Eating Behavior in Continuous Ambulatory Peritoneal Dialysis and Hemodialysis Patients Britta Hylander, MD, PhD, Britta Barkeling, BSe, and Stephan Rossner, MD, PhD • Three groups consisting of 12 subjects each (continuous ambulatory peritoneal dialysis [CAPO] patients, hemodialysis patients, and healthy controls) matched for age, sex, and body weight were invited to a test meal for the study of hunger, fullness, and food preferences. They were served an excess portion of hash served on a plate placed on a hidden scale (UVIKTOR"), which was connected to a computer registering the eating process on-line. The patients filled in visual analogue scales (VAS) concerning appetite and food preferences before and after the test meal. Mean total intake of food (±SO) was significantly higher for healthy controls (357 ± 175 g) compared with hemodialysis patients (295 ± 105 g), which in turn was higher than in CAPO patients (206 ± 70 g). Eating velocity was lower in both dialysis groups compared with controls. CAPO patients experienced less hunger and desire to eat compared with hemodialysis patients and controls. The reason for the low eating drive in CAPO patients despite their great need for protein and calories is unknown, but might be explained by gastric retention, insufficient dialysis, metabolic effects of the high sugar load from the dialysate, or combinations of these factors. © 1992 by the National Kidney Foundation, Inc. INDEX WORDS: Appetite; continuous ambulatory peritoneal dialysis; eating behavior; hemodialysis; universal eating monitor.
P
ROTEIN AND ENERGY malnutrition is common in both continuous ambulatory peritoneal dialysis (CAPO) and hemodialysis patients, and contributes to increased mortality and morbidity.I-3 Malnutrition increases the risk for infections, surgical complications, and in-hospital stay. The dialysis treatment per se is associated with loss of amino acids and proteins to the dialysate. 4 •5 A strong inverse correlation has been found between the serum albumin concentration and mortality.3 Hunger and appetite are low in CAPO patients, 1,6 which can partly be explained by insufficient dialysis. 7,8 Another explanation for poor food intake in CAPO patients has been that the feeling of fullness from the dialysate in the abdomen reduces the drive to eat. The contribution of glucose from the dialysate as a caloric source could also explain the lack of hunger and appetite. 2 ,9 Renal failure patients are often nauseated and may experience altered senses of taste. 10 Furthermore, effects of medication, mental depression, or poor physical activity may all From the Department 0/ Medicine, Section o/Nephrology, and Health Behaviour Research, Karolinska Hospital, Stockholm, Sweden. Received April 23, 1992; accepted in revised/orm July 21, 1992. Supported by the Swedish National Foundation/or Kidney Patients. Address reprint requests to Britta Hylander MD, PhD, Section o/Nephrology, Division o/Medicine, Karolinska Hospital, Box 60500, S-J04 01 Stockholm, Sweden. © 1992 by the National Kidney Foundation, Inc. 0272-6386/92/2006-0011 $3.00/0
592
contribute to reduce appetite and hence food intake in both hemodialysis and CAPO patients. 10 We undertook the following study to compare food intake and eating motivation in a group of dialysis patients and matched healthy controls. PATIENTS AND METHODS Three matched groups of each 12 subjects (eight men, four women) were studied. Both dialysis groups were matched as regards their clinical condition, other concomitant diseases, and medication. No subject in any group was diabetic. Hypertension was diagnosed in all but one patient. No patient had chronic obstructive pulmonary disease. Patients in the CAPO group were studied after having been on CAPO for more than 3 months. The mean time on CAPO was 10 ± 5 (SO) months (range, 3 to 18). Their mean age was 60 ± 14 years (range, 37 to 79). Their initial diagnoses were glomerulonephritis (5), nephrosclerosis (3), pyelonephritis (I), congenital aplasia (I), and others (2). Mean body mass index (BMI) was 23.0 ± 2.8 kg/m 2 (reference value, 20 to 25 kg/m2). The 12 hemodialysis patients had been on hemodialysis for more than 3 months and their mean time on dialysis was 23 ± 28 months (range, 3 to 108). The mean age was 60 ± 13 years. Their initial diagnoses were glomerulonephritis (5), nephrosclerosis (3), cystic disease (2), and others (2). Mean BMI was 22.7 ± 3.4 kg/m2. All patients received hemodialysis treatment three times a week. Six patients were treated with erythropoietin. In the third group of healthy matched controls, the mean age was 60 ± 14 years and the mean BMI was 23.4 ± 1.8 kg/m2. No CAPO patient had experienced peritonitis within 5 months preceding the study. The dialysate volumes varied between 2.0 I and 2.5 1. All patients had four exchanges of dialysate per day. Body weights were noted at both "VIKTOR" test meals on an electric scale.
