- Split View
-
Views
-
Cite
Cite
Joseph R. Sharkey, The Interrelationship of Nutritional Risk Factors, Indicators of Nutritional Risk, and Severity of Disability Among Home-Delivered Meal Participants, The Gerontologist, Volume 42, Issue 3, 1 June 2002, Pages 373–380, https://doi.org/10.1093/geront/42.3.373
- Share Icon Share
Abstract
Purpose: This study examines the direct and indirect relationships between individual components of nutritional risk and increased severity of disability among a large and diverse sample of homebound older adults. Design and Methods: Using routinely collected nutrition and function data, structural equation modeling of recursive and nonrecursive models examined the interrelationships of nutritional risk factors, indicators of nutritional risk, and disability severity among 1,010 home-delivered meals program participants in Wake County, NC. Results:The equally good fit for both the recursive and nonrecursive structural models revealed that specific nutritional risk factors were directly and indirectly associated with indicators of nutritional risk and increased severity of disability. The nonrecursive model also revealed significant reciprocal associations of increased disability with unintended weight change and medication use. Implications: The findings from this study acknowledge aspects of the complex direct and indirect relationships between nutrition and function among homebound older persons. This knowledge will help service providers with the development of effective elderly nutrition programs with nutritional and functional status outcomes.
Decision Editor: Laurence G. Branch, PhD
Although subgroups of community-living older adults (namely women, the poor, Blacks, those with limited education, and the homebound) have been shown to be at increased risk for poor nutrition and functional disability (Coulston, Craig, and Voss 1996; Fried and Walston 1999; Institute of Medicine 2000; Ponza, Ohls, and Millen 1996; Quandt and Chao 2000; Sharkey and Haines 2001; Sharkey, Haines, and Zohoori 2000), there is a limited understanding of the interrelationship between components of nutritional risk and severity of disability. The largest single nutrition program providing long-term home- and community-based care to older Americans is the Elderly Nutrition Program (ENP; Schlenker 1998). Even though this program serves vulnerable older adults in both group (congregate site) and home settings (home-delivered), dramatic changes in demographics are contributing to an ever-increasing demand for the home-delivered meal component of the ENP (Moyer and Balsam 1996; Ponza et al. 1996; Schlenker 1998; Wellman 1999).
Accompanying this shift in demand are the emergence of a new ENP paradigm, with a focus on outcomes (both short- and long-term) that help older persons maintain adequate nutritional status and remain independent and at home (Wellman 1999), and two recent reports that suggest an association between nutritional risk and disability among home-delivered meal participants. Both studies analyzed information that is routinely collected from ENP participants by the nutrition service providers: the Nutrition Screening Initiative's (NSI's) DETERMINE Checklist (NSI 1996), Activities of Daily Living or ADL (Katz, Ford, Moskowitz, Jackson, and Jaffe 1963), and Instrumental Activities of Daily Living or IADL (Lawton and Brody 1969). One study, using categorical indicators for nutritional risk level and disability, which were derived from summary measures, reported that being at high nutritional risk increased the odds for severe disability (Sharkey and Haines 2001). A second study, using a categorical indicator for disability, found a direct association between specific nutritional risk factors and impairment in any self-care ADL (Sharkey and Haines 2000). Little is known about the interrelationship between individual nutritional risk factors and increased severity of disability; and it is this understanding that can help move ENP services toward the development of effective ENP programs with nutritional and functional status outcomes.
The conceptual model (see Fig. 1) for the pathway from nutritional risk factors to disability was adapted from the work of Anderson and consists of three interrelated components: nutritional risk factors, indicators of nutritional risk, and the consequences of nutritional risk (Anderson 1990). Nutritional risk factors, which are extrinsic factors that influence nutritional status and place an older person at increased risk for poor nutrition, include multiple medications, social isolation /eating alone most of the time, oral/dental problems, difficulty in the preparation of meals, difficulty in shopping for groceries, economic status/lacking enough money for food, and having an illness or condition that caused a diet change (Goodwin 1989). Nutritional risk factors are directly linked to indicators of nutritional risk (i.e., for poor dietary intake), such as eating few meals; consuming limited portions of fruits, vegetables, or milk products; and experiencing an unintentional weight change. One of the more severe consequences is functional disability, that is, requiring assistance in the performance of basic self-care tasks of daily living. Through the use of structural equation modeling of path analysis, the present study extends the nutrition and disability literature by examining the direct and indirect relationships between individual components of nutritional risk and increased severity of disability among a large and diverse sample of homebound older men and women. By taking the total relationship (i.e., direct and indirect effects) between individual nutritional risk factors, indicators of nutritional risk, and disability severity into account, this may further extend the application of ENP routinely collected data and provide insight into potential areas of research and intervention common to both poor nutrition and functional decline.
