Article Text

Original research
Rural-urban and regional variations in aspects of caregiving, support services and caregiver health in the USA: evidence from a national survey
  1. Steven A Cohen1,
  2. Neelam H Ahmed2,
  3. Kerri A Ellis3,
  4. Hayley Lindsey4,
  5. Caitlin C Nash1,
  6. Mary L Greaney1
  1. 1Department of Public Health, University of Rhode Island, Kingston, Rhode Island, USA
  2. 2School of Public Health, Brown University, Providence, Rhode Island, USA
  3. 3College of Nursing, University of Rhode Island, Kingston, Rhode Island, USA
  4. 4Department of Psychology, University of Rhode Island College of Health Sciences, Kingston, Rhode Island, USA
  1. Correspondence to Dr Steven A Cohen; steven_cohen{at}uri.edu

Abstract

Objectives Due to substantial regional variability in available caregiving services and supports, culture and health status among informal caregivers in the USA, the study objective was to explore how rural-urban differences in aspects of caregiving—caregiving intensity, distance to care recipient, caregiver burden, caregiver health and caregiving support—vary by US Census region (Northeast, South, Midwest and West) after accounting for other social determinants of health.

Design This study was a secondary analysis of multiwave, cross-sectional study data.

Setting The data were collected on a representative sample of informal, unpaid caregivers to older adults.

Participants A sample of n=3551 informal caregivers from the National Study of Caregiving identified by older adult care recipients from waves 1 (2011) and 5 (2015) of the National Health and Aging Trends Study.

Primary and secondary outcome measures Primary outcome measures were caregiving intensity (provided support for/with the number of activities of daily living (ADLs) and instrumental ADL (IADLs)) caregiver assisted with, hours of caregiving per month), caregiver burden (physical, emotional and financial), support services sought (types and total number), caregivers’ self-reported health and health status (individual comorbidities and a total number of comorbidities). Analyses were stratified by US Census region and rural-urban status, as defined by the US Census Bureau, of census tract of caregiver residence.

Results Urban caregivers provided higher levels of ADL support in the Northeast (beta=0.19, 95% CI 0.03, 0.35) and West (beta=0.15, 95% CI 0.05,0.26) regions. Urban caregivers provided significantly higher levels of ADL support (p=0.020), IADL support (p=0.033) and total ADLs plus IADLs (p=0.013) than rural caregivers. Caregivers living in the South had higher amounts of monthly hours spent caregiving, ADL support, IADL support and combined ADLs plus IADLs and were more likely to have obesity, report poor or fair health, have heart conditions and experience emotional difficulty from caregiving (all p<0.001).

Conclusions Study findings underscore caregiving’s multifaceted and complex nature and identify important urban-rural and regional differences in caregiving in the USA. Healthcare providers and healthcare organisations can have an important role in identifying and mitigating the negative impacts of caregiving on caregivers’ overall health. Interventions and support should be tailored to caregivers’ demographic backgrounds, addressing regional differences.

  • epidemiology
  • geriatric medicine
  • health equity
  • surveys and questionnaires
  • caregivers

Data availability statement

Data may be obtained from a third party and are not publicly available. Data can be requested. See www.nhats.org for details.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • This is a cross-sectional; therefore, causality cannot be inferred.

  • This is a secondary analysis of previously collected data, so the analysis was limited to information already collected, such as gender (collected as a dichotomous variable) and did not account for the clustered nature of the data.

  • Participants’ census tract and state of residence determined rural and urban status, which does not address specific elements of rural-urban status that may impact caregiving and caregiver burdens, such as socioeconomic status, geographical isolation, access to public transportation and access to basic services.

  • One strength is that the study uses a large sample size of caregivers identified by their care recipients and an array of measures to assess caregiving-related responsibilities, caregiver burden, use of caregiving supports, and caregiver health and quality of life.

  • Another strength is that the study was among the first to examine critical nuanced variability by geography in well-established rural-urban differences in caregiving.

Introduction

In the USA, individuals aged 65 years and older comprise the most rapidly increasing age group.1 In 2022, 17% of US adults were 65 years and above, and this percentage will increase to 21% by 2050.2 3 As the US population steadily ages, more individuals rely on friends and family for assistance and care. Informal caregiving is common in the USA to counter the deficit of medical and long-term care resources while also allowing older adults to age in place.4 Although each caregiver’s experience differs, the tasks associated with this role are often similar. Caregiving consists of an array of responsibilities, such as aiding with housework, transportation and self-care.5 Informal caregiving is unpaid and time-consuming, affecting the caregiver’s well-being and lifestyle in several ways.6 Benefits of informal caregiving for the caregiver include reduced mortality7 and increased life satisfaction.8 Conversely, caregiving also poses many adverse effects on the caregiver’s health.9 10 These negative impacts include increased risk of cardiovascular diseases and stroke.11 Additionally, caregiving has been associated with mental health disorders such as anxiety and depression.12 However, it should be noted that not all caregivers experience the same set of benefits and harms.

