Mini Nutritional Assessment in geriatric rehabilitation: inter-rater reliability and relationship to body composition and nutritional biochemistry.

Abstract

Aim: To determine the inter-rater reliability of the Mini Nutritional Assessment (MNA) and relationship with body composition and nutritional biochemistry among older Australians undergoing rehabilitation.

Methods: Thirty-eight adults aged [greater than or equal to]65 years were prospectively and consecutively recruited from an Australian rehabilitation ward. Two dietitians independently administered the 18-item MNA to determine inter-rater reliability. MNA classifications (well nourished, at risk of malnutrition, malnourished) were compared with body composition (using dual-energy X-ray absorptiometry) and serum albumin. These analyses were also performed for the short-form version of the MNA (six items).

Results: In this cross-sectional study, inter-rater reliability of the 18-item MNA score, estimated by the intraclass correlation coefficient, was 0.833, while inter-rater reliability estimated by the weighted kappa index was 0.53. The two raters reached agreement on MNA classification for 26 of 38 cases (68%). Women classified as malnourished/at risk of malnutrition using the 18-item MNA had lower total body fat (11 kg vs 29 kg, P < 0.01) and per cent body fat (25% vs 40%, P < 0.01), compared with women classified as well nourished. Similar findings were not apparent for men, although men classified as malnourished/at risk of malnutrition had lower serum albumin (32 g/L vs 36 g/L, P = 0.04) compared with men classified as well nourished. Similar findings were evident for the short-form version of the MNA.

Conclusion: The MNA was found to be useful for identifying older women with lower body fat in the Australian rehabilitation setting. The 18-item MNA score has substantial inter-rater reliability, and fair inter-rater reliability when used according to the classifications. Inclusion of subjective and self-reported items in surveys can be problematic for optimal reliability as can the use of such items in a subject population that is experiencing rapid progress in recovery.

Key words: aged, body composition, Mini Nutritional Assessment.

INTRODUCTION

Poor nutrition is common among older adults undergoing rehabilitation, with reports of prevalence between 14% and 63%. (1-4) Older adults who are malnourished on admission to rehabilitation often have poorer morbidity and mortality outcomes than those who are well nourished. (5,6)

Malnutrition often goes unrecognised due in part to a lack of standardised diagnostic criteria. (7-9) To enable the treatment of malnutrition, rapid and effective methods of early identification and evaluation of treatment are required that include evaluation of dietary intake, anthropometric measurements and detection of nutritional risk factors.

The Mini Nutritional Assessment (MNA) is a simple, rapidly administered tool designed to evaluate the potential risk of malnutrition of older adults in the hospital setting. (10,11) The development of the MNA was a collaborative effort between investigators from France, the USA and Switzerland in the late 1990s. The MNA incorporates a screening section and an assessment section. The short-form MNA (screening section) consists of six items and classifies individuals as not at risk of malnutrition or possibly malnourished. For individuals classified as possibly malnourished, the complete MNA (assessment section) is required. The complete MNA (18 items), which also incorporates the items of the short-form MNA, assesses anthropometry, dietary intake, global assessment and subjective assessment, and classifies individuals as well nourished, at risk of malnutrition or malnourished. The comprehensive nutrition screening and assessment tool has been tested among a diverse collection of older adults, including the active healthy (10) and inpatients of acute care, (12) and has been used extensively in these settings and in rehabilitation facilities. (2,4,13)

The complete MNA has been shown to have a high level of inter-rater reliability among older adults in a Spanish residential care facility. (14) Guigoz et al. demonstrated high sensitivity (96%) and specificity (98%) of the MNA in a sample of 600 men and women aged 60-90 years from France, Mexico and Spain in residential care when compared with a 'clinical status assessment' (nutritional assessment performed by two independent, trained clinicians). (10)

