Estimation of Calorie and Protein Intake in Aged Patients: Validation of a Method Based on Meal Portions Consumed

Transcription

1 Journal of Gerontology: MEDICAL SCIENCES 2002, Vol. 57A, No. 1, M52 M56 Copyright 2002 by The Gerontological Society of America Estimation of Calorie and Protein Intake in Aged Patients: Validation of a Method Based on Meal Portions Consumed Gilles Berrut, 1 Anne Marie Favreau, 2 Emmanuela Dizo, 2 Beatrice Tharreau, 2,3 Corrine Poupin, 2,3 Michel Gueringuili, 4 Phillipe Fressinaud, 3 and Patrick Ritz 3,5 1 Unité de Gérontologie Clinique, 2 Service de Diététique, and 3 Service de Médecine B, Centre Hospitalier Universitaire, Angers, France. 4 Service de Médecine Hôpital d Ancenis, and 5 Inserm EMI-U 00.18, Angers, France. Background. Malnutrition is highly prevalent in hospital, mainly in geriatric, wards. Weight loss results from a negative energy balance, a situation where energy intake does not match energy requirements. Estimates of patient calorie consumption are not performed routinely because of technical difficulties. We performed three studies to investigate the meal-portion (MP) method as a tool for estimating calorie and protein intakes in clinical situations. Methods. The MP method was designed to estimate calorie and protein consumption from the portion of the food items actually eaten by the patient, which is evaluated at the time plates and dishes are cleared away. Study 1 tested accuracy of the MP method in 50 meals by comparison to food weighing. Study 2 evaluated the validity of estimates obtained by a physician, a member of nursing staff, and a dietician in 30 elderly patients. Study 3 evaluated the robustness and feasibility of the method by comparing estimates obtained by nursing staff (after 1 year of practice with no additional training) and that of a dietician. Results. Comparison of estimates and true values (obtained by weighing) showed a mean difference of 2 kcal/ 0.8 g of protein from evaluations of one-half portions of food (50 meals) and 7 kcal/ 1.0 g of protein from onequarter portions of food; the difference was only significant for protein and one-quarter portions (p.03). When evaluations were performed by observers of different professional categories (nursing staff, physicians, and dieticians) on actual meals consumed by 30 elderly people afflicted with disease, no statistical differences were shown. This interobserver agreement remained, regardless of the cognitive or physical status of the patient. A third study, performed after 1 year of no additional training, showed that the MP method is robust, but prone to clerical errors. Conclusions. Valid estimates of calorie and protein consumption can be obtained with the MP method, quoting in one-half portions. Quality controls are required both at the food production site (to avoid propagation of errors arising from food composition) and in data collection (to eliminate clerical mistakes). These results suggest that the MP method could be a tool for estimating calorie and protein intakes in many clinical situations. M ALNUTRITION is a major health care problem, especially in elderly persons afflicted with disease staying in institutions or hospitals (1). In France, it is estimated that fewer than 5% of elderly people living at home are malnourished, while this percentage rises to 40% to 60% in those who are hospitalized (2). Consequences of malnutrition, in terms of morbidity and mortality (1) and hospital expenditure (3), are well known. Whatever the mechanism(s) responsible for weight loss, malnutrition results from an energy intake that is inadequate to compensate for energy requirements (4), regardless of the energy expenditure level. Anorexia is frequently associated with disease and probably plays an important role. In other clinical situations, such as with wound healing (pressure sores, burns, etc.), protein requirements are increased but seldom met by protein intake. An estimate of protein and calorie intake is therefore required in malnourished patients and patients at risk of malnutrition. Precise measurements (weighing food consumed or recall of food intake over several days) cannot be obtained in all patients for logistical reasons (5). It would be helpful if there was an available method that met the following criteria: simplicity (possible use in all patients), precision and robustness (capable of being used even by those with minimal training), and accuracy. A method based on the estimation of the portions consumed in each part of a meal (meal portions [MP]) has been already described by Ödlund and colleagues (6) and Bourdel-Marchasson and colleagues (7), but not validated. In this method, when plates and dishes are being cleared away, the nursing staff evaluates the portion of the meal that has been eaten. The estimate can be made using one-half or one-quarter portions, a procedure originally designed for the estimate of food intake. If written food preparation procedures are available, and the quantity of ingredients used in the preparations is known, macronutrient and caloric content can then be calculated. The combination of portion estimates and known preparation compositions permits an estimate of calorie and protein intake. This method is therefore very simple and requires only a limited amount of training. The present study is aimed at testing the validity of the MP M52

