QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Davis, M. A. Articles by Quiroga, S. S. Search for Related Content PubMed PubMed Citation Articles by Davis, M. A. Articles by Quiroga, S. S.

---

Article

Living Arrangements Affect Dietary Quality for U.S. Adults Aged 50 Years and Older: NHANES III 1988–1994¹

Department of Epidemiology and Biostatistics, School of Medicine, University of California, San Francisco, CA 94143-0560

²To whom correspondence should be addressed.

	ABSTRACT

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
The number and proportion of older U.S. adults who live alone have increased dramatically in the past three decades, and there is concern that these individuals may have particularly poor dietary quality. We examined the association of four living arrangements (living with a spouse only, with a spouse plus someone else, with someone other than a spouse or living alone) with dietary quality (the number of low nutrients out of a possible 15, with low defined as <67% of the recommended dietary allowance) among 6525 U.S. adults aged 50–64 y and those 65 y in the third National Health and Nutrition Examination Survey (NHANES III 1988–1994). Among non-Hispanic Caucasian adults, those who lived with a spouse only had better dietary quality, with significant differences ranging from 0.8 to 1.5 fewer low nutrients compared with those with other living arrangements. Effects of living arrangements on dietary quality were also seen among non-Hispanic African-Americans, Mexican-Americans, and those of "other" races, but differences were significant only for African-American men aged >65 y living with a spouse plus others (1.6 additional low nutrients compared with those living with a spouse only). Energy intake was strongly associated with dietary quality, but did not account for the associations between living arrangements and dietary quality. Although middle-aged and older adults with living arrangements other than living with a spouse only (including those living alone) tended to have poorer dietary quality, the effects varied substantially across age, gender and ethnic categories.

KEY WORDS: • older adults • living arrangements • dietary quality • nutrient intake • NHANES III

	INTRODUCTION

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
In the United States, both the number and proportion of older adults who live alone have increased dramatically since 1960 (U.S. Department of Commerce 1986 ). In 1996, of the 31.7 million persons aged 65 y in the U.S., 24% of those aged 65–74 y lived alone and 41% of those aged 75 y lived alone (Saluter and Lugaila 1998 ). Changes in marital status and living arrangements with advancing age are more prevalent for older women than for older men because older women are more likely to be widowed (Saluter and Lugaila 1998 ). Thus, women 65 y are more likely to live alone (41 vs. 17% for men), whereas older men are more likely to live with their spouses (73 vs. 41% for women) (Saluter and Lugaila 1998 ). In general, the ability to remain independent and live alone is associated with a high quality of life among older adults (Cantor and Little 1985 ). However, there is also concern that older adults living alone may be particularly vulnerable to poverty, social isolation, inadequate dietary intake and adverse health outcomes (Commonwealth Fund 1989 , 1992a and 1992b , U.S. Senate 1991 , USDA 1994 ).

Studies of dietary intakes of older adults in the United States have shown low intakes of energy and several nutrients (Murphy et al. 1990 , Ryan et al. 1992 , USDA 1997 ), leading to concern about dietary adequacy of older adults compared with middle-aged and younger adults. Evidence addressing the issue of dietary quality of older adults living alone is fragmented, inconsistent and inconclusive. Many studies have included only women, and there has been little research examining the dietary risk of ethnic minorities or the oldest old in relation to their living arrangements. Some studies suggest a negative dietary influence of living alone, eating alone or social isolation (Garofalo and Hynak-Hankinson 1995 ), whereas other studies have not found these factors to be associated with dietary quality (McIntosh et al. 1989 , Posner et al. 1994 ). Most studies are based on small samples of older adults, so that generalization of the findings is limited.

Although there is some evidence to suggest that dietary quality and living arrangements may affect health and mortality differently for middle-aged compared with older adults (Davis et al. 1992 , Murphy et al. 1996 ), there has been little investigation concerning differences in the association of living arrangements with dietary quality for older adults compared with middle-aged and younger adults. A number of eating behaviors that are associated negatively with dietary quality have been found to occur more prevalently in adults aged 55–64 y, with the prevalence of these negative eating behaviors decreasing with age (65–74 y and 75 y). It is therefore important to examine the association of living arrangements with dietary quality among middle-aged adults to identify potential risk factors for poor dietary quality that may continue as the cohort ages.

National dietary data from the first and second National Health and Nutrition Surveys (NHANES)⁴ have shown that living arrangements were more strongly associated with dietary patterns for older men than for older women (those older men not living with a spouse were at higher risk of poor dietary intake, regardless of whether they lived alone). Men of low income, not living with a spouse, were at highest risk of poor dietary patterns (Davis et al. 1985 , Ryan et al. 1989 ). Analysis of data from the 1977–1978 Nationwide Food Consumption Survey (NFCS) also indicated a gender difference in the association of living arrangements with dietary quality for older adults. Older men living alone were more likely to consume poor quality diets compared with those living with a spouse (Davis et al. 1990 ). In addition, Murphy et al. (1990) found that differences in energy intake accounted for much of the association of living arrangements with dietary quality; this finding was consistent with those reported from the 1987–1988 NFCS (USDA 1994 ).

