The Relationship Between Living Arrangement and Preventive Care Use Among Community-Dwelling Elderly Persons

Abstract

Objectives. We sought to examine the relationship between living arrangements and obtaining preventive care among the elderly population.

Methods. We obtained data on 13 038 community-dwelling elderly persons from the 2002 to 2005 Medical Expenditure Panel Survey and used multivariate logistic regression models to estimate the likelihood of preventive care use among elderly persons in 4 living arrangements: living alone (38%), living with one's spouse only (52%), living with one's spouse and with one's adult offspring (5%), and living with one's adult offspring only (5%). Preventive care services included influenza vaccination, physical and dental checkup, and screenings for hypertension, cholesterol, and colorectal cancer.

Results. After we controlled for age, gender, race, education, income, health insurance, comorbidities, self-reported health, physical function status, and residence location, we found that elderly persons living with a spouse only were more likely than were those living alone to obtain all preventive care services, except for hypertension screening. However, those living with their adult offspring were not more likely to obtain recommended preventive care compared with those living alone. These results did not change when the employment status and functional status of adult offspring were considered.

Conclusions. Interventions to improve preventive care use should target not only those elderly persons who live alone but also those living with adult offspring.

Ensuring the timely use of preventive care services among adults who are 65 years and older is a public health priority as the elderly population in the United States continues to grow. Strong evidence indicates that annual influenza vaccination and early detection of certain conditions such as hypertension, hypercholesterolemia, heart disease, and many forms of cancers are effective in reducing mortality and associated disability among elderly men and women.^1,2 Based on the strength of scientific evidence and the net benefit assessment of various preventive health services, the United States Preventive Services Tasks Force (USPSTF) has developed recommendations on clinical preventive care that include screening tests, counseling, and preventive medications for adults 65 years and older.³ Unfortunately, adherence to USPSTF recommendations is below expectation, and the available research documents wide variation in preventive service use. Prior studies have documented variation in preventive care use by sociodemographic characteristics such as race/ethnicity, education, and income,^4,5 as well as health-related factors including health status, psychological distress, and health beliefs.^6,7 Other studies have examined community characteristics including urban versus rural status, racial/ethnic composition, and median income level.^8,9

One important factor that has not been studied but may influence preventive care use among the elderly population is living arrangement—cohabitation with other individuals in a household unit. The United Nations has identified living arrangements of older persons as one of the most pressing concerns of the aging population.¹⁰ About 1 in 3 community-dwelling elderly persons 65 years or older, and 1 in 2 aged 80 years and older, live alone in the United States.¹¹ Previous studies found that elderly persons living alone are more likely to use publicly subsidized home care services to meet instrumental activities of daily living (IADL).¹²

In addition, some evidence suggests that with all else being equal, living arrangement may be better than marital status at predicting health services use, including physician office visits.¹³ Given these findings, it is reasonable to expect that living arrangements may be related to elderly persons' use of preventive care according to USPSTF recommendations. Because elderly persons' living arrangement can be considered a proxy measure of family-related resources for accessing preventive services, we hypothesized that even after we controlled for health and other characteristics, community-dwelling elderly men and women who live alone would be less likely to adhere to recommended preventive care than would elderly men and women who live with their spouse, children, or both.

METHODS

Data for this cross-sectional study were from the Household Component of the 2000 to 2005 Medical Expenditure Panel Surveys (MEPS). MEPS is a series of longitudinal surveys based on clustered and stratified samples of households that provide nationally representative estimates of health care use, insurance coverage, and sociodemographic characteristics for the US noninstitutionalized population sponsored by the Agency for Healthcare Research and Quality (AHRQ).¹⁴ Data were collected in 5 interview rounds over 2 years using computer-assisted in-person interviews. For all households, a knowledgeable reference person was chosen to provide information for all household members.