Because there is currently no 'gold standard' measure of nutritional status, it is important that findings of clinical tools demonstrate a consistent and relatively strong relationship to measures that have traditionally represented nutritional status such as body composition and nutritional biochemistry. (15,16) One of two studies has demonstrated such a relationship when a comparison of the complete MNA was made with a direct measure of body composition, dual-energy X-ray absorptiometry (DXA). Persson et al. examined 60 women and found that those classified by the complete MNA as having protein energy malnutrition, or being at risk of developing protein energy malnutrition, had significantly lower total body fat and per cent body fat than those classified as well nourished. (17) In the study performed by Persson et al., total body fat was reported as 12 kg for those classified as having protein energy malnutrition, and this was significantly different from both those at risk of protein energy malnutrition (23 kg, P < 0.05) and those classified as well nourished (29 kg, P < 0.05). (17) Similar findings were observed for comparisons between groups for per cent fat mass. (17) Per cent fat mass was reported as 34% for those classified as having protein energy malnutrition, and this was significantly different from both those at risk of protein energy malnutrition (47%, P < 0.05) and those classified as well nourished (44%, P < 0.05). Chumlea et al. investigated the body composition of 61 community-dwelling male and female volunteers aged 20-78 years, and found no difference in body composition according to the classifications of the MNA for either gender. (18)

Given the apparent broad applicability, overall feasibility and increasing popularity of the MNA, it is important to determine whether the MNA is reliable in a rehabilitation context, and whether the findings are consistent with routinely used measures of nutritional status. It is also important to determine these parameters in the Australian setting.

METHODS

All patients admitted for rehabilitation in a 15-bed rehabilitation ward at the Repatriation General Hospital, Adelaide, South Australia, between May and September 2003 (n = 80) were considered for participation. Identification of new admissions to the rehabilitation ward was via electronic notification from admitting staff and confirmed via regular visits by the first author (SAN) to the ward. Patients were excluded if they were: younger than 65 years, unable to provide written informed consent, terminally ill, medically unstable, unable to be recruited within four days of admission, or previously admitted during the study period. The Repatriation General Hospital Research and Ethics Committee approved the study protocol.

Two dietitians (SAN and MDM) who are familiar with the use of the complete MNA in the clinical setting performed the MNA independently and blind to previous nutritional assessments and results of DXA. The short-form MNA score was calculated as the sum of items A through F, with a score of [greater than or equal to]12/14 indicating no risk of malnutrition and a score of <12/14 indicating possible malnutrition. The complete MNA was administered to all participants, including those who were found not to be at risk of malnutrition by the short-form MNA. The complete MNA was calculated as the sum of items A through R, with a score [greater than or equal to]24/30 indicating that the participant is well nourished, 17-23.5/30 at risk of malnutrition and <17/30 malnourished. (10)

Body weight was measured to the nearest 0.01 kg using a calibrated weigh chair (A & D Mercury Pty Ltd, Adelaide, Australia) with the patient wearing light clothing and no shoes. Due to the practical difficulties in obtaining valid heights in older rehabilitation patients without discomfort, (15) knee height was recorded to the nearest 0.1 cm for all participants using a Ross knee height sliding caliper (Ross Laboratories, Columbus, OH, USA), as described by Chumlea et al. (19) Knee height was converted to height using equations for adults aged 65 years and over, (19) and estimated body mass index (eBMI) was calculated by dividing weight (kg) by height ([m.sup.2]). Mid-arm circumference (MAC) was measured using a steel, metric tape measure (KDS, Tokyo, Japan) midway between the tip of the acromion and the olecranon, and was recorded to the nearest 0.1 cm. Calf circumference was measured at the widest point of the calf muscle using the same steel, metric tape measure and recorded to the nearest 0.1 cm. Both MAC and calf circumference were measured on the right side of the body unless affected by disease or disability.