2 FOOD INTAKE IN ELDERLY PERSONS WITH DISEASE M53 method in terms of its precision, accuracy, and robustness. Three studies are reported with the following objectives: Study 1, meals mimicking a reduced food intake were prepared in which the evaluation of calorie consumption (CC) and protein consumption (PC) with the MP method was compared with the true CC and PC; Study 2, staff from different backgrounds (nursing, dietetics, or medical) were compared in their evaluations; and, in Study 3, consistency of the estimate (via the MP method) was assessed after 1 year of using the method without additional training. METHODS Study 1 The aim of Study 1 was to test the validity of CC and PC estimated by the MP method, by comparing the results with those derived from direct food weighing. The second aim was to estimate the superiority of either the one-half- or one-quarter-portion method. Fifty meals were prepared that mimicked the reduced food intake observed in elderly persons afflicted with disease. Complete meals were cooked in the kitchen of the Centre Hospitalier Universitaire in Angers, France, and laid on similar trays. Written procedures and recipes were available, so that almost the exact quantity of ingredients used was known. Each meal was composed of a starter (uncooked vegetables, soup, or pork); a main course with meat, eggs, or fish, accompanied by a carbohydrate or vegetables; a dairy product; fruit; and approximately 50 g ca of bread. Each course of every meal was weighed. A senior dietician (A.M.F.) removed part of each course to simulate consumption by the elderly participants, so that the 50 meals ranged between 67 kcal/1 g of protein and 590 kcal/26 g of protein (the mean composition of the meals was 860 kcal/32 g of protein). The MP Method The MP method is described in Figure 1. Meals in French hospitals traditionally comprise a starter (or a soup), a main course (meat or fish plus vegetables or carbohydrates), a dairy product, a dessert, and a bread roll. The MP method consists of evaluating the portions consumed. On visualizing the food item as being in four parts, the quotations are as follows: 0 if less than a one-quarter of the item is consumed (i.e., between three quarters and the whole item remains in the dish); 1/2 if between one quarter and three quarters of the food item is consumed; and 1 if more than three quarters of the item is consumed. The same principle is applied for the quotation in one-quarter portions. For each meal item, a quotation was ascribed by two independent observers. The first observer quoted in one-half-portion meals 1 to 25, and in one-quarter-portion meals 26 to 50. The second observer did the reverse (one-quarter portions for meals 1 to 25, and so on). The calorie and protein contents of the meals were calculated from the food composition tables (8) and the recipes used. The portions (0, 1/4, 1/2, 3/4, and 1), multiplied by the true content, yielded the calorie and protein content after estimation with the MP method. Figure 1. An example of the decription of the meal-portion (MP) method. Food items shown on tray (top left) are as follows: the starter is celery (no. 1 on bottom panel, none of which was eaten, hence, quoted 0 in the grid on the top right panel); meat (no. 2, almost all was eaten, quoted as 1 ); couscous is the carbohydrate (no. 3, quoted as 1 ); the dairy product is cheese (no. 4, half of which was eaten, quoted as 1/2 ); dessert is an orange (no. 5, quoted as 1/2 ); the bread roll (no. 6, quoted as 1/2 ); and the orange peels are no. 7. On the grid, there is room for comments (e.g., the patient fasted because of a computed-tomography scan in the afternoon). From the quotations in the grid, the recipes, and the food composition table, it is possible to calculate energy and protein intake. Study 2 This study compared the estimates obtained by staff of different backgrounds (nursing staff, dieticians, and physicians). This study was performed in a clinical ward, where 30 elderly patients afflicted with disease, aged 70 years and older, were recruited in consecutive order of admission There were no exclusion criteria. For each participant, a breakfast and lunch were evaluated, and the one-half-portions method was chosen. The different staff did not share information concerning their estimates. The medical and nursing staffs were trained by a dietician, during a 2-hour session. Each participant was evaluated by a Mini-Mental State Examination (MMSE) (9), an index of activities in daily living (10), and a measurement of plasma albumin. The calorie and protein consumptions were evaluated as previously described in Study 1. Study 3 This study tested the robustness and feasibility of the MP method in the medical ward after 1 year with no training, other than that provided in Study 2. Estimates obtained by the nursing staff were compared with those acquired by a dietician (considered as the reference). For 1 month, all patients in the clinical ward had their meals (including snacks) evaluated by the one-half-portions method. One meal per patient was evaluated each day (e.g., breakfast on Mondays, lunch on Tuesdays, and so on). Forty-nine patients partici-

