Revisiting the concept of malnutrition in older people
Cheryl Chia-Hui Chen
DNSc, GNPAssistant Professor, National Taiwan University School of Nursing & National Taiwan University Hospital, Taipei, Taiwan
Ya-Ying Bai
RN, BSNGraduate Student, National Taiwan University School of Nursing, Taipei, Taiwan
Guan-Hua Huang
PhDAssociate Professor, National Chiao Tung University Institute of Statistics, Hsinchu, Taiwan
Siew Tzuh Tang
RN, DNScAssociate Professor, Chang Gung University School of Nursing, Kwei-Shan Tao-Yuan, Taiwan
Submitted for publication: 24 April 2006 Accepted for publication: 9 September 2006
Correspondence: Siew Tzuh Tang Associate Professor
Chang Gung University School of Nursing 259 Wen-Hwa 1st Road Kwei-Shan Tao-Yuan Taiwan Telephone: 8863 2118800 ext. 3417 E-mail: sttang@mail.cgu.edu.tw C H E N C C - H , B A I Y - Y , H U A N G G - H & T A N G S T ( 2 0 0 7 )
C H E N C C - H , B A I Y - Y , H U A N G G - H & T A N G S T ( 2 0 0 7 ) Journal of Clinical Nursing 16, 2015–2026
Revisiting the concept of malnutrition in older people
Aim. The aim of this study is to revisit the concept of malnutrition in older people, trace the new development and test the use of refined framework with empirical data.
Background. Malnutrition in older people is a common and significant problem worldwide. Continuing Chen’s earlier work in 2001, a refinement was conducted and a prospective study was designed to test the use of this refined framework. Design. A cohort study of 114 hospitalized older patients in Northern Taiwan. Methods. The sample consists of 114 older patients aged 65 years and older, who were admitted for the cardiac and orthopaedic services at a tertiary 2300-bed hospital. From March to August 2004, assessed by one trained nurse, participants completed a structured face-to-face interview evaluating their age, visual/hearing impairments, oral health, cognitive status, comorbidities, medication use, social economic status, functional status, social support, depressive symptoms and nutri-tional status within 48 hours of admission. Participants who stayed >5 days were reassessed before discharge (n ¼ 70). The data from admission were the main focus of this report.
Results. Regression analysis revealed that that more medication taken, female gender, lower functional status (beta ¼ 0Æ34, P < 0Æ001) and higher depressive symptoms were independent predictors of poor nutritional status, with the full model accounting for 48Æ2% of the variance. The result is in-line with the original theoretical underpinnings and it suggests that this refined framework detailing sub-concepts and measurable indices appears to fit the empirical data and suitable for clinical use.
Conclusion. The findings lend support to the use of this framework in managing malnutrition in older people.
Relevance to clinical practice. Nurses have an essential role in providing care for older people a framework like this would provide a road map guiding the inter-vention efforts.
Key words: concept development, elder care, nursing, nursing care, nutrition, theory-practice gap
Introduction
Finding ways to deliver care anchored in the best possible evidence remains a continuing challenge. Further refinements of a conceptual framework based on empirical data are of importance in moving health care forward. A concept analysis of malnutrition in older people has been published in 2001 (Chen et al.) to delineate on the key elements of the conceptual framework including critical attributes, anteced-ents and consequences, leading to refinement of the frame-work. While these key elements remain theoretically meaningful to the development of malnutrition in older people, new empirical data have been available to their constituent sub-elements, enabling the detail of the frame-work to be revised.
In 2001, a concept analysis of malnutrition in older people was first published in the Journal of Advanced Nursing, in which the critical attributes of malnutrition in older people were identified to include insufficient dietary intake, muscle wasting, weight loss, poor appetite and downward trajectory. Loss, chronic illness, dependency and loneliness were pro-posed as the antecedents of malnutrition in older people (Chen et al. 2001). As the key elements of the framework were proposed and opened up to scrutiny, it has received much attention and interest and appears to have face validity to those in the field working with older people (20 citations since 2001). This paper aims to refine the framework further in reflecting a synthesis of the current literature and to make the framework more distinct than the original. It also seems timely to report on a prospective cohort study designed primarily from this refined framework to examine its empirical values.
Background & conceptualization
Ensuring proper nutritional status for older people has been shown to reduce healthcare costs and enhance quality of life. Studies, however, have revealed that malnutrition is present in 10–60% of hospitalized older patients, depending on the definition used and the population studied (Sullivan
et al. 1999, Van Nes et al. 2001, Guigoz et al. 2002). The variance in this figure is the result of the differences in the study method employed, the type of setting in which the study was conducted and the measures used for defining malnutrition. The seemingly complex and ambiguous nature of the concept places a barrier on the development of targeted interventions for preventing and treating malnutri-tion. This gap in the literature is serious, because malnu-trition has been documented to occur frequently and it leads to increased morbidity and mortality (Institute of Medicine 2000, Liu et al. 2002). There is clearly a lack of understanding on the complex factors having an impact on malnutrition in older people. To build on the previous work and adding the new insight to move forward, this study aims to refine and operationalize these proposed conceptual elements and test the validity of these elements with real world data. As the first preliminary step, the critical attributes of malnutrition in older people were used as a guide for choosing nutrition measures. The antecedents of malnutrition in older people, including loss, chronic illness, dependency and loneliness were operationalized to test their relationships with the development of malnutrition in older people. The process of operationalizing these antecedents was described in the following text.