The third National Health and Nutrition Examination Survey (NHANES III, 1988–1994) [National Center for Health Statistics (NCHS) 1994 ] constitutes the most current national data available to address issues of living arrangements and dietary quality among older U.S. adults. This survey offers new opportunities for examining the association of these variables because unlike previous cycles of NHANES, no upper age limits were imposed, and older African-American and Mexican American populations were oversampled (Burt and Harris 1994 ).

The goals of our analysis of the NHANES III data are as follows: 1) to examine gender, age and ethnic differences in the association of living arrangements and dietary quality among U.S. adults 50 y old; 2) to investigate whether certain factors, e.g., age, race/ethnicity, education, income, employment, body mass index (BMI), physical activity, energy intake, skipped meals, nutrient supplement use, alcohol consumption, smoking and health status, accounted for the association of living arrangement with dietary quality; 3) to analyze the relative importance of these factors compared with living arrangements in the association with dietary quality.

	SUBJECTS AND METHODS

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
Sample.

The NCHS of the Centers for Disease Control conducted NHANES III from 1988 through 1994 (NCHS 1994 ) and designed it to provide national estimates of the health and nutritional status of the U.S. civilian, noninstitutionalized population. An interview was conducted in the participant’s home, followed by a clinical examination in a mobile examination center. For this study, we selected participants who were 50 y old, who reported for themselves, and who provided complete living arrangement and 24-h dietary recall data. We chose this age range so that we could compare the association of living arrangements with dietary quality for older adults (65 y) compared with middle-aged adults (50–64 y). The sample size for analysis included 3435 women and 3090 men. The protocol was approved by the University of California San Francisco, Committee on Human Research.

Living arrangements assessment.

The household interview collected information on the number of persons in the household and marital status. We used this information to categorize survey participants into one of the following four living arrangements: living alone; living with a spouse and no one else; living with a spouse plus at least one other person; and living with persons other than a spouse. Other persons in the household included unrelated individuals, relatives such as child, grandchild, parent or other relatives as well as the spouse of a relative. Information was not available on the relationship of all persons in a household (only the relationship of the sample person to the head of household). Therefore it was not possible to identify the relationship of all of the people in the household in more detail than "other than spouse." Persons living with a spouse included both those who reported that they were married and those who reported that they were living as married.

Dietary assessment.

The examination collected a 24-h recall of all foods and beverages consumed the previous day. The survey used an automated, interactive dietary interview and coding system. Portion sizes were quantified using abstract food models, shape charts and measuring aids such as rulers, cups and spoons. Dietary recalls were collected for every day of the week; weekend days are underrepresented, whereas Fridays are overrepresented. Nutrient intakes for each participant were calculated using the gram amounts of the food consumed and the USDA Survey Nutrient Database. For further details of these procedures see the description by Briefel et al. (1997) .

We compared daily nutrient intake totals (from food) with the recommended dietary allowance (RDA) for each of 15 nutrients (protein, thiamin, riboflavin, niacin, folate, vitamins A, C, E, B-6 and B-12, iron, zinc, calcium, phosphorus and magnesium). For calcium, we used the recently published Adequate Intake of 1200 mg/d; we also used the new RDA for phosphorus, magnesium and the B-vitamins [Institute of Medicine (IOM) 1997 and 1998 ]. For all other nutrients, we used the RDA from 1989 (NRC 1989 ). We developed a score for each participant that reflected the number of nutrients that were <67% of the RDA. Because all nutrients are necessary for optimum health, a scale that reflects the number of low nutrients has a meaningful interpretation in evaluating overall nutritional adequacy. We chose intake below two thirds of the RDA as a conservative cut-off to define a low nutrient intake; this allowed us to account for underreporting as well as for the likelihood that nutrients in a 1-d diet have a broader distribution than those reflecting usual intake (Life Science Research Office 1986 ).

Gender, age and ethnicity.

Our analytic strategy allowed the associations of living arrangement and dietary quality to vary by gender and age group (55–64 y, 65 y) by conducting separate analyses in the four gender-age groups. These analyzes also allowed the living arrangement-dietary quality associations to vary by ethnicity by including interactions of ethnicity and living arrangement.

We classified participants into one of four ethnic categories: non-Hispanic Caucasian, non-Hispanic African-American, Mexican American and other race (including Native American, Asian and other Hispanics), using self-reported race and ethnic identification.

Potential confounding variable assessment.

Other confounding variables included education (years completed); employed (yes or no); clinically measured BMI (kg/m²); 1-d energy intake; skipping a meal the day before due to lack of food or money to buy food (yes or no); vitamin/mineral supplement consumption in the past month (yes or no); smoking status (current smoker or not a current smoker); and self-perceived general health status (excellent, very good, good, fair or poor). Alcohol consumption was dichotomized as "none" or "any" regular consumption, on the basis of responses to the food-frequency question regarding beer, wine and liquor consumption. Because we were not specifically investigating health effects of alcohol consumption, we did not further expand the number of categories. We also included leisure-time physical activity in the analyses, creating a variable based on the self-reported total number of times in the past month that the subject performed moderate exercise. We counted only exercise performed at an intensity of at least three times the resting metabolic rate and trichotomized the variable as no moderate exercise vs. 1–30 times/mo vs. at least 31 times/mo (or approximately daily). We chose these categories to reflect the current recommendation that all adults should accumulate 30 min of moderate-intensity physical activity on most, preferably all days of the week (Pate et al. 1995 ).

