Doubled-Up Households, Self-Management Behaviors, Diabetes Preventive Care Services, and Hospital Use in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) 2015–2020

The prevalence of individuals in doubled-up households increased steadily in the years following the Great Recession in 2007 (27.7% in 2007 to 30.0% in 2011) and remain high (1–3). In fact, data from the American Housing Survey show that doubled-up households have increased in recent years as adult children are delaying leaving their parents’ home to start their own households, living with aging parents is more commonplace, and/or households accommodate other adult relatives or other nonrelatives (4,5). The definition of doubled-up households can vary depending on government agency but generally include households having one or more adults in addition to the head of household and spouse or partner. The U.S. Department of Housing and Urban Development tracks doubled-up households to identify stress in the housing market and determine vulnerable households at risk for housing insecurity. In the U.S., Hispanic/Latino households are more likely to be doubled up than non–Hispanic/Latino White households (2). Hispanics and Latinos, in general, were particularly hard hit by the housing market crash in 2007 and represent diverse heritages with varying health risks (6–10). Given that the economic crisis did not affect all racial or ethnic groups in the same way, a better understanding of those most at risk can be helpful in identifying communities in most need of health care access and housing assistance.

The composition of doubled-up households and the reasons for doubling up can be heterogeneous. This heterogeneity may reflect, in simple terms, an involuntary housing arrangement (e.g., a means to manage escalating housing costs and an indication of housing and financial insecurity) or a voluntary housing arrangement (e.g., caring for an ill parent and reflecting a consolidation of financial resources or for social considerations) (11,12). Household members managing chronic health conditions like diabetes can require consistent access to health care providers and need a support network to assist in management efforts. When positive, these social networks can facilitate sharing of information and other resources that have health benefits (13,14). However, in many cases, doubled-up households can also signify a less visible form of housing insecurity and, therefore, may confer a high risk for unhealthy behaviors or poor health outcomes (15). Many studies show that housing-insecure individuals have less access to preventive health care, less use and postponement of medications, delayed needed medical care, and increased acute care use (2,15–18). Although living doubled up can stabilize housing costs, putting less strain on financial budgets for basic needs (13,19), this housing condition may not reduce barriers to accessing additional resources for health-promoting behaviors like healthy eating and well-care visits (15). The health benefits or risks of living in a doubled-up household are unclear and warrant further investigation.

The aims of this study were to determine whether doubled-up status was associated with 1) self-care behaviors and preventive care services, and 2) an increased risk of hospital use among participants with diabetes in the Hispanic Community Health Survey/Study of Latinos (HCHS/SOL). We hypothesized that residents of doubled-up households would be more likely to smoke cigarettes and drink alcohol, less likely to engage in preventive care services for their diabetes, and have increased risk of emergency department (ED) visits or hospitalizations compared with residents not living doubled up. Furthermore, we hypothesized that health risks would be highest among residents living doubled up with unrelated adults and lowest among residents living doubled up with adult children.

The HCHS/SOL is a community-based, prospective cohort study of 16,415 self-identified Hispanic/Latino people aged 18–74 years residing in one of four U.S. cities (Bronx, NY; Chicago, IL; Miami, FL; San Diego, CA). The goals of the HCHS/SOL, sample design, and cohort selection have been previously described (20,21). Briefly, a baseline health examination took place in 2008 to 2011. A second health examination (V2) similar to baseline took place in 2014–2017, whereby a clinical examination was conducted that included comprehensive biological (e.g., anthropometrics, blood sample collection), behavioral (e.g., tobacco use assessed by self-report), and sociodemographic (e.g., socioeconomic status, nativity) assessments. All participants of the HCHS/SOL cohort with diabetes (defined as either fasting plasma glucose [FPG] level ≥126 mg/dL [7 mmol/L] for >8 h fasting; FPG ≥200 mg/dL [11.2 mmol/L] for ≤8 h fasting; a 2-h post-load glucose level (2-h oral glucose tolerance test [OGTT] ≥200 mg/dL [11.2 mmol/L]; HbA_1C level ≥6.5% [48 mmol/mol]; self-reported use of hypoglycemic agents; or self-report of diabetes at V2), with complete covariate information including information to determine doubled-up status at V2, were included in this analysis (n = 3,474). Follow-up information from the V2 examination (during 2014–2017) to annual follow-up year 8 (March 2015–January 2020) on hospital use was used in this analysis. The institutional review board at each field center approved the study. All participants gave written informed consent before participation in the study.