Moreover, the area of residence contributes to the prevalence of health disparities in a population. Given the large size of the USA and the diverse demographics, environmental factors, and resources across the country, health disparities are complex and vary dramatically by geography.13 Two major geographical factors influencing health disparities in the USA are rural-urban status and the US region. Rural-urban health disparities are not uniform across the USA and vary substantially by region. An example of a persistent disparity by the US region is the ‘southern mortality penalty’, whereby health and mortality outcomes in rural areas are consistently lower in rural compared with urban areas of the southern USA, while those associations are reversed in other regions.14 15 Therefore, treating rural or urban populations as a monolith may mask critical regional variation across rural-urban gradients. Furthermore, the theory of fundamental causes, which has been used in caregiving research, suggests that health inequalities arise due to pervasive disparities in social determinants of health.16 17 Understanding the differences in health disparities across multiple contexts in the USA is critical to creating and implementing effective policies and programmes designed to reduce inequities across diverse caregiver subgroups and contexts throughout the USA.

Not only do rural populations in the USA have poorer health than their urban counterparts, but these areas also have higher rates of unpaid, informal caregiving.11 18 19 Between 2004 and 2016, more older adults in rural areas of the USA received familial care than paid care, whereas urban adults primarily relied on formal care.20 Rural caregivers experience many barriers to providing care and protecting their health that exacerbate existing disparities. Their caregiving responsibilities can require long-distance travel for appointments. A study using 2012 Behavioural Risk Factor Surveillance System (BRFSS) data, which is collected in the USA, determined that rural caregivers had more financial barriers than their urban counterparts, 38.1% vs 31.0%, respectively.21 Rural US caregivers are more likely to use their savings to cover out-of-pocket expenses, further augmenting the financial burden.21–23

There are distinct differences in the support available for rural and urban caregivers depending on socioeconomic status, demographics and geography in the USA. For example, employed caregivers in rural areas are less likely to have paid leave, access to employee assistance programmes (EAPs), and the ability to work from home, all heightening caregiver burden.24 Another critical aspect of caregiving is the time commitment, especially for employed caregivers or those with children living at home. An analysis of the 2018 BRFSS dataset found that rural caregivers in the USA have 23% greater odds of providing long-term (2+ years) caregiving than their urban counterparts.25 Similarly, a study using a convenience sample of rural and urban caregivers ascertained that rural caregivers in the USA spend an average of 11 hours of caregiving a week. In comparison, urban caregivers dedicate an average of 4 hours per week to caregiving.26 Overall, rural caregivers’ general health-related quality of life (HRQoL) is poorer than that of their urban counterparts.27 When assessing caregiver burden, the impacts on caregiver health and well-being, and healthy behaviours, each increase of one unit on the role burden scale for rural caregivers decreased by 1.15 units on the health behaviour scale, indicating fewer healthy behaviours.26 However, another study found no significant difference between caregiver burden and rural/urban status among caregivers in the USA.28

Nonetheless, regional variation in rural versus urban caregiving exists across the USA, and studies have assessed the extent of these differences. An analysis of data from the 2020 Cornell National Social Survey, a cross-sectional survey collected annually with a randomised sample of 1000 US adults, found that familial caregiving is more prevalent in Northeast and Southern regions compared with Midwestern and Western regions. More specifically, living in the West region was associated with a 37% reduced likelihood of informal caregiving than Northeastern residence.29 While there are differences in the overall likelihood of caregiving across US regions, there are also critical regional differences between caregiving resources. An analysis of the Centers for Medicare and Medicaid Services Home Health Compare data indicating ratings of healthcare providers and facilities for Medicare based on four factors: years of Medicare certification, rates of full service, ownership status and distance between agency and patients’ homes was undertaken. The West, Northeast and South regions have lower home healthcare availability than those in the Midwest region.30

Although there are well-established, complex variations in rural-urban caregiving and regional differences in caregiving, their joint effects are unknown. This exploratory study aimed to determine whether rural-urban differences in caregiving intensity, caregiver burden, caregiver health and seeking caregiving support, vary by the US region after accounting for other social determinants of health.

Methods

Study design

To address this research objective, a cross-sectional study was conducted using two waves of caregiving data. The parent study was the National Health and Aging Trends Study (NHATS), which collects information on a nationally representative sample of Medicare beneficiaries ages 65 and older. NHATS consists of annual, in-person interviews that collect detailed information on the disablement process and its consequences.31

Patient and public involvement

This is a secondary analysis of previously collected data in an ongoing longitudinal study.

