, & Kasper, 2015).
The literature recognizes that older adults typically rely on a network of caregivers (Andersson & Monin, 2018; Keating, Otfinowski, Wenger, Fast, & Derksen, 2003), but relatively little attention has been paid to how these networks develop over time for older adults with and without dementia. Previous research examining prototypical late life illness and functioning trajectories, however, has identified dementia/age-related frailty as involving a longer period of low and declining function and high care needs relative to other illness trajectories (Freedman et al., 2006; Lunney, Lynn, Foley, Lipson, & Guralnik, 2003). A better understanding of whether and how network development differs by dementia status could prove useful in timing the introduction of strategies to enhance skills within caregiving networks and extending the reach of such efforts beyond a single “primary” caregiver.
We use data from the National Health and Aging Trends Study (NHATS) covering the period 2011–2015 to examine (a) the association of dementia with network size, the types of assistance provided, task sharing, and the composition of caregiving networks in 2015, and (b) how the types of caregivers, task sharing, and network composition for those with and without dementia changed between 2011 and 2015. Given the prior evidence of greater demands on caregivers, larger caregiving networks, and a longer period of high care needs for older adults with dementia, we expect differences by dementia status in task sharing, network composition, and potentially the ways and pace at which dementia networks change over time.
Background and Conceptual Framework
Most research on caregiving for older adults has focused on a single “primary caregiver” or the experiences of caregivers at a point in time (Andersson & Monin, 2018; Cornwell, Schumm, Laumann, Kim, & Kim, 2014; Keating et al., 2003; Szinovacz & Davey 2013). Szinovacz & Davey (2013), focusing on change in parental care among adult children over a 2-year period, found that more than one quarter of adults identified as the primary caregiver were replaced by a sibling, and two thirds of networks added or dropped sibling caregivers. Miller and McFall (1991) found that increases in network size and care intensity over a 2-year period were associated with increases in the care recipient’s number of activity limitations but not with primary caregiver burden. Allen, Lima, Goldscheider, & Roy (2012) found that more than half of adults aged 70 or older with a primary informal caregiver transitioned over 2 years to a different primary caregiver. Having other informal caregivers in the network was associated with a twofold likelihood of transition to a different informal caregiver, whereas having formal helpers increased the likelihood of a transition to a formal primary caregiver or an institutional care setting. None of these studies specifically examined whether the characteristics of individuals’ caregiving networks changed over time or differences in caregiving networks by dementia status.
The latter two studies were informed in part by the task-specific model of caregiver selection developed by Litwak (1985), which provides a generalized conceptual framework for caregiving network development based on a combination of the fit between services required and the potential caregivers available: relatives, friends, neighbors, and paid caregivers. The model incorporates as a special case the hierarchical compensatory model (Barker, 2002; Cantor, 1979), which posits an ordered preference for caregivers based on the primacy of the relationship between the caregiver and older recipient (Messeri, Silverstein, & Litwak, 1993). Like the hierarchical compensatory model, the task-specific model has been used to explain observed caregiving patterns in cross-sectional samples, not to examine differences between networks for those with and without dementia. Service characteristics include whether a task requires ongoing contact and substantial time or resources, benefits from shared social and life experience of caregiver and recipient, and can reasonably be divided among caregivers (Messeri et al., 1993). Key factors for caregiver-task fit thus are proximity, long-term commitment, commonality of lifestyle or social roles, and size of the group. Spouses typically meet all requirements for fulfilling ongoing and time-consuming basic tasks, such as personal care assistance or meal preparation, and proximate children or other relatives may supplement or replace their efforts. Friends or neighbors are more likely to provide social support or assistance with intermittent or ad hoc tasks, but may be the optimal choice to undertake or share some tasks, for example, when a spouse or other relatives are incapable or not present for tasks requiring proximity. The model predicts that networks will adjust when current network configuration is, or becomes, suboptimal to meet recipient needs, depending on care resources available, as indicated by the results in Allen and colleagues (2012). The model also can be extended to paid caregiving, which can supply technical skills or increase network size and capacity in terms of time and resources, for example, providing night or 24-hr care, although we do not explicitly do so in this analysis.
Consistent with both the hierarchical compensatory and task-specific model predictions for caregiver fit, research has uniformly indicated that spouses and children are the main source of assistance to older adults with disabilities, followed by other relatives, neighbors, or friends, depending on the potential caregivers available (Spillman & Black, 2005; Spillman & Pezzin, 2000; Wolff & Kasper, 2006). In 2011, 60% of caregivers to older adults in settings other than nursing homes were spouses or children; 8% were other household members; 22% were other relatives, social network members, or other unrelated persons; and only 10% were unrelated paid caregivers (Freedman & Spillman, 2014).
We focus on the task-side of Litwak’s framework to characterize types of tasks performed by caregivers and the composition of the caregiving network as a whole and examine how task sharing and network composition change over time for those with and without dementia. To do so we group specific tasks into four care domains: medical activities, household activities, mobility or self-care activities—the three major domains associated with the need for proximity and ongoing time commitment—and transportation, which may be intermittent or ad hoc. We examine task sharing across network members (whether multiple caregivers undertake tasks in the same domain) and the extent to which caregivers are specialists (helping in a single domain) or generalists (helping in multiple domains). For generalist caregivers, we further distinguish more intensely engaged caregivers who help with activities in all three major domains of care (medical activities, household activities, and mobility or self-care) from those who help with activities in other domain combinations, but not all three major domains.