Loss
As proposed by Chen et al. (2001) and further refined and developed in Fig. 1, the antecedent construct of loss involved four concepts including loss of optimal body composition, loss of sensory function, loss of oral health and loss of role function. These more concrete concepts must be defined in measurable ways and hypotheses stating specific and observ-able relationship must be derived from the proposition of the conceptual framework before action can occur. Therefore, four sub-concepts within the construct of loss were further defined and operationalized as the following:
Advanced age is associated with a remarkable number of changes in body composition, including reduction in lean
body mass and increase in the body fat. This age-related loss in muscle mass has been termed ‘sarcopenia’ and it is a direct cause of an age-related decrease in muscle strength. The most accurate measures for body composition include hydrostatic weighing, dual-energy X-ray absorptiometry (DXA), air-displacement plethysmography (ADP) and multi-compart-ment models (Weyers et al. 2002, Fields & Hunter 2004). However, given the complexity and limited clinical implica-tion of these measures and the high correlaimplica-tion between age and body composition changes, a decision was made to use chronological age as a proxy to illustrate the impact of loss of optimal body composition on nutrition. In terms of loss of sensory function, although smell, taste, visual and hearing sensory losses have all been suggested to be detrimental to nutrition (Chen et al. 2001), several studies have indicated
that loss of smell and taste did not appear predictive nutritional status (Griep et al. 2000). In contrast, visual and hearing losses were suggested to be more significant (Payette et al. 1995, Valentijn et al. 2005). Visual and/or hearing losses, therefore, were chosen to represent the concept of loss of sensory function in this study. Loss of oral function was measured by general oral health functioning. Specific dental health indices such as the number of decay, missing and filling teeth (DMFT) index were not used given that general oral health function has been suggested to have a significant impact on nutrition (de Groot et al. 2000). Loss of role function was represented by the cognitive status given that cognition; ‘knowing what to do’ is essential to maintain the role function, particular for older people (Sarna et al. 1993, Knight 2000). Loss Chronic illness Dependency Loneliness Malnutrition in the elderly
Loss of optimal body composition
Loss of sensory function
Loss of oral health
Loss of role function
Co-morbidities Polypharmacy Financial dependency Functional dependency Social support Depression
Insufficient dietary intake Muscle wasting
Weight loss Poor appetite Downward trajectory
Age as proxy
Visual / hearing impairment
Oral health Cognitive status Age in years Presence of impairment GOHAI / # of teeth MMSE
# of co-morbidities Co-morbidity list
# of medication Med count
SES Income / edu / gender
Functional status Barthel index / ADLs
Availability of support Satisfaction support SSQ-SF Depressive symptoms GDS-15 Nutritional status Mini-Nutritional Assessment (MNA) % of weight loss A n tecedents Concept
Conceptual Conceptual – Theoretical Empirical indices
GOHAI = Geriatric oral health assessment index, MMSE = Mini-mental state examination, SSQ - SF = Social support questionnaire-short form, GDS-15 = Geriatric depression scale-15
Figure 1 Refined conceptual-theoretical-empirical structure framework of malnutrition in older hospitalized patients. GOHAI, General Oral Health Assessment Index; MMSE, mini-mental state examination; SSQ-SF, Social Support Questionnaire-Short Form; GDS-15, Geriatric Depression Scale-15.
Chronic illness
Comorbidities and polypharmacy were suggested as the two sub-concepts within the construct of chronic illness in Chen’s original work. Accordingly, polypharmacy, defined as the concomitant use of multiple drugs, was measured by a simple count of medications (Hanlon et al. 2001). Comorbidities were assessed by counting the diagnoses of 19 common chronic illnesses in older people. In short, numbers of comorbidities and medication taken were operationally chosen as two indices to represent the concepts.
Dependency
Within the construct of dependency, financial dependency as characterized by socioeconomic status (SES) was examined in this study. A national survey indicated that nutritional intake was significantly different between persons with low and high SES (Guthrie & Lin 2002). Therefore, income, educational level and gender were used as empirical indices for SES in this study. Gender was studied because of its importance in nutritional status and in the degree of SES impact on health at large. For example, a weaker socioeconomic gradient in health and mortality has been observed for female gender (Duncan et al. 2002). However, a feminization of poverty is also being suggested. Women have been considered to be less well off financially and to suffer more intensely because they are discriminated against and fewer opportunities are open for them, particularly in the East (World Health Organiza-tion 2001). Therefore, the impact of gender was studied within the context of SES in this study. Functional depend-ency was measured by activities of daily living (ADL)-related functional status.
Loneliness
The construct of loneliness, as suggested in Chen’s original work, was measured by the presence of social support and depressive symptoms. Depressive symptoms have been sug-gested as one of the most important factors impacting nutrition in older people (Wilson et al. 1998, Chen et al. 2005). On the other hand, empirical evidence on the impact of social support on nutrition is inconclusive and in need for closer examination.
In summary, age in years, visual/hearing impairment, oral health, cognitive status, number of comorbidities, number of medications taken, social economic status (education, income and gender), functional status and presence of social support and depressive symptoms were examined as the predictors of nutritional status, which was measured by the
mini-nutri-tional assessment (MNA) in a sample of 114 older patients who had been admitted for cardiac and orthopaedic services at a tertiary medical centre located in Northern Taiwan.
Objectives & methods
The objective of this study was to evaluate the use of the proposed conceptual framework using empirical data. Specifically, the relative importance of age, visual/hearing impairments, oral health, cognitive status, comorbidities, medication use, social economic status, functional status, presence of social support and depressive symptoms in predicting nutritional status of hospitalized older patients was cross examined. As illustrated in Fig. 1, these predictors were selected from an exhaustive literature synthesis (Payette et al. 1995, de Groot et al. 2000, Institute of Medicine 2000, Chen et al. 2005), along with the initial concept development by Chen et al. (2001).