Methods
Sample and Data
The cross-sectional data came from a chart review conducted in February and March 2000 of all 1,026 participants, ages 60 years and older, in a home-delivered meals program in Wake County, NC. Using data from the Wake County home-delivered meals program offered us the advantage of one of the largest and most diverse home-delivered meals programs in North Carolina to operationalize the conceptual model of nutrition and function interrelationships. Complete information on demographics, nutritional risk screening, and functional status, which were routinely collected through self-report by interviewer-administered standardized questionnaires, was extracted for 1,010 current program participants. Sixteen participants were excluded from collection because they were missing all demographic, nutritional risk, and/or functional status information. With a mean age of 78.6 (standard deviation of 8.4, range of 60–103 years), almost 47% of the study sample was minority (99% African American); 73% was female; 67% reported economic need (an income ≤125% of the poverty level); and 55% lived alone.
Measures
The three general categories from the conceptual model provided the measures for analysis: nutritional risk factors, indicators of nutritional risk, and disability severity. Nutritional risk factors were drawn from the NSI Checklist (NSI 1996) and the nutrition-related tasks of the IADL (Lawton and Brody 1969). The NSI Checklist is routinely used by providers of ENP services as a screening tool for the identification of risk factors for malnutrition (Chernoff 2001; Joseph et al. 1997; Ponza et al. 1996) and the IADL as a measure of higher level function. The following nutritional risk factors were selected from the NSI Checklist: presence of an illness or condition that caused a change in the kind or amount of food eaten (Illness); tooth or mouth problems that make it hard to eat (Oral); not always have enough money to buy needed food (Money); eat alone most of the time (Alone); and take three or more different prescribed or over-the-counter drugs a day (Medications). Since the NSI Checklist includes a single risk factor that combines two separate activities (i.e., not always physically able to shop for food and/or cook), the decision was made a priori to substitute the two separate nutrition-related IADL tasks for this single (and combined) risk factor. The IADL provided these two nutritional risk factors: meal preparation (Prepare) and grocery shopping (Shop). Three indicators of nutritional risk were selected from the NSI Checklist: consume few servings of fruit or vegetables or milk products (Servings); eat fewer than two meals per day (Meals); and without wanting to, lost or gained 10 pounds in the last 6 months (Weight). Each of the risk factors and indicators of nutritional risk consisted of a dichotomous variable (1 = affirmative response, 0 = negative response).
Disability severity reflected the inability to perform tasks of self-care, known as ADL (Katz et al. 1963), without assistance. The individual tasks were eating, getting dressed, bathing, using the toilet, and getting in and out of bed. A count variable was constructed on the total number of ADL disabilities. This assumed that a greater number of disabilities would indicate a level of more severe disability. Sample prevalence for the nutritional risk factors, indicators of nutritional risk, and disability severity is shown in Table 1 .
Data Analysis
Because all the variables considered as nutritional risk factors and indicators of nutritional risk in this analysis were ordinal and assumed to be approximations of underlying latent continuous indicators (Bollen 1989; West, Finch, and Curran 1995), a method of analysis that calculated and used the appropriate matrix of correlations was selected. The correlation between two ordinal variables is called a polychoric correlation. A correlation between a continuous variable, such as age and disability severity, and one that is ordinal is a polyserial correlation. Finally, a correlation between two continuous variables is a Pearson product correlation. PRELIS 2.30 (Joreskog and Sorbom 1996b) was used to estimate a matrix of correlations (polychoric, polyserial, and Pearson product) and an asymptotic covariance matrix, which served as the appropriate weight matrix in LISREL 8.30 path analysis using Weighted Least Squares (WLS; Joreskog and Sorbom 1996a).
An initial path analysis of a recursive model was based on the unidirectional conceptual pathway in Fig. 1, with the inclusion of age in each of the structural equations. As suggested by Joreskog and Sorbom 1996a, model respecification was minimal and included the elimination of insignificant direct paths and the addition of paths based on careful inspections of the nonfitting model (Kühnel, 2001), combined with theoretical or substantive knowledge (Bollen 1989). Although there are sample size concerns for WLS (Bollen 1989), the present study sample of 1,010 seems large enough for convergence. Indicators of model fit (χ2, Root Mean Square Error of Approximation [RMSEA], p Value for Test of Close Fit [RMSEA < .05], Incremental Fit Index [IFI], Normed Fit Index [NFI], Non-Normed Fit Index [NNFI], Goodness-of-Fit Index [GFI], and Adjusted Goodness-of-Fit Index [AGFI]) were used to evaluate the fit of the model.
Results
Table 1 reports the sample prevalence for each of the study variables. It is noteworthy that almost 66% of the sample reported no ADL disability, 27% one or two disabilities, and 7% at least three ADL disabilities. Table 2 shows the correlation coefficients (polychoric correlations, polyserial correlations, and Pearson product correlations) for bivariate correlations between all analytical variables, which were calculated with PRELIS 2.30. The magnitude of bivariate correlation was small between most sets of variables. For the dietary intake variables, the correlation coefficient between food servings and the number of meals was of moderate magnitude. The largest correlation, moderately large in magnitude, was indicated for the two nutrition-related IADL variables (preparing meals and shopping for food).