Given the greater demands of dementia caregiving and the larger caregiving networks previously observed for older adults with dementia, we hypothesize a greater degree of task sharing and greater presence of generalist caregivers taking on a broad array of responsibilities. To underpin hypotheses regarding network change over time, we draw on the prior evidence on prototypical late life illness and functioning trajectories indicating that dementia is characterized by years of low and declining function and high care needs as opposed to other disabling conditions characterized by a shorter course of declining function and rising care needs with periodic exacerbations or abrupt declines in function only shortly before death (Freedman et al., 2006). Thus, we expect that the longer duration of high and increasing care needs for those with dementia may require earlier marshaling of a network of committed caregivers that adjusts as needed but remains relatively stable over time. Conversely, the shorter course of accelerating care needs for those with disabilities arising primarily from physical illness may imply the need for faster-paced adjustments to network size and complexity.
Method
Data and Analysis File Structure
NHATS, a nationally representative longitudinal study of Medicare enrollees aged 65 or older, collects annual information on the full caregiving networks of older adults receiving assistance. NHATS obtains reports of help in the last month with routine activities and identifies all persons assisting with each activity. This information allows both characterization of networks for older adults in cross-section at each survey round and examination of changes in network characteristics over time.
NHATS began in 2011 with a cohort of 8,245 Medicare beneficiaries aged 65 or older living in all settings, replenished in 2015 to restore national cross-sectional representation. Of 8,334 NHATS sample members in 2015, we selected the 3,627 surviving respondents from the 2011 cohort who were living in settings other than a nursing home. We eliminated those receiving no assistance (n = 2,266) or assistance from only unrelated paid caregivers (n = 79), and 24 older adults who were nonrespondents in any intervening year between 2011 and 2015. The resulting analysis cohort is 1,288 older adults receiving care from family members or unpaid caregivers in 2015 and continuously followed between 2011 and 2015. This selection provides five annual observations for each cohort member, a total of 6,440 person-year observations on the care history of individuals receiving assistance from family or unpaid caregivers in 2015, many of whom began to receive care during the observation period. The sample design offers an observable history of care for persons with and without dementia in 2015 and the ability to directly control for care duration.
We derived data on older adults from the NHATS annual sample person (SP) file which contains responses during in-person interviews regarding participants’ living situation, demographics, socioeconomic status, physical and cognitive functioning, help with routine activities, and who provided assistance. In 2015, about 18% of the analysis sample members had a proxy respondent, of whom about one quarter had a proxy only in 2015, another quarter had a proxy in all 5 years, and nearly half had a proxy in 2–4 rounds.
We constructed measures of the full caregiving network for each NHATS respondent in each round, using the NHATS 2015 other person (OP) file, a cumulative file over all waves (Kasper & Freedman, 2016). The OP file contains a roster of unique persons identified by the older adult, their relationship to the older adult, whether they assisted the older adult, and the activities with which they helped in each round. The roster includes spouses, all children, all household members, and up to five social network members, all of whom may or may not be identified as a caregiver in a given round, as well as any other persons the older adult identified as having provided assistance. Thus, the 2015 OP file contains a complete catalog for the years 2011–2015 of roster members in the current or any previous round, whether they provided assistance, and the activities with which they helped.
Older Adult Characteristics
Key older adult characteristics for this analysis are the types of assistance received and the presence of dementia. We examined help in the month prior to each NHATS interview with any of seven mobility or self-care activities or with any of four household activities or keeping track of medications (for health or functioning reasons). Within the sample of older adults reporting these types of assistance, we also examined help with medical care-related activities and transportation.
Our dementia measure is a classification of probable dementia in 2015 based hierarchically on (a) reported diagnosis by a physician in 2015 or any previous survey round; (b) a proxy-reported score of 2 or higher on the AD8 Dementia Screening Interview in 2015 or a previous round; or (c) a score less than or equal to 1.5 SD below the mean in at least 2 of 3 domains evaluating memory, orientation, and executive function. Details including programming language and analyses of cross-sectional specificity and sensitivity are available in Kasper, Freedman, and Spillman (2013).
Sociodemographic characteristics included are age (69–74, 75–84, and 85 or older), sex, marital status (not married vs married), number of children (none, 1–2, 3+), and education (less than high school, high school or vocational education beyond high school, any college) (Freedman & Spillman, 2014; Freedman, Kasper, Spillman, & Plassman, 2018; Sharp & Gatz, 2011). We also included the number of consecutive NHATS survey rounds (1–5) in which care was received from 2011 to 2015 as a proxy for the length of the assistance episode observed in 2015. A value of 1 indicates that the older adult reported assistance only in 2015, a value of 2 indicates assistance in 2015 and 2014, and so on.