Data were collected by one trained research nurse who used structured instruments to obtain information and measures on demographics, visual/hearing impairments, oral health, cog-nitive status, comorbidities, medication use, functional status, social support, depressive symptoms and nutritional status. All participants were interviewed within 48 hours of admis-sion (n ¼ 114) and before discharge if length of stay was over 5 days (n ¼ 70). The interview schedule consisted of four sections: (1) demographics; (2) descriptive data; (3) predictor scales and (4) anthropometric measures including weight, knee height and mid-arm and calf circumference. Participants also underwent a brief oral health examination involving counting the remaining teeth and assessing the fit of denture. Because the pool of participants at the discharge assessment was limited (n ¼ 70), this paper only reports results obtained at admission assessment (n ¼ 114).
Sample/participants
A hospital sample was recruited. The study population was defined as older patients aged 65 years and above who had been admitted to cardiac and orthopaedic floors of a tertiary medical centre in Taipei, Taiwan during the months of March to August, 2004. Every older patient admitted to these two floors was invited to participate. Patients were not recruited if their mini-mental state examination (MMSE) scores were <20 (n ¼ 43), expected length of stay <3 (n ¼ 235), isolated within infection control protocol (n ¼ 8), intubated; unable to communicate or under critical conditions (n ¼ 133). Consent by proxy was not used in this study due to both its methodological and ethical problems. Of 203 eligible patients, 114 completed baseline assessments during
the study period (1 March 2004 to 31 August 2004). The nonparticipants (n ¼ 89) did not differ significantly from the participants in age (p ¼ 0Æ82), gender (p ¼ 0Æ53) or educa-tion (p ¼ 0Æ99). The reasons for nonparticipaeduca-tion included ‘not interested’ (n ¼ 13), ‘not feeling well’ (n ¼ 41), ‘in pain or discomfort’ (n ¼ 25) and for ‘privacy protection’ (i.e. do not wish to give signature for any causes; n ¼ 10).
Instruments
Reliability of data collection was maintained by standardized protocols, use of validated instruments and training of research staff. The psychometric properties of instruments were briefly described as the following:
Demographics & sensory impairment data
A demographic form was designed to collect the data in-cluding age, gender, marital status, living status, income, education, occupation and ethnic group. Additionally, sen-sory impairments were assessed by the self-report of visual and hearing impairments. Participants who presented with hearing or visual difficulties during the data collection were automatically coded as having hearing or visual impairments. Oral health
A 12-item Chinese version of General Oral Health Assessment Index (GOHAI) was used to assess oral health. The GOHAI is designed to assess the dimensions of oral function (eating and speaking), pain, discomfort, worry and oral health-related social functioning (Atchison & Dolan 1990). Reliability esti-mates (Cronbach’s alpha ¼ 0Æ83–0Æ79; inter-rater reliability a¼ 0Æ61) and established content, discriminant and construct validity have been reported for both English and Chinese versions (Calabrese et al. 1999, Wong et al. 2002a). The sum scores ranged from 12–60 with a higher score indicative of better perceived oral health. Information on number of remaining teeth and fit of denture were also obtained. Cognitive status
The 30-item Chinese MMSE was used to measure cognitive status. The MMSE is the most widely used cognitive assess-ment tool in Taiwan. The final score is the sum of the scores from 30 items and it ranged from 0–30 (Folstein et al. 1975). Satisfactory test-retest (r ¼ 0Æ89–0Æ93), inter-rater reliability (r ¼ 0Æ83) and content and discriminant validity have been reported (Guo et al. 1988).
Comorbidities
The history and number of comorbidities was elicited from the medical record. A standardized comorbidity checklist was
used to assess common chronic illnesses including myocardial infarction, angina, congestive heart failure (CHF), hyperten-sion, diabetes, hyperlipidaemia, arthritis, dementia, stroke, asthma or lung disease, renal disease, Parkinson’s disease, osteoporosis, hip fracture, pressure sore, cancer and others. Medication use
Using a standardized protocol, medication review was con-ducted by the trained research nurse with the data from the medical record and information from the patients. The number and type of prescription and over-the-counter med-ications taken currently by participants was documented. Functional status
The 10-item Chinese version of Barthel Index (BI) was used to measure functional status. Satisfactory reliability (Cron-bach’s a ¼ 0Æ87–0Æ91; Kappa ¼ 0Æ70–0Æ88) and validity have been reported for the BI (Dai et al. 1999, Sainsbury et al. 2005). The BI comprises 10 ADLs with the sum scores range from 0–100 and higher score indicating better functional status (Mahoney & Barthel 1965).
Social support
The six-item Chinese Social Support Questionnaire-Short Form (SSQ-SF) was used to measure social support. In six common situations, subjects were asked to list up to nine people who can be counted on and specify overall degree of satisfaction with such support. The tool has been used with older population: the coefficient a ranged from 0Æ90–0Æ93 and established construct and content validity have been reported (Sarason et al. 1987). The SSQ-SF was translated into Chi-nese for the purpose of this study. The translation procedure incorporated two iterations of translations from English to Chinese by the first author and the back translation from Chinese to English by an independent researcher blind to the original English SSQ-SF. Additionally, five bilingual experts in nursing, psychology, public health and gerontology were asked to evaluate the conceptual equivalence of Chinese version SSQ-SF with original SSQ-SF and satisfactory results have been obtained (CVI ¼ 0Æ80–1Æ00). The Cronbach’s a of Chinese SSQ-SF was 0Æ85 in the present study.
Depressive symptoms
The 15-item Chinese version of Geriatric Depression Scale Short-Form (GDS-15) was used to measure the presence of depressive symptoms. The 95% confidence intervals for sensitivity and specificity of the long-form GDS (GDS-30) are 84–93% (Yesavage et al. 1983). The GDS-15 has demon-strated a high correlation with the long-form GDS (Gerety et al. 1994, Wong et al. 2002b).