Our study sample was limited to 13 038 elderly men and women who resided in 4 living arrangements: living alone, living with one's spouse only, living with one's spouse and at least 1 adult offspring 18 years or older, and living with at least 1 adult offspring but no spouse. This subsample constituted 86.7% of all elderly persons in the MEPS sample. Our analyses therefore excluded 2726 elderly persons who were living in other household arrangements because they were an extremely heterogeneous group. If additional variables were constructed to capture this heterogeneity, sample sizes would be insufficient.

Variables

Preventive care.

Respondents were asked questions about time elapsed since they last received specific preventive care by a doctor or other health professional: within the past year, within the past 2 years, within the past 3 years, within the past 5 years, more than 5 years ago, or never. Based on responses to these questions and national recommended guidelines, separate dichotomous variables were created to indicate whether each of the following 6 services were obtained within the recommended timeframe: (1) influenza vaccination within the past year, (2) screening or testing for hypertension within the past 2 years, (3) screening or testing for cholesterol within the past 5 years, (4) screening or testing for colorectal cancer (either fecal occult blood test within the past year or sigmoidoscopy within the past 5 years), (5) routine physical check-up within the past 2 years, and (6) dental checkup within the past year. Similar to previous studies,^6,15 we included routine physical check-ups and dental check-ups as recommended by expert groups.^16,17

Based on USPSTF and other expert recommendations, the selected services therefore are representative of general preventive care for all individuals 65 years and older. Although not an exhaustive list, these preventive services were those captured in the MEPS. Furthermore, because these services are important for primary and secondary prevention of diseases, we did not consider prior diagnosis of a particular condition as an exclusion criterion from obtaining a related preventive service.

Living arrangement.

MEPS respondents were asked to report the relationships between all household members and a reference person who was the MEPS key informant. Using the resulting “relationship grid,” we constructed 3 versions of our key independent variable—living arrangement of the elderly population. For version 1, living arrangement was grouped into the 4 basic categories described previously. Although adult offspring could include the elderly person's adult children and children's spouse (in-laws), the majority (89%) of elderly persons living with adult offspring lived with only 1 other person.

We included versions 2 and 3 of the living arrangements for secondary analyses to also examine employment status and functional status of the adult offspring, respectively. Version 2 of the living arrangement variable was created to examine the relationship between the employment status of the adult offspring living with the elderly person and the elderly person's preventive care use. If at least 1 adult offspring was employed, it was reasonable to assume that adult offspring could potentially act as a financial resource for the elderly person living in the household. Alternatively, it was possible that employed children may have difficulty taking time off to assist an elderly parent to a doctor's appointment. Version 2 had the following 6 categories: living alone, living with one's spouse only, living with one's spouse and at least 1 employed adult offspring, living with one's spouse and unemployed adult offspring, living with at least 1 adult offspring who was employed (but no spouse), and living with all unemployed adult offspring (but no spouse).

Similarly, version 3 of the elderly persons' living arrangement variable was created to evaluate the association between functional status among adult offspring and their elderly parents' preventive care use. Functional limitation of adult offspring was defined as adult offspring needing help or supervision with at least 1 activity of daily living (ADL) or at least 1 instrumental activity of daily living (e.g., shopping, food preparation, housekeeping). If at least 1 adult offspring had a functional limitation, it is reasonable to assume that the parents would most likely have been the caregivers rather than net receivers of family resources. Version 3 had the following 6 categories: living alone, living with one's spouse only, living with one's spouse and at least 1 adult offspring who had functional limitations, living with one's spouse and functional adult offspring, living with at least 1 adult offspring who had functional limitations (but no spouse), and living with functional adult offspring (but no spouse).

Control variables.