3 M54 BERRUT ET AL. pated. Nursing staffs were not informed of the aim of the study. RESULTS Statistics Results are expressed as means SD, because all data were normally distributed (not shown). A comparison of means was performed by Student s t test or analysis of variance (ANOVA), where applicable. The agreement between methods was tested according to the Bland and Altman method (11). Correlations were sought with a Spearman rank test. Significance was accepted at the.05 level. Study 1 Analyzed meals represented kcal and 13 5 g of protein. Table 1 displays the CC and PC estimates by MP methods (one-half and one-quarter portions). CC estimates did not differ significantly from true values obtained by weighing. PC estimated by one-half portions did not differ from the true value, but PC estimated by one-quarter portions differed significantly from the true value (p.03). Estimates obtained by the one-half and one-quarter portions did not differ between them, both for CC (p.69) and PC (p.76). Figure 2 shows the agreement between estimates of CC by one-quarter or one-half portions and those obtained by weighing. The mean difference is 2 kcal for the one-half portions (p not significant [NS]; 95% confidence limits [CLs], 4, 1 kcal) and 7 kcal for the one-quarter portions (p NS; 95% CLs, 9, 4 kcal). Residuals were evenly distributed between extremes of CC, and there was no significant correlation between residuals and mean CC (R 2.04, p NS for one-half portions, and R , p NS for one-quarter portions, respectively). Figure 3 displays the agreement between estimates of PC by quarter and half portioning and those obtained by weighing. The mean difference was 0.8 g for the one-half portions (p NS; 95% CLs, 1.3, 0.2 g) and 1.0 g for the one-quarter portions (p.03; 95% CLs, 1.4, 0.5 g). Residuals were evenly distributed between extremes of PC and were not correlated with mean PC values (R 2.05, p NS for one-half portions, and R 2.016, p NS for onequarter portions, respectively). Table 1. Calorie (CC) and Protein (PC) Consumption Estimated Using the One-half and One-quarter Portions Method, Compared With True Values (via Food Weighing) in 50 Prepared Meals Mean SD p Value True CC, kcal One-half portion CC, kcal One-quarter portion CC, kcal True PC, g 13 5 One-half portion PC, g One-quarter portion PC, g Notes: Results are from Study 1 (see text for methods). P comparison to true values (Student s t test). Figure 2. Agreement between estimates of calorie consumption (meal-portion method) and true values (food weighing) according to Bland and Altman (11). A, one-half portions; B, one-quarter portions. Study 2 Thirty elderly patients, aged 84 6 years, were recruited. Eighteen were living at home before admission to the hospital. Their mean body mass index (BMI) was kg/m 2 and plasma albumin was g/l. The participants scored an average of 4 (range, 2 6) on the activitiesof-daily-living (ADL) scale and 19 (range, 5 29) on the MMSE. For each participant, a coefficient of variation (CV) was calculated from the three estimates of CC or PC obtained by the different staffs. CVs were not correlated with age, BMI, MMSE score, ADL index, or plasma albumin. Estimates of CC for breakfast and lunch did not differ between nursing staffs ( kcal), dieticians ( kcal), and medical personnel ( kcal; ANOVA, p.69). The same was true for PC (respectively, g, g, and g; ANOVA, p.68). Study 3 Forty-nine patients were studied during an average stay of 7 days (range, 1 19 days). Therefore, a similar number of breakfasts, lunches, afternoon snacks, and dinners were evaluated both by the nursing staff and a dietician (E.D.). Table 2 shows the number of cases where observers disagreed in their portion quotations. Disagreements were more frequent for lunch and dinner and involved all possible items in each meal. For breakfast, quotation disagreements arose in 5 out of 12 cases, whereas the evaluation was missing in 7 other cases (clerical mistakes). For lunch, er-

4 FOOD INTAKE IN ELDERLY PERSONS WITH DISEASE M55 Figure 3. Agreement between estimates of protein consumption (meal-portion method) and true values (food weighing) according to Bland and Altman (11). A, one-half portions; B, one-quarter portions. rors arose primarily from missed evaluations or from quoting items that were not actually given (60 of 85 discordances). However, the number of true discordances (quoting portions differently) was limited (25 of the 298 estimates). The same profile was observed for the dinners. DISCUSSION The present study concerned the validity of estimating CC and PC by a simple method, the MP method, which uses quotations in quarters or halves of the portion of food items consumed by patients. When compared with a direct weighing method, CC and PC were accurate, although the CP that was estimated from the one-quarter portions showed a small, although significant, bias. It appears that estimates using the one-half-portion method are robust and reproducible across different personnel categories. Table 2. Agreement Between Estimates Carried Out by Nursing Staff and a Dietician With the One-half Portion Method No. Meals No. Items No. Disagreements Breakfast Lunch Snack Dinner Notes: No. meals total number of meals studied; no. items total number of food items quoted; no. disagreements number of evaluations differing between the two observers. The main objective of this study was to show that a very simple tool can be used for the estimate of CC and PC in elderly patients afflicted with disease. These patients are at a high risk of malnutrition (1), and reduced energy intake plays an important role. Study 1 has addressed the validity of the MP method over a range of reduced CCs ( kcal), and results were very close to those observed in the clinical ward ( kcal for breakfast and lunch). Validity