Nutritional status
The 18-item Chinese version of MNA was used to measure nutritional status, given its inclusion of all five proposed crit-ical attributes among the items. The summed score could be used either as a categorical or interval variable. The score can categorize older people as (1) ‡24, well-nourished; (2) 23Æ5– 17, at risk and (3) <17, malnourished. High levels of sensi-tivity (96%), specificity (98%) and satisfactory inter-rater reliability (Kappa coefficient ¼ 0Æ65–0Æ42) have been reported (Guigoz et al. 1996, Chan et al. 2002). Additionally, weight loss history in the prior three-month was also collected.
Ethical consideration
The study was approved by the Research Ethics Review Board at the National Taiwan University Hospital. The process of obtaining consent for participation was lengthy, as the investigators were mindful of the vulnerability of older population. All eligible participants were approached by the research nurse to explain the nature and purpose of the study and invited to participate. The voluntary and confidential nature of the study was braced. At any time, if participants appear stressed or in discomfort, the assessment was termin-ated. No assessment was terminated for stress or discomfort reasons, but few participants’ assessments were break into two sessions to reduce responder burden (n ¼ 3). The period of time needed to collect data was approximately 25– 40 minutes. Participants with identified health concerns, such as depressive symptoms, were referred to their care providers when consent was granted (n ¼ 8).
Data analysis
Data were analysed usingSASSAS, version 8Æ0 (SAS Institute, Inc.,
Cary, NC, USA). Data were double-entered and comparative analyses were completed for accuracy. Descriptive and bivariate analyses including t-test for two group data,ANOVAANOVA
for three or more groups and Pearson correlation for continue variables were computed, to provide an overview showing unadjusted bivariate relationship among variables. Of note, scores on depressive symptoms were fairly skewed (skew-ness ¼ 2Æ9) in terms of their distributions on normality and therefore non-parametric test-Spearman rank-order correla-tion was applied.
As many factors could be used as either a continuous or categorical variable, care was given to use continuous variables whenever possible to avoid losing information. The proposed framework was then tested by regression models with MNA scores as the dependent variable. As shown in Fig. 1, independent variables were entered in block
and the order of antecedent construct being entered was decided as loss, chronic illness, dependency and loneliness. To evaluate the unique variance of MNA scores explained by each antecedent construct, partial F-test was used (Klein-baum et al. 1998). Prior to analyses, data were assessed for congruence with regression assumption. Normality and homoscedasticity were evaluated by residual scatterplots. The residuals were distributed evenly. Therefore, assump-tions for regression were reasonably met. Significance was set at p < 0Æ05.
Results
The sample was evenly distributed by gender, with 56Æ14% (n ¼ 64) female. The majority were living with others (93Æ86%; n ¼ 107) and one-third (31Æ58%, n ¼ 36) were widow. The sample represented diverse educational, income and ethnic backgrounds. Ages ranged from 65 to 90 years with a mean of 75Æ22 years and standard deviation (SD) of 6Æ29 years. The detailed demographics of the sample are shown in Table 1.
Descriptive analysis
The BMI ranged from 15Æ22 to 35Æ17 kg/m2with a mean of 24Æ07 and SD of 4Æ10. Using the criteria recommended by the Department of Health in Taiwan (2002), 11 subjects (9Æ65%) had BMI < 18Æ5 kg/m2and were classified as underweight.
Conversely, 57 subjects (50%) having BMI > 24 kg/m2were considered overweight. In addition, 19 subjects (16Æ67%) reported having a weight loss equal to or >5% in the last 3 months. In terms of MNA scores, five subjects (4Æ39%) scored <17 and were classified as malnourished and 49 subjects (42Æ98%) scored 17–24 and were at risk of malnutrition; the remaining 60 (52Æ63%) were considered well-nourished.
Bivariate analysis
Age in years, visual/hearing impairments, oral health, cogni-tive status, number of comorbidities, CHF diagnosis, number of medications taken, functional status and depressive symp-toms appeared to be significantly associated with MNA nutritional status all in the expected direction, which provi-ded tangible evidence for the validity of Chen’s framework (Tables 2 and 3). Conversely, number of teeth, marital status, income and certain comorbidities were later removed from the model building given its dismal association. It is note worthy that income has not been a sensitive index for financial security, particularly for Taiwanese older people.
This might be due to the fact that children are supposed to support their aged parents in Chinese culture.
Meanwhile, female gender (p ¼ 0Æ21), education (p ¼ 0Æ20) and scores of social support (p ¼ 0Æ21), although they did not reach the significant level at p < 0Æ05, were forced into the regression model to examine the merits of their impact. Additionally, within the sensory impairments, only visual impairment (p ¼ 0Æ02) was retained in further model building. As to the concern for multicollinearity, number of comorbidities was later removed from the regres-sion model given its less desirable tolerance and high correlation with number of medications (tolerance ¼ 0Æ52; r ¼ 0Æ44).
Table 1 Sample demographic characteristics (n ¼ 114)
No. %
Age group
65–74 years 56 49Æ1
75–84 years 48 42Æ1
85 years and above 10 8Æ8
Gender Male 50 43Æ9 Female 64 56Æ1 Ethnicity Taiwanese 86 75Æ4 Mainlander 15 13Æ2 Hakka 12 10Æ5 Native Taiwanese 1 0Æ9 Married status Widowed 36 31Æ3 Married 74 64Æ9 Single/separate 4 3Æ5 Education level Illiterate 27 23Æ7 Six grade 46 40Æ4 Nine grade 15 13Æ2 ‡12 grade 26 22Æ8 Monthly income <5000 NTD* 22 19Æ3 5000–10 000 20 17Æ5 10 001–20 000 10 8Æ8 ‡20 001 16 14Æ0
Prefer not to tell 46 40Æ4
Religion Buddhist 38 12Æ0 Daoism 39 30Æ0 Protestant/catholic 8 56Æ0 Other 29 1Æ6 Sensory impairments Visual impaired 41 36Æ0 Hearing impaired 19 16Æ7
Both visual/hearing impaired 10 8Æ8
Neither impaired 64 56Æ1
*New Taiwan Dollar, 32 NTD ¼ 1USD ¼ 0Æ56GRP.