Control variables included elderly persons' sociodemographic and health characteristics (Table 1). The elderly persons' demographic characteristics in this analysis included age (calculated based on the beginning of the survey period in which the respondent participated), gender, race/ethnicity, and interview language (English versus other language during the MEPS interview). The elderly persons' socioeconomic status was based on highest educational attainment, household income and poverty status (based on gross annual household income as a percentage of the federal poverty levels for the year of the survey¹⁴), Supplemental Security Income status, and health insurance. Health status among elderly men and women was based on self-assessed overall health status, functional limitations as indicated by the need for help or supervision with ADLs and IADLs, and the number of diagnosed conditions captured in MEPS that a person had among the following: angina, asthma, coronary heart disease, diabetes, emphysema, hypertension, heart attack, and stroke. Finally, we included residence within or outside a metropolitan statistical area.

TABLE 1.

Living Arrangement, by Characteristics of Elderly Persons 65 Years and Older (N = 13 038): Medical Expenditure Panel Survey (MEPS), United States, 2002–2005

		Living Arrangement
	Total, %	Alone, Weighted %	Spouse Only, Weighted %	Spouse and Adult Offspring, Weighted %	Adult Offspring Only, Weighted %	P	Fa
Age, y						<.001	34.19
65–69	28	27	63	6	4
70–74	24	29	61	6	4
75–79	22	39	52	3	6
80–84	15	49	40	4	7
≥ 85	11	65	24	2	9
Gender						<.001	292.78
Men	43	24	67	7	2
Women	57	49	40	3	8
Race/ethnicity						<.001	12.69
Non-Hispanic White	84	37	55	4	4
Non-Hispanic Black	7	53	33	4	10
Hispanic	6	35	42	10	13
Asian	2	29	54	8	9
Other	1	55	38	4	3
Language used in interview						<.001	5.57
English	97	35	46	9	10
Non-English	3	39	52	4	5
Highest educational attainment						<.001	23.99
Less than high school	26	45	41	5	9
High school diploma or GED	50	38	53	4	5
College degree	24	30	63	4	3
Poverty levelb						<.001	51.67
Poor	10	60	32	3	5
Near poor	7	66	28	2	4
Low income	20	49	44	2	5
Middle income	29	35	54	5	6
High income	34	22	66	7	5
Receives Social Security Income						<.001	17.44
No	96	37	53	5	5
Yes	4	54	29	6	11
Health insurance						<.001	28.83
Medicare and any private insurance	63	34	58	5	3
Medicare only or Medicare and other public insurance	37	46	42	4	8
No. of chronic health conditionsc						.004	2.79
0	25	34	56	5	5
1	34	38	53	4	5
2	21	40	50	4	6
3–8	20	40	50	5	5
Self-reported health						<.001	4.74
Excellent	14	35	56	4	5
Very good	30	38	54	4	4
Good	33	38	52	4	6
Fair	17	40	47	6	7
Poor	6	34	53	5	8
Limitations in ADL						<.001	11.3
None	6	37	53	5	5
At least 1	94	41	44	4	11
Limitations in IADL						<.001	14.3
None	12	35	55	5	5
At least 1	88	54	32	4	10
Metropolitan statistical aread						.055	2.72
Yes	23	36	55	5	4
No	77	38	51	5	6

Note. GED = general equivalency diploma; ADL = activities of daily living; IADL = instrumental activities of daily living. All statistical estimates were calculated using appropriate sample weights. Among the 13 038 individuals, 38% lived alone, 52% lived with their spouse only, 5% lived with their spouse and at least 1 adult offspring, and 5% lived with at least 1 adult offspring only.

^aTests of association between living arrangement and elderly adults' characteristics were corrected for the MEPS survey design to produce an F score.

^bPoor was defined as household annual income at or below 100%, near poor as 101% to 125%; low income as 125% to 200%; middle income as 201% to 400%; and high income as over 400% of the federal poverty level according to the survey year.¹⁴

^cChronic health conditions included angina, asthma, coronary heart disease, diabetes, emphysema, hypertension, heart attack, and stroke.

^dA metropolitan statistical area is defined as an area having 50 000 or more inhabitants.