American Journal of Kidney Diseases, Vol XX, No 6 (December), 1992: pp 592-597
593
EATING BEHAVIOR IN DIALYSIS PATIENTS
The following blood tests were determined in the two patient groups: hemoglobin, serum urea, creatinine, standard bicarbonate, and albumin. In all CAPO patients, these values were obtained within 2 days of the test meal to reflect the average dialysis situation. In hemodialysis patients, the serum urea and creatinine values were obtained after dialysis as a posttreatment value, whereas the albumin concentration was calculated as the mean of the values postdialysis before the test and predialysis after the test. This value was taken to reflect an average estimate of the degree of hydration. Hemoglobin and bicarbonate values were obtained from the blood test before dialysis closest to the test meal.
VIKTOR The lunch intake was measured by VIKTOR, a universal eating monitor of the type first developed by Kissileff et alII and later modified by us. 12·14 The VIKTOR equipment consists of a hidden scale built into a table and connected to a microcomputer, which registers the amount of food eaten from the plate placed on top of the scale. The computer (LUXOR ABC 8(0) is equipped with an analog/digital converter to convert the readings from the scale to the computer. Before each experiment, the whole system is calibrated by an interactive calibrating process with known weights. Aforkholder provides signals to the computer to start or stop data readings when the fork is removed or returned to the holder. As soon as the person under study removes the fork from the fork-holder and starts to eat, a weight reading is stored three times per second for subsequent analysis. The readings are also shown graphically on-line as an eating curve on the computer monitor. All computer programs were written in BASIC. After each test meal, the computer analyzes the eating data and calculates the total intake (grams), duration of consumption (minutes), rate of consumption (g/min), and the relative rate of consumption (1m = the intake of food during the first half of the meal minus the intake of the second half of the meal divided by the total intake of food). 1m values greater than 0.1 indicate a decelerated eating pattern, that is, an eater who slows down the eating rate toward the end of the meal. II This indicates strong internal satiety signals. An 1m value less than 0.1 indicates that the subject is a nondecelerated eater with weaker internal satiety signals.
Eating Motivation Assessments of subjective feelings of hunger, appetite, fullness, prospective consumption, food choice, and food preferences were made by a model system developed by Blundell et ai, translated and adapted to Swedish conditions by US . 14 The assessments are described below. Visual analogue scales. Visual analogue scale (VAS) ratings are made on lOO-mm horizontal scales, by which subjects answer questions with a cross on each scale. The scales are anchored with the description in brackets: "How strong is your desire to eat?" (not strong at all to very, very strong), "How hungry do you feel?" (not hungry at all to very, very hungry), "How full do you feel?" (not full at all to very, very full), " How much food do you think you could eat?" (none to a very big portion). The last question is also referred to as measuring " prospective consumption." The pleasantness of
the meals was checked by using a further VAS at the end of each meal where the subjects rated the question "How pleasant have you found your food?" (not at all pleasant to very, very pleasant). A f ood preference checklist. Subjects tick off the food items out of a list of 32 items they would like to eat immediately. Tbe food items represent four food groups (eight from each group): high-protein, high-fat, high-carbohydrate, and lowenergy items. The lists are used to amplify measures of general motivation (V AS), that is, the more hungry a person is, the more food items he ticks of[ A f orced-choice (food preference) list. SUbjects are forced to chose between 32 pairs of high-carbohydrate and high-protein items. Half of the high-carbohydrate foods are sweet and half are savory. This procedure is designed to reveal a specific high preference for carbohydrates or proteins. Examples of food pairs are roast chicken breast or a baked potato with a small knob of butter; panfried codfiIIet or a slice of jam-filled sponge cake. The list gives a preference quota. The quota is used to determine whether a high (32:0 to 29:3), medium (28: 4 to 24:8), low (23:9 to 19: 13), or no (18: 14 to 14: 18) preference for either macronutrient is present.