Body composition was assessed as soon as patients could be booked into the bone densitometry clinic following administration of the MNA. The whole-body scans were performed and analysed by a senior radiographer blind to the findings of the MNA at the Repatriation General Hospital using the Lunar DPX-1-Q model (Lunar Corp, Chalfont St Giles, UK). Participants were required to lie supine on a scan table for approximately 20 minutes while the machine, positioned above the participant, performed a series of transverse scans from head to toe, emitting X-rays at a constant strength. Fat-free mass, total body fat and per cent body fat were determined. Orthopaedic metal was excluded from the DXA analysis of participants with endogenous prostheses. (20)

Serum albumin levels measured from blood samples taken by hospital phlebotomists on admission to the rehabilitation ward were recorded from the hospital's computerised database. The laboratory indicated that values 31-44 g/L were to be considered normal.

Statistical analysis was performed using SPSS version 10.0 software (SPSS for Windows 1999, SPSS Inc., Chicago, IL, USA). Inter-rater reliability was calculated for the complete MNA score, the overall MNA classification and the individual items of the complete MNA. For the complete MNA score, inter-rater reliability was estimated by the intraclass correlation coefficient (ICC) mixed-effects model, while for the MNA classification, weighted kappa index was used. Inter-rater reliability of individual MNA items was estimated by kappa index and weighted kappa index for nominal and ordinal items respectively, according to Cohen. (21) Kappa or weighted kappa indices are classified as 'substantial' (0.81-1.00), 'moderate' (0.61-0.80), 'fair' (0.41-0.60), 'slight' (0.11-0.40) or 'virtually none' (<0.11). (22)

Because of insufficient numbers of participants classified as malnourished by the complete MNA, participants classified by the complete MNA as at risk of malnutrition or malnourished were combined, thereby creating two classifications (well nourished and at risk of malnutrition/malnourished) instead of three. This is common practice in the MNA literature (4,18,23) and is justified given that at-risk patients are likely to require some form of nutrition support to prevent transition to malnutrition. The Mann-Whitney U-test was performed to determine differences in median (interquartile range, IQR) fat-free mass, total body fat, per cent body fat and serum albumin between the two classifications of the short-form MNA and the two classifications of the complete MNA. Men and women were analysed separately due to recognised gender differences in body composition. (24)

Receiver-operating curve analyses were performed to further explore the relationship between MNA and body composition. In the absence of clinically relevant DXA cutoffs, the values for fat-free mass, total body fat and percentage body fat were split according to whether individuals fell above or below the gender-specific median. The area under the curve (95% CI) and the table of coordinates were observed for fat-free mass, total body fat and per cent body fat to determine whether the MNA scores providing the most satisfactory sensitivity and specificity values replicated the MNA cut-offs reported by Guigoz et al. (10)

RESULTS

Thirty-eight of 68 (56%) eligible participants provided informed consent for the reliability study. Four of these participants were excluded from the validity study because they could not lie comfortably in a supine position for measurement of body composition. Median (IQR) age of the 38 participants was 84 (78-88) years, and 18 (47%) were female. There was no significant difference in age between men and women. Reasons for undergoing rehabilitation were: hip and knee replacement (n = 6, 16%), trauma fracture (n = 15, 38%), neurological injury (n = 5, 13%), general deconditioning (n = 11, 30%) or amputation (n = 1, 3%).

Median (IQR) number of days between administration of the first and second MNA was 2 (1-4) days. Using the complete MNA, the first rater (SAN) classified 17 (45%), 19 (50%) and 2 (5%) participants as well nourished, at risk of malnutrition and malnourished, respectively. The second rater (MDM) classified 24 (63%), 13 (34%) and 1 (3%) participant as well nourished, at risk of malnutrition and malnourished, respectively. The two raters reached agreement on MNA classification (well nourished, at risk of malnutrition, malnourished) for 26 of 38 cases (68%), and 82% if use only the well-nourished and combined at-risk/malnourished categories. The inter-rater reliability of the complete MNA score estimated by the ICC mixed-effects model was 0.83 (95% CI 0.68-0.91, P < 0.01), while the inter-rater reliability of MNA classification (malnourished, at risk of malnutrition or well nourished) estimated by the weighted kappa index was 0.53 (Table 1). The inter-rater reliability of individual items estimated by the kappa index and the weighted kappa index for nominal and ordinal items respectively ranged from -0.07 (self-rated nutritional status) to 0.91 (eBMI) (Table 1).