Analysis

We calculated bivariate statistics to examine the correlation between living arrangement and other variables, including preventive service use. All standard errors and statistical tests were adjusted for the MEPS survey design with the second-order correction proposed by Rao and Scott,¹⁸ resulting in an F score. To determine the association between living arrangement and preventive service use while we controlled for other factors, we conducted multivariate logistic regression analyses for each recommended preventive service. We also estimated models for all 3 versions of the living arrangement variable. Although all of the results are discussed, only the results for versions 1 and 2 of the living arrangement variable are presented. All statistical estimates were calculated using appropriate sample weights, and standard errors were adjusted for the complex sample design of MEPS.

RESULTS

Table 2 shows the weighted percentage of the study sample obtaining preventive care as recommended, and Table 1 shows weighted sample means of all variables by living arrangement.

TABLE 2.

Weighted Percentages of Elderly Adults Aged 65 Years and Older (N = 13 038) Using Preventive Care Services, by Living Arrangement: Medical Expenditure Panel Survey (MEPS), United States, 2002–2005

		Living Arrangement
Preventive Care Usea	Total, %	Alone, Weighted %	Spouse Only, Weighted %	Spouse and Adult Offspring	Adult Offspring Only, Weighted %	P	Fb
Influenza vaccination	72	73	77c	74	63c	<.001	11.33
Hypertension screenings	92	95	96	95	93	.144	1.84
Cholesterol screenings	85	86	90c	87	84	<.001	12.19
Colorectal cancer screenings	54	52	61c	50	43c	<.001	21.79
Routine physical check-up	86	87	89c	87	84	.003	4.85
Routine dental check-up	53	48	62c	49	35c	<.001	44.82

Note. All statistical estimates were calculated using appropriate sample weights. Among the 13 038 individuals, 38% lived alone, 52% lived with their spouse only, 5% lived with their spouse and at least 1 adult offspring, and 5% lived with at least 1 adult offspring only.

^aUse of preventive care service according to US recommendations included (1) influenza vaccination within the past year, (2) screening for hypertension within the past 2 years, (3) screening for cholesterol within the past 5 years, (4) screening for colorectal cancer (either fecal occult blood test within past year or sigmoidoscopy within the past 5 years), (5) routine physical check-up within the past 2 years, and (6) dental checkup within the past year.

^bA test of association between living arrangement and preventive care service use corrected for the MEPS design to produce the F score.

^cThe proportion was significantly different from the proportion of those living alone at P = .05.

Bivariate Analysis

Over one half of our study sample lived with their spouse only, and more than one third lived alone. Adherence rates for recommended preventive care among elderly men and women were highest for hypertension screening and lowest for colorectal cancer screening and routine dental check-up. These average percentages differed significantly across the 4 living arrangements among all preventive care services except hypertension screening. Compared with living alone, those living with their spouse only had higher average percentages of all preventive services except hypertension screening. Those living with adult offspring only had lower average percentages of influenza vaccination, colorectal cancer screening, and routine dental check-up compared with those living alone. Furthermore, those living with their spouse and adult offspring had statistically similar average percentages of using preventive care services as did those living alone.

All sample characteristics, except for residence in a metropolitan statistical area, varied significantly by living arrangement. Elderly persons in younger cohorts (ages 65–69 and 70–74 years) disproportionately lived with their spouse only or their spouse and adult offspring, whereas elderly persons in older cohorts (ages 80–84 and ≥ 85 years) disproportionately lived alone or with adult offspring only. Similarly, more men lived with their spouse only or their spouse and adult offspring, whereas more women lived alone or with adult offspring only. Those residing in poor and near-poor households disproportionately lived alone, whereas those residing in middle- and high-income households disproportionately lived with their spouse only. Those reporting fair and poor health disproportionately lived with adult offspring only. A greater number of elderly persons having limitations in at least 1 ADL or IADL lived alone or with adult offspring only, whereas a greater number of elderly persons with no limitations in ADL or IADL lived with their spouse only.