Data sources

Data were abstracted from waves 1 (2011) and 5 (2015) of the National Study of Caregiving (NSOC). The NSOC is a periodic national survey that gathers information on family and unpaid caregivers to participants in the parent NHATS.31 No respondents were in both waves. To be eligible for inclusion in the NSOC samples, NHATS respondents may identify up to five individuals (‘helpers’) who assisted them with personal care and/or household tasks.32 NSOC respondents were linked with corresponding data from the NHATS, a nationally representative sample of Medicare beneficiaries ages 65 and older in the US. Data on NHATS participants were collected annually using in-person interviews designed to collect detailed information on health, the disablement process and other aspects of daily life.33 These helpers were then eligible for potential inclusion in NSOC if they helped a potentially eligible NHATS participant with any activity listed below. People who were paid for their caregiving services were not eligible for inclusion in the NSOC sample.

Informal, unpaid caregivers identified by NHATS participants were contacted by the NHATS study team and asked to complete a 30 min telephone survey assessing aspects of caregiving and health, including caregiving activities, duration and intensity, support services sought and used, effects on caregiver participation in everyday activities and demographics. Participants were excluded if they received compensation for their caregiving services. The principal investigator of the current study (SAC) was granted access to the NHATS/NSOC restricted files through the University of Michigan Center on the Demography of Aging (MiDCA) NHATS Restricted Data Repository to obtain restricted geographical information on caregivers.34 The resultant sample size was n=3551 eligible informal caregivers of NSOC respondents.

Exposure measures

Primary exposure variables were rural/urban status and US Census region of each eligible NSOC respondent in the sample. Respondents’ rural/urban classification was based on their census tract of residence. US Census region (Northeast, South, Midwest and West) depended on the state where each respondent resided at the time of the survey. Table 1 provides a list of US states by US Census region. The US Census Bureau provides classifications of regions, which are defined as geographical groupings of states that subdivide the USA, based on historical usage and are used today to provide standardised regional comparisons on health, economic and population data.35 Included covariates were respondent’s age, gender, race (black, Asian, other or missing vs white), currently employed (yes vs no) and education (at least high school vs less than high school). All exposure variables were gleaned from the US Census data classifications provided in the NHATS/NSOC data sets. All variables were obtained through the restricted data files from the Michigan Center on the Demography of Aging.34

Table 1

List of US states by US Census region

Outcome measures

Based on prior research, key outcome measures available in the NSOC survey that were used in this analysis fall into one of several categories: basic demographics,36 aspects of caregiving,37 38 caregiver burden,39 support services sought,40 and health and quality of life.41 A summary of all major outcome variables assessed can be found in online supplemental appendix table 1. Basic demographic characteristics included age, gender, relationship to care recipient, race (white, black, Asian, other), Hispanic/Latinx origin (yes vs no), education and employment status. In most cases, variables with three or more categories except race were dichotomised for the multivariable models (see below) due to small cell sizes. An additional sensitivity analysis was conducted using age categories (18–39, 40–64, 65+) to assess the potential for non-monotonic associations between age and key outcome measures of interest.

Intensity of caregiving: Seven measures of caregiving intensity were considered. These measures included (1) the number of minutes from the caregiver’s residence to the care recipient’s residence, (2) living with the care recipient (yes vs no), (3) hours spent caregiving per month and (4) years spent caregiving. In addition, two other measures of caregiving were also assessed: (5) a total number of activities of daily living (ADLs) and (6) instrumental ADL (IADL) assisted with and (7) a sum of ADLs and IADLs. The assessed ADLs included assisting with personal care, getting around, lifting from a seated position, holding steady, leaning on for support, teeth care, foot care and wound care. The assessed IADLs included assisting with chores, shopping, bills, medications, insurance and several others. For each ADL and IADL examined, respondents were coded as ‘yes’ if they reported assisting with the task frequently or always, and otherwise were coded as ‘no’.

Caregiving burden: Three measures were used for caregiver burden: Each caregiver was asked if they experienced any of the following due to caregiving activities: (1) physical burden, (2) emotional burden and/or (3) financial burden. Each burden (physical, emotional, financial) was assessed using the question: ‘Helping a spouse or partner who has health problems can be difficult./Helping older relatives can be difficult. Is helping ‘XX’ {financially, emotionally, physically} difficult for you? Possible responses were ‘yes’ or ‘no’.

Caregiving support: For support sought, 16 aspects of caregiving support were examined. These included talking to friends and family, using any caregiving support service, receiving formal caregiver training, finding information from a variety of sources (eg, government, employer, place of worship), helping with assistive technology, mobile device and home safety and hired a paid helper to supplement informal caregiving responsibilities (see online supplemental appendix table 1 for the complete list.) In addition, two summary variables were also created based on support seeking—sought any supports and the total number of supports sought—for a total of 18 caregiving support variables.