For a subsample of persons with reported income (n = 5835), we included the poverty index ratio in the analysis. NCHS calculated the poverty index ratio as the ratio of total family income to a poverty threshold for the year of the interview. We used this ratio as a continuous variable in our analyses.

Statistical analysis.

We assessed the association of living arrangements and other factors with dietary quality (the number of nutrients <67% of the RDA) using linear regression analysis and the GENMOD routine in SAS (SAS Institute 1997 ). The regression models included indicator variables to describe the living arrangement groups, interactions of ethnicity and living arrangement, and terms to adjust for the effects of sociodemographic, health behavior and health variables. Because standard linear regression models for the number of nutrients <67% of the RDA can produce values out of the range 0–15, we used a transformation of the count outcome as the response in regression models. Specifically, the outcome for the regression models shown in Tables 3 and 4 was the arcsine of the square root of the proportion of nutrients (out of 15) that were <67% of the RDA. This is a standard transformation for counts and proportions [see, for example, Draper and Smith (1981) ], and all predicted values based on this response are valid values. We used the ethnicity-living arrangement interaction terms to calculate ethnicity-specific associations of living arrangement and dietary quality. We calculated adjusted means and associated 95% confidence intervals (CI) using the fits of the models on the arcsine scale and then transformed these values back to the original count scale for Tables 3 and 4 . We computed CI for the differences in diet quality among living arrangement groups using the Wald approach. We calculated variance estimates of the differences in proportions using Taylor series methods (Walter 1985 ). We compared the magnitudes of the associations of living arrangement with dietary quality with the associations of other variables such as ethnicity with dietary quality by comparing the adjusted means and differences between means.

View this table: [in this window] [in a new window]   Table 3. Differences in number of nutrients below 67% of the RDA from multiple linear regression models for those living with spouse plus others, with others than spouse or living alone compared with those living with spouse only, by age group (50–64 y and 65 y), gender and ethnicity: NHANES III, 1988–199412

View this table: [in this window] [in a new window]   Table 4. Differences in number of nutrients below 67% of the RDA from multiple linear regression models for sociodemographic, health behavior and health characteristics for women and men by age group (50–64 y; 65): NHANES III, 1988–199412

	RESULTS

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
Table 1 presents the distributions of the demographic and health characteristics by gender for each living arrangement category. Persons who lived with a spouse only were more likely to be non-Hispanic Caucasian, had the most education, were more likely to consume alcohol and were less likely to smoke than those in other living arrangement groups. Women who lived with a spouse only were more likely to exercise regularly and to have higher self-reported health than those in other groups. Men who lived with a spouse only were more likely to take vitamin/mineral supplements.

Adults living with a spouse plus someone else were younger, were more likely to be employed and had a higher energy intake than those in other living arrangement groups. Women living with a spouse plus someone else had higher BMI, whereas men had better self-reported health status. Persons who lived with someone other than a spouse were somewhat more likely to be non-Hispanic African-American, had the least education, were less likely to exercise and had lower energy intake than persons in other living arrangement groups. Women who lived with someone other than a spouse were more likely to smoke and report fair/poor health than women in other living arrangement groups.

Persons who lived alone were older than those in the other living arrangement groups, with women living alone having the oldest mean age (71 y). They were less likely to be employed and had the lowest BMI. More men who lived alone smoked and were in fair/poor health.

For all adults aged 50 y, the mean number of nutrients (from food) less than two thirds of the RDA was 4.7 (SEM = 0.7) for women and 3.6 (SEM = 0.06) for men. Overall, 12% of women and 20% of men reported diets with no low nutrients, whereas only 1.5 and 0.5%, respectively, reported diets with all 15 nutrients less than two thirds of the RDA. Nutrients most likely to be low were calcium, magnesium, zinc, folate and vitamin E.

Table 2 presents the mean number and percentage of low nutrients for each living arrangement category by gender. Among the four living arrangements, both women and men living with a spouse had the fewest number of low nutrients in their 1-d diets (4.4 and 3.3, respectively) compared with persons of the same sex in the other living arrangement categories. Both women and men living with someone other than a spouse had the greatest number of low nutrients (5.4 and 4.6, respectively). Persons living alone, or living with a spouse plus someone else, reported diets of intermediate nutrient quality.

Because we were interested in possible age, gender and race/ethnicity differences in the association of living arrangements and dietary quality, we stratified these data by gender and two age groups (50–64 y and 65 y). Within the age and gender subgroups, we then examined the data for interactions between living arrangements and race/ethnicity. Likelihood ratio test statistics for the interaction of ethnicity and living arrangement with 9 df in each gender-age group were as follows: 23.6 (P = 0.005) for women 50–64 y; 23.1 (P = 0.006) for men 50–64 y; 23.6 (P = 0.005) for women 65 y; 15.8 (P = 0.071) for men 65 y. Because the interactions of living arrangements and race/ethnicity were significant for three of the four age-gender groups, the analyses for Tables 3 and 4 were stratified by age group and gender and included interactions of race/ethnicity and living arrangements.

Table 3 gives the age-group–, gender- and ethnicity-specific (non-Hispanic Caucasian, non-Hispanic African-American, Mexican-American and other) mean number of low nutrients by living arrangement category. Within each ethnic group, those living with spouse only comprise the comparison group. The data are adjusted for all potential confounding variables from Table 2 .