Median (IQR) number of days between administration of the MNA and measurement of body composition was 5 (2-6) days. In most cases, body composition was measured within eight days of administration of the MNA (n = 33), although one participant was transferred to another ward and body composition could not be measured for 34 days because the patient was medically unstable. Endogenous prostheses were present in 26 (77%) participants. Median (IQR) body weight and fat-free mass were significantly higher for men compared with women (weight 73.8 kg (68.5, 81.4) vs 62.4 kg (51.0, 73.8); fat-free mass 49.0 kg (45.3, 52.8) vs 36.3 kg (33.1, 40.8)). Median (IQR) body weight, eBMI, fat-free mass, total body fat and per cent body fat of men classified by the short-form MNA as possibly malnourished was not different from those classified as not at risk of malnutrition. Serum albumin was significantly higher for men classified by the short-form MNA as not at risk of malnutrition compared with those classified as possibly malnourished (P = 0.04). Similarly, body weight, eBMI, fat-free mass, total body fat and per cent body fat of men classified by the complete MNA as at risk of malnutrition or malnourished was not different from those classified as well nourished, while the relationship with serum albumin remained significant (Table 2). All female participants classified as possibly malnourished by the short-form MNA were also classified as at risk of malnutrition or malnourished by the complete MNA, and had significantly lower weight, eBMI, total and per cent body fat compared with those classified as not at risk of malnutrition by the short-form MNA or well nourished by the complete MNA (Table 2).

Total body fat was the only variable that provided a significant better-than-chance result of detecting malnutrition, area under the curve 0.74 (95% CI 0.56-0.91). The coordinate providing the highest sensitivity and specificity was 22.3, closely corresponding to the classification in the literature used to distinguish between individuals who are well nourished in comparison with those who are at risk of malnutrition (MNA score 23.5/30).

DISCUSSION

In a sample of 38 older adults admitted to rehabilitation, 'substantial' inter-rater reliability of the complete MNA score was observed; however, only 'fair' inter-rater reliability was observed for agreement between the two dietitians according to the well-nourished, at-risk and malnourished classifications of the MNA. Items in the areas of anthropometric and global assessment generally performed better than those of subjective and dietetic assessment, which were more subjective. We also found that body composition, according to DXA, was significantly different across the MNA classifications (short-form and complete) in older women admitted for rehabilitation but not in men, while serum albumin was significantly different across the MNA classifications (short-form and complete) for men only.

Inter-rater reliability was higher in the areas of anthropometric and global assessment, and lower in the areas of dietetic and subjective assessment, which is consistent with previous findings by Bleda et al. (14) The kappa indices we obtained for individual items of the MNA were generally lower than those found by Bleda et al., (14) although the items that had low reliability in their study were also low in the present study. This finding highlights the importance of the MNA being administered by assessors with comprehensive training, especially with regards to the method and style of questioning for the subjective and self-reported items. The 'fair' agreement reached between the two dietitians, demonstrated by the kappa of 0.53, should be acknowledged, as the MNA classifications, rather than the total score, are generally the system used for determining treatment. Despite this, there may have only been very small differences in scoring by the two dietitians, with some of the differences perhaps arising due to the subjective questions that may have been influenced by a rapidly improving subject population. This makes sense when the data are investigated further, with the dietitian who completed the MNA second being less likely to score the subject as at risk or malnourished. The short-form MNA items did not demonstrate better reliability than the additional items of the complete MNA, with eight of the 12 complete MNA items resulting in kappa coefficients greater than 0.34 (the lowest kappa coefficient of the short-form MNA items).