Table 2 Bivariate relationship between predictors and mini-nutritional assessment (MNA) scores (n ¼ 114)
Characteristic MNA, mean (SD) Test statistics p-values Gender Female 23Æ1 (3Æ4) 1Æ26 0Æ21 Male 23Æ9 (3Æ3) Education Illiterate 22Æ7 (3Æ8) 1Æ47 0Æ23 Six grade 23Æ9 (3Æ0) Nine grade 23Æ9 (2Æ0) ‡12 grade 22Æ3 (4Æ8) Income, monthly <5000 NTD* 23Æ4 (2Æ4) 0Æ04 0Æ99 5001–10 000 23Æ4 (2Æ3) ‡10 001 23Æ2 (4Æ6) No to tell 23Æ5 (3Æ5) Marital status Widow 22Æ8 (4Æ1) 1Æ04 0Æ38 Married 23Æ6 (3Æ0) Single/separate 24Æ5 (4Æ4) Selected comorbidities Visual impaired Yes 22Æ4 (3Æ5) 2Æ33 0Æ02 No 23Æ9 (3Æ3) Hearing impaired Yes 22Æ2 (3Æ8) 1Æ65 0Æ10 No 23Æ6 (3Æ3)
Congestive heart failure
Yes 21Æ6 (4Æ2) 2Æ34
0Æ02
No 23Æ7 (3Æ2)
Kidney function impairment
Yes 22Æ4 (4Æ0) 1Æ47
0Æ14
No 23Æ6 (3Æ8)
Any cancer other than of the skin
Yes 22Æ0 (3Æ8) 1Æ52
0Æ13
No 23Æ6 (3Æ3)
*New Taiwan Dollar, 32 NTD ¼ 1USD ¼ 0Æ56GRP;
t-test; ANOVA ANOVA.
Table 3 Bivariate correlation between continue predictors and mini-nutritional assessment scores (n ¼ 114)
Continued variables Mean ± SD Correlation p-values
Age in years 75Æ2 (6Æ3) 0Æ25 0Æ01
Geriatric oral health assessment index
51Æ5 (4Æ9) 0Æ19 0Æ05 Number of remaining teeth 10Æ3 (9Æ3) 0Æ09* 0Æ33 Mini-mental state examination 26Æ3 (3Æ2) 0Æ33 <0Æ001 Number of comorbidities 4Æ8 (2Æ0) 0Æ34 <0Æ001 Number of medications taken 4Æ8 (3Æ2) 0Æ24 0Æ01 Barthel Index 94Æ4 (12Æ3) 0Æ56 <0Æ001 Social Support
Questionnaire-Short Form
60Æ1 (16Æ4) 0Æ12 0Æ21 Geriatric Depression
Scale-Short Form
1Æ7 (2Æ3) 0Æ29* 0Æ0018
Multivariate analysis
Results of multivariate analysis are shown in Table 4. In the model building, each antecedent construct comprising sub-concepts were entered the model as a group in the following sequential. At step 1, the construct of loss, measured by age, visual impairment, oral health and cognitive status was first put into the model and explained 19Æ1% of the variance in MNA scores. At step 2, the construct of chronic illness, measured by CHF diagnosis and number of medications taken was then added and explained extra 4Æ7% of the variance (p ¼ 0Æ04). At step 3, the construct of dependency, measured by education, gender and functional status was added and explained extra 17Æ1% of the variance in MNA scores (p < 0Æ001). At step 4, the construct of loneliness, measured by social support and depressive symptoms was added lastly and explained extra 7Æ3% of the variance (p ¼ 0Æ001). As a whole, the model explained 48Æ2% of the variance in MNA nutritional scores and in each step, every construct contributed gradually to the prediction of nutri-tional status.
Comparing four regressions in Table 4, it is worthy noting that regression coefficients for the construct of loss, although were significant at first, their significance gradually became null after adding other constructs. Conversely, the constructs of chronic illness, dependency and loneliness did
not alter their significance through four models. This indicates that the construct of loss by itself was a significant predictor of MNA scores, but its association with MNA scores was diluted when other three constructs entering the model.
Discussion
Data from our study indicated that higher depressive symp-toms, lower functional status, female gender and higher medication taken were independently associated with poor nutritional status after multiple adjustments. By using regression analyses, the changes on every single predictor when adding new variables were visible, which provide a better contrast and in-depth understandings on potential confounding effects among variables. The findings also lend support to the validation of this refined conceptual structure, in which each construct contributed to the prediction of nutritional status, as shown in Fig. 1.
Specifically, in our study, depressive symptoms appear to be most significant even after adjusting for many risk factors. The prevalence of depression in the community-dwelling older people is reported as 8–16% (Blazer 2002). One of the leading factors reported to be associated with malnutrition and weight loss in older people is depressive symptoms (Huffman 2002, Callen & Wells 2005, Chen et al. 2005).
Table 4 Regression models with betas shown at each step (n ¼ 114) Step predictors
MNA scores, standardized beta weights (p-values)
Step I Step II Step III Step IV
Loss Age in years 0Æ18 (0Æ04) 0Æ17 (0Æ06) 0Æ08 (0Æ35) 0Æ13 (0Æ11) Visual impair 0Æ18 (0Æ04) 0Æ12 (0Æ18) 0Æ06 (0Æ44) 0Æ04 (0Æ62) GOHAI score 0Æ11 (0Æ20) 0Æ10 (0Æ25) 0Æ07 (0Æ32) 0Æ04 (0Æ61) MMSE score 0Æ26 (0Æ004) 0Æ28 (0Æ002) 0Æ18 (0Æ06) 0Æ11 (0Æ24) Chronic illness CHF diagnosis 0Æ08 (0Æ40) 0Æ01 (0Æ89) 0Æ03 (0Æ68) # meds taken 0Æ18 (0Æ07) 0Æ19 (0Æ04) 0Æ19 (0Æ03) Dependency Education 0Æ10 (0Æ31) 0Æ04 (0Æ64) Female gender 0Æ15 (0Æ09) 0Æ18 (0Æ03) BI score 0Æ43 (<0Æ001) 0Æ34 (<0Æ001) Loneliness SSQ-SF score 0Æ04 (0Æ62) GDS-15 score 0Æ32 (<0Æ001) R2change (P)* 0Æ05 (0Æ04) 0Æ17 (<0Æ001) 0Æ07 (0Æ001) Cumulative R2 0Æ19 0Æ24 0Æ41 0Æ48 Adjusted R2 0Æ16 0Æ20 0Æ36 0Æ43
*p-value is based on partial F-test.