Network Member Characteristics
We developed network characteristics for each older person starting at the individual caregiver level by selecting all roster members from the 2015 OP file (n = 11,063) associated with the 1,288 older adults in our cohort. First, we identified for each survey round whether an OP file member provided assistance. Next, we categorized the activities each caregiver assisted with in each survey round into the four domains of assistance identified earlier: medical, household activities, mobility or self-care, and transportation. Medical activities are attending medical visits with the older adult and helping the older adult keep track of medications. Household activities are laundry, shopping, preparing hot meals, or paying bills and banking. Mobility or self-care activities are getting around outside, getting around inside, getting out of bed, eating, bathing/showering or washing up, getting to or using the toilet, or dressing. Transportation is going places outside of the home.
At each survey round, we distinguished between caregivers who were specialists in a single activity domain or generalists assisting in multiple domains. We further defined two mutually exclusive categories of generalist caregivers: (a) those assisting in all major domains (who may also assist with transportation) and (b) those assisting in other domain combinations.
Network Measures
Collectively, network measures characterize caregiving activities in terms of sharing tasks and scope of assistance (e.g., specializing). To derive these measures, we summarized the round-by-round caregiver-level information to the older adult level. In each survey round, we identified the total number of caregivers; the number of caregivers providing assistance in each activity domain; and the number who are specialists or generalists by type. For each of these counts in each round, we constructed three-level indicators: no caregivers, one caregiver, or multiple caregivers. Task sharing occurs when multiple caregivers engage in tasks in the same domain. They may undertake different “assignments” of activities within a domain (e.g., different household activities—one cooking, one helping with banking and paying bills) or jointly take on the same activity (e.g., alternating meal preparation). In any given year, an older adult’s network may include any combination of specialists and generalists of one or both types. Finally, as a global characterization of network composition, we constructed a three-level measure for each round: specialist caregivers only, generalist caregivers only (combining the two types of generalist caregivers), or both.
Statistical Approach
We first developed descriptive estimates profiling differences in characteristics of older adults with and without dementia in the analysis sample and their caregiving networks in 2015. We then estimated multinomial logit models to examine whether network development over time differed for older adults with and without dementia. The first model assesses the probability of having (a) one or (b) multiple specialist caregivers (Model 1) relative to none. The second and third models estimate the probability of having (a) one or (b) multiple generalist caregivers of the two types of generalists, relative to having none (Models 2 and 3). The fourth model assesses the probability that a network is composed of (a) only generalist caregivers or (b) a combination of specialist and generalist caregivers, relative to having only specialist caregivers. In each model, key explanatory variables are an indicator of dementia in 2015, a time variable to capture change over time (valued 0 in 2011, the base year, to 4 in 2015), and an interaction between dementia and time to test whether change over time differed by dementia status. To differentiate the effects of dementia from differences in other characteristics of those with and without dementia, models control for the older adult characteristics described earlier.
We produced estimates using weights that properly represent the analysis sample (persons from the 2011 cohort receiving care in 2015) and statistical software designed to adjust standard errors for survey design (DeMatteis, Freedman, & Kasper, 2016). To account for the multiple observations on each older adult, we created an adjusted stratum variable by appending a unique identifier for each respondent to the stratum variable provided with the data. This effectively creates a unique stratum for each sample member and adjusts for unmeasured heterogeneity at the individual level. We used SAS (version 9.4) survey procedures for descriptive estimates and performed tests of differences between those with and without dementia. Multinomial logit models were estimated using Stata svy: mlogit commands (version 15.1). Results are presented as relative risk ratios (RRRs), which indicate how the relative risk of being in a particular outcome group (e.g., one or multiple caregivers), relative to the relative risk of being in the omitted group (no caregivers) changes with the value of an explanatory variable. Because goodness of fit measures are not available for svy commands, we also estimated the models using mlogit, with person weights and clustered at the individual level to obtain a Wald chi-squared statistic assessing the explanatory power of the model relative to an empty model with no independent variables. We used predicted values produced through Stata margins commands to examine network change over time, adjusted for differences in characteristics of those with and without dementia.
Results
Descriptive Analysis
Consistent with previous findings in the literature, our estimates indicate that older adults with dementia were older, more commonly unmarried, and had lower levels of education than those without dementia (Table 1). Most notably, episodes of care were longer for those with dementia, with 58% having received help in all five NHATS rounds, compared with 31% of older adults without dementia.
Table 1.
Characteristics of Care Recipients With and Without Dementia in 2015 (Older Adults From 2011 Cohort Receiving Care in 2015)
| Dementia | No dementia | |
|---|---|---|
| Percent | Percent | |
| All | 100.0 | 100.0 |
| Age | ||
| 69–74 | 12.6* | 23.0 |
| 75–84 | 40.7* | 47.9 |
| 85+ | 46.7* | 29.0 |
| Female | 63.3 | 67.6 |
| Unmarried | 64.6* | 54.1 |
| Number of children | ||
| None | 8.1 | 8.6 |
| 1–2 | 34.0 | 39.6 |
| 3+ | 58.0 | 51.8 |
| Education | ||
| Less than high school | 41.8* | 23.3 |
| High school/trade | 30.2 | 35.8 |
| Some college | 28.0* | 40.8 |
| Not white/non-Hispanic | 28.4 | 22.5 |
| Number of NHATS rounds in current (2015) assistance episode | ||
| 1 | 13.1* | 30.0 |
| 2 | 9.5* | 16.6 |
| 3 | 10.3 | 11.3 |
| 4 | 9.2 | 11.4 |
| 5 | 58.0* | 30.7 |
Notes: National Health and Aging Trends Study (NHATS) 2015 continuing sample from 2011 cohort (n = 1,288).