The full recursive model simultaneously estimated a system of four structural equations: (1) Disability severity was regressed on the three indicators of nutritional risk (weight change, eating few meals, and consuming limited servings), three nutritional risk factors (medications, eating alone, and difficulty shopping for food), and age; (2) Weight was regressed on two indicators (meals and servings), four nutritional risk factors (illness that caused a diet change, medications, alone, and difficulty preparing meals), and age; (3) Servings was regressed on meals, three nutritional risk factors (lack of money for food, shop, and prepare), and age; and (4) Meals was regressed on five nutritional risk factors (money, oral problems, alone, prepare, and shop) and age. Although the chi-square statistic was satisfactory considering the large sample size (χ2[13, N = 1,010] = 26.19, p = .016), other goodness-of-fit statistics indicated a close fit of the model in relation to the degrees of freedom (Browne and Cudeck 1993): RMSEA = .032 (90% confidence interval of .013, .049); p Value for Test of Close Fit = .96; IFI = 1.00; NNFI = .98; NFI = .99; GFI = 1.00; and AGFI = .99. Fig. 2 shows the significant pathway estimates for the full recursive model.
Fig. 2 shows the direct and indirect associations of nutritional risk variables with increased severity of disability. Indirect associations occur when one variable is associated with a second, which in turn is associated with a third. As such, the first variable is associated with the third via the second, regardless of whether it has a direct association with the third variable (Maruyama 1998). The indirect effects of a variable may be mediated by more than one intervening variable, with the total effects of a variable to be the sum of direct and indirect effects (Bollen 1989).
Eating few meals and consuming few servings both served as intervening variables. Lacking money for food was not directly associated with disability severity; however, the model suggested indirect influences of money on consuming few servings (β = .16; t = 2.58) and disability severity (β = .04; t = 2.15). These associations were mediated by meals, which were both directly and indirectly (β = −.05; t = −2.04) associated with increased severity of disability. Meals also mediated the influence of eating alone (β = .12; t = 2.17) and difficulty preparing meals (β = .31; t = 2.51) on consuming few servings, and preparing meals on disability severity (β = .05; t = 1.98). Eating few servings mediated the effect of consuming fewer than two meals on weight change (β = .14; t = 2.31).
The complexities of the interrelationships were evidenced by several variables that demonstrated opposite signs for direct and indirect effects. The direct relationship between shopping for groceries and servings was positive, and the indirect association mediated by meals was negative; the direct association between money and servings was negative, and the indirect association mediated by meals was positive; the direct association between meals and disability severity was positive, and the indirect association mediated by servings was negative (β = −.05; t = −2.04); and the direct association between alone and disability severity was negative, and the indirect association mediated by meals was positive. When the indirect pathways from eating alone to disability severity (through consuming few meals) were constrained to zero, the direct association between eating alone and disability severity remained negative (β = −.36; t = −8.85). This is plausible when considering that older persons who are more dependent in basic activities of self-care (e.g., ADL) would probably have an aide or caregiver present, especially at mealtime, and thus not eat alone. Further examination of the total indirect effects of the intervening variables for eating few meals and consuming few servings revealed that having difficulty preparing meals and lacking enough money for food were significantly associated with increased severity of disability (β = .05, t = 1.98 and β = .04, t = 2.15, respectively).
After the recursive model was fitted, two reciprocal pathways and a fifth structural equation were added. The reciprocal pathways between disability severity and medication use and weight change were justified on the basis of published research that has suggested a link between impairments in physical function and increased medication use and an unintended weight change (Chrischilles et al. 1992; Fried and Walston 1999). In the fifth structural pathway, medication use was regressed on disability severity (reciprocal pathway), illness or condition that caused a diet change, and eating alone. Because reciprocal pathways were modeled for the relationship of disability severity with medications and weight change, the disturbances of medications and disability severity and weight change and disability severity were correlated (Bollen 1989). This model provided a close fit (χ2[14, N = 1,010] = 26.11, p = .025), RMSEA = .029 (90% confidence interval of .010, .047); p Value for Test of Close Fit = .98; IFI = 1.00; NNFI = .98; NFI = .99; GFI = 1.00; and AGFI = .99. Fig. 3 shows the significant pathway estimates for the nonrecursive model and illustrates the significance of both reciprocal pathways.
An increase in the severity of disability was directly associated with an increase in weight change and an increase in medication use. Additionally, an illness or condition that caused a diet change and eating alone were both directly associated with medication use. With the inclusion of the reciprocal pathways, medication use and eating few meals were no longer directly associated with disability severity; and difficulty with meal preparation was now associated with eating few meals at the 6% level (β = .32, t = 1.87). The signs of several coefficients were opposite of the recursive model (e.g., age) and increased in magnitude (e.g., servings with disability severity, shopping for groceries with disability severity, and medications with weight change). Consuming few servings was no longer directly associated with weight change; and difficulty in shopping for food, eating alone, and lacking money for food were no longer associated with eating few meals. Absent a direct association, the results indicated significant indirect associations, with disability severity as the intervening variable, for several sets of variables: eating alone, age, and shopping for food with weight change (β = −.16, t = −2.28; β = −.13, t = −4.50; and β = .46, t = 3.13, respectively); age and shopping for food with medications (β = −.23, t = −7.78; β = .36, t = 4.37, respectively); and consuming few servings with medications (β = −.09, t = −2.61). Only considered were indirect effects where intervening pathways were significant. If nonsignificant pathways were considered, significant indirect associations were observed for illness and medications with disability severity.