Caregiver health: Nine total measures were assessed using several self-reported measures: (1) reporting fair or poor health status, (2) having obesity, (3) having a history of heart attack and having any of the following: (4) heart disease, (5) hypertension, (6) arthritis, (7) osteoporosis, (8) diabetes, lung disease, cancer, hearing loss, vision loss or chronic pain. A summary heart condition measure was created for heart attack, heart disease and hypertension. In addition, a 10th variable was obtained from the summary count of all health conditions (termed ‘comorbidities’) from the conditions above (see online supplemental appendix table 1 for the complete list.)

All 38 variables were analysed in the bivariate analyses. However, 13 representative variables were used in the multivariable models based on the caregiving literature due to the high number of outcome variables. Five variables were based on intensity of caregiving: minutes to care recipient,42 43 caregiving hours per month,6 44 45 total ADLs,46–48 total IADLs46 49 and sum of ADLs+IADLs.47 All three caregiver burden variables were used- physical,50–53 emotional51 52 54 and financial.32 55 56 The sum of all caregiving support sought32 and four caregiver health measures—general health,57 58 having obesity,59–62 having any heart condition63 64 and a sum of all comorbidities65 66—were included, as well.

Statistical analysis

Descriptive statistics were obtained for all study variables, including means or medians and measures of dispersion (SD, IQR, etc) for all continuous and discrete variables. Frequencies (sample size and per cent of the sample) were ascertained for categorical variables. Bivariate analyses were assessed for all study variables of interest—demographic characteristics, aspects of caregiving, caregiver burden, support services sought, and health and quality of life— by US Census region and rural/urban status using χ2 analyses for categorical outcomes and either t-tests of Wilcoxon rank sum tests for continuous and discrete variables.

Multivariable modelling was used to assess how the magnitude and direction of rural-urban differences in caregiving attributes and caregiver health varied by US Census region. For multivariable models, only the 13 selected outcome variables identified above were included due to the large volume of initial variables. These variables were minutes to care recipient, hours per month, total ADLs, total IADLs, sum of ADLs+IADLs, total number of supports sought, total comorbidities, having obesity, reporting poor or fair health, any heart condition, and physical, emotional, and financial burdens due to caregiving.

A two-stage analysis was used. First, generalised linear models (GLMs) with an appropriate, variable-specific link function were used to adjust for several key explanatory demographic variables for each major outcome explored—the intensity of caregiving, caregiver burden, support services sought, and health and quality of life. Models were adjusted for age, gender, race (black, Asian, other or missing vs white), currently employed (yes vs no) and education (at least high school vs less than high school). These GLM models produced a set of predicted values—probabilities or likelihood values for categorical outcomes and predicted values for count or continuous variables—for each outcome modelled, adjusted for the demographic variables above.

In the second stage, each predicted value from the first stage of models was modelled on the main predictor values: rural/urban status and US Census region. This stage was completed in two ways: stratification and interaction terms. Data were first stratified by region and predicted values were modelled against rural/urban status to explore and identify variable patterns of rural/urban differences by region. Then, interaction models were used to test for the statistical significance of regional differences in the rural/urban associations, using combined data, predicted values were modelled against rural/urban status, US Census region and a combined rural/urban status×US Census region interaction term, with ‘urban Northeast’ being the reference group.

Model assumptions were checked to ensure that the procedures complied with requirements for using GLM. Model fit was assessed using log-likelihood criteria, F-tests and Akaike information criteria, where applicable. Missing values were assumed to be at random. Cell counts of less than n=11 needed to be combined to maintain data confidentiality in accordance with the Michigan Center on the Demography of Aging requirements. SAS V.9.4 (Cary, NC) and IBM SPSS V.28 (Armonk, NY) were used for all statistical analyses.

Results

Demographics by rural-urban status

Descriptive statistics of demographics for the study sample (n=3551) are found in table 2. Of the overall sample, 1677 (47.2%) respondents were from NSOC wave 1 (2011) and 1874 (52.8%) were from wave 5 (2015). A greater proportion (64.8%) of the sample in rural areas was female than in urban areas (60.8%) (p=0.017). Rural areas had a higher proportion of white caregiver respondents (55.3%) compared with urban areas, caregivers (53.1%). Similarly, there was a greater proportion of urban caregivers self-reporting as black (9.3%) or Asian (1.1%), compared with caregivers in rural areas (5.6% and 0.2%, respectively). A significantly greater proportion of urban caregivers were of Hispanic/Latinx origin (12.2%) than rural caregivers (5.5%) (p=0.005). Urban caregivers were about 70% more likely to have at least a college degree than rural caregivers. There were no significant differences in the frequencies of individual health issues or total comorbidities by rural/urban status.