Mid-arm circumference and eBMI demonstrated 'substantial' inter-rater reliability, while calf circumference and self-reported weight loss demonstrated poor reliability. It is possible that the length of time between MNA assessments may have contributed to the low reliability of these items. Calf circumference can be altered as a result of fluid changes as postsurgical inflammation and swelling diminishes. Sixteen per cent of our sample were admitted to rehabilitation following a hip or knee replacement. Weight history was not routinely documented in the medical case notes, and therefore, the MNA item relating to recent weight loss was dependent solely on patient recall and, hence, prone to error. Similarly, MNA items requiring patients to rate their nutritional status and health may be influenced from day to day by external factors. To explore whether the length of time between the two MNA assessments (<2 vs [greater than or equal to]2 days) was a contributing factor, we repeated the kappa analyses for those MNA items with below moderate reliability (kappa or weighted kappa [less than or equal to]0.40) (unpublished data, 2003). For those items with sufficient numbers across the available MNA item categories, the findings did not appear to result in any changes in kappa values; however, a larger sample size is required to confirm this finding.

Validation of the complete MNA using body composition as measured by DXA has not previously been reported in the rehabilitation setting, but has been attempted in a sample of 28 men and 60 women aged 65-96 years from a Swedish geriatric inpatient ward (17) and 24 men and 37 women aged 20-78 years living in the community in the USA. (10) Persson et al. found that among the female geriatric inpatients in Sweden, mean [+ or -] SD body fat was significantly lower in those classified by the complete MNA as malnourished (11.8 [+ or -] 8.7 kg) compared with those classified as at risk of malnutrition (22.5 [+ or -] 6.8 kg) or well nourished (28.8 [+ or -] 8.6 kg) (P < 0.05). (17) All of the men in the study by Persson et al. were classified by the MNA as at risk of malnutrition, thereby preventing comparison of body composition between MNA classifications. (17) Chumlea et al. found that body composition did not differ between MNA categories in men or women, despite classification of individuals as both at risk of malnutrition and well nourished, allowing comparison between groups. (18) While the study by Persson et al. was limited to older adults, (17) Chumlea et al. included individuals aged 20-78 years. (18) Interestingly, the most consistent difference in response to the MNA between individuals classified as at risk of malnutrition or malnourished and those classified as well nourished in the study by Chumlea et al. (18) was self-reported recent weight loss. Given that participants were younger, healthy free-living volunteers, weight loss is speculated to most likely have occurred as a result of self-induced dieting rather than disease-related malnutrition.

The results of the present study for women concur with those of Persson et al. (17) While Persson et al. did not find any significant differences in body composition in men according to MNA classification, (17) this was because all men were classified as at risk of malnutrition, preventing comparisons between groups. In the present study, no difference in body composition was found in men according to MNA classification despite use of both MNA categories (11 classified as at risk of malnutrition or malnourished; seven classified as well nourished). This result remained even when total fat mass was adjusted for body weight.

In order to investigate these results further, we compared the results of the short-form MNA, as well as the results of the complete MNA, with body composition to see whether there was a difference in reporting in men and women for the two forms of the MNA. The short-form MNA classified the same number of women as possibly malnourished as the complete MNA classified as at risk of malnutrition or malnourished. One more man was classified as possibly malnourished by the short-form MNA compared with at risk of malnutrition or malnourished by the complete MNA. Thus, in the present study, the short-form MNA generally produced a classification that was consistent with the classification of the complete MNA, providing some evidence to suggest that the short-form MNA can replace the complete MNA if the intent is to identify those who are likely to benefit from nutrition counselling and intervention. Where resources are poor, there may be a need to only identify severely malnourished rehabilitation patients. In this situation, the complete 18-item MNA should be performed.