Multivariate Analysis

Table 3 shows the results from 6 multivariate logistic regression models, 1 for each preventive service. Multivariate regression models that controlled for other factors (i.e., age, gender, race/ethnicity, primary language, education level, household income, health insurance, chronic conditions, self-reported health, functional status, and residence location) indicated that elderly men and women living with their spouse only were more likely than were those living alone to use all preventive care services according to guidelines, except for hypertension screening. (Table 3 shows only adjusted odds ratios for the living arrangement variable. Complete results are available upon request from the authors.) Elderly persons living with their spouse and adult offspring had similar odds as those living alone to obtain all types of preventive care. Compared with elderly men and women living alone, those living with adult offspring only had similar odds of obtaining routine physical check-ups and screening for hypertension and cholesterol but had lower odds of obtaining influenza vaccination, colorectal cancer screening, and dental check-ups.

TABLE 3.

Adjusted Odds Ratios (AORs) for Living Arrangements From Multivariate Logistic Regression Models on Recommended Preventive Care Use Among Elderly Adults Aged 65 Years and Older (N = 13 038): Medical Expenditure Panel Survey, United States, 2002–2005

Living Arrangement	Influenza Vaccination, AOR (95% CI)	Hypertension Screening, AOR (95% CI)	Cholesterol Screening, AOR (95% CI)	Colorectal Cancer Screening, AOR (95% CI)	Routine Physical Check-Up, AOR (95% CI)	Routine Dental Check-Up, AOR (95% CI)
Alone (Ref)	1.00	1.00	1.00	1.00	1.00	1.00
Spouse only	1.33** (1.13, 1.55)	1.28 (0.95, 1.71)	1.49** (1.23, 1.79)	1.17* (1.02, 1.34)	1.25* (1.05, 1.50)	1.31** (1.14, 1.49)
Spouse and adult offspring	1.15 (0.85, 1.55)	1.13 (0.68, 1.88)	1.00 (0.73, 1.38)	0.77 (0.58, 1.02)	0.97 (0.69, 1.36)	0.80 (0.59, 1.07)
Adult offspring only	0.65** (0.49, 0.85)	0.80 (0.48, 1.32)	0.90 (0.63, 1.27)	0.77* (0.60, 0.98)	0.80 (0.59, 1.08)	0.60** (0.47, 0.76)

Note. CI = confidence interval. Preventive care services included (1) influenza vaccination within the past years, (2) hypertension screening within the past 2 years, (3) cholesterol screening within the past 5 years, (4) colorectal cancer screening (either fecal occult blood test within the past year or sigmoidoscopy within the past 5 years), (5) routine physical check-up within the past 2 years, and (6) dental checkup within the past year. All multivariate regression models were adjusted for living arrangement and the following variables: age, gender, race/ethnicity, language, education, poverty status, Social Security income, health insurance, number of chronic health conditions, self-perceived health, limitations in activities of daily living and instrumental activities of daily living, and metropolitan statistical area.

*P ≤ .05; **P ≤ .001.

To further examine whether living with adult offspring was a potential resource for elderly persons to obtain preventive care, we conducted additional multivariate regression models using versions 2 and 3 of the living arrangement variables that controlled for the employment status and functional status of adult offspring. We found that, with respect to preventive care use, the difference between elderly persons living with employed adult offspring and those living with unemployed adult offspring was minimal (Table 4). When there was a difference, living with an employed offspring was associated with lower odds of receiving recommended preventive care than was living with unemployed offspring. Similarly, we found no evidence that elderly persons living with adult offspring who had limitations in ADLs or IADLs differed from those living with functional adult offspring with respect to preventive service use (data available upon request).

TABLE 4.