Table 2

Demographic characteristics of analytical sample by rural-urban status

Caregiving and caregiver health by rural-urban status

Additional descriptive statistics for aspects of caregiving can be found in table 3 and online supplemental appendix table 2. On average, urban caregivers assisted with 0.55 more IADLs than rural caregivers (p<0.001). Urban caregivers (27.2%) were more likely than rural caregivers (21.8%) to use the computer to shop or pay bills on behalf of their care recipient (p=0.003), assist with obtaining health insurance plans (23.5% vs 19.4%) (p=0.003), deal with other health insurance issues (31.1% vs 26.6%) (p=0.020) and administer injections, such as insulin (8.4% vs 6.6%) (p=0.019). There were no statistically significant differences by rural/urban status for ADL caregiving. Urban caregivers were more likely to reside with their care recipients (57.3%) than their rural counterparts (52.0%) (p<0.001). In terms of seeking caregiving support, rural caregivers were more likely to find information on caregiving support from the government (16.6% vs 11.6%, p<0.001) and to find information from their employer (3.6% vs 1.8%, p=0.013) than urban caregivers.

Table 3

Summary of selected aspects of caregiving (type of activity, caregiver burden and support sought) and caregiver health by rural-urban status

Multivariable results: overall sample

First-stage regression model (controlling for confounders to produce adjusted outcomes) results are shown in online supplemental appendix table 3 for each major outcome. Caregivers identifying as black (beta=28.2, 95% CI 16.8, 39.7), Asian (beta=52.8, 95% CI 2.0, 103.7) and those of other (beta=42.6, 95% CI 10.0, 75.3) or missing race (beta=14.3, 95% 3.9, 24.8) had significantly higher caregiving hours per month than those identifying as White. On average, female caregivers provided 19.9 more hours per month of caregiving than male caregivers (95% CI 10.7, 29.0). Caregivers who were employed (beta=35.4, 95% CI 25, 9, 44.9) and those with at least a high school education (beta=18.1, 95% CI 4.4, 31.9) provided significantly fewer hours than those not employed and with a less than high school education. The sensitivity analysis using age as a categorical variable (18–39, 40–64 and 65+) showed significant monotonic associations for the risk of poor or fair health and physical difficulty with caregiving. For emotional difficulty with caregiving, those aged 40–64 were significantly more likely to report emotional difficulty with caregiving than those aged 18–39; there was no significant association comparing those aged 18–39 to those 65+. Similar results—a significantly higher likelihood of any heart condition—were evident among 40–64 year olds, but not for those 65+.

Caregivers identifying as black provided significantly higher levels of ADLs (OR 1.20, 95% CI 1.08, 1.34), more total ADLs and IADLs (OR 1.22, 95% CI 1.11, 1.34) and were more likely to have obesity (OR 1.66, 95% CI 1.37, 2.00), report poor or fair health (OR 1.41, 95% CI 1.13, 1.76) and have a heart condition (OR 2.01, 95% CI 1.66, 2.45) compared with caregivers identifying as white. Although caregivers identifying as black were more likely to report that caregiving is an emotional burden (OR 1.38, 95% CI 1.14, 1.66), they were less likely to report that caregiving is a financial burden (OR 0.79, 95% CI 0.63, 0.98) compared with those identifying as white. Increasing age was associated with significantly lower odds of reporting that caregiving is a physical burden (OR 0.993, 95% CI 0.987, 0.999), but significantly higher odds of a financial burden due to caregiving (OR 1.02, 95% CI 1.01, 1.03).

Adjusted results by US Census region

Model-adjusted outcomes by region comparing rural and urban caregivers are shown in online supplemental appendix table 4. Compared with rural caregivers, urban caregivers had significantly higher levels of IADL and combined ADL+IADL caregiving in the Northeast, Midwest and West regions. Likewise, urban caregivers also provided higher levels of ADL caregiving in the Northeast (beta=0.19, 95% CI 0.03, 0.35) and West (beta=0.15, 95% CI 0.05, 0.26) regions. The likelihood of having obesity was significantly higher in urban than in rural caregivers living in the Northeast (beta=2.1%, 95% CI 0.7%, 3.6%) and Midwest (beta=1.2%, 95% CI 0.1%, 2.2%). The likelihood of having a heart condition was higher among urban caregivers in the Midwest compared with rural caregivers (beta=3.2%, 95% CI 0.4%, 5.9%), but the association was reversed in the West, where the likelihood of having a heart condition was higher among rural caregivers (beta=−4.5%, 95% CI −8.5%, −0.5%). The frequency of reporting that caregiving is financially difficult was significantly higher among rural caregivers than urban caregivers in both the Northeast (beta=−0.8%, 95% CI −1.4%, −0.1%) and West (beta=−0.7%, 95% CI −1.0%, –0.3%). In the South, there were no significant rural/urban differences for any of the outcomes examined, however. The predicted levels of each major outcome (eg, caregiver attributes, support services sought and health outcomes) from these models are displayed by region and rural-urban status in online supplemental figure 1A–D, and a summary of differences can be found in online supplemental figure 2.