The first column of Table 3 shows the mean number of low nutrients (<67% of the RDA) for persons living with a spouse only for age-gender-ethnic groups. Non-Hispanic African-American men in both age groups who lived with a spouse only had significantly more low nutrients compared with non-Hispanic Caucasian men in this same living arrangement category; those 50–64 y old had 1.54 more low nutrients (95% CI = 0.34, 2.74), whereas those 65 y had 0.94 more low nutrients (95% CI = 0.06, 1.83).

Among all of the groups of persons living with a spouse only, non-Hispanic Caucasian men and Mexican-American men 50–64 y old had the fewest number of low nutrients (2.5), whereas women 65 y in the "other" ethnic group had the most low nutrients (5.6). Women tended to have more low nutrients than men in both age groups. Men 65 y living with a spouse only had more low nutrients than those 50–64 y. The pattern was similar for women, with the exception of non-Hispanic African-American women.

The rows of Table 3 show that within each of the four ethnic groups, men and women aged 50–64 y who lived with a spouse only generally had better dietary quality than men and women in the other living arrangement groups, although most differences were significant only for non-Hispanic Caucasians. Among non-Hispanic Caucasians 50–64 y, significantly more low nutrients were observed for women living with someone other than a spouse and those living alone; men with all three other living arrangements had significantly more low nutrients compared with those 50–64 y living with a spouse only. For adults 65 y, significantly more low nutrients between living with a spouse only and the three other types of living arrangements were observed for non-Hispanic Caucasian men living alone and for non-Hispanic African-American men living with a spouse plus someone else. In contrast, Mexican-American women and "other" women and men in this age group, who were living with a spouse plus some one else, living with someone other than a spouse or living alone, had fewer low nutrients than those living with a spouse only, although these differences were not significant.

Table 4 presents the gender/age-group associations of several other sociodemographic, health behavior and health variables from Table 2 with dietary quality (the number of nutrients <67% of the RDA), after controlling for all other variables in the model. Examining the magnitude of the differences that are significant in both Tables 3 and 4 , we see that for all age and gender groups, energy intake made the largest difference in dietary quality. Increasing energy intake from 1500 kcal/d (6.27 MJ/d) to 2500 kcal/d (10.45 MJ/d) was associated with a decrease of >2 low nutrients/d for men 50–64 y and >3 low nutrients/d for women 50 y and for men 65 y. A number of other factors accounted for a difference of 1.5 or greater in the number of low nutrients, but none was as large a difference as for energy intake. Significant differences of at least 1.5 low nutrients were observed for smoking for women 65 y (1.7 more low nutrients than women of the same ages who did not smoke); living arrangement for non-Hispanic African-American men 65 y (living with a spouse plus someone else, 1.6 more low nutrients compared with non-Hispanic African-American men 65 y living with a spouse only ); an ethnic-living arrangement effect of 1.5 more low nutrients for non-Hispanic African-American men 50–64 y living only with a spouse compared with the same age group of non-Hispanic Caucasian men living only with a spouse.

Some living arrangements were associated with changes of 1.5 low nutrients or more, but did not reach significance due to the smaller sample sizes for the gender-, age-group– and ethnicity-specific analyses: "other" men 65 y who lived alone (1.7 fewer low nutrients than "other" men 65 y who lived with a spouse only); Mexican-American women 65 y who lived with a spouse plus someone else (1.6 fewer low nutrients than Mexican-American women 65 y who lived with a spouse only); "other" men 50–64 y who lived with someone other than a spouse (1.6 more low nutrients compared with their spouse only comparison group).

Education, exercise, alcohol consumption, smoking and taking vitamin or mineral supplements were all associated with dietary quality, but none consistently accounted for an increase or decrease of >1 low nutrient for any of the four age-gender groups. Employment and self-reported health status were not consistently associated with dietary quality across the four age-gender groups, whereas BMI and skipping meals were not significant predictors of dietary quality for any gender-age group (data not shown).

Because 8% of the self-respondents (n = 690) in the survey did not report income data, we did not include this measure in the analyses in Table 3 or 4 . The means of the nutrient quality variable differed between those who reported income data (mean = 4.19; SEM = 0.05) compared with those who did not report income data (mean = 4.59; SEM = 0.15). Therefore, analyses of the subsample who provided income information may yield misleading results. To avoid this bias and retain the representative sample, we analyzed the full sample and adjusted for socioeconomic status using variables such as education and ethnicity instead of income. Regression analyses of the subsample who reported income showed that the covariates in our final models explained 36% of the variance in income values, suggesting that our models using the full sample adjusted adequately for income.

However, we also assessed the relationship of income (poverty index ratio) with living arrangement and dietary quality for the subsample of 5835 self-respondents who had valid income data (3030 women and 2805 men) as in Tables 3 and 4 (data not shown). Although income was associated with dietary quality, adjusting additionally for income produced little change in the estimated regression coefficients and thus no meaningful differences in the patterns of associations of living arrangements with dietary quality.

Overconsumption as well as underconsumption can affect dietary quality. Therefore, we examined the percentage of energy from fat and from saturated fat as markers of overconsumption that might be related to chronic diseases among older adults. Men averaged 34 ± 10% of energy from fat and 11 ± 4% of energy from saturated fat; corresponding percentages for women were 32 ± 11% and 11 ± 5%. Women living alone averaged 1.3% less energy from fat than women living with a spouse only and women with a spouse plus others averaged 0.5% more energy from saturated fat than women living with a spouse only. Men living alone also averaged 0.5% more energy from saturated fat than men living with a spouse. Thus, although these differences were significant, the magnitude of the differences in the proportion of energy from fat and saturated fat among living arrangement groups was small.