*p value <.05 for differences between care recipients with and without probable dementia.
Task sharing
By construction, all analysis sample members had at least one caregiver in 2015, but older adults with dementia more often had multiple caregivers than those without dementia (77.8% vs 71.7%; Table 2). Older adults with dementia also more often had caregivers who were sharing responsibilities for medical care, household activities, and mobility or self-care activities. The difference in caregiver task sharing for older adults with and without dementia was most pronounced for medical assistance (32.8% vs 16.4%) and mobility or self-care assistance (39.4% vs 22.5%). Those without dementia more often had multiple caregivers assisting with transportation (45.1% vs 36.9%).
Table 2.
Distribution of Care Recipients With and Without Dementia by the Number of Caregivers and Assistance Domain in 2015 (Older Adults From 2011 Cohort Receiving Care in 2015)
| Number of caregivers by assistance domain | Dementia | No dementia |
|---|---|---|
| Percent | Percent | |
| Assistance in any domain | ||
| 1 | 22.2* | 28.3 |
| 2+ | 77.8* | 71.7 |
| Medical-related assistance | ||
| 0 | 15.0* | 35.1 |
| 1 | 52.2 | 48.5 |
| 2+ | 32.8* | 16.4 |
| Household activity assistance | ||
| 0 | 1.4* | 5.5 |
| 1 | 41.8* | 50.2 |
| 2+ | 56.8* | 44.3 |
| Mobility or self-care assistance | ||
| 0 | 29.9* | 42.1 |
| 1 | 30.6 | 35.5 |
| 2+ | 39.4* | 22.4 |
| Transportation assistance | ||
| 0 | 23.8 | 23.2 |
| 1 | 39.3* | 31.7 |
| 2+ | 36.9* | 45.1 |
Notes: National Health and Aging Trends Study 2015; persons from 2011 cohort receiving care from a family or unpaid caregiver in 2015 (n = 1,288).
*p value <.05 for difference between care recipients with and without probable dementia.
Network composition
Older adults with dementia more commonly had no specialist caregivers in their networks than those without dementia (41.7% vs 33.1%), although the proportion with multiple specialist caregivers was similar in the two groups (30.1% vs 30.5%; Table 3). In addition, older adults with dementia more often had generalist caregivers who helped in all major domains—43.5% had one and 13.6% had multiple, compared with 30.2% and 5.6%, respectively, of those without dementia. The prevalence of generalist caregivers who helped with activities in other domain combinations did not differ by dementia status.
Table 3.
Distribution of Care Recipients With and Without Dementia by the Number of Specialist and Generalist Caregivers and Network Composition in 2015 (Older Adults From the 2011 Cohort Receiving Care in 2015)
| Number of specialist and generalist caregivers in networks | Dementia | No dementia |
|---|---|---|
| Percent | Percent | |
| Specialist (single-domain) caregivers | ||
| 0 | 41.7* | 33.1 |
| 1 | 28.2* | 36.4 |
| 2+ | 30.1 | 30.5 |
| Generalist (multiple-domain) caregivers | ||
| All major domainsa | ||
| 0 | 42.9* | 64.2 |
| 1 | 43.5* | 30.2 |
| 2+ | 13.6* | 5.6 |
| Other domain combinationsb | ||
| 0 | 35.1 | 34.9 |
| 1 | 42.7 | 47.2 |
| 2+ | 22.3 | 17.9 |
| Network composition | ||
| Specialist caregivers only | 6.5* | 13.6 |
| Generalist caregivers only | 41.7* | 33.1 |
| Both | 51.8 | 53.4 |
Notes: National Health and Aging Trends Study 2015; persons from 2011 cohort receiving care from a family or unpaid caregiver in 2015 (n = 1,288).
aHelps with all three major domains (medical, household, and mobility/self-care activities); may help with transportation.
bHelps with more than one of the four domains, but not all three major domains.
*p value <.05 for differences between care recipients with and without probable dementia.
The final measure in Table 3 shows overall network composition with respect to specialist and generalist caregivers. About half of older adults with or without dementia had both specialist and generalist caregivers in their networks, but networks consisting exclusively of generalist caregivers were more common for persons with dementia (41.7% vs 33.1%), and networks consisting exclusively of specialist caregivers were less common (6.5% vs 13.6%).
Regression Analysis
Task sharing
Multinomial logit regression results (Table 4 and Supplementary Table 1) assess the association of dementia with how task sharing in caregiving networks changed over time. We present RRRs for having one or multiple specialist caregivers (Model 1), and generalist caregivers who assist in all major domains (Model 2) or in other domain combinations (Model 3). Considering Model 2 first, the relative risk in 2011 of having one generalist caregiver helping with all major activity domains (vs no such caregivers) for those with dementia in 2015 was 2.74 times the relative risk for those without dementia. The time parameter (RRR = 1.43) indicates an increasing probability over time for those without dementia in 2015. Finally, the relative risk for the interaction between dementia and time indicates that the time effect was less steep (RRR = .883) for those with dementia. That is, older adults with dementia had a higher initial relative risk of having one generalist caregiver in all major domains in 2011, and a slower rate of increase in relative risk over time than those without dementia. Similarly, the relative risk of multiple generalist caregivers helping in all major domains was far higher (RRR = 4.97) for those with (vs without) dementia, with an increase over time for those without dementia (RRR = 1.86). The interaction term again indicates slower increase for those with dementia but is not statistically significant at conventional levels (p value = .07).