Discussion
The use of path analysis allowed the simultaneous examination of the interrelationship of individual nutritional risk factors, indicators of nutritional risk, and disability severity among a large sample of home-delivered meal participants. This provides further evidence of the link between two disparate areas of information that are routinely collected by community-based nutrition service providers; namely, nutritional risk and self-care disability. These findings extend previous work (Jensen, Kita, Fish, and Heydt 1997; Sharkey et al. 2000) and appear to be the first report to examine the direct and mediating effects of the putative nutritional risk factors and indicators of nutritional risk on the severity of disability among home-delivered meal participants. Furthermore, this is the first report that we are aware of that examines reciprocal relationships in the association between nutritional risk and disability among community-dwelling older persons.
In particular, the results suggest that increased severity of disability is a function of medication use and difficulty shopping for food through direct associations, and lacking enough money for food and having difficulty preparing meals through indirect associations. It is believed that these same four characteristics (i.e. medication use, difficulty shopping for food, lacking enough money for food, and difficulty preparing meals) place older persons at increased risk for food insecurity, which has been referred to as an uncertainty or inability to acquire or consume an adequate quality or sufficient quantity of food in socially acceptable ways (Blumberg, Bialostosky, Hamilton, and Briefel 1999; Lee and Frongillo 2001). That is, the four nutritional risk characteristics associated with increased risk for food insecurity are also associated with increased severity of disability.
Considering that other studies have suggested the importance of an unintentional weight change as an indicator of the presence of poor nutritional status (Brunt, Schafer, and Oakland 1999a; White 1991) and a predictor of a change in physical function (Tully and Snowdon 1995), we found that medication use and the consumption of few servings of fruits or vegetables or milk products were both positively associated with an unintentional weight change. Posner, Jette, Smith, and Miller 1993 and Brunt and colleagues 1999a, Brunt and colleagues 1999b have suggested that eating fewer than two meals per day and consuming few servings of fruits or vegetables or milk products were predictive of a diet deficient in key nutrients. Our findings indicate that three of the risk factors were directly linked to eating fewer meals: lacking enough money for food, eating alone, and having difficulty with meal preparation. Eating fewer than two meals per day was a pivotal correlate of consuming few servings of fruits or vegetables or milk products, significant as both a direct effect and intervening variable for eating alone. Of particular importance are our findings of significant indirect associations of lacking money for food and having difficulty preparing meals with increased severity disability; and eating alone with consuming few servings of micronutrient rich fruits or vegetables or milk products. We believe that this last group of findings, along with the previously mentioned significant indirect associations with severity of disability, would have gone unobserved in analyses of direct associations (e.g., ubiquitous ordinary least squares or logistic regression models). It is important to note that there is not agreement in the literature on the relationship between eating alone/social isolation and poor dietary intake, nor on how to define poor dietary intake (Brunt et al. 1999a, Brunt et al. 1999b; Posner et al. 1993; Weimer 1998). However, the studies are consistent in not examining interrelationships among variables, in particular the indirect effects of one variable on the outcome through an intervening variable. It is through such an examination that the findings of the present study become more salient. For example, we found eating alone to have a direct and indirect association with two variables that have been shown to be predictive of poor nutrient intake (Brunt et al. 1999b; Posner et al. 1993): eating fewer than two meals per day and consuming few servings of fruits or vegetables or milk products. Our findings provide additional evidence, after controlling for the indirect effects of indicators of nutritional risk, that dietary intake-related nutritional risk factors (e.g., lacking enough money for food, difficulty shopping for food, and difficulty with meal preparation) are significantly associated with increased ADL disability. All three are considered barriers to adequate household and individual food supplies (Quandt and Rao 1999). It is the provision of adequate food, which supplies the needed nutrients and energy, which is essential to health and well-being (Institute of Medicine 2000). Dietary intake is modifiable, whether it is the number and type of meals or the nutrient density of each meal, and provides an appropriate target for future interventions that seek to prevent or reverse an inadequate diet and functional decline.
It is important to note that many of the significant relationships observed in the recursive model disappeared with the inclusion of reciprocal pathways in the nonrecursive model (Fig. 3). Interestingly, both reciprocal pathways (i.e., from Disability Severity to Medications and from Disability Severity to Weight Change) were significant. This suggested that, after controlling for covariates such as the presence of an illness or conditions that caused a diet change and alternate pathways, an increase in the number of ADL disabilities was directly linked to an increase in medication use and unintentional weight change. Having difficulty shopping for food maintained a direct association with both increased disability and decreased consumption of fruits or vegetables or milk products, with larger coefficients in the nonrecursive model. Through intervening variables, difficulty shopping for groceries was associated with an unintentional weight change and increased medication use. Because having difficulty shopping for groceries may reflect multiple unrelated factors (e.g., problems with vision, mobility, endurance, breathing, and transportation and community characteristics), the areas of intervention become more complex and challenging. According to Wolfe, Olson, Kendall, and Frongillo 1996, community characteristics include accessibility and availability of grocery stores, neighbors, transportation services, food programs, and other services for older persons. Understandably, several of the factors related to difficulty shopping for food may also underpin the difficulty older people experience with meal preparation (e.g., vision, mobility/standing, and endurance).