Test Meal The meal served was a microwave-heated excess portion of an industrially produced Swedish hash with a standard composition and an energy content of 150 kcalflOO g consisting of diced meat, onions, and potatoes that were mixed and fried (Table 1).
Procedure The participants were informed before coming to the laboratory that ratings of subjective feelings of hunger, fullness, appetite, and food preferences in connection with the meals were to be recorded, but no information of how food intake would be measured was given before the meals. The patients were instructed to eat their ordinary breakfast in their homes between 7:00 and 8:00 AM. The interval between breakfast and lunch was 4 to 5 hours. Patients were instructed not to drink or eat between breakfast and the test lunch. The CAPO patients came for lunch twice to the eating laboratory after breakfast with the same food composition, once with dialysate in their abdomen and once without. This order was at random and unknown to the VIKTOR investigator. The test occasions were about 1 week apart. In CAPO patients, the last dialysate was drained approximately 30 minutes before one of the test meals. Ouring the alternate meal, all patients had standard lactate dialysate (2.0 to 2.5 L) in the abdominal cavity, instilled approximately 30 minutes before the meal. The hemodialysis patients and the healthy controls came once Table 1. Composition of "Swedish Hash Test Meal"
Per 100 9 Energy %
Energy (kcal)
Protein (g)
Fat (g)
150
6 16
43
7
Carbohydrates (g)
15 41
594
HYLANDER, BARKELlNG, AND ROSSNER
for lunch. When the patients arrived at the laboratory, they were given the test meal together with a glass of water (200 mL). They were asked to eat as much as they wanted of the food to pleasant fullness and to drink the water when they wanted throughout the meal. Immediately before and after lunch, repeated VAS, food preference checklists, and forced-choice food preference lists were filled in by the patients.
Data Analysis Statistical analyses of the eating data and motivational ratings between groups were performed using one-way analysis of variance (ANOVA) and appropriate post hoc analyses. Student's t test was used for the within-group analysis. KruskalWallis' one-way ANOVA by ranks was used to evaluate forcedchoice food preference lists and food preference checklists between groups, and Wilcoxon's signed-rank test was used for analysis within groups. When CAPD patients were compared with the other groups, data from the eating occasion with the dialysate-filled abdomen were used. All results are expressed as the mean ± SD if not otherwise indicated. The study was approved by the Karolinska Ethics Committee.
RESULTS
Lunch Meal Intake
Fifteen CAPD patients were invited, but only 12 completed two lunch meals. The three patients who dropped out had been transplanted after the first test meal. All invited hemodialysis patients and healthy controls came for their lunches as scheduled. The total intake of food for CAPD patients with dialysate in the abdomen was 206 ± 70 g and without dialysate 190 ± 76 g (NS). The hemodialysis patients consumed 295 ± 105 g and the controls 357 ± 175 g. Comparison of the intake offood of the three groups with ANOVA showed a significant difference between the groups (P < 0.05). Further analysis indicated that these differences were between controls and CAPD patients (P < 0.05, Fisher). Other variables related to eating behavior, such as duration of consumption, rate of consumption, and relative rate of consumption, are shown in Table 2. There were no significant differences between dialysis groups in duration of consumption and relative rate of consumption. The rate of consumption was significantly different between groups (dialysis and controls) (P < 0.05, ANOVA). These differences were found between controls and CAPD patients (P < 0.05, Fisher) and between controls and hemodialysis patients (P < 0.05, Fisher), but not between CAPD and hemodialysis patients.