A possible explanation for the lack of difference found in the body composition of men according to MNA classification, and a limitation of the present study, is the time delay between the MNA assessment and measurement of body composition. The present study is also limited by a lack of reference values available for classifying individuals as adequately, over- or undernourished according to body composition as measured by DXA. While some components of body composition were found to be significantly different between MNA categories, it is unknown whether these differences equate to actual differences in nutritional status. It is also important to acknowledge that measurement of fat and fat-free mass using DXA has been demonstrated to be less accurate in severely malnourished subjects when tissue is close to the bone, (25) although the impact of this is likely to be minimal in the present study, where the mean body mass index was 25.7 kg/[m.sup.2]. The lack of a gold standard for defining malnutrition also limits our ability to confidently diagnose poor nutritional status, and it is possible that use of a single parameter for defining malnutrition may have resulted in misclassification of individuals. The sample size for the present study was also relatively small, but similar in size to some previous work. (12,18) Further studies with larger numbers of participants, particularly men, are required. An increase in the number of participants would also allow the analyses to be performed using the three conventional MNA categories: well nourished, at risk of malnutrition and malnourished.

The unexpected direction of association observed for albumin between women classified as well nourished according to the MNA, compared with those classified as at risk of malnutrition or malnourished, is worthy of note. Although this finding could be related to validity of the MNA, this is unlikely as the MNA classifications demonstrated good validity on all measures of body composition for women. It is more likely a reflection of the inappropriate use of serum albumin as a measure of nutritional status in this clinical group, whereby a large proportion have endured stress caused by surgery or infection that may result in albumin levels fluctuating through mobilisation to the intravascular space. (26)

In conclusion, the MNA appears to have 'fair' inter-rater reliability, when used according to the MNA classifications, among older Australians in the rehabilitation setting. The short-form and complete MNA both, however, provide evidence of a relationship with body composition in older women in the rehabilitation setting. The findings of the present study would suggest that, in the rehabilitation setting, introduction of the short-form MNA as a means for identifying risk of malnutrition is adequate; however, if multiple assessors are to be involved, then comprehensive training and monitoring, and possibly revision of individual items, will be required to achieve adequate reliability.

REFERENCES

1 Beck E, Patch C, Milosaljevic M et al. Implementation of malnutrition screening and assessment by dieticians: malnutrition exists in acute and rehabilitation settings. Aust J Nutr Diet 2001; 58: 92-7.

2 Thomas D, Zdrowski C, Wilson M et al. Malnutrition in subacute care. Am J Clin Nutr 2002; 75: 308-13.

3 Visvanathan R, Penhall R, Chapman I. Nutritional screening of older people in a sub-acute care facility in Australia and its relation to discharge outcomes. Age Ageing 2004; 33: 260-65.

4 Neumann SA, Miller MD, Daniels LA, Crotty M. Nutritional status and clinical outcomes of older patients in rehabilitation. J Hum Nutr Diet 2005; 18: 129-36.

5 Muhlethaler R, Stack A, Minder C, Freg B. The prognostic significance of protein and energy malnutrition in geriatric patients. Age Ageing 1995; 24: 193-7.

6 Compan B, Di Catsri A, Plaze J, Arnaud-Battandier F. Epidemiological study of malnutrition in elderly patients in acute, sub-acute and long-term care using the MNA. J Nutr Health Aging 1999; 3: 146-51.

7 McWhirter JP, Pennington CR. Incidence and recognition of malnutrition in hospital. BMJ 1994; 308: 945-8.

8 Morley JE. Why do physicians fail to recognize and treat malnutrition in older persons. J Am Geriatr Soc 1991; 39: 1139-40.

9 Mowe M, Bohmer T. The prevalence of undiagnosed protein-calorie undernutrition in a population of hospitalized elderly patients. J Am Geriatr Soc 1991; 39: 1089-92.

10 Guigoz Y, Vellas B, Garry P. Mini nutritional assessment: a practical assessment tool for grading the nutritional state of elderly patients. Facts Res Gerontol 1994; 4 (Suppl. 2): 15-60.

11 Guigoz Y, Vellas B, Garry P. Assessing the nutritional status of the elderly: the Mini-Nutritional Assessment as part of the geriatric evaluation. Nutr Rev 1996; 54: S59-65.