Whether participants lived doubled up was determined by self-report using the HCHS/SOL Household Roster Survey. The doubled-up categories were guided by definitions used in the American Housing Survey (4). The households were grouped on the basis of hypothesized risk of housing insecurity, where not living doubled up had the lowest risk, and living doubled up had the highest risk of related health consequences. Participants were determined to not live doubled up if they lived alone, or as a couple, or a couple with children aged <18 years, or as a single parent with children aged <18 years and no other adults living in the home. All other participants were considered to live doubled up and then were further categorized into one of the following three categories: 1) living with adult children: at least one single adult child aged ≥18 years (e.g., son, daughter, niece, nephew, stepson, stepdaughter) and no other related adults or unrelated adults living in the home; 2) living with related family: all other relatives >18 years old (e.g., parents, in-laws, sibling, cousins, grandparents, grandchildren) but no unrelated adults living in the home; or 3) living with any unrelated adults in the household (e.g., friends, family friends, partner of friends, child of friends) regardless of other inhabitants.

Sociodemographic characteristics included age category (18–44 years; 45–64 years; ≥65 years), sex (male; female), education (less than high school; some high school; more than high school), annual household income (in US$: <20,000; 20,000–50,000; >50,000), employment (retired; unemployed; part-time; full-time), nativity (U.S.-born including the 50 states and the District of Columbia versus foreign-born or U.S. territory–born), health insurance status (yes; no), homeowner status (renter; owner), and Hispanic/Latino background (Mexican; Cuban; Puerto Rican; Dominican; Central American; South American; mixed/other [participants who chose “other” or “more than one background”]). Home ownership was determined at baseline HCHS/SOL visit, and participants were categorized as homeowner or renter. Participants who did not own or rent their housing but had some other arrangement were not included in analysis (n = 83). All covariates were measured at V2.

Whether blood glucose level or feet were checked by self, family member, or friend in the past year were measured by questionnaire at V2. Diabetes preventive care services were defined as the participant indicating affirmatively that any of the following four diagnostic tests had been conducted for diabetes-related complications in the past year: 1) dilated eye examination to determine whether diabetes has affected the retina; 2) urine test to determine whether diabetes has affected the kidneys; 3) feet were checked by a doctor, nurse, or other health professional for sores or lesions to test for foot complications and nerve damage; and 4) HbA_1c was checked by a doctor, nurse, or other health professional to determine average blood glucose level in the previous 3 months. We tested the cross-sectional association between doubled-up status and the diabetes self-management behaviors and diabetes preventive care services measured at V2.

After V2, participants were followed up annually through telephone interview and asked about all-cause hospital use since the last annual follow-up interview. The two types of hospital use variables used in this analysis were 1) ED visits only, and 2) hospitalizations only. To calculate incidence per person-year, we analyzed the cumulative frequency of each type of these hospital-based use services between the V2 visit date and the last follow-up date at the point of this analysis (November 2021).

We computed descriptive statistics (e.g., proportions) of study variables across all doubled-up categories and by Hispanic/Latino background. We used the χ² test to determine differences in prevalence of demographic, diabetes self-management behaviors, preventive care services, and health care use. Poisson regression models were used to estimate incidence rate ratios (IRRs) with 95% CIs for diabetes self-management behaviors and preventive care services at V2. Poisson regression models with count of event as an outcome, and log follow-up time (time from in-person V2 date to last annual follow-up date) as an offset was used to estimate rates (the average number of events per 100 person-years) and IRRs of hospital use. Multivariable models adjusted for all sociodemographic covariates, including additional adjustment for hospital use 1 year prior to V2 (yes versus no). All analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC), SUDAAN software, release 11.0 (RTI International, Research Triangle Park, NC), and R, version 3.6.3 (The R Foundation, Vienna, Austria). All analyses accounted for sample weights and the complex two-stage area household probability sampling design (strata, block groups) in HCHS/SOL and nonresponse at V2 (20). Survey weights adjusted for nonresponses at V2 and were trimmed and calibrated to the 2010 U.S. Census characteristics by age, sex, and Hispanic/Latino background in each field center’s target population. Statistical significance was determined at the P < 0.05 level.