GOHAI, General Oral Health Assessment Index; MMSE, mini-mental state examination; CHF, congestive heart failure; SSQ-SF, Social Support Questionnaire-Short Form; GDS, Geriatric Depression Scale.
Unfortunately, only 22% of cases characterized as major depression are diagnosed and receive adequate treatment (Hirschfeld et al. 1997). A mounting body of work has identified management of depressive symptoms as one of the future key areas in the health of older people. Nurses have intimate and ongoing contact with older patients and therefore are best positioned to screen for depressive symp-toms and referring high-risk older people to care. Routine screening for depressive symptoms should be reinforced given that many effective treatments exist for older people with depressive symptoms and its strong link with malnutrition and other related disabilities.
Functional status measured by BI has emerged as another important predictor after multiple adjustments in our study. Functional status, a cardinal component of the care of older people has been identified in many studies as an important predictor to malnutrition in older people (Unosson et al. 1991, Chen 2005). The importance of identifying impair-ments of functional status lies in the fact that they may be amenable to intervention. Prior studies using specialized geriatric units or geriatric consultation services have demon-strated promising effects in improving functional status of older patients during their hospitalization (Thomas 2002). Many nursing interventions including feeding assistance, mobilizing plan for patients, daily visit/orientation protocol and making environmental modifications, have been the essential components of these successes (Wanich et al. 1992, Inouye et al. 2000). Studies are warranted to test the effectiveness of these hospital-based nursing interventions in improving nutritional status and related outcomes in the hospitalized older people.
Although cumulative research points to a robust associ-ation between SES (income and educassoci-ation in particular) and health, the magnitude of the effect of SES on health seems to vary across social groups. For example, a weaker socioeco-nomic gradient in health and mortality has been observed for retired elders and women (Beckett 2000, Duncan et al. 2002). In our study, income and education were not significantly related to nutritional status. Only gender remained significant in the final regression model, in which the nutritional status of female subjects was slightly worse than that of males, after multiple adjustments. Gender difference has been observed in the nutrition literature in terms of food preference, intake pattern and risk profile for malnutrition (Bates et al. 1999, Quandt & Chao 2000, Locher et al. 2005). There are two possibilities for this observed gender difference. First, females are biologically different from males and are more susceptible to poor nutritional status. Studies have suggested that a number of fundamental metabolic differences existed be-tween older males and females (Bates et al. 1999, Dionne
et al. 1999). Secondly, the interaction effects of age, educa-tion with income and/or other predictors like social support or depressive symptoms might be more complicated for women (Castel et al. 2006). For example, gender differences have been reported in the exposure and reactivity to stressors and social support, giving rise to the assumption that the pathogenic and interactive effect of these factors is different in males and females (Hops 1996). Therefore, the possibility of gender serving as an adjusting factor in our model cannot be excluded. Given the limited sample size of our study, such interaction effects cannot be tested. Future study with diverse sample profiles might want to investigate on these hypotheses.
Polypharmacy, defined as the number of medications taken, emerged as a significant predictor of nutritional status in our study. This finding is consistent with the Belgian study, in which Griep et al. (2000) studied nutritional status in a group of retirement home-living older people and found that number of medications contributed negatively to the MNA scores. Another larger cohort study has reported that a great number of medications were associated with increased risk of weight loss in older people (>4Æ54 kg in a year), after extensive adjustment for age, visual impair-ment, comorbidities, cognitive impairment and depressive symptoms (Agostini et al. 2004). Similarly, the relationship between number of medications taken and MNA nutritional status was consistent and remained after extensive adjust-ment in our study. Present medication prescribing is predi-cated on the disease-specific benefits and individual adverse effects. Our study finding suggests that more careful consideration should be directed toward the compromises between the benefits and harms of total medication con-sumption in older patients.
The study has limitations. First, the sample size is small and data were collected at one site so the generalizibility is limited. Secondly, the possibility of undetected confounders cannot be excluded. Furthermore, although many comorbi-dities have been considered, information on all chronic illness and severity of illness were not available. Nevertheless, the relationship between functional status, number of medica-tions taken, depressive symptoms and MNA nutritional status was fairly consistent and remained after extensive adjustment. These results mandate heightened attention to these prevalent geriatric issues, namely polypharmacy, depressive symptoms, compromised ADL functional status and their impact on nutrition and related disability. A mounting body of work suggests that malnutrition or weight loss is one of most important contributors to frailty, which has important implications for both prevention and treatment (Walston 2004). In agreement with our study, these findings
emphasize the importance of timely recognition of high-risk older people who seem to be on the edge of falling into the cascade of illness and frailty.