12 Murphy M, Brooks C, New S, Lumbers M. The use of the Mini-nutritional assessment (MNA) tool in elderly orthopaedic patients. Eur J Clin Nutr 2000; 54: 555-62.

13 Visvanathan R, Macintosh C, Callary M, Penhall R, Horowitz M, Chapman I. The nutritional status of 250 older Australian recipients of domiciliary care services and its association with outcomes at 12 months. J Am Geriatr Soc 2003; 51: 1007-11.

14 Bleda M, Bolibar I, Pares R, Salva A. Reliability of the Mini Nutritional Assessment (MNA) in institutionalised elderly people. J Nutr Health Aging 2002; 6: 134-7.

15 Omran M, Morley J. Assessment of protein energy malnutrition in older persons, part I: history, examination, body composition, and screening tools. Nutrition 2000; 16: 50-63.

16 Omran M, Morley J. Assessment of protein energy malnutrition in older persons, part II: laboratory evaluation. Nutrition 2000; 16: 131-40.

17 Persson M, Brismar K, Katzarski K, Nordenstrom J, Cederholm T. Nutritional status using Mini Nutritional Assessment and Subjective Global Assessment predict mortality in geriatric patients. J Am Geriatr Soc 2002; 50: 1996-2002.

18 Chumlea C, Hall G, Lilly F, Siervogel R, Guo S. The Mini Nutritional Assessment and body composition in healthy adults. Nestle Nutr Workshop Ser Clin Perform Program 1999; 1: 13-22.

19 Chumlea W, Roche A, Steinbaugh M. Estimating stature from knee height for persons 60-90 years of age. J Am Geriatr Soc 1985; 33: 116-20.

20 Madsen O, Egsmose C, Lorentzen J, Lauridsen U, Sorensen O. Influence of orthopaedic metal and high-density detection on body composition as assessed by dual-energy x-ray absorptiometry. Clin Physiol 1999; 19: 238-45.

21 Cohen A. Coefficient of agreement for nominal scales. Educ Psychol Meas 1960; 19:3-11.

22 Shrout PE. Measurement of reliability and agreement in psychiatry. Stat Methods Med Res 1998; 7: 301-17.

23 Miller D, Perry H, Morley J. Associations among the Mini Nutritional Assessment instrument, dehydration, and functional status among older African Americans in St. Louis, Mo., USA. Nestle Nutr Workshop Ser Clin Perform Program 1999; 1: 79-86.

24 Hansen R, Raja C, Aslani A, Smith R, Allen B. Determination of skeletal muscle and fat-free mass by nuclear and dual-energy absorptiometry methods in men and women aged 51-84 years. Am J Clin Nutr 1999; 70: 228-33.

25 Tothill P, Nord RH. Limitations of dual-energy x-ray absorptiometry. Am J Clin Nutr 1995; 61: 398-400.

26 James WP, Hay AM. Albumin metabolism: effect of the nutritional state and the dietary protein intake. J Clin Invest 1968; 47: 1958-72.

Sonja A. NEUMANN, (1) Michelle D. MILLER, (2) Lynne A. DANIELS, (2) Michael AHERN (3) and Maria CROTTY (1)

Departments of (1) Rehabilitation and Aged Care, (2) Nutrition and Dietetics, and (3) Immunology, Allergy and Arthritis and Centre for Clinical Change and Health Care Research, Flinders University, Adelaide, South Australia, Australia

S.A. Neumann, BNutrDiet Hons, APD, Former Honours Student

M.D. Miller, PhD, APD, Lecturer

L.A. Daniels, PhD, APD, Associate Professor

M. Ahern, PhD, FRACP, Associate Professor

M. Crotty, PhD, FAFRM[RACP], Professor

Correspondence: M.D. Miller, G4 FMC Flats, Flinders Drive, Bedford Park, Adelaide, SA 5042, Australia. Email: [e-mail]