Considering that others have not found a significant relationship between two of the nutritional risk factors (e.g., an illness or condition that caused a diet change and oral/tooth or mouth problems) and dietary inadequacy (Brunt et al. 1999a, Brunt et al. 1999b; Payette, Gray-Donald, Cyr, and Boutier 1995; Posner et al. 1993), we are not surprised with their lack of association with any of the three indicators of nutritional risk (e.g., eating fewer than two meals, consuming few servings of fruits or vegetables or milk products, and unintended weight change) in the present study. It is important to note that prior reports found eating fewer than two meals per day, consuming few servings of fruits or vegetables or milk products, and unintended weight change to be directly associated with an inadequate nutritional intake (Brunt et al. 1999a, Brunt et al. 1999b; Payette et al. 1995; Posner et al. 1993). We found having difficulty shopping for food, having difficulty with meal preparation, and eating fewer than two meals per day to have significant links with these three indicators of nutritional risk.
It is important to point out that if a chi-square difference test were calculated, that is, the difference in the chi-square estimators for the restricted (recursive) and unrestricted (nonrecursive) models with degrees of freedom equal to their difference in degrees of freedom (Bollen 1989), the result would indicate no difference between the two models. However, since the goodness-of-fit statistics indicate an equally close fit for both models and both reciprocal pathways of the nonrecursive model are statistically significant, both models offer valuable insights into the interrelationship of components of nutritional risk and severity of disability among home-delivered meal participants. These models require further exploration, which would include validation in independent samples of comparable size.
Limitations of the Study
Although cross-sectional research is important in establishing associations between nutritional risk factors, indicators of nutritional risk, and disability severity in older adults who receive home-delivered meals, it will take prospective studies to help determine the direction of association. Because we were examining data routinely collected by the ENP provider of home-delivered meals, our only source of information on nutritional risk was the NSI Checklist. We found the current design of the Checklist to pose several limitations. First, compound risk factors (e.g., eating few fruits, vegetables, or milk products; lost or gained weight) prevent the identification of specific problem areas, such as eating few fruits or eating few vegetables or eating few milk products or unintentionally losing weight or unintentionally gaining weight. The Administration on Aging has recognized this problem, and a new format of separate and multiple-part questions is currently being pilot tested (U.S. Administration on Aging 2000). Second, the wording of items, such as "illness" and "oral," may yield an insensitive metric that lacks significant predictive value. We also recognize that there are important areas of nutritional and functional risk that are not assessed with the current questionnaires: food security, income adequacy, depressive symptoms, appetite, chronic conditions and burden of disease, and numbers and types of medications. Many of these are linked to physical function and dietary inadequacy (Lee and Frongillo 2001). It is important to note that the effect of these limitations extends beyond this study to the community-based service providers who may rely on the information for planning, policy, and interventions. We also recognize that the reciprocal relationships (e.g., severity of disability and medication use and severity of disability and weight change) call for further examination, because they could reflect an underlying acute illness, chronic conditions, and injuries that cause weight change, physical disability, and increased medication use (Salive and Guralnik 1997; Tully and Snowdon 1995).
Implications of the Study
This research has implications for both researchers and nutrition service policy makers and planners. There are two important implications to researchers interested in functional outcomes: (a) Structural equation modeling is an important methodology for the understanding of direct and indirect relationships between disease, risk factors, and functioning (e.g., nutritional, psychological, social, and physical functioning); and (b) ignoring the simultaneous effect of variables on each other by failing to model reciprocal pathways may introduce some confounding and lead to erroneous estimates of the impact of individual variables (Zohoori and Savitz 1997). Although both the recursive and nonrecursive models in the present study provided an equally good fit, the inclusion of significant reciprocal pathways altered the results and interpretation from the recursive model.
Prior findings from single equation analyses (Sharkey et al. 2000; Sharkey and Haines 2000) would infer that only the nutritional risk factors associated with functional status (e.g., weight change and medication use) would merit ENP monitoring. Much has been theorized of the role of poor nutritional intake and unintentional weight change in the acceleration of functional decline (Anderson 1990; Fried and Walston 1999; Reuben, Greendale, and Harrison 1995). The findings in the present study are the first to report on the complex direct and indirect associations between individual nutritional risk factors, indicators of nutritional risk, and severity of disability, which suggest that some of the nutritional risk factors are directly associated with indicators of nutritional risk and indirectly associated with functional status. It will be the knowledge of both indirect and direct causal pathways that will help target interventions for short- and long-term outcomes. This will inform ENP service providers who concentrate on helping older persons through the targeting of appropriate services; the monitoring of change in weight and disability severity (improved, declined, and stable); and estimation of service effects. Additionally, these findings also focus attention on the individual nutritional risk factors and indicators of nutritional risk, and away from the use of cumulative scores or categories of nutritional risk level (U.S. Administration on Aging 2000). This accompanies a shift in ENP emphasis to identify performance outcomes, especially in the area of program activities and outcomes that contribute to the health and independence of older adults (Splett and Weddle 1999).