was tested against the exact weighing of food with PC and CC being derived from food composition tables (8). Estimates of CC are accurate to 2 kcal (i.e., 0.6%) for onehalf-portions methodology or to 7 kcal (2.1%) in the onequarter-portions methodology. Precision of the method is good, as judged from the 95% CLs. This is remarkable in view of the simplicity of the procedure. The one-quarter-portions method was seen to induce a small but significant bias in PC estimates. Because the onehalf-portions method was simpler and probably faster to quote, we performed Studies 2 and 3 using that method. It is interesting to note that estimates of CC and PC are independent of the clinical status of the patients, or of their autonomy and cognitive performance. The MP method is therefore adaptable to any elderly patient, whether at home, in a hospital, or in a nursing home. Two more points are worth future development. The first concerns the correspondence between portions and calories or protein. It assumes excellent control over the procedures involved in food production. Written procedures and recipes are needed, and quality controls must be performed regularly, both of which are important features of the International Organization for Standardization 9002 certification. Future developments are required to improve the ease with which portions can be converted into nutritional information, so as they may be implemented in a large number of patients. The second point relates to the validity of CC and PC estimates where values are closer to recommended intakes. The present study cannot address this point because the study s aim focused on reduced energy and protein intakes. However, the absence of a correlation between residual and mean values in the Bland and Altman (11) analyses suggests that accuracy and precision might not deteriorate in higher values. Estimates of CC and PC were very similar between observers of different professional backgrounds. This provides strong support for the general use of the MP method, after initial training that outlines a clear definition of the portions. This training must be continued as shown in Study 3. Very few errors are observed in simple meals (breakfast or snacks) where food items are easy to evaluate (bread rolls, number of cakes, and standardized butter or jam portions). More errors arise, however, in complex meals (lunch and dinner) where portions are not standardized. It is worth noting that most errors are clerical (i.e., observers quoting food items that were not distributed and forgetting to quote others). It is possible that the implementation of a quality-control procedure in this data collection would result in an improvement. Malnutrition is a major health care problem in elderly persons afflicted with disease and is responsible for increased morbidity and mortality (1) and significant hospital

5 M56 BERRUT ET AL. expenditure (3). No tool has proved efficient in screening patients at risk of malnutrition. Reduced energy intake is a determinant of weight loss and is correlated with later complications (12). The merit of the MP method for screening patients at risk of malnutrition remains to be established. In conclusion, estimates of CC and PC derived by the MP method appear to be valid in situations of reduced intake, such as in malnourished elderly persons afflicted with disease. The method is robust and reproducible between observers. Quality-control procedures must be implemented both in the food production and in the method itself in order to optimize the estimates. The present study suggests that the MP method could be carried out on a large scale. Acknowledgments We thank the nursing staff for their collaboration, Line Godiveau for her secretarial assistance, and Miriam Ryan for correcting the English. Address correspondence to Prof. P. Ritz, MD, PhD, Service de Médecine B, Centre Hospitalier Universitaire, F Angers, Cedex 1, France. patrick.ritz@wanadoo.fr References 1. Morley JE. Protein-energy malnutrition in older subjects. Proc Nutr Soc. 1998;57: Collectif Inserm. Carences nutritionnelles. Etiologie et dépistage. Les éditions INSERM. Paris: Inserm; Tucker HN, Miguel SC. Cost containment through nutrition intervention. Nutr Rev. 1996;54: Toth MJ. Energy expenditure in wasting diseases: current concepts and measurement techniques. Curr Opin Clin Nutr Metab Care. 1999; 2: Barnes KE, Hodkinson HM. Quantification of dietary intake in longstay geriatric patients: do we need seven days observation? Eur J Clin Nutr. 1988;42: Ödlund OA, Osterbert P, Hadell K, et al. Energy enriched hospital food to improve energy intake in elderly patients [abstract]. JPEN J Parenter Enteral Nutr. 1996;20: Bourdel-Marchasson I, Barateau M, Rondeau V, et al. A multi-center trial of the effects of oral nutritional supplementation in critically ill older inpatients. Nutrition. 2000;16: Feinberg M, Favier JC, Ireland-Ripert J, eds. Repertoire Général des Aliments. Paris: Lavoisier; Folstein MF, Folstein SE, Mc Hugh PR. Mini Mental State: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12: Katz S, Down TD, Cash HR, Grotz RC. Progress in development of the index of ADL. Gerontologist. 1970;10: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1: Sullivan DH. The utility of an admission assessment to predict in-hospital nutrient intake. J Am Geriatr Soc. 1994;42: Received September 27, 2000 Accepted March 6, 2001 Decision Editor: John E. Morley, MB, BCh