	DISCUSSION

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
The most consistent finding from our analysis of the association of living arrangements with dietary quality among persons aged 50 y in NHANES III was the favorable dietary pattern of non-Hispanic Caucasian adults living with a spouse only compared with those with other living arrangements. Analysis of NHANES I and II for persons aged 65–74 y old included three categories of living arrangements (living with a spouse, living with someone other than a spouse or living alone) (Davis et al. 1985 , Ryan et al. 1989 ). For both NHANES I and II, those living with a spouse had more favorable dietary quality than persons who lived with someone other than a spouse or who lived alone, (Davis et al. 1985 , Ryan et al. 1989 ), suggesting that merely living with someone was not the main criterion for more favorable dietary intake; in fact, the presence of a spouse had more significance. To test this hypothesis, we split the category of living with a spouse into two categories: living with a spouse only and living with a spouse plus someone else. We expected that if living with a spouse resulted in favorable dietary quality, then those living with a spouse only and those living with a spouse and someone else would have comparable favorable dietary quality, and living arrangement groups without a spouse present would have less favorable dietary quality. However, our findings suggest a more complex effect in which the presence of another person in the husband-wife household decreases dietary quality.

The NHANES III sample design allowed us to examine for the first time with a national dietary data set whether non-Hispanic African-American or Mexican-American race/ethnicity affected the association of living arrangements with dietary quality, and whether there was a different pattern of association between living arrangements and dietary quality among different race/ethnic groups. Like non-Hispanic Caucasian adults, non-Hispanic African-American adults also tended to have better diets if they lived only with a spouse, as did Mexican-American adults in the younger age group. However, surprisingly, the association was much less consistent among older Mexican-Americans and was even reversed for some age and gender categories. Although the findings did not reach significance, Mexican-American women aged 65 y consistently had poorer diets if they lived with a spouse only. Persons of "other" races who were 65 y also reported poorer diets if they lived only with a spouse, although due to the lack of homogeneity in this ethnic category (it includes Native Americans, Asians and other Hispanics), as well as the small sample size, it would be premature to make recommendations based on these analyses. However, further investigation of the effect of living arrangements on dietary quality for these ethnic groups should be undertaken.

Unfortunately, the NHANES III data do not provide information on the relationship of all persons in the household; consequently, we were unable to identify whether there were household composition constellations that were specifically associated with poor dietary quality. Living arrangements are influenced by life circumstances such as marital status, childbearing patterns, education and income, as well as personal and cultural influences and preferences, and current financial resources, health behaviors and health status that affect whether a person lives alone or with whom they live (Davis et al. 1996 ). Clearly with the NHANES III data, we were able to analyze the effect of only some of these factors in accounting for the associations of living arrangement and dietary quality that we observed, and it is important for future research to elaborate these issues. We did not find that marital status, ethnicity, education, income, employment, self-reported health status, exercise, BMI, alcohol consumption, smoking, energy consumption, taking vitamins or mineral supplements or skipping meals accounted for the observation that both middle-aged and older adults who lived with a spouse only had better dietary quality than those in other living arrangements. However, it is possible that our measures were not sensitive or specific enough to measure these variables adequately. We did not observe that middle-aged or older adults who lived alone were consistently at higher risk of poor dietary quality than those who lived with someone. Instead, we found that living alone was a risk factor for poor dietary quality in some age/gender/ethnicity subgroups, but there was no consistent pattern.

National dietary data collected in the 1970s (NHANES I, II, and NFCS 1977–1978) observed that older men living alone, but not older women living alone, were at higher dietary risk than those in other living arrangements (Davis et al. 1985 and 1990 , Ryan et al. 1989 ). Lower energy intake of those living alone was the main contributor to lower dietary quality (Davis et al. 1990 ). However, more recent national data have not shown a higher dietary risk for those living alone, thus supporting the current findings from NHANES III. Data from the NFCS 1987–1988 suggest that for adults 55 y, those living alone compared with those living with someone consume less energy and fewer nutrients, although they did not necessarily have lower nutrient densities (USDA 1994 ). Data from the Continuing Survey of Food Intakes by Individuals (CSFII) 1989–1991 did not show a difference in dietary quality between those living alone and those living with someone for persons aged 60 y (Weimer 1998 ). Because analyses of these national data sets have used a variety of living arrangement categories, it is difficult to make comparisons. However, there does not seem to be a systematic pattern of older adults living alone being at the highest risk of poor dietary quality.

NHANES III did not restrict the upper age limit of the sample to 74 y as in NHANES I and II; thus, we were able to examine the effects of a wider range of ages on the association of living arrangements and dietary quality. We performed separate analyses for middle-aged (50–64 y) and older (65 y) adults, and found generally consistent results for non-Hispanic adults, although differences were more often significant for the middle-aged adults. Because the NHANES III data are cross-sectional, it was not possible to examine the aging vs. cohort hypotheses. It is important for future longitudinal studies to address this issue to understand how or whether dietary risk patterns that are present in middle age continue as the cohort ages.