Because disability status, the ability of an older person to function independently in the community, has been found to be a strong predictor of further declines in function, nursing home placement, and mortality (Dunlop, Hughes, and Manheim 1997; Fried and Guralnik 1997; Guralnik, Fried, and Salive 1996; Wolinsky, Callahan, Fitzgerald, and Johnson 1993) it is of paramount importance that home- and community-based service providers have the ability to identify targets of intervention to prevent or reverse functional decline. This may well include identification of chronic health conditions and aggressive intervention or service to augment the meal program. In particular, these interventions or services may include strategies to ameliorate difficulty with meal preparation and shopping for groceries.
As formal services for older adults, such as the ENP's Home-Delivered Meal Program, struggle with providing programs to contribute to good nutrition and independent living, an understanding of the individual risk indicators common to both poor nutrition and functional disability becomes increasingly important. This study has added to that understanding. The goal will be to identify the modifiable determinants of nutritional and functional status, and develop programs to help older persons live independently in the community.
Variable | n | % | Mean (SD) |
Nutritional Risk Factorsa | |||
Medications (take at least three prescribed or over-the-counter drugs per day) | 771 | 76.3 | |
Alone (eat alone most of the time) | 594 | 58.8 | |
Oral (tooth or mouth problems that make it hard to eat) | 207 | 20.5 | |
Prepareb (require assistance to prepare meals) | 873 | 86.4 | |
Shopb (require assistance to shop for groceries) | 935 | 92.6 | |
Money (times when lack enough money for food) | 252 | 24.9 | |
Illness (illness or condition that caused a change in the kind or amount of food eaten) | 501 | 49.6 | |
Indicators of Nutritional Riska | |||
Servings (eat few fruits or vegetables or milk products) | 224 | 22.2 | |
Meals (eat fewer than 2 meals per day) | 84 | 8.3 | |
Weight (lost or gained 10 lbs in the last 6 months without trying) | 259 | 25.6 | |
Disability Severityc | 0.66 (1.14) | ||
0 ADL Disability | 666 | 65.9 | |
1 ADL Disability | 167 | 16.5 | |
2 ADL Disabilities | 102 | 10.1 | |
3 ADL Disabilities | 24 | 2.4 | |
4 ADL Disabilities | 34 | 3.4 | |
5 ADL Disabilities | 17 | 1.7 |
Variable | n | % | Mean (SD) |
Nutritional Risk Factorsa | |||
Medications (take at least three prescribed or over-the-counter drugs per day) | 771 | 76.3 | |
Alone (eat alone most of the time) | 594 | 58.8 | |
Oral (tooth or mouth problems that make it hard to eat) | 207 | 20.5 | |
Prepareb (require assistance to prepare meals) | 873 | 86.4 | |
Shopb (require assistance to shop for groceries) | 935 | 92.6 | |
Money (times when lack enough money for food) | 252 | 24.9 | |
Illness (illness or condition that caused a change in the kind or amount of food eaten) | 501 | 49.6 | |
Indicators of Nutritional Riska | |||
Servings (eat few fruits or vegetables or milk products) | 224 | 22.2 | |
Meals (eat fewer than 2 meals per day) | 84 | 8.3 | |
Weight (lost or gained 10 lbs in the last 6 months without trying) | 259 | 25.6 | |
Disability Severityc | 0.66 (1.14) | ||
0 ADL Disability | 666 | 65.9 | |
1 ADL Disability | 167 | 16.5 | |
2 ADL Disabilities | 102 | 10.1 | |
3 ADL Disabilities | 24 | 2.4 | |
4 ADL Disabilities | 34 | 3.4 | |
5 ADL Disabilities | 17 | 1.7 |
Note: ADL = activity of daily living.
From the Nutrition Screening Initiative's DETERMINE Checklist, with number and percentage of sample reporting yes to each nutritional risk factor and indicator.
From the Instrumental Activities of Daily Living.
Require assistance to bathe, dress, eat, use the toilet, and get in and out of bed.