VAS
The motivation to eat as measured by VAS differed between groups before the meals (P < 0.05, ANOVA). CAPD patients had less hunger (P < 0.05, Fisher) and less desire to eat (P < 0.05, Fisher) than both controls and hemodialysis patients. Eating lunch caused a large and significant decrease in ratings of desire to eat, hunger, and prospective consumption, and a significant increase in rated fullness in all groups compared with ratings before lunch (paired t test, at least P < 0.01). After the lunch meals, there were no differences between groups in motivation to eat, fullness, etc. The lunch meal was also equally well liked by the groups (Table 3). There was a tendency for CAPD patients to be less hungry and have less desire to eat before lunch with dialysate in the abdomen compared with empty abdomen, but these differences were not statistically significant. After the lunch meals, there were no differences in motivation to eat in CAPD patients with or without dialysate in the abdomen. Food Preference Checklist
Eating lunch caused a significant decrease in the number of item chosen from the food preference list for all groups, compared with before lunch (Wilcoxon's rank test, at least P < 0.01). There were no differences in the number of items chosen between these groups, either before or after the meals. Dialysate in the abdomen did not affect the number of items chosen by CAPD patients. Forced-Choice (Food Preference) List
The forced-choice (food preference) list demonstrated relative preferences for high-protein food items in favor of high-carbohydrate foods before lunch in all groups. The median preference ratio (carbohydrate to protein items) before meals were similar in all groups: 6:26 for CAPD patients with both full and empty abdomen, 6:26 for hemodialysis patients, and 8:24 for controls. After lunch, the relative preferences changed significantly to preferences for high-carbohydrate food items for hemodialysis patients (median ratio, 18: 14) and controls (median ratio, 24:8) (Wilcoxon's rank test, P < 0.05 for both groups), but no
595
EATING BEHAVIOR IN DIALYSIS PATIENTS
Table 2. Characteristics of Food Intake in Dialysis Patients and in Controls CAPO (Full)
Intake of food (g) Duration of consumption (min) Rate of consumption (g/min) Relative rate of consumption (1m)
206 8.9 24 0.09
CAPO (Empty)
± 70 ± 3.4 ±8 ± 0.21
190 8.0 25 0.18
± ± ± ±
Hemodialysis
76 3.0 6 0.16
295 12.4 24 0.09
± ± ± ±
Controls
105 3.6 8 0.13
356 11.0 32 0.14
± ± ± ±
175 4.7 10 0.14
NOTE. Values are means ± SO. See text for statistical analysis.
change was seen for CAPD patients. Their median ratios were 8:24 with full abdomen and 10: 22 with empty abdomen (Table 4). Laboratory Values Mean values for hemoglobin, serum creatinine and serum urea, albumin, and standard bicarbonate are shown in Table 5. DISCUSSION
Uremia is known to be associated with poor appetite and malaise. A study of the eating behavior of these patients may disclose possible dietary mechanisms for the clinically observed low eating drive and the body weight loss during dialysis therapy, and can thus help to design adequate dietary advice. 4 ,15 In our study, there was a significantly lower food consumption in CAPD patients compared with controls, which was also lower than in hemodialysis patients, although of borderline sigTable 3. Motivation to Eat Measured by VAS in Humans
Oesire to eat Before meal After meal Hunger Before meal After meal Fullness Before meal After meal Prospective consumption Before meal After meal Palatability
CAPO (Full)
CAPO (empty)
Hemodialysis
Controls
27 ± 26 4±2
36 ±30 5±3
53 ±26 4±3
53 ± 18 3±2
33± 28 3±2
38± 30 4±2
54 ± 22 4±3
57 ± 23 4±3
36 ± 28 82 ± 15
36 ± 36 84 ± 15
27 ± 22 83± 14
23 ±20 82 ± 21
39 ± 18 6±6 66 ± 24
40 ± 20 8±7 77 ± 24
51 ± 18 11 ± 9 69 ± 29
56 ± 18 15 ± 26 74 ± 22
NOTE. Values are means ± SO. See text for statistical analysis and description of scale anchors.