Table 4.
Relationship Between Dementia, Time, and the Likelihood of Having One or More Specialist and Generalist Caregivers Over Time, 2011–2015 (Relative Risk Ratios)
| Number of caregivers by type (relative to none of that type) | Specialist caregivers | Generalist caregivers | |
|---|---|---|---|
| All major domainsa | Other domain combinationsb | ||
| Model 1 | Model 2 | Model 3 | |
| 1 caregiver | |||
| Probable dementia | 0.806 | 2.740* | 1.195 |
| Time | 1.019 | 1.426* | 1.117* |
| Interaction | 0.988 | 0.883* | 0.928 |
| 2+ caregivers | |||
| Probable dementia | 0.856 | 4.970* | 1.432 |
| Time | 1.130* | 1.856* | 1.335* |
| Interaction | 0.963 | 0.786 | 0.889 |
Notes: Multinomial logit regression using National Health and Aging Trends Study 2015; persons from 2011 cohort receiving care from a family or unpaid caregiver in 2015 (n = 6,440 observations in 1,288 clusters at the cohort member level). Key covariates are dementia in 2015; time, valued 0 for 2011 to 4 for 2015; and an interaction between the two. Other covariates are age, sex, marital status, number of children, education, race/ethnicity, and number of survey rounds receiving care (see Table 1).
aHelps with medical, household, and mobility/self-care activities.
bHelps in multiple domains, but not all three major domains.
*p value <.05.
Considering specialist caregivers in Model 1, there was no difference by dementia in the relative risk of having either one or multiple specialist caregivers. The relative risk of having multiple specialist caregivers increased with time but did not differ by dementia. Finally, the relative risk of having one or multiple generalist caregivers assisting in other domain combinations (Model 3) increased with time for those without dementia (more steeply for having multiple such caregivers), but no dementia differential is evident for having one or multiple caregivers of this type or change over time.
Network composition
Table 5 and Supplementary Table 2 present results for models examining how network composition changed between 2011 and 2015. The relative risk of having only generalist caregivers in 2011 was higher by a factor of 1.81, and the relative risk of having both specialist and generalist caregivers was higher by a factor of 1.72 for those with (vs without) dementia in 2015. Again, the RRRs associated with time are greater than one, indicating that networks including generalist caregivers increase over time for those without dementia, but the interaction terms indicate no differential change over time for those with dementia.
Table 5.
Relationship Between Dementia, Time, and Network Composition and Presence of Specialist and Generalist Caregivers Over Time, 2011–2015 (Relative Risk Ratios)
| Network composition (relative to specialist caregivers only) | |
|---|---|
| Generalist caregivers only | |
| Probable dementia | 1.813* |
| Time | 1.346* |
| Interaction | 1.025 |
| Both generalist and specialist caregivers | |
| Probable dementia | 1.724* |
| Time | 1.501* |
| Interaction | 0.973 |
Notes: Multinomial logit regression using National Health and Aging Trends Study 2015; persons from 2011 cohort receiving care from a family or unpaid caregiver in 2015 (n = 6,440 observations in 1,288 clusters at the cohort member level). Key covariates are dementia in 2015; time, valued 0 for 2011 to 4 for 2015; and an interaction between the two. Other covariates are age, sex, marital status, number of children, education, race/ethnicity, and number of survey rounds receiving care (see Table 1).
*p value <.05.
Predicted Trajectories
To illustrate model results, we mapped predicted values for the five survey rounds, from 2011 to 2015 (see Supplementary Table 3 for all predicted values by year and dementia status).
Task sharing
The shape and level of trajectories for specialist caregivers are similar for those with and without dementia, with the percentage of older adults having one specialist caregiver falling over time and the trajectories for having multiple specialist caregivers rising (Supplementary Figure 1). A smaller percentage of those with dementia have either one or multiple specialist caregivers, but the differences by dementia status are small.
Conversely, the percent with one generalist caregiver assisting in all major domains (Supplementary Figure 2) rises over time for those with and without dementia, but is higher throughout for those with dementia (rising from 25% in 2011 to nearly 40% in 2015) compared with those without dementia (rising from 13% in 2011 to 31% in 2015). The trajectories for having multiple generalist caregivers assisting in all major domains also are similar in shape but higher for those with dementia (rising from about 3% to nearly 9%), compared with those without dementia, with a relatively constant differential of 2–3 percentage points.
There is almost no difference by dementia status in the shape and level of trajectories for having either one or multiple network members who are generalist caregivers assisting in other combinations of domains (Supplementary Figure 3). Although the trajectories for having one such caregiver are essentially flat, however, the trajectories for having more than one rise from about 10% to 20% between 2011 and 2015.