Adjusted Odds Ratios (AORs) for Living Arrangements and Employment Status of Adult Offspring From Multivariate Logistic Regression Models on Recommended Preventive Care Use Among Elderly Adults Aged 65 Years and Older (N = 13 038): Medical Expenditure Panel Survey, United States, 2002–2005

	Influenza Vaccination, AOR (95% CI)	Hypertension Screening, AOR (95% CI)	Cholesterol Screening, AOR (95% CI)	Colorectal Cancer Screening, AOR (95% CI)	Routine Physical Check-Up, AOR (95% CI)	Routine Dental Check-Up, AOR (95% CI)
Alone (Ref)	1.00	1.00	1.00	1.00	1.00	1.00
Spouse only	1.32** (1.13, 1.55)	1.28 (0.95, 1.72)	1.49** (1.23, 1.79)	1.17* (1.02, 1.34)	1.26* (1.05, 1.50)	1.3** (1.14, 1.49)
Spouse and unemployed adult offspring	1.33 (0.77, 2.32)	1.71 (0.66, 4.42)	1.25 (0.68, 2.30)	1.23 (0.79, 1.91)	1.19 (0.64, 2.20)	0.91 (0.51, 1.61)
Spouse and employed adult offspring	1.08 (0.77, 1.53)	0.99 (0.56, 1.76)	0.92 (0.63, 1.32)	0.64** (0.47, 0.87)	0.9 (0.61, 1.33)	0.76 (0.53, 1.07)
Unemployed adult offspring only	0.82 (0.53, 1.27)	0.52 (0.25, 1.11)	0.71 (0.41, 1.23)	0.62* (0.43, 0.89)	0.64 (0.41, 1.00)	0.8 (0.55, 1.15)
Employed adult offspring only	0.59** (0.44, 0.81)	1.03 (0.54, 1.97)	1.01 (0.66, 1.53)	0.83 (0.62, 1.11)	0.88 (0.59, 1.30)	0.54** (0.41, 0.72)

Note. CI = confidence interval. Preventive care services included (1) influenza vaccination within the past year, (2) hypertension screening within the past 2 years, (3) cholesterol screening within the past 5 years, (4) colorectal cancer screening (either fecal occult blood test within the past year or sigmoidoscopy within the past 5 years), (5) routine physical check-up within the past 2 years, and (6) dental checkup within the past year. All multivariate regression models were adjusted for living arrangement and the following variables: age, gender, race/ethnicity, language, education, poverty status, Social Security Income, health insurance, number of chronic health conditions, self-perceived health, limitations in activities of daily living and instrumental activities of daily living, and metropolitan statistical area.

*P ≤ .05; **P ≤ .001.

DISCUSSION

An important Healthy People 2010 objective is ensuring adequate and timely access to health care for individuals of all ages.¹⁷ Toward that goal, the USPSTF made evidence-based recommendations on a number of preventive care services for all adults 65 years and older. Using nationally representative data from the 2002 to 2005 MEPS, we examined the association between household living arrangement and the elderly's adherence to these recommendations. We hypothesized that living with others (a spouse, adult offspring, or both) would improve elderly persons' adherence to recommended preventive care. This hypothesis was only partially supported.

Consistent with our hypothesis, our findings indicated that compared with those living alone, elderly adults living with their spouse were more likely to use preventive care. By contrast, elderly adults living with adult offspring, even with the presence of a spouse, were similar to those living alone with respect to obtaining recommended preventive care. For some services, living with adult offspring actually indicated a lower likelihood of adhering to preventive care guidelines. Further analyses accounting for employment status and functional status of adult offspring produced similar findings. These findings did not change even when many of the elderly adults' sociodemographic and health characteristics were held constant.

Our findings raise an obvious question for future research: Among elderly persons, why does living with one's spouse aid in the timely use of preventive services whereas living with adult offspring does not, when compared with elderly persons living alone? Prior studies have found that spouses often play important roles in providing emotional and instrumental support and influencing an individual's health behaviors and illnesses.^19,20 Perhaps because of shared life and health experiences or being in the same age cohort, elderly adults and their spouses may help each other in accessing preventive care by reminding each other of health examinations or assisting each other in traveling to health care facilities together. Moreover, parent–child relationships may differ from spousal relationships in terms of the social support and caregiving provided.