In the population of Hispanic/Latino adults with diabetes (Table 1), 55.7% of participants lived doubled up when interviewed at V2. Compared with those not doubled up, a higher percentage of doubled-up participants had annual household incomes >$20,000 and had full-time employment but fewer were homeowners. The prevalence of doubled-up participants by sex, education, and age category was similar except doubled-up households had a slightly higher proportion of middle-aged members (45–64 years old). Participants doubled up with other related family had a larger proportion of family members 18–44 years old, males, and born in the U.S. but fewer had an income of <$20,000 a year or were retired than in other doubled-up arrangements. Small but distinct patterns of doubled-up living were observed by Hispanic/Latino background (Fig. 1A) and study site (city) (Fig. 1B). A larger proportion of participants of Dominican, Central American, and Mexican backgrounds lived in doubled-up households than participants of Cuban, Puerto Rican, or South American background (P < 0.0001). In Miami, a smaller proportion of participants lived in doubled-up households than participants residing in the Bronx, San Diego, or Chicago (P < 0.0001).

In adjusted multivariable analysis (Table 2), the aggregate living doubled up was associated with a lower likelihood of checking feet for sores but not in adjusted models. When disaggregated, doubled-up subcategories were not associated with either of the diabetes self-management behaviors. Living doubled up with other unrelated adults was associated with lower likelihood of having a dilated eye examination (IRR 0.83; 95% CI 0.72, 0.96) in adjusted models compared with living doubled up with adult children. Living doubled up was associated with lower likelihood of having urine checked for kidney function (IRR 0.94; 95% CI 0.89, 0.99) but not with any of the other diabetes preventive services compared with participants not living doubled up.

The average number of ED visits (n = 22 p25.3 per 100 person-years) and hospitalizations (5.4 per 100 person-years) during follow-up was higher for participants living doubled up than participants not living doubled up (Table 3) but not significantly different. The average number of ED visits per 100 person-years was lowest for participants living doubled up with adult children (n = 22 per 100 person-years) and highest for participants living doubled up with other related adults (n = 27.8 per 100 person-years) adjusting for covariates (IRR 1.33; 95% CI 1.00, 1.78). A different, however not statistically significant, pattern to that of ED visits was observed for hospitalizations, where the average number of hospitalizations per 100 person-years was highest for participants living doubled up with adult children (n = 6.3 per 100 person-years) and lowest for participants living doubled up with other unrelated adults (n = 4.6 per 100 person-years). There were no significant differences in risk of hospitalizations associated with doubled-up status. The mean follow-up time for the cohort was 2.2 years (mean 814 days; SD 228 days; median 860 days).

There are no studies, to our knowledge, that examined the relationship between living doubled up and diabetes self-management behaviors, diabetes preventive services, or hospital use among Latinos or Hispanics with diabetes. In our study, we show that living doubled up has a pattern of lower likelihood of engaging in diabetes self-management behaviors or preventive services, but the pattern was only statistically significant in adjusted analyses for having urine checked for kidney function by a doctor. However, living doubled up with other related adults was associated with a 33% increased risk of ED visits compared with living doubled up with adult children. Participants living doubled up with unrelated adults were less likely to have a dilated-eye examination by their health care provider in the past year compared with participants living doubled-up with adult children. Living doubled up was not associated with any other diabetes preventive care services. The average hospital use per person-year was lowest among participants living doubled up with adult children and highest for participants living doubled up with other unrelated adults; however, these results did not reach statistical significance.

Our results are a first step to uncovering variability in diabetes health risk associated with different types of doubled-up households among Latino/Hispanic populations. Although some of these results may appear counterintuitive, there may be many reasons for doubling up, some of which may have benefits for the household members and others that may signify housing insecurity (e.g., household crowding) and related health risks (11,22–26). The mechanisms that lead to heterogeneity in health risk associated with different housing arrangements observed in our study are unclear. In the U.S., Hispanic/Latino older adults are more likely than non–Hispanic/Latino White older adults to live in multigenerational homes with their adult children (27). In our study, there was variability in doubled-up status by Hispanic/Latino background and by study site (Fig. 1A and B). These differences may reflect how different subgroups of Latino/Hispanics in different cities and under different housing costs and regulations experience doubling up, perhaps in response to a national housing crisis like the one that precipitated the Great Recession in 2007 or within the context of multiple coexisting conditions like poverty and immigration status (28), for example. In the case of hospital use, it may be that living with older adult children is helpful in supporting healthy behaviors and managing health conditions, which may be particularly important for preventing outcomes requiring an ED visit and hospitalization. Of course, in this study, we were unable to determine with the data available the reasons for doubling up or the nature of the relationships of residents of the household. More examination of these relationships is necessary to better understand how risk is distributed under these housing conditions.