Network composition
Figures 1 and 2 show the resulting change over time in the inclusion of specialist and generalist caregivers for those with and without dementia, respectively. The patterns are generally similar for the two groups, with networks made up only of specialist caregivers becoming less common over time, those with both specialist and generalist caregivers becoming more common, and those including only generalist caregivers relatively stable. Despite these similarities, important differences are evident. For older adults with dementia, networks consisting entirely of specialist caregivers are the least common network type throughout, decreasing from 28% in 2011 to 10% in 2015, whereas the most common network type, including both specialist and generalist caregivers, rises from 36% to 50% of networks. Conversely, for older adults without dementia, networks consisting entirely of specialists are initially the most common type, but decline by 23 percentage points, from 38% to 15% of networks, whereas mixed networks rise 20 percentage points, from 32% of networks in 2011 to 52% in 2015.
Figure 1.
Network composition, older adults with dementia.
Figure 2.
Network composition, older adults without dementia.
Discussion
We have taken a first step toward understanding how caregiving networks for a cohort of older adults with and without dementia develop over time by examining trajectories over the 5-year period ending in 2015. Consistent with cross-sectional studies (Kasper et al., 2015), we found that older adults with dementia had larger networks in 2015 than older adults without dementia. Except for transportation, older adults with dementia more often had multiple caregivers sharing responsibility within the same activity domain. This was true particularly for medical activities and mobility or self-care but also for household activities, all of which are types of assistance associated with a need for long-term caregiver commitment and proximity (Litwak, 1985; Messeri et al., 1993). The scope of activities undertaken by caregivers also differed by dementia status: older adults with dementia were more likely to rely on generalist caregivers, consistent with the greater intensity of dementia caregiving (Kasper et al., 2015).
This study provides novel insight into how caregiving networks change over time. We found that trajectories in network composition were qualitatively similar for those with and without dementia. Consistent with our expectation that the long course of care for those with dementia would require earlier assembly of a network of committed caregivers, however, in each year, older adults with dementia had a greater probability of having one or multiple generalist caregivers assisting in all major domains, suggesting that these caregivers account for their larger networks. Considered from the network composition perspective, dementia care networks including generalist caregivers, were far more common initially and increased relatively slowly over time. Conversely, for those without dementia, specialist-only networks initially were most common but were rapidly outstripped over time by networks including generalists.
Study results extend knowledge regarding the formulation of caregiving networks beyond change in a “primary” caregiver (Miller & McFall, 1991; Szinovacz & Davey, 2013). In particular, findings suggest that networks for those with and without dementia appear to adjust to improve or maintain fit with the recipient’s needs, as predicted by the task-specific model (Messeri et al., 1993), albeit at a more rapid pace for those without dementia. For example, our finding that networks with only specialist caregivers declined and those with both specialists and generalists increased steadily among older adults with and without dementia may indicate that specialist caregivers take on activities in additional domains or are added to generalist networks as older adults’ needs increase. The relative stability over time of networks with only generalist caregivers, which were more common for those with dementia, may mask a more complicated process with cycling into and out of this network configuration, consistent with the finding in Allen and colleagues (2012) that likelihood of a change to another “primary” informal caregiver was twofold for older adults with other informal caregivers in the network. Further caregiver-level research would be needed to identify such flows in caregiving careers.
This study has limitations. We focused on a sample of individuals receiving care in 2015 and examined network change for those with and without dementia over a 5-year history. A prospective approach to examining network development from initiation of help might offer different insights and shed light, for example, on whether and when specialist caregivers take on additional tasks, whether networks change after dementia diagnosis, and how paid care is incorporated. Our sample of older adults receiving family or unpaid care in 2015 excludes those receiving only paid care in 2015, and we did not differentiate paid from unpaid care within networks. Understanding how paid caregiving fits within the broader context of evolving network configurations merits attention and is a logical next step of this work.
Our innovation is the focus on caregiving networks and how they develop with respect to task sharing and specialization, rather than on a primary caregiver. As seen in longitudinal studies, change over time in the person identified as the primary caregiver is common and may be associated with poorer care coordination and older adult outcomes (Allen et al., 2012; Miller & McFall, 1991; Szinovacz & Davey, 2013). Moreover, primary caregivers have been defined using various criteria, including hours of care provided, type of help (e.g., personal care vs household tasks), recipient identification of “who helps the most,” or care coordination responsibility. This designation may often be unclear within multiple-caregiver networks like those prevalent in our study, and this lack of clarity may lead to uncertainty or disagreements about who fills the primary caregiver role (Marcum, Ashida, & Koehly, 2018). More generally, focusing on how networks develop and function with respect to task sharing and organization rather than a primary-caregiver-centered approach may provide new insights into network dynamics and the implications for design of interventions to improve older adult and caregiver health, well-being, and other outcomes, including service utilization, cost savings, and cost effectiveness (Andersson & Monin, 2018).
Interventions assessing or supporting caregivers and/or involving them as part of the health care team have shown promise for improving caregiver and older adult outcomes at critical junctures, such as hospital discharge, and for individuals with dementia, stroke, or other complex needs (e.g., Boult et al., 2011; Callahan et al., 2006; Coleman, Parry, Chalmers, & Min, 2006; Counsell et al., 2007; Elliott, Burgio, & DeCoster, 2010; Naylor et al., 2004; Nichols et al., 2017). Such interventions typically have focused on identifying the presence and capacity of a single “primary” caregiver, however, without explicitly considering how the larger caregiving network may work better together (National Academies of Sciences, Engineering, and Medicine, 2016), or, alternatively, how to mitigate challenging circumstances that inadvertently reduce coordination and disease management (Andersson & Monin, 2018). Our findings support the need for training on managing multiple task domains and—for dementia caregivers in particular—task-sharing skills. More generally, our findings suggest the value of considering the complexity, heterogeneity, and change over time of caregiving networks in the design of new approaches to improving care for older adults and supporting the caregiving networks they depend on.