Regional variability in rural-urban differences in the model-adjusted outcomes presented above are displayed as predicted values by rural/urban status and US Census region in figure 1. Irrespective of US Census region, urban caregivers provided significantly higher levels of ADLs (p=0.020), IADLs (p=0.033) and total ADLs plus IADLs (p=0.013) than rural caregivers. Caregivers living in the South region had higher amounts of hours per month spent caregiving, ADLs, IADLs and combined ADLs plus IADLs (all p<0.001), as well as a higher likelihood of having obesity, poor or fair general health, having heart conditions (all p<0.001) and experiencing emotional difficulty from caregiving (p<0.001). Two interaction terms were significant. The first, ADLs+IADLs, the rural×south term (p=0.026) was significant and positive. This significant term indicates that rural caregivers living in the South provided significantly higher levels of ADL and IADL caregiving tasks compared with other caregiver subgroups. The second interaction term was for financial difficulties stemming from caregiving, the rural×midwest (p=0.025) term was significant and negative, suggesting those caregivers (rural and midwestern) were significantly less likely than others to have difficulties from caregiving.

Figure 1

Indicators of which main effects and rural×region interactions were statistically significant (X=significant at p<0.05, blue=significant negative association, pink=significant positive association).

Discussion

Variability by region

This study identified several important rural-urban differences in attributes of informal caregiving that varied by US Census region, but the observed patterns were not universal for all measures across regions. The frequency of reporting that caregiving is a financial burden was significantly higher among rural caregivers than urban caregivers in Northeast and West, but not in the South or Midwest, consistent with previous research.21 67 The reasons for these observed associations remain unclear. Rural caregivers in the Northeast and West US Census regions may be fundamentally different in terms of coping with caregiving stressors, or perhaps a lack of resources relative to their urban counterparts in those regions. Previous research has suggested that rural caregivers in South and Midwest regions may be more likely to use coping mechanisms such as positive reframing, seeking social support through their peer or community networks or problem-solving to reduce the negative impact of stressors rather than avoiding them,21 which may occur less frequently in rural areas of the highly developed Northeast or West regions. Future research should explore these differences further.

Overall, urban caregivers provided, on average, approximately 10 more hours of care per month than rural caregivers. This difference overall was not significant (87.3 vs 77.5, p=0.085), but a significant difference was observed in the West region. However, the current study found regional variation in rural-urban differences in other aspects of caregiving. For example, urban caregivers in the Northeast, Midwest, and West regions provided greater assistance with IADLs and combined IADLs+ADLs than their rural counterparts. This finding contrasts with prior research that found that rural caregivers provide more caregiving hours than urban caregivers.25 Although the reasons for these differences are unclear, it may be due to recent demographic trends over the past several decades where younger adults are moving out of rural areas into more urbanised areas than older adults. So, it may be a matter of self-selection, where care recipients who require more care may be moved out of necessity to more urban areas to receive appropriate care from their families than healthier older adults who may be more able to remain in rural areas since they require less care. However, understanding the underlying mechanisms that promote such variation is important to address these differences and ensure that those caregivers in need of the most support receive the support they need.

Similarly, a recent analysis of data from the Health and Retirement Study examining caregiving in formal settings and family homes from 2004 to 2016 among older adults 65+ with two or more reported limitations in ADLs and/or IADLs found that a greater proportion of rural respondents that urban respondents received caregiving from family and that a greater proportion of urban respondents than rural respondents relied exclusively on formal care and a lower proportion solely on family care.20

Variability across demographic and socioeconomic groups

This study also underscores a significant observation observed in previous studies: caregivers identifying as black, Asian or from diverse racial backgrounds, as well as female caregivers, tend to spend more time on caregiving tasks. Race generally refers to individuals belonging to different groups based on physical and/or biologically determined features, whereas ethnicity often refers to region of origin and emphasises cultural characteristics, although there is substantial overlap in these concepts.68 In the NHATS/NSOC data set, the sole measure of ‘ethnicity’ is Hispanic/Latinx origin. Prior studies also have identified differences in caregiving by race, Hispanic/Latinx origin and education.12 35 Subsequently, this increased workload exposes these caregivers to a higher risk of burn-out and mental health issues among these vulnerable populations.