Whatever the reason for doubling up, the national data show that since the U.S. economic downturn precipitated by the housing market collapse in 2007, the number of people living in doubled-up households has steadily increased for all racial and ethnic groups (2,3). Doubling up is not as extreme a state of housing insecurity as street or shelter homelessness but can be a marker for risk of homelessness (29). Studies show that chronic conditions like diabetes (30) and hypertension (31) are more common among people experiencing homelessness. Furthermore, people experiencing homelessness also tend to have more difficulty managing chronic conditions, especially when complications arise (32). Identifying which types of doubled-up households are associated with heath risk can provide an earlier opportunity for intervention. Conversely, living doubled up can be associated with benefit and provide insights to resilience among populations.

The interpretation of these study results should consider that the collection of data on doubling up and health outcomes were cross-sectional. As such, the temporal relationship between doubled-up status and health outcomes cannot be determined. The mechanisms that help elucidate how Hispanic/Latino residents living doubled up are either protected against or at increased risk for engaging in diabetes self-management behaviors, engaging in diabetes preventive care services, or using the hospital or health care system remain unclear and require additional investigation. Furthermore, in our analyses, we assumed that doubled-up status remained unchanged from V2 through the end of data collection. Fluctuations in household composition are possible but unable to be determined with the data available. Last, the hospital use outcome was self-reported. As with all self-reported data, there is the possibility of recall bias. At the time of this analysis, there was no objective verification of self-reported hospital use.

Hispanic/Latino adults with diabetes are more likely than their White adult counterparts to have complications from diabetes and less likely to get diabetes preventive care services (33,34). Studies have shown characteristics of housing to be an important social determinant of diabetes health, including the results presented in this study examining doubling up in Hispanic/Latino households (35,36). There are 22 million doubled-up households in the U.S. and 18% are Hispanic/Latino households (9% of not doubled-up households are Hispanic/Latino) (2). Results from studies like ours provide needed data on the association of housing conditions and health that could inform housing policy aimed at providing stable housing options for at-risk families. As it relates to housing policy, policy makers and housing advocates alike have called to expand the Department of Housing and Urban Development definition of homelessness to include many more families and children that are living doubled up and at risk for homelessness, therefore expanding eligibility for federal housing assistance (37). Some estimates show an increase in the number qualifying as homeless by a factor of five if individuals living doubled up are included in the definition (38). In the current U.S. housing environment, many Hispanic/Latino families are at risk for housing insecurity. The rise in housing insecurity resulting from the economic downturn caused by the COVID-19 pandemic may warrant careful monitoring, particularly among already housing-vulnerable subgroups, including Hispanic/Latino populations (28,39). Federally Qualified Health Centers receiving Health Resources and Services Administration funding are already required to report annually on housing insecurity among their patient population, including data on doubled-up status (40). Other health care settings where Hispanic/Latino adults with diabetes receive trusted care should add housing characteristics such as doubled-up status to social-needs screening to identify residents in need of connecting with housing services and more targeted diabetes self-management and preventive services.

Funding. This work was supported by the National Heart, Lung, and Blood Institute (NHLBI) (grant R03HL140265) and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health (grants P30DK111022 and R01DK121896 to E.C.C.), and the Hispanic Community Health Study/Study of Latinos carried out as a collaborative study supported by contracts from the NHLBI to the University of North Carolina (contract N01-HC65233), University of Miami (contract N01-HC65234), Albert Einstein College of Medicine (contract N01-HC65235), Northwestern University (contract N01-HC65236), and San Diego State University (contract N01-HC65237). The following institutes, centers, and offices contribute to the HCHS/SOL through a transfer of funds to the NHLBI: National Center on Minority Health and Health Disparities, the National Institute of Deafness and Other Communications Disorders, the National Institute of Dental and Craniofacial Research, the NIDDK, the National Institute of Neurological Disorders and Stroke, and the Office of Dietary Supplements.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. E.C.C., J.L., R.S.K., and S.H. contributed to the acquisition, analysis, and interpretation of the study data, and drafted the manuscript. M.E.Y., K.M.P., L.C.G., A.L.G., R.K, M.C.-F., M.J.O., M.D.G., and C.R.I. were major contributors to interpreting the housing and diabetes data within the context of Hispanic/Latino health and writing and critically revising the manuscript. All authors read and approved the final manuscript prior to journal submission. E.C.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.