Funding
This work was supported by the National Institute on Aging of the National Institutes of Health (R01AG047859 and U01-AG032947).
Author Contributions
B. Spillman designed the study, conducted the data analysis and interpretation of results, and wrote the article. V. Freedman, J. Kasper, and J. Wolff collaborated in design of the study, interpretation of results, and review and revisions of the article.
Conflict of Interest
None reported.
Supplementary Material
References
- Allen S. M., Lima J. C., Goldscheider F. K., & Roy J (2012). Primary caregiver characteristics and transitions in community-based care. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 67, 362–371. doi:10.1093/geronb/gbs032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Andersson M. A., & Monin J. K (2018). Informal care networks in the context of multimorbidity: Size, composition, and associations with recipient psychological well-being. Journal of Aging and Health, 30, 641–664. doi:10.1177/0898264316687623 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Barker J. C. (2002). Neighbors, friends, and other nonkin caregivers of community-living dependent elders. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 57, S158–S167. doi:10.1093/geronb/57.3.s158 [DOI] [PubMed] [Google Scholar]
- Boult C., Reider L., Leff B., Frick K. D., Boyd C. M., Wolff J. L., . . . Scharfstein D. O (2011). The effect of guided care teams on the use of health services: Results from a cluster-randomized controlled trial. Archives of Internal Medicine, 171, 460–466. doi:10.1001/archinternmed.2010.540 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Callahan C. M., Boustani M. A., Unverzagt F. W., Austrom M. G., Damush T. M., Perkins A. J., . . . Hendrie H. C (2006). Effectiveness of collaborative care for older adults with Alzheimer disease in primary care: A randomized controlled trial. JAMA, 295, 2148–2157. doi:10.1001/jama.295.18.2148 [DOI] [PubMed] [Google Scholar]
- Cantor M. H. (1979). Neighbors and friends: An overlooked resource in the informal support system. Research on Aging, 1, 434–463. doi:10.1177/016402757914002 [Google Scholar]
- Coleman E. A., Parry C., Chalmers S., & Min S. J (2006). The care transitions intervention: Results of a randomized controlled trial. Archives of Internal Medicine, 166, 1822–1828. doi:10.1001/archinte.166.17.1822 [DOI] [PubMed] [Google Scholar]
- Cornwell B., Schumm L. P., Laumann E. O., Kim J., & Kim Y. J (2014). Assessment of social network change in a national longitudinal survey. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 69 (Suppl. 2), S75–S82. doi:10.1093/geronb/gbu037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Counsell S. R., Callahan C. M., Clark D. O., Tu W., Buttar A. B., Stump T. E., & Ricketts G. D (2007). Geriatric care management for low-income seniors: A randomized controlled trial. JAMA, 298, 2623–2633. doi:10.1001/jama.298.22.2623 [DOI] [PubMed] [Google Scholar]
- DeMatteis J., Freedman V. A., and Kasper J. D (2016). National Health and Aging Trends Study development of round 5 survey weights. NHATS Technical Paper No. 14. Baltimore, MD: Johns Hopkins University School of Public Health; Retrieved June 1, 2019, from https://www.nhats.org/scripts/TechnicalWeighting.htm. [Google Scholar]
- Elliott A. F., Burgio L. D., & Decoster J (2010). Enhancing caregiver health: Findings from the resources for enhancing Alzheimer’s caregiver health II intervention. Journal of the American Geriatrics Society, 58, 30–37. doi:10.1111/j.1532-5415.2009.02631.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- Freedman V. A., Hodgson N., Lynn J., Spillman B. C., Waidmann T., Wilkinson A. M., & Wolf D. A (2006). Promoting declines in the prevalence of late-life disability: Comparisons of three potentially high-impact interventions. Milbank Quarterly, 84, 493–520. doi:10.1111/j.1468-0009.2006.00456.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- Freedman, V. A., Kasper, J. D., Spillman, B.C., & Plassman, B.L. (2018). Short-term changes in the prevalence of probable dementia: An analysis of the 2011–2015 National Health and Aging Trends Study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 73, S48–S56. doi:10.1093/geronb/gbx144 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Freedman V. A., & Spillman B. C (2014). Disability and care needs among older Americans. Milbank Quarterly, 92, 509–541. doi:10.1111/1468-0009.12076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kasper J. D., & Freedman V. A (2016). National Health and Aging Trends Study user guide: Rounds 1, 2, 3, 4 & 5 beta release. Baltimore, MD: Johns Hopkins University School of Public Health; Retrieved June 1, 2019, from www.NHATS.org [Google Scholar]
- Kasper J. D., Freedman V. A., & Spillman B. C (2013). Classification of persons by dementia status in the National Health and Aging Trends Study. Technical Paper No. 5. Baltimore, MD: Johns Hopkins University School of Public Health; Retrieved June 1, 2019, from https://www.nhats.org/scripts/TechnicalDementiaClass.htm [Google Scholar]