nificance. Theoretically, an increased intraabdominal pressure by the dialysate could explain this reduced food intake. However, we have previously measured intragastric pressures in CAPD patients with and without dialysate and found no differences in intragastric pressure. 16 Delayed gastric emptying in uremia, in particular in the CAPD patient group, is another alternative that might explain the differences between dialysis patients and controls. We cannot rule out this mechanism as a causal factor. The smaller food intake in CAPD patients might further be explained by the continuous glucose load suppressing hunger and appetite. 9 Normally, food preferences switch from protein to carbohydrates during the VIKTOR test meal. 12 In contrast to controls and hemodialysis patients, this did not happen in CAPD patients, which could reflect an effect of the high continuous carbohydrate load from the dialysate. Low standard bicarbonate levels are associated with poor appetite, but standard bicarbonate levels were 23 ± 1.5 mmol/L in CAPD patients and 22 ± 3.0 mmol/L in hemodialysis patients (normal range, 21 to 26). Poor intake could thus not be explained by acidosis. The differences between CAPD and hemodialysis patients in serum albumin concentrations seem to reflect the fact that sampling conditions were not identical, rather than clinically significant differences between the two paTable 4. Changes in Food Preferences Before and After Meal
Before meal After meal
CAPO (Full)
CAPO (Emtpy)
Hemodialysis
Controls
6:26 8:24
6:26 10:22
6:26 18:14
8:24 24:8
NOTE. Forced-choice (food preferences) between (carbohydrate:protein) items. See text for statistical analysis.
HYLANDER, BARKELlNG, AND ROSSNER
596
Table 5. Laboratory Parameters in CAPO and Hemodialysis Patients
CAPO patients Hemodialysis patients
Hemoglobin (gIL)
Creatinine ("moI/L)
Urea (mmoI/L)
Albumin (gIL)
Standard Bicarbonate (mmol/L)
94 ± 20.4 100.0 ± 16.3
814 ± 279 505 ± 127
20.9 ± 4.4 11.4 ± 2.0
33.5 ± 5.4 41.2 ± 4.3
23 ± 1.5 22 ± 3.0
tient groups. The higher hemoglobin levels in hemodialysis patients could partly be attributed to erythropoietin treatment in six of these patients, but the minor nonsignificant difference could probably not explain their higher food intake and more pronounced appetite. The slow consumption rate in association with low motivation to eat in CAPD patients remains difficult to explain, but might relate to the subjective feeling of fullness either due to delayed emptying from the stomach or to the dialysate despite unchanged intragastric pressures. The finding of an equally slow rate of consumption in hemodialysis patients implies that the uremic status per se contributes to the low motivation to eat. The loss of hunger and appetite and the development of anorexia during dialysis have numerous possible explanations such as inadequate dialysis, unpalatable or inadequate diets, medications, psychosocial factors, acidosis, depression, low physical activity, etc. 17- 19 At the
time of this study, Kt/V analyses were not regularly included in the clinical evaluation of these patients. We therefore do not know whether the food consumption data, in particular from the CAPD patients, reflect an inadequate dialysis. Despite all these possible factors relating eating behavior, food preferences, or eating motivation to the clinical state of the uremic patients, these relationships have not been much studied. Disturbances in taste threshold for sour and sweet in predialysis patients and improvement after dialysis have been demonstrated. 2o Low serum zinc levels could also be a contributing factor in loss of taste, but this was not measured in our patients. 21 Further study on factors affecting eating motivation is warranted. Even if adequate nutritional advice, based on theoretical calculations of the needs of these patients, is designed, this is oflittle help unless we understand what motivates the uremic patient to actually consume such a recommended diet.