Unfortunately, due to small subgroup sample sizes, we were unable to further stratify our age groupings to distinguish the older age groupings, including the oldest old and still include the analysis for ethnicity. Analyses of NFCS 1977–1978 did compare the association of living arrangements with dietary quality for groups of U.S. adults from 55 to 85+ y of age, and did not find an increase in the risk pattern of living arrangements by age (Davis et al. 1990 , Murphy et al. 1990 ).

We did not observe any consistent gender patterns in the association of living arrangements with dietary quality across age/ethnicity subgroups, although more of the associations reached significance for the men. This finding supports previous reports for older adults, using national dietary data collected in the 1970s (NHANES I and II, NFCS, 1977–1978), in which living arrangements were observed to have a larger effect on dietary quality for men than women (Davis et al. 1985 and 1990 , Ryan et al. 1989 ). Researchers who analyzed the NFCS 1987–1988 (USDA 1994 ), however, did not observe a gender difference in the association of living arrangements and dietary patterns for this age group. Different dietary methodologies and different measures among surveys and across time periods, as well as differences in living arrangement analyses, could confound the actual gender patterns in regard to living arrangements and dietary quality.

Income has been shown previously to be associated with dietary quality in national surveys (Davis et al. 1985 and 1990 , Murphy et al. 1990 , Ryan et al. 1989 ,Weimer 1998 ); this association was also seen in NHANES III. However, we also observed that the association of living arrangement with dietary quality was independent of income.

In this study, we found that most older Americans reported 1-d diets that were low (<67% of the RDA) in multiple nutrients: over half of the women and about one third of the men consumed diets that were simultaneously low in four or more nutrients. Because two thirds of the RDA is a conservative cut-off point for defining a low nutrient (well below the average requirement for most nutrients), it seems likely that many of these individuals are not consuming an adequate diet. Data from other national surveys, such as the recent 1994–1996 CSFII, also have identified several nutrients likely to be low in the diets of older adults (USDA 1997 ). We found that calcium, magnesium, zinc, folate and vitamin E were the nutrients most likely to be low, whereas the CSFII data identified calcium, magnesium, zinc and vitamin E as nutrients with mean intakes below the RDA (USDA 1997 ). Because we used the new Dietary Reference Intakes for calcium, magnesium and folate, our estimates of low nutrients reflect more current recommendations than do those published from the CSFII. However, the NHANES III nutrient intake data do not consider the increased bioavailability of fortification folate, as recommended by the Food and Nutrition Board (IOM 1998 ); thus, we may have overestimated the number of persons with low folate intakes.

We also observed that the association of living arrangements and dietary quality was not explained by dietary quantity. When we controlled for energy intake (Table 3) , the patterns of association between living arrangements and nutrient intake remained. This is in contrast to previous analyses using the 1977–1978 NFCS (Murphy et al. 1990 ) in which the primary factor that predicted nutrient intake of older Americans was the quantity of the diet (as measured by energy intake). This change implies that many older Americans in the mid-1990s chose foods of high nutrient density; thus, the link between quantity and quality is no longer as strong as in previous decades.

A limitation of the NHANES III survey is the availability of only a single day of dietary data for most participants. Because 1 d of data generally does not reflect usual intake (LSRO 1986 ), it is likely that the number of low nutrients for some participants was incorrect. Random misclassification will tend to attenuate the association of nutrient quality with living arrangements. In addition, because the distribution of 1-d intakes is broader than the distribution of usual intakes, the number of low nutrients was overestimated for some survey participants. However, the ranking of individuals by the number of low nutrients is less likely to have been affected by the shape of the intake distribution, and thus there would be little effect on the associations reported in Tables 3 and 4 . In addition, it is known that at least energy intake, and probably intake of vitamins and minerals as well, is underreported on 24-h recalls (Briefel et al. 1997 , Johnson et al. 1998 ), providing another source of overestimation of the number of low nutrients. Because underreporting varies with participant characteristics (e.g., age, BMI and literacy), overestimation of the number of low nutrients may be greater for some individuals than for others. By adjusting for age, BMI and education, we have adjusted in part for this differential underreporting. Although we cannot entirely eliminate the possibility of a bias in dietary reporting that is associated with living arrangement, each person independently reported the previous day’s diet, and thus we would not expect underreporting to vary by living arrangement.

Because this report focused on those who had self-reports, the generalizability of our findings is limited to adults who are able to respond for themselves. It is therefore possible that the dietary quality reported here may reflect the dietary patterns of the healthier persons in the sample.

It is important to keep in mind that living arrangements and health are part of an ongoing process of change that is particularly germane to older adults. It is likely that the living arrangement, dietary quality, health and well-being of older individuals reflect past life circumstances and expectations regarding the future, neither of which can be measured in a simple survey such as NHANES.

In conclusion, because many older Americans do not report consuming nutritionally adequate diets, it is important to identify factors such as living arrangements that allow food assistance and education programs to target persons who would benefit most from these programs. We found that middle-aged and older non-Hispanic adults who lived with a spouse only tended to have better dietary quality than persons with other living arrangements. However, because the association of dietary quality and living arrangements varied by age/gender/ethnicity subgroups, it is important for future research to examine whether certain household composition characteristics differ in their effect on dietary quality for various race/ethnicity subgroups.