Conclusion
The results of this study have allowed the research team to scrutinize the sub-element concepts and empirical indicators of the framework presented in 2001. The changes made represent the results of an up-to-date literature review and an empirical testing about what antecedes the development of malnutrition in hospitalized older people. It is, therefore, the detail of the framework that has been further refined. The refined framework now reflects a critical synthesis of the literature and is more distinct than the original that was first proposed. Although some of the contents have been refined, the basic proposition of it remains the same. Malnutrition in older people is a multi-dimensional concept encompassing physical, psychosocial and cultural elements. It is worth noting that although the original concept analysis was based mostly on western literature, the resulting framework appears to be appropriate for empirical use in Taiwan. Risk profile is fairly similar across regions. Given that malnutrition in older people is significant globally, a framework like this will have international implications for guiding nurses in design-ing or implementdesign-ing hospital-based screendesign-ing or intervention programmes.
This process of revisiting the concept of malnutrition in older people has highlighted that more needs to be under-stood about the links outside of these individual factors affecting nutrition in older people, given that only 48Æ2% of the variance was explained by these individual factors. For example, the way in which biology and individual and social action interact needs to be studied further. The importance of construct of loss, although it is heavily emphasized in the literature, needs to be studied further given its minimal prediction on nutrition after other constructs being added. It would be premature to suggest that this refined framework represents a final version. Research is currently underway to validate and refine the framework further. This phase will provide a larger, longitudinal dataset to ensure that the framework is not just comprehensive and generalizable but also relevant to healthcare providers who are implementing evidence-based practice.
Acknowledgements
The study was supported by National Taiwan University College of Medicine New Faculty Development Fund, Grant # NTUCM 9218–92A136, awarded to Dr Chen.
Contributions
Study design: CC, ST; data collection and analysis: CC, YB, GH and manuscript preparation: CC, YB, GH, ST.
References
Agostini JV, Han L & Tinetti ME (2004) The relationship between number of medications and weight loss or impaired balance in older adults. Journal of American Geriatrics Society 52, 1719– 1723.
Atchison KA & Dolan TA (1990) Development of the geriatric oral health assessment index. Journal of Dental Education 54, 680– 687.
Bates CJ, Prentice A, Finch S (1999) Gender differences in food and nutrient intakes and status indices from the National Diet and Nutrition Survey of people aged 65 years and over. European Journal of Clinical Nutrition 53, 694–699.
Beckett M (2000) Converting health inequalities in later life-an artifact of mortality selection. Journal of Health and Social Behavior 41, 106–119.
Blazer DG (2002) The prevalence of depressive symptoms. Journal of Gerontology: Biological Sciences 57, M150–M151.
Calabrese JM, Friedam PK, Ross LM & Jones JA (1999) Using the GOHAI to assess oral health status of frail homebound elders: reliability, sensibility and specificity. Special Care in Dentistry 19, 214–219.
Callen BL & Wells TJ (2005) Screening for nutritional risk in community-dwelling old-old. Public Health Nursing 22, 138– 146.
Castel H, Shahar D, Harman-Boehm I (2006) Gender differences in factors associated with nutritional status of older medical patients. Journal of the American College of Nutrition 25, 128–134. Chan L-C, Lao Senyeong Chin H-ML & Lee M-S (2002) Nutritional
status assessment and predictors of community-dwelling and institutionalized elderly in Northern Taiwan [Chinese]. Nutritional Science Journal 27, 147–158.
Chen CC-H (2005) A framework for studying nutritional health of community-dwelling elders. Nursing Research 54, 13–21. Chen CC-H, Schilling LS & Lyder CH (2001) A concept analysis
of malnutrition in the elderly. Journal of Advanced Nursing 36, 131–142.
Chen CC-H, Chang CK, Chyun DA & McCorkle R (2005) Dynamics of nutritional health in a community sample of American elders. Advances in Nursing Science 28, 376–389.
Dai Y-T, Yip P-K, Huang G-S & Lou M-F (1999) Cognitive function of elderly patients [Chinese]. Formosan Journal of Medicine 3, 279–286.
Department of Health (2002) The Guideline for Adult Obesity. Available at: http://www.doh.gov.tw/statistic (accessed 25 January 2006).
Dionne I, Despres JP, Bouchard C & Tremblay A (1999) Gender difference in the effect of body composition on energy metabolism. International Journal of Obesity 23, 312–319.
Duncan GJ, Daly MC, McDonough P & Williams DR (2002) Optimal indicators of socioeconomic status for health research. American Journal of Public Health 92, 1151–1157.
Fields DA & Hunter GR (2004) Monitoring body fat in the elderly: application of air-displacement plethysmography. Current Opi-nion in Clinical Nutrition and Metabolic Care 7, 11–14. Folstein MF, Folstein SE & McHugh PR (1975) ‘Mini-mental
state’ a practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research 12, 189–198.
Gerety MB, Williams JW, Mulrow CD, Cornell JE, Kadri AA, Rosenberg J (1994) Performance of case-finding tools for depres-sion in the nursing home: influence of clinical and functional characteristics and selection of optimal threshold scores. Journal of the American Geriatrics Society 42, 1103–1109.
Griep MI, Mets TF, Collys K, Ponjaert-Kristoffersen I & Massart DL (2000) Risk of malnutrition in retirement homes elderly persons measured by the mini-nutritional assessment. The Journal of Gerontology: Medical Sciences 55A, M57–M63.
de Groot CPGM, van Staveren WA & de Graaf C (2000) Determi-nants of macronutrient intake in elderly people. European Journal of Clinical Nutrition 54(Suppl. 3), S70–S76.
Guigoz Y, Vellas B & Garry PJ (1996) Assessing the nutritional status of the elderly: the mini nutritional assessment as part of the ger-iatric evaluation. Nutrition Review 54, S59–S65.
Guigoz Y, Lauque S & Vellas BJ (2002) Identifying the elderly at risk for malnutrition: the mini-nutritional assessment. Clinics in Geriatric Medicine 18, 737–757.
Guo N-W, Liu H-C, Wong P-F, Liao K-K, Yan S-H, Lin K-O, Chang C-Y & Hsu T-C (1988) Chinese version and norms of the mini-mental state examination [Chinese]. Journal of Rehabilitation Medicine Association 16, 52–59.