Differences in health status among caregivers were identified in the current study. Caregivers identifying as black were more likely to have obesity, report their health as poor or fair and have a heart condition than caregivers identifying as white. These differences are consistent with prior research69 and indicate the need to address health disparities and the social determinants of health. Differences in obesity among caregivers also were identified by US Census region. Urban caregivers in the Northeast and Midwest were more likely to have obesity than their rural counterparts. Additionally, the likelihood of having a heart condition was higher among urban than rural caregivers in the Midwest compared with rural caregivers; however, in the West, the likelihood of having a heart condition was higher among rural caregivers than urban caregivers. Increasing age was also associated with an increased likelihood of financial difficulty due to caregiving, which is not unexpected due to older adults being less likely to be employed. It does, however, indicate the need for assistance for this population. Increasing age was associated with significantly lower odds of stating that caregiving was physically difficult and higher odds of reporting financial difficulties. The rationale for these unexpected findings is unclear and merits further research. Furthermore, the sensitivity analysis for age showed several non-monotonic associations between health conditions, aspects of caregiving and age groups that could be explored further.

Despite some geographical differences across US Census regions, urban caregivers generally reported higher levels of caregiving responsibilities and some worse health outcomes than their rural counterparts, which is somewhat consistent with prior research.70 There is suggestive evidence that, although rural caregivers in some areas may experience greater financial burden than urban caregivers,21 urban caregivers may experience worse health outcomes and have higher levels of overall burden due, in part, to a relative lack of social support compared with rural caregivers.71 A recent study by Crouch et al found that rural caregivers were healthier than their urban counterparts after adjustment for other factors.72 Results from a qualitative study of Australian caregivers suggested that urban-rural differences in caregiving strain and impacts on health may be partially attributable to higher levels of resilience among rural caregivers.73 Understanding the underlying mechanisms for these observed overall differences in caregiver strain and caregiving intensity by rural-urban status remain largely unknown.

It should be noted that the observed regional variability in differences in caregiver health and quality of life by rural-urban status may not be directly attributable to caregiving itself. Comparing caregivers to non-caregivers to explore such differences using NSOC data was not possible since the NSOC sample consisted exclusively of caregivers. However, it was possible to compare a limited number of corresponding health outcomes by caregiving status and rural-urban status using the 2018 BRFSS database.74 In the sample, 30.2% of urban non-caregivers had obesity, compared with 35.5% of rural non-caregivers, 34.9% of urban caregivers and 34.4% of rural caregivers. Similarly, 29.0% of urban non-caregivers reported poor HRQoL, compared with 32.6% of rural non-caregvers, 35.1% of urban caregivers and 33.8% of rural caregivers. Further analyses to quantify to what extent health issues can be directly attributed to caregiving, and how those effects differ by geography is warranted, but are beyond the scope of the current analysis.

Although not the central focus of the study, it was observed that black caregivers were more likely to report that caregiving is emotionally difficult than white caregivers. However, black caregivers were less likely than white caregivers to report that caregiving was financially difficult. This difference could be due to more black caregivers living with their care recipient. A recent analysis of national data determined that black respondents without dementia were more likely than white respondents to live with their adult children (38.9%, 49.7% and 23.3%, respectively).75 This study also found that black and Hispanic respondents were more likely than white respondents to receive financial assistance from their children (28.0%, 25.9% and 15.0%, respectively). The authors attribute these differences to personal, family/cultural factors and systemic racism across the lifespan, although more research is needed to gain a more thorough understanding of these observations.

Research and policy implications

The regional disparities in caregiving tasks and challenges necessitate localised strategies. Urban caregivers, especially in certain US Census regions, were found to have higher levels of caregiving tasks. Healthcare providers, including medical doctors, nurses, physical and occupational therapists, medical assistants, certified nursing assistants, and healthcare organisations, should collaborate with community organisations to develop region-specific support networks, educational workshops and respite care services. Conversely, rural caregivers in particular US Census regions faced elevated financial difficulties. This highlights the importance of addressing financial stress as a potential trigger for mental health issues.76 77 Healthcare providers and organisations can provide financial planning resources, connect caregivers with relevant support programmes and engage in open discussions about the economic impact of informal caregiving.78

Many caregivers likely need support, which should be available in various settings, including work (for employed caregivers), home, community and healthcare organisations. Offered support should be culturally sensitive, tailored to the caregiver and recognise the caregiving context.79 Healthcare providers can offer tailored emotional support and appropriate referrals to help caregivers develop coping mechanisms and time-management skills. Recognising emotional disparities among informal caregivers from different racial backgrounds emphasises the need for precise interventions.80