In addition to focusing on living arrangements, programs may use other sociodemographic and behavioral characteristics to identify groups of middle-aged and older adults that are at higher risk for poor dietary quality, i.e., women, persons who are >65 y old and non-Hispanic African-American men. In addition, persons having less education, and those who smoke, drink alcoholic beverages, do not exercise and have low energy intake are at greater risk. All of these characteristics should be considered when designing programs to improve dietary quality among older Americans.

	ACKNOWLEDGMENTS

 
The authors thank Christine Choy for assistance with typing and editing the manuscript.

	FOOTNOTES

 
¹ Supported by grant 5R37AG05284 from the National Institute on Aging.

³ Current address: Cancer Research Center of Hawaii, University of Hawaii, Honolulu, HI 96813.

⁴ Abbreviations used: BMI, body mass index; CI, confidence interval; CSFII, Continuing Survey of Food Intakes by Individuals; IOM, Institute of Medicine; NCHS, National Center for Health Statistics; NFCS, Nationwide Food Consumption Survey; NHANES, National Health and Nutrition Examination Survey; RDA, Recommended Dietary Allowance.

Manuscript received January 27, 2000. Initial review completed April 27, 2000. Revision accepted May 15, 2000.

	REFERENCES

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Binder D.A. On the variances of asymptotically normal estimators from complex surveys. Int. Stat. Rev. 1983;51:279-292

2. Briefel R. R., Sempos C. T., McDowell M. A., Chien S., Alaimo K. Dietary methods research in the third National Health and Nutrition Examination Survey: underreporting of energy intake. Am. J. Clin. Nutr. 1997;65:1203S-1209S[Abstract/Free Full Text]

3. Burt V. L., Harris T. The Third National Health and Nutrition Examination Survey: contributing data on aging and health. The Gerontologist 1994;34:486-490[Abstract]

4. Cantor M., Little V. Aging and social care. Binstock R. H. Shanas E. eds. Handbook of Aging and the Social Sciences 1985 Van Nostrand Reinhold Company New York, NY.

5. (The) Commonwealth Fund, Commission on Elderly People Living Alone A survey of elderly Hispanics 1989 The Commonwealth Fund New York, NY.

6. (The) Commonwealth Fund, Commission on Elderly People Living Alone Improving the well-being of elderly people living alone: the unfinished agenda 1992a Commonwealth Fund New York, NY.

7. (The) Commonwealth Fund Commission on Elderly People Living Alone Problems facing elderly Americans living alone 1992b The Commonwealth Fund New York, NY.

8. Davis M. A., Moritz D. J., Neuhaus J. M., Barclay J. D. The influence of sociodemographic characteristics and morbidity on the likelihood of living alone among older U.S. adults who become unmarried: a comparison of women and men. J. Women Aging 1996;8:3-17

9. Davis M. A., Murphy S. P., Neuhaus J. M., Lein D. Living arrangements and dietary quality of older U.S. adults. J. Am. Diet. Assoc. 1990;90:1667-1672[Medline]

10. Davis M. A., Neuhaus J. M., Moritz D. J., Segal M. R. Living arrangements and survival among middle-aged and older adults in the NHANES I Epidemiologic Follow-up Study. Am. J. Public Health 1992;82:401-406[Abstract/Free Full Text]

11. Davis M. A., Randall E., Forthofer R. N., Lee E. S., Margen S. Living arrangements and dietary patterns of older adults in the United States. J. Gerontol. 1985;40:434-442[Abstract/Free Full Text]

12. Draper N. R., Smith H. Applied Regression Analysis 2nd ed. 1981:239 Wiley New York, NY.

13. Garofalo J. A., Hynak-Hankinson M. T. New Jersey’s nutrition screening initiative: activities and results. J. Am. Diet. Assoc. 1995;95:1422-1424[Medline]

14. Institute of Medicine Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D, and Fluoride 1997 National Academy Press Washington. DC.

15. Institute of Medicine Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B₆, Folate, Vitamin B₁₂, Pantothenic Acid, Biotin, and Choline 1998 National Academy Press Washington, DC.

16. Johnson R. K., Soultanakis R. P., Matthews D. E. Literacy and body fatness are associated with underreporting of energy intake in U.S. low-income women using the multiple-pass 24-hour recall: a doubly labeled water study. J. Am. Diet. Assoc. 1998;98:1136-1140[Medline]

17. Liang K. Y., Zeger S. L. Longitudinal data analysis using generalised linear models. Biometrika 1986;73:13-22[Abstract/Free Full Text]

18. Life Science Research Office Guidelines for Use of Dietary Data 1986 FASEB Bethesda, MD.

19. McIntosh W. A., Shifflett P. A., Picou J. S. Social support, stressful events, strain, dietary intake, and the elderly. Med. Care 1989;27:140-153[Medline]

20. Murphy S. P., Davis M. A., Neuhaus J. M., Lein D. Factors influencing the dietary adequacy and energy intake of older Americans. J. Nutr. Educ. 1990;22:284-291

21. Murphy S. P., Davis M. A., Neuhaus J. M., Lein D. Dietary quality and survival among middle-aged and older adults in the NHANES I Epidemiologic Follow-up Study. Nutr. Res. 1996;16:1641-1650

22. National Center for Health Statistics Plan and operation of the Third National Health and Nutrition Examination Survey, 1988–94. Vital Health Stat. 1 1994 U.S. Government Printing Office Washington, DC.