Variable | n | % | Mean (SD) |
Nutritional Risk Factorsa | |||
Medications (take at least three prescribed or over-the-counter drugs per day) | 771 | 76.3 | |
Alone (eat alone most of the time) | 594 | 58.8 | |
Oral (tooth or mouth problems that make it hard to eat) | 207 | 20.5 | |
Prepareb (require assistance to prepare meals) | 873 | 86.4 | |
Shopb (require assistance to shop for groceries) | 935 | 92.6 | |
Money (times when lack enough money for food) | 252 | 24.9 | |
Illness (illness or condition that caused a change in the kind or amount of food eaten) | 501 | 49.6 | |
Indicators of Nutritional Riska | |||
Servings (eat few fruits or vegetables or milk products) | 224 | 22.2 | |
Meals (eat fewer than 2 meals per day) | 84 | 8.3 | |
Weight (lost or gained 10 lbs in the last 6 months without trying) | 259 | 25.6 | |
Disability Severityc | 0.66 (1.14) | ||
0 ADL Disability | 666 | 65.9 | |
1 ADL Disability | 167 | 16.5 | |
2 ADL Disabilities | 102 | 10.1 | |
3 ADL Disabilities | 24 | 2.4 | |
4 ADL Disabilities | 34 | 3.4 | |
5 ADL Disabilities | 17 | 1.7 |
Variable | n | % | Mean (SD) |
Nutritional Risk Factorsa | |||
Medications (take at least three prescribed or over-the-counter drugs per day) | 771 | 76.3 | |
Alone (eat alone most of the time) | 594 | 58.8 | |
Oral (tooth or mouth problems that make it hard to eat) | 207 | 20.5 | |
Prepareb (require assistance to prepare meals) | 873 | 86.4 | |
Shopb (require assistance to shop for groceries) | 935 | 92.6 | |
Money (times when lack enough money for food) | 252 | 24.9 | |
Illness (illness or condition that caused a change in the kind or amount of food eaten) | 501 | 49.6 | |
Indicators of Nutritional Riska | |||
Servings (eat few fruits or vegetables or milk products) | 224 | 22.2 | |
Meals (eat fewer than 2 meals per day) | 84 | 8.3 | |
Weight (lost or gained 10 lbs in the last 6 months without trying) | 259 | 25.6 | |
Disability Severityc | 0.66 (1.14) | ||
0 ADL Disability | 666 | 65.9 | |
1 ADL Disability | 167 | 16.5 | |
2 ADL Disabilities | 102 | 10.1 | |
3 ADL Disabilities | 24 | 2.4 | |
4 ADL Disabilities | 34 | 3.4 | |
5 ADL Disabilities | 17 | 1.7 |
Note: ADL = activity of daily living.
From the Nutrition Screening Initiative's DETERMINE Checklist, with number and percentage of sample reporting yes to each nutritional risk factor and indicator.
From the Instrumental Activities of Daily Living.
Require assistance to bathe, dress, eat, use the toilet, and get in and out of bed.
Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
1. Meals | 1.00 | |||||||||||
2. Servings | .50 | 1.00 | ||||||||||
3. Weight | .09 | .14 | 1.00 | |||||||||
4. Disability | −.05 | −.05 | .10 | 1.00 | ||||||||
5. Illness | −.08 | .04 | .16 | −.02 | 1.00 | |||||||
6. Oral | .06 | .08 | .01 | .07 | .00 | 1.00 | ||||||
7. Money | .19 | −.05 | −.20 | −.05 | −.02 | .11 | 1.00 | |||||
8. Alone | .06 | .04 | −.03 | −.20 | .11 | −.07 | −.08 | 1.00 | ||||
9. Meds | −.19 | −.02 | .21 | .17 | .26 | .02 | −.13 | .09 | 1.00 | |||
10. Age | −.10 | −.07 | −.10 | .01 | −.13 | .02 | −.21 | .02 | −.21 | 1.00 | ||
11. Prepare | .11 | .04 | .14 | .25 | .09 | −.09 | −.03 | .06 | .22 | .01 | 1.00 | |
12. Shop | −.08 | .15 | .26 | .28 | .22 | .06 | −.11 | .30 | .25 | .11 | .63 | 1.00 |
Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
1. Meals | 1.00 | |||||||||||
2. Servings | .50 | 1.00 | ||||||||||
3. Weight | .09 | .14 | 1.00 | |||||||||
4. Disability | −.05 | −.05 | .10 | 1.00 | ||||||||
5. Illness | −.08 | .04 | .16 | −.02 | 1.00 | |||||||
6. Oral | .06 | .08 | .01 | .07 | .00 | 1.00 | ||||||
7. Money | .19 | −.05 | −.20 | −.05 | −.02 | .11 | 1.00 | |||||
8. Alone | .06 | .04 | −.03 | −.20 | .11 | −.07 | −.08 | 1.00 | ||||
9. Meds | −.19 | −.02 | .21 | .17 | .26 | .02 | −.13 | .09 | 1.00 | |||
10. Age | −.10 | −.07 | −.10 | .01 | −.13 | .02 | −.21 | .02 | −.21 | 1.00 | ||
11. Prepare | .11 | .04 | .14 | .25 | .09 | −.09 | −.03 | .06 | .22 | .01 | 1.00 | |
12. Shop | −.08 | .15 | .26 | .28 | .22 | .06 | −.11 | .30 | .25 | .11 | .63 | 1.00 |
Note: Variables for Disability and Age are continuous. All other variables are ordinal (0, 1).
Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
1. Meals | 1.00 | |||||||||||
2. Servings | .50 | 1.00 | ||||||||||
3. Weight | .09 | .14 | 1.00 | |||||||||
4. Disability | −.05 | −.05 | .10 | 1.00 | ||||||||
5. Illness | −.08 | .04 | .16 | −.02 | 1.00 | |||||||
6. Oral | .06 | .08 | .01 | .07 | .00 | 1.00 | ||||||
7. Money | .19 | −.05 | −.20 | −.05 | −.02 | .11 | 1.00 | |||||
8. Alone | .06 | .04 | −.03 | −.20 | .11 | −.07 | −.08 | 1.00 | ||||
9. Meds | −.19 | −.02 | .21 | .17 | .26 | .02 | −.13 | .09 | 1.00 | |||
10. Age | −.10 | −.07 | −.10 | .01 | −.13 | .02 | −.21 | .02 | −.21 | 1.00 | ||
11. Prepare | .11 | .04 | .14 | .25 | .09 | −.09 | −.03 | .06 | .22 | .01 | 1.00 | |
12. Shop | −.08 | .15 | .26 | .28 | .22 | .06 | −.11 | .30 | .25 | .11 | .63 | 1.00 |
Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
1. Meals | 1.00 | |||||||||||
2. Servings | .50 | 1.00 | ||||||||||
3. Weight | .09 | .14 | 1.00 | |||||||||
4. Disability | −.05 | −.05 | .10 | 1.00 | ||||||||
5. Illness | −.08 | .04 | .16 | −.02 | 1.00 | |||||||
6. Oral | .06 | .08 | .01 | .07 | .00 | 1.00 | ||||||
7. Money | .19 | −.05 | −.20 | −.05 | −.02 | .11 | 1.00 | |||||
8. Alone | .06 | .04 | −.03 | −.20 | .11 | −.07 | −.08 | 1.00 | ||||
9. Meds | −.19 | −.02 | .21 | .17 | .26 | .02 | −.13 | .09 | 1.00 | |||
10. Age | −.10 | −.07 | −.10 | .01 | −.13 | .02 | −.21 | .02 | −.21 | 1.00 | ||
11. Prepare | .11 | .04 | .14 | .25 | .09 | −.09 | −.03 | .06 | .22 | .01 | 1.00 | |
12. Shop | −.08 | .15 | .26 | .28 | .22 | .06 | −.11 | .30 | .25 | .11 | .63 | 1.00 |
Note: Variables for Disability and Age are continuous. All other variables are ordinal (0, 1).
I thank Vivien Keys and Meals on Wheels of Wake County, Inc., for their interest in exploring the relationship between nutrition and function as a way of identifying future service needs and for access to routinely collected data. This article benefited from the thoughtful comments of Dr. Kenneth A. Bollen, Department of Sociology at the University of North Carolina at Chapel Hill on earlier drafts.
References
Anderson S. A.,
Blumberg S. J., Bialostosky K., Hamilton W. L., Briefel R. R.,
Browne M., Cudeck R.,
Brunt A., Schafer E., Oakland M. J.,
Brunt A. R., Schafer E., Oakland M. J.,
Chernoff R.,
Chrischilles E. A., Foley D. J., Wallace R. B., Lemke J. H., Semla T. P., Hanlon J. T., et al.
Coulston A. M., Craig L., Voss A. C.,
Dunlop D. D., Hughes S. L., Manheim L. M.,
Fried L. P., Guralnik J. M.,
Fried L. P., Walston J.,
Goodwin J. S.,
Guralnik J. M., Fried L. P., Salive M. E.,
Institute of Medicine
Jensen G. L., Kita K., Fish J., Heydt D. F. C.,
Joseph H. M., Rogers B. L., Houser R. F., Scharf J., Millen B. E., Balsam A. L.,
Katz S., Ford A. B., Moskowitz R. W., Jackson B. A., Jaffe M. W.,
Kühnel, S. (2001). The didactic power of structural equation modeling. In R. Cudeck, S. du Toit, & D. Sörbom (Eds.), Structural equation modeling: Present and future (pp. 79–96). Lincolnwood, IL: Scientific Software International.
Lawton M., Brody E.,
Lee J. S., Frongillo E. A.,
Moyer W. R., Balsam A.,
Nutrition Screening Initiative (NSI). (1996). Keeping older Americans healthy at home. Washington, DC: Author.
Payette H., Gray-Donald K., Cyr R., Boutier V.,
Ponza M., Ohls J. C., Millen B. E.,
Posner B. M., Jette A. M., Smith K. W., Miller D. R.,
Quandt S., Chao D.,
Quandt S. A., Rao P.,
Reuben D., Greendale G. A., Harrison G. G.,
Salive M. E., Guralnik J. M.,
Schlenker E. D.,
Sharkey J., Haines P.,
Sharkey J. R., Haines P. S.,
Sharkey J. R., Haines P. S., Zohoori N.,
Splett P. L., Weddle D. O.,
Tully C. L., Snowdon D. A.,
U.S. Administration on Aging. (2000). Building the network on aging toolkit: Summary report of the U.S. Administration on Aging Symposium. Washington, DC: Author.
Weimer, J. (1998, October). Affecting nutrient intake of the elderly (Agricultural Economic Report No. 769). Washington, DC: U.S. Department of Agriculture, Economic Research Service, Food and Rural Economics Division.
Wellman N. S.,
West S. G., Finch J. F., Curran P. J.,
White J.,
Wolfe C. S., Olson C. M., Kendall A., Frongillo E. A., Jr.
Wolinsky F. D., Callahan C. M., Fitzgerald J. F., Johnson R. J.,