Understanding the characteristics that predispose elderly adults to live with their adult offspring or that predispose adult offspring to live with their elderly parents may also help interpret these findings. For example, elderly adults who transition into living with their adult offspring for health-related or economic reasons may have great difficulty in obtaining preventive care regardless of the presence of adult offspring. Adult offspring who live with their elderly parents may have characteristics that make them less able to provide assistance and resources to their parents in obtaining timely preventive care. Some adult offspring may even have characteristics that demand time and resources from elderly parents, thereby making access to preventive care more challenging. Although this study has provided some evidence that employed children may have difficulty in assisting an elderly parent to a doctor's appointment for preventive care, further examination of intergenerational resource exchange and caregiver burden may explain these findings.

We also found that the prevalence of elderly adults getting colorectal cancer screenings and routine dental check-ups were generally low compared with other preventive services; these findings are consistent to previous studies.⁶ Reasons for low rates of colorectal cancer screening may be include patients' concerns about painful procedures or stigma associated with this particular screening test. Low rates of routine dental check-ups may be attributed to the fact that these procedures are frequently not covered by health insurance, including Medicare. Furthermore, USPSTF recommendations currently have not established the appropriate age at which screenings should be discontinued, except for colorectal cancer screening. In October 2008, USPSTF revised their guidelines to no longer recommend colorectal cancer screening in adults older than 75 years. Additionally, analysis found that living arrangement was a significant factor associated with colorectal cancer screening among adults between the ages of 65 and 75 years. Clinical decisions about preventive screenings in older adults should take into account the potential costs and benefits of such tests to the patient.

Limitations

There are limitations to this study. Data on preventive care use were self-reported and therefore subject to recall bias. Although there is evidence that supports the reliability of self-reported influenza vaccination²¹ and colorectal screening,²² evidence is mixed on the reliability of other self-reported preventive care services, especially those involving blood tests.²³ Furthermore, although we examined general preventive care services that were captured in the MEPS for all individuals 65 years and older, other services could be investigated in future research, such as mammography screening for elderly women.

In addition, our analyses excluded elderly persons living with individuals other than their spouse and adult offspring, because elderly adults living in other household arrangements are an extremely heterogeneous group (e.g., living with nonrelatives or with children younger than 18 years) and constructing additional variables to capture this heterogeneity would result in insufficient sample sizes. Limiting our study sample therefore prevents us from extending the interpretation of our findings to elderly adults living in those arrangements that we have excluded from our sample.

Finally, because of the cross-sectional and nonexperimental design of this study, a causal relationship between household living arrangement and preventive care use cannot be assumed. Although we controlled for a number of potential confounders including a number of sociodemographic characteristics and health-status measures, our findings may be a reflection of some other underlying factors such as unmeasured functional or health status that may have affected both living arrangement and preventive care use. Unobserved factors that predispose adult offspring to live with their elderly parents or predispose elderly individuals to live with their adult offspring may prevent the elderly adult's use of preventive care. Further research should investigate whether our findings are the results of causation or selection.

Conclusions

To our knowledge, ours is the first study to investigate the potential association between living arrangement and recommended preventive care. In doing so, our study can inform policies designed to improve the elderly population's adherence to preventive care guidelines and make service delivery to this group more effective. Despite the limitations, we provide evidence that the prevalence of elderly adults' adherence to recommended preventive care, especially colorectal cancer screenings and routine dental check-ups, remains below national goals. Preventive care should remain a priority among the elderly population, even among those with poor health, to guard against secondary diseases and promote overall health. In addition, our findings call attention to the importance of recognizing elderly adults' living arrangement as an important factor for designing public health programs to improve preventive care use among the elderly population. Because the presence of adult offspring cannot be considered by default a resource for elderly individuals to obtain preventive care, educational and outreach interventions should target not only those elderly adults who live alone but also those living with adult offspring. Furthermore, research efforts should be taken to understand how elderly individuals benefit (or do not benefit) from living situations involving their adult offspring.