Guthrie JF & Lin BH (2002) Overview of the diets of lower-and higher-income elderly and their food assistance options. Journal of Nutrition Education and Behavior 34(Suppl.1), S31–S41. Hanlon JT, Schmader KE, Rudy CM & Weinberger M (2001)
Sub-optimal prescripting in older inpatients and outpatients. Journal of American Geriatrics Society 49, 200–209.
Hirschfeld RM, Keller MB, Panico S, Arons BS, Barlow D & Davidoff F (1997) The national depressive and manic-depressive association consensus statement on the undertreatment of depression. Journal of American Medical Association 277, 333–340.
Hops H (1996) Intergenerational transmission of depressive symp-toms: gender and developmental considerations. In Interpersonal Factors in the Origin and Course of Affective Disorders (Mundt C, Goldstein MJ, Hahlweg K & Fiedler P eds). Gaskell, London, pp. 113–129.
Huffman GB (2002) Evaluating and treating unintentional weight loss in the elderly. American Family Physician 65, 640– 650.
Inouye SK, Bogardus ST, Baker DI, Leo-Summers LM & Cooney LM (2000) The hospital elder life program: a model of care to prevent cognitive and functional decline in older hospitali-zed patients. Journal of American Geriatrics Society 48, 1697– 1706.
Institute of Medicine (2000) The Role of Nutrition in Maintaining Health in the Nation’s Elderly. National Academia Press, Wash-ington, DC.
Kleinbaum DG, Kupper LL, Muller KE & Nizam A (1998) Applied Regression Analysis and Other Multivariable Methods, 3rd edn. Duxbury Press, Pacific Grove, CA.
Knight MM (2000) Cognitive ability and functional status. Journal of Advanced Nursing 31, 1459–1468.
Liu L, Bopp MM, Roberson PK & Sullivan DH (2002) Under-nutrition and risk of mortality in elderly patients within one year of hospital discharge. Journal of Gerontology: Medical Sciences 57, M741–M746.
Locher JL, Ritchie CS, Roth DL, Baker PS, Bodner EV & Allman RM (2005) Social isolation, support and capital and nutritional risk in an older sample: ethnic and gender differences. Social Science and Medicine 60, 747–761.
Mahoney FI & Barthel DW (1965) Functional evaluation: the Barthel Index. Maryland State Medical Journal 14, 61–65.
Payette H, Gray-Donald K, Cyr R & Boutier V (1995) Predictors of dietary intake in a functionally dependent elderly population in the community. American Journal of Public Health 85, 677– 683.
Quandt SA & Chao D (2000) Gender differences in nutritional risk in among older rural adults. Journal of Applied Gerontology 19, 138–150.
Sainsbury A, Seebass G, Bansal A & Young JB (2005) Reliability of the Barthel Index when used with older people. Age and Ageing 34, 228–232.
Sarason I, Sarason B, Shearin E & Pierce GR (1987) A brief measure of social support: practical and theoretical implications. Journal of Social and Personal Relationships 4, 497–510.
Sarna L, Lindsey AM, Dean H, Brecht M & McCorkle R (1993) Nutritional intake, weight changes, symptoms distress and func-tional status over time in adults with lung cancer. Oncology Nursing Forum 20, 481–489.
Sullivan DH, Sun S & Walls RC (1999) Protein-energy under-nutrition among elderly hospitalized patients: a prospective study. Journal of American Medical Association 281, 2013– 2019.
Thomas DR (2002) Focus on functional decline in hospitalized older adults. Journal of Gerontology: Medical Sciences 57A, M567– M568.
Unosson M, Ek AC, Bjurulf P & Larsson J (1991) Demographical, sociomedical and physical characteristics in relation to malnutri-tion in geriatric patients. Journal of Advanced Nursing 16, 1406– 1412.
Valentijn SA, van Boxtel MP, van Hooren SA, Bosma H, Beckers HJ, Ponds RW & Jolles J (2005) Change in sensory functioning predicts change in cognitive functioning: results from a 6-year follow-up in the Maastricht aging study. Journal of the American Geriatrics Society 53, 374–380.
Van Nes MC, Herrmann FR, Gold G, Michel JP & Rizzoli R (2001) Does the mini nutritional assessment predict hospitalization outcomes in older people? Age and Ageing 30, 221–226. Walston J (2004) Frailty – The search for underlying causes. Science
of Aging Knowledge Environment 2004, pe4.
Wanich CK, Sullivan-Marx EM, Gottlieb GL & Johnson JC (1992) Functional status outcomes of nursing intervention in hospitalized elderly. Image: Journal of Nursing Scholarship 24, 201–207.
Weyers AM, Mazzetti SA, Love DM, Gomez AL, Kraemer WJ & Volek JS (2002) Comparison of methods for assessing body composition changes during weight loss. Medicine and Science in Sports and Exercise 34, 497–502.
Wilson MMG, Vaswani S, Liu D, Morley JE & Miller DK (1998) Prevalence and causes of undernutrition in medical outpatients. American Journal of Medicine 104, 56–63.
Wong MC, Liu JK & Lo EC (2002a) Translation and validation of the Chinese version of GOHAI. Journal of Public Health Dentistry 62, 78–83.
Wong MTP, Ho TP, Ho MY, Yu CS, Wong YH & Lee SY (2002b) Development and inter-rater reliability of a standardized verbal
instruction manual for the Chinese Geriatric Depression Scale – Short Form. International Journal of Geriatric Psychiatry 17, 459–463. World Health Organization (2001) The Status of Women’s Health:
Western Pacific Region. World Health Organization Publication, Manila, Philippines.
Yesavage JA, Brink TL & Rose TL (1983) Development and vali-dation of a geriatric depression scale. Journal of Psychiatric Research 17, 31–49.