23. National Research Council Recommended Dietary Allowances 10th ed. 1989 National Academy Press Washington, DC.

24. Pate R. R., Pratt M., Blair S. N., Haskell W. L., Macera C. A., Bouchard C, Buchner D, Ettinger W, Heath G. W., King A. C., Kriska A., Leon A. S., Marcus B. H., Morris J., Paffenbarger R. S. Jr, Patrick K., Pollock M. L., Rippe J. M., Sallis J., Wilmore J. H. Physical activity and public health: a recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. J. Am. Med. Assoc. 1995;273:402-407[Abstract/Free Full Text]

25. Posner B. M., Jette A., Smigelski C., Miller D., Mitchell P. Nutritional risk in New England elders. J. Gerontol. 1994;49:M123-M132[Abstract]

26. Ryan A. S., Craig L. D., Finn S. C. Nutrient intakes and dietary patterns of older Americans: a national study. J. Gerontol. 1992;47:M145-M150[Abstract]

27. Ryan A. S., Martinez G. A., Wysong J. L., Davis M. A. Dietary patterns of older adults in the United States, NHANES II 1976–1980. Am. J. Hum. Biol. 1989;1:321-330

28. Saluter A. F., Lugaila T. A. Marital status and living arrangements: March 1996. Current Population Reports, Population Characteristics, Series P- 20, no. 496 1998 U.S. Department of Commerce Washington, DC.

29. SAS Institute Inc SAS/Stat Software: changes and enhancements through release 6.12 1997 SAS Institute Cary, NC.

30. U.S. Department of Agriculture How Does Living Alone Affect Dietary Quality? 1994 U.S. Government Printing Office Washington, DC.

31. U.S. Department of Agriculture Data Tables: Results from USDA’s 1994–96 Continuing Survey of Food Intakes by Individuals and 1994–96 Diet and Health Knowledge Survey 1997 USDA Riverdale, MD.

32. U.S. Department of Commerce Marital status and living arrangements: March 1985. Current Population Reports, Population Characteristics, Series P- 20, no. 410 1986 U.S. Department of Commerce Washington, DC.

33. U.S. Senate Special Committee on Aging, American Association of Retired Persons, Federal Council on the Aging, U.S. Administration on Aging Aging America: Trends and Projections, DHHS publication FcoA 91–28001 1991 U.S. Government Printing Office Washington, DC.

34. Walter K. Introduction to Variance Estimation, ch. 6 1985 Springer-Verlag New York, NY.

35. Weimer, J. (1998) Factors Affecting Nutrient Intake of the Elderly. Agricultural Economic Report no. 769: U.S. Department of Agriculture, Washington, DC.

This article has been cited by other articles:

				P. P. Vitaliano, J. Zhang, H. M. Young, L. W. Caswell, J. M. Scanlan, and D. Echeverria Depressed Mood Mediates Decline in Cognitive Processing Speed in Caregivers Gerontologist, April 1, 2009; (2009) gnp004v4. [Abstract] [Full Text] [PDF]

				J. L. Locher, C. S. Ritchie, C. O. Robinson, D. L. Roth, D. Smith West, and K. L. Burgio A Multidimensional Approach to Understanding Under-Eating in Homebound Older Adults: The Importance of Social Factors Gerontologist, April 1, 2008; 48(2): 223 - 234. [Abstract] [Full Text] [PDF]

				K. Joutsenniemi, T. Martelin, L. Kestila, P. Martikainen, S. Pirkola, and S. Koskinen Living arrangements, heavy drinking and alcohol dependence Alcohol Alcohol., September 1, 2007; 42(5): 480 - 491. [Abstract] [Full Text] [PDF]

				J. L. Locher, C. O. Robinson, D. L. Roth, C. S. Ritchie, and K. L. Burgio The Effect of the Presence of Others on Caloric Intake in Homebound Older Adults J. Gerontol. A Biol. Sci. Med. Sci., November 1, 2005; 60(11): 1475 - 1478. [Abstract] [Full Text] [PDF]

				L. K. Wen, M. D. Shepherd, and M. L. Parchman Family Support, Diet, and Exercise Among Older Mexican Americans With Type 2 Diabetes The Diabetes Educator, November 1, 2004; 30(6): 980 - 993. [PDF]

				J. H Ledikwe, H. Smiciklas-Wright, D. C Mitchell, G. L Jensen, J. M Friedmann, and C. D Still Nutritional risk assessment and obesity in rural older adults: a sex difference Am. J. Clinical Nutrition, March 1, 2003; 77(3): 551 - 558. [Abstract] [Full Text] [PDF]

				J. R Sharkey, L. G Branch, N. Zohoori, C. Giuliani, J. Busby-Whitehead, and P. S Haines Inadequate nutrient intakes among homebound elderly and their correlation with individual characteristics and health-related factors Am. J. Clinical Nutrition, December 1, 2002; 76(6): 1435 - 1445. [Abstract] [Full Text] [PDF]

				J. Hippisley-Cox, C. Coupland, M. Pringle, N. Crown, and V. Hammersley Married couples' risk of same disease: cross sectional study BMJ, September 21, 2002; 325(7365): 636 - 636. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Davis, M. A. Articles by Quiroga, S. S. Search for Related Content PubMed PubMed Citation Articles by Davis, M. A. Articles by Quiroga, S. S.