The current study also found that urban caregivers were more likely to use a computer to assist their care recipient with shopping/paying, dealing with health insurance (obtaining plans, handling issues), administering injections and assisting with exercises. These differences may be due to urban caregivers in the present study being about 70% more likely to have at least a college degree than rural caregivers. Other research has also determined that populations in rural areas tend to have less formal education.81 It also is possible that these differences are due to computer and/or internet access. Public libraries provide computer and internet access,82 which can be an important venue for providing caregivers with computer and internet search skills and increasing health literacy skills.83 Many older adults believe that public libraries can help them learn new things (72% of people aged 52–70, 65% of people aged 71–80).84 However, it is important to mention that access to and use of public libraries may not be feasible for informal caregivers who may be homebound due to caregiving demands. Although rural caregivers were more likely than urban caregivers to seek caregiving information from the government (16.6% vs 11.6%) and employers (3.6% vs 1.8%), it is notable that the percentage of respondents seeking information from these sources was very limited, suggesting that caregivers may need to be made aware of available information sources. EAP may be limited in rural areas,24 and often, these programmes focus on substance use, EAP should increase their focus to include caregiver burden and stress.85

Limitations and strengths

This study has several limitations to consider. First, the study is cross-sectional; therefore, causality cannot be inferred. Second, data were self-reported, which may introduce recall bias. Third, since the NSOC survey was telephone based, there may be an increased likelihood of social desirability bias. Another important consideration is the age of the data; at the time of analysis, the oldest data from wave 1 was over 10 years old. Next, respondents’ census tract and the state where they resided when completing the survey determined rural and urban status. Although the analysed data were provided in the NHATS/NSOC data sets, the rural-urban measure used in the study does not address specific elements of rural-urban status that may impact caregiving and caregiver burdens, such as socioeconomic status, geographic isolation, access to public transportation and access to basic services. It is important to note that this is a secondary analysis of previously collected data. While the NSOC and NHATS samples are robust concerning the scope and breadth of topics addressed and questions asked, several issues arose. One example is gender, which was collected in the initial NSOC surveys as a binary ‘female’ versus ‘male’ variable. This approach does not allow the possibility of non-binary gender responses. Therefore, it was not possible to distinguish the associations with gender-based (social) and sex-based (biological) factors. The analysis did not account for the clustered nature of the data, where in a few instances multiple NSOC respondents could provide care for one NHATS respondent. Due to restrictions on data use, such an approach would not be feasible as it would lead to numerous empty cells in the adjusted models and extensive missing data. Also, only 13 representative variables were selected for use in multivariable modelling, their inclusion was based on prior research. It is possible that additional patterns of associations could be observed had the remaining variables been analysed using multivariable modelling. Lastly, a state-level analysis would provide more granularity and nuance to the results, but due to the relatively small sample sizes in some states, it was not possible to do this for all 50 US states and the District of Columbia.

The study also has several strengths, including the large sample size and use of a range of measures to assess caregiving-related responsibilities (IADLs and ADLs), caregiver burden, use of supports for caregiving, and caregiver health and quality of life. This is the first study, to date, that has explored differences in these aspects of caregiving by rural-urban status and the US region simultaneously to assess geographic variability in the strength and direction of rural-urban differences that would otherwise be masked had the sample been analysed as a whole.

Conclusions

In conclusion, the study findings underscore caregiving’s multifaceted nature and identify important urban and rural differences in caregiving in the USA. Caregiver burden, stress and overall health status must be addressed, and healthcare providers and healthcare organisations can have an important role in identifying and mitigating the negative impacts of caregiving on caregivers’ overall health. Healthcare providers play a crucial role in addressing challenges faced by informal caregivers and easing their burdens and can empower informal caregivers to navigate their caregiving journeys with resilience and maintain their well-being. They could also briefly assess caregivers’ health and well-being to identify caregivers needing support. These assessments could occur at medical appointments of the caregiver or the care recipient. Healthcare providers could then offer referrals to services and interventions when available. Healthcare organisations and organisations serving older adults (eg, Departments of Public Health and senior centres) could let caregivers know of existing programmes and develop online educational and support materials available online and in person. Offered interventions should be tailored based on caregivers’ demographic backgrounds, addressing regional disparities and offering comprehensive support strategies.

Data availability statement

Data may be obtained from a third party and are not publicly available. Data can be requested. See www.nhats.org for details.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants but an Institutional Review Board exempted this study (University of Rhode Island, protocol number 1907678-1). Participants gave informed consent to participate in the study before taking part.

References

Supplementary materials

Footnotes

  • Contributors SAC conceived the idea for the manuscript, conducted the data analysis and wrote the methods and results sections. NHA conducted the literature search, drafted the introduction and provided extensive and critical editing of the manuscript. KAE wrote portions of the discussion and conclusion sections and provided critical editing of the manuscript. HL assisted in the creation of the figures and tables and provided critical editing of the manuscript. CCN contributed to the exploratory data analysis and provided critical editing of the manuscript. MLG wrote the initial draft of the discussion section and provided extensive editing of the whole manuscript. The guarantor of the study is SAC. He accepted full responsibility for the finished work and/or the conduct of the study, had access to the data and controlled the decision to publish.

  • Funding This work was generously supported by the Rhode Island Foundation (#8432_20210966).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.