Table 1 Kappa index of agreement (a) for the complete 18-item Mini
Nutritional Assessment (MNA) (10) score and each MNA item (n = 38)

Area                   MNA item                          Kappa (b)

Anthropometric         Estimated body mass index (c)      0.91
  assessment           Mid-arm circumference (c)          0.83
                       Calf circumference                 0.37
                       Weight loss (c)                    0.36
Global assessment      Independence at home               0.87
                       Number of medications per day      0.47
                       Psychological stress               0.45
                       Mobility (c)                       0.34
                       Neuropsychological problems (c)    0.45
                       Pressure sores/skin ulcers         0.38
Dietetic assessment    Number of meals per day (c)        0.50
                       Serves of high-protein foods (c)   0.40
                       Fruit and vegetable intake         0.66
                       Fluid intake (c)                   0.14
                       Mode of feeding                   -0.04
                       Appetite (c)                       0.54
Subjective assessment  Self-rated nutritional status     -0.07
                       Self-rated health (c)              0.27
Total score                                               0.83
MNA classification                                        0.54

(a) Median (IQR) days between repeat MNA measures was 1.5. (1-4)
(b) Kappa or weighted kappa indices are classified as 'substantial'
(0.81-1.00), 'moderate' (0.61-0.80), 'fair' (0.41-0.60), 'slight' (0.11-
0.40) or 'virtually none' (<0.11). (22)
(c) Weighted kappa.
IQR = interquartile range.

Table 2 Median (IQR) body weight, estimated body mass index (eBMI),
serum albumin and body composition as measured by dual energy X-ray
absorptiometry in men and women aged [greater than or equal to]65 years
classified as at risk of malnutrition or malnourished (n = 19) and well
nourished (n = 15) by the complete 18-item Mini Nutritional
Assessment (10) according to the Mann-Whitney U-test of association

                       At risk of
                       malnutrition/
                       malnourished   Well nourished
                       (score < 24/   (score [greater than or equal to]
                       30)            24/30)

Men                    n = 11         n = 7
  Body weight (kg)*    70 (67-79)     78 (70-83)
  eBMI (kg/[m.sup.2])  24 (22-27)     25 (24-29)
  Fat-free mass (kg)*  50 (47-53)     48 (43-53)
  Total body fat (kg)  19 (14-24)     24 (20-28)
  % Body fat           29 (23-36)     30 (28-36)
  Serum albumin (g/L)  32 (27-33)     36 (35-39)
Women                  n = 8          n = 8
  Body weight (kg)*    52 (44-63)     67 (60-78)
  eBMI (kg/[m.sup.2])  21 (18-25)     28 (25-30)
  Fat-free mass (kg)*  36 (33-41)     37 (35-41)
  Total body fat (kg)  11 (7-22)      29 (20-40)
  % Body fat           25 (16-36)     40 (34-46)
  Serum albumin (g/L)  34 (32-38)     30 (29-32)

                       P-value  Total

Men                             n = 18
  Body weight (kg)*     0.21    73 (68-81)
  eBMI (kg/[m.sup.2])   0.13    25 (23-27)
  Fat-free mass (kg)*   0.72    49 (45-53)
  Total body fat (kg)   0.18    20 (15-27)
  % Body fat            0.54    30 (23-36)
  Serum albumin (g/L)   0.01    33 (29-36)
Women                           n = 16
  Body weight (kg)*     0.04    61 (51-71)
  eBMI (kg/[m.sup.2])   0.01    25 (21-29)
  Fat-free mass (kg)*   0.72    36 (33-41)
  Total body fat (kg)  <0.01    21 (10-30)
  % Body fat           <0.01    36 (24-40)
  Serum albumin (g/L)   0.05    32 (29-35)

eBMI: body mass index calculated using stature estimated from knee
height. (19)
* Significant difference across gender for weight (P < 0.01) and fat-
free mass (P < 0.01).
IQR = interquartile range.