- Kasper J. D., Freedman V. A., Spillman B. C., & Wolff J. L (2015). The disproportionate impact of dementia on family and unpaid caregiving to older adults. Health Affairs (Project Hope), 34, 1642–1649. doi:10.1377/hlthaff.2015.0536 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kaye H. S., Harrington C., & LaPlante M. P (2010). Long-term care: Who gets it, who provides it, who pays, and how much? Health Affairs (Project Hope), 29, 11–21. doi:10.1377/hlthaff.2009.0535 [DOI] [PubMed] [Google Scholar]
- Keating N., Otfinowski P., Wenger C., Fast J., & Derksen L (2003). Understanding the caring capacity of informal networks of frail seniors: A case for care networks. Ageing and Society, 23, 115–127. doi:10.1017/S0144686X02008954 [Google Scholar]
- Litwak E. (1985). Helping the elderly: The complementary roles of informal networks and formal systems. New York, NY: Guilford Press. [Google Scholar]
- Lunney J. R., Lynn J., Foley D. J., Lipson S., & Guralnik J. M (2003). Patterns of functional decline at the end of life. JAMA, 289, 2387–2392. doi:10.1001/jama.289.18.2387 [DOI] [PubMed] [Google Scholar]
- Marcum C. S., Ashida S., & Koehly L. M (2018) Primary caregivers in a network context. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences. doi:10.1093/geronb/gbx165. Advance Access publication January 03, 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Messeri P., Silverstein M., & Litwak E (1993). Choosing optimal support groups: A review and reformulation. Journal of Health and Social Behavior, 34, 122–137. doi:10.2307/2137239 [PubMed] [Google Scholar]
- Miller B., & McFall S (1991). Stability and change in the informal task support network of frail older persons. Gerontologist, 31, 735–745. doi:10.1093/geront/31.6.735 [DOI] [PubMed] [Google Scholar]
- National Academies of Sciences, Engineering, and Medicine (2016). Programs and supports for family caregivers of older adults. In Schulz R., & Eden J. (Eds.), Families caring for an aging America (pp. 159–209). Washington, DC: The National Academies Press. [PubMed] [Google Scholar]
- Naylor M. D., Brooten D. A., Campbell R. L., Maislin G., McCauley K. M., & Schwartz J. S (2004). Transitional care of older adults hospitalized with heart failure: A randomized, controlled trial. Journal of the American Geriatrics Society, 52, 675–684. doi:10.1111/j.1532-5415.2004.52202.x [DOI] [PubMed] [Google Scholar]
- Nichols L. O., Martindale-Adams J., Zhu C. W., Kaplan E. K., Zuber J. K., & Waters T. M (2017). Impact of the REACH II and REACH VA dementia caregiver interventions on healthcare costs. Journal of the American Geriatrics Society, 65, 931–936. doi:10.1111/jgs.14716 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sharp E. S., & Gatz M (2011). Relationship between education and dementia: An updated systematic review. Alzheimer Disease and Associated Disorders, 25, 289–304. doi:10.1097/WAD.0b013e318211c83c [DOI] [PMC free article] [PubMed] [Google Scholar]
- Spillman B. C., & Black K (2005). Staying the course: Trends in family caregiving. AARP Public Policy Institute; Issue Paper No. 2005–17. Washington, DC: AARP. [Google Scholar]
- Spillman B. C., & Pezzin L. E (2000). Potential and active family caregivers: Changing networks and the “sandwich generation”. Milbank Quarterly, 78, 347–374, table of contents. doi:10.1111/1468-0009.00177 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Spillman B. C., Wolff J. L., Freedman V. A., & Kasper J. D (2014). Informal caregiving for older Americans: An analysis of the 2011 National Study of CaregivingReport to the Department of Health and Human Services, Assistant Secretary for Planning and Evaluation, Office of Disability, Aging, and Long-Term Care Policy. Washington, DC: US Department of Health and Human Services. Retrieved June 1, 2019, from http://aspe.hhs.gov/daltcp/reports/2014/NHATS-IC.pdf [Google Scholar]
- Szinovacz M. E., & Davey A (2013). Prevalence and predictors of change in adult-child primary caregivers. International Journal of Aging & Human Development, 76, 227–249. doi:10.2190/AG.76.3.d [DOI] [PubMed] [Google Scholar]
- Wolff J. L., & Kasper J. D (2006). Caregivers of frail elders: Updating a national profile. Gerontologist, 46, 344–356. doi:10.1093/geront/46.3.344 [DOI] [PubMed] [Google Scholar]
- Wolff J. L., & Spillman B. C (2014). Older adults receiving health care assistance and their family caregivers: Prevalence, characteristics, and hours of care. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 69 (Suppl. 1), S65–S72. doi:10.1093/geronb/gbu119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wolff J. L., Spillman B. C., Freedman V. A., & Kasper J. D (2015). A national profile of family and unpaid caregivers who assist older adults with health care activities. JAMA Internal Medicine, 176, 372–379. doi:10.1001/jamainternmed.2015.7664 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.