REFERENCES I. Lindholm B, Bergstrom J: Protein and amino acid metabolism in patients undergoing continuous ambulatory peritoneal dialysis (CAPD). Clin Nephrol 30:S59-S63, 1988 (suppl I) 2. Kopple JD, Blumenkrantz MJ: Nutritional requirements for patients undergoing continuous ambulatory peritoneal dialysis. Kidney Int 24:S-295-S-302, 1983 (suppl 16) 3. Lowrie EG, Lew NL: Death risk in hemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 15:458-482, 1990 4. Blumenkrantz MJ, Gahl GM, Kopple JD, et al: Protein losses during peritoneal dialysis. Kidney Int 19:593-602, 1981 5. Lindholm B, Bergstrom J: Nutritional aspects ofCAPD, in Gokal R (ed): Continuous Ambulatory Peritoneal Dialysis. London, England, Churchill Livingstone, 1986, pp 228-264 6. Popovich RP, Moncrief JW, Nolph KD, et al: Continuous ambulatory peritoneal dialysis. Ann Intern Med 88:444456, 1978 7. Lindsay RM, Spanner E: A hypothesis: The protein catabolic rate is dependent upon the type and amount of treatment in dialyzed uremic patients. Am J Kidney Dis 13:382389, 1989
8. Lysaght MJ, Pollock CA, Hallet MD, et al: The relevance of urea kinetic modelling to CAPD. ASAIO Trans 35:784788, 1989 9. Bouma SF, Dwyer JT: Glucose absorption and weight change in 18 months of continuous ambulatory peritoneal dialysis. J Am Diet Assoc 84:194-197, 1984 10. Barany P, Pettersson E, Ahlberg M, et al: Nutritional assessment in anemic hemodialysis patients treated with recombinant human erythropoietin. Clin Nephrol 35:270-279, 1991 II. KissileffHR, KIingsbergG, Van Itallie TB: A universal eating monitor for continuous recording of solid or liquid consumption in man. Am J PhysioI238:RI4-R22, 1988 12. Barkeling B, Rossner S, Bjorwell H: Effects of a highprotein meal (meat) and a high-carbohydrate meal (vegetarian) on satiety measured by automated computorized monitoring of subsequent food intake, motivation to eat and food preferences. Int J Obesity 14:743-751, 1990 13. Blundell JE, Rogers PJ, Hill AJ: Evaluating the satiating power offoods: Implications for acceptance and consumption, in Solms J (ed): Chemical Composition and Sensory Properties of Food and Their Influence on Nutrition. London, England, Academic, 1988, pp 1-15
EATING BEHAVIOR IN DIALYSIS PATIENTS
14. Hill AJ: Investigation of some short-term influences on hunger, satiety and food consumption in man. Thesis, Department of Physiology, University of Leeds, England, 1986 15. Blagg C: Importance of nutrition in dialysis patients. Am J Kidney Dis 17:458-461,1991 16. Hylander B, Burkart J, Dalton C, et al: Effect of intraperitoneal fluid volume changes on esophageal pressures. Studies in patients on continuous ambulatory peritoneal dialysis. Am J Kidney Dis 17:307-310,1991 17. Marckmann P: Nutritional status of patients with hemodialysis and peritoneal dialysis. Clin Nephrol 29:75-78, 1988
597 18. Young GA, Kopple JD, Lindholm B, et al: Nutritional assessment of CAPD patients. An international study. Am J Kidney Dis 17:462-471,1991 19. Bergstrom J: Anorexia and malnutrition in hemodialysis patients. J Blood Purif (in press) 20. Burge JC, Park HS, Whitlock CP, et al: Taste acuity in patients undergoing long-term hemodialysis. Kidney Int 15:49-53, 1979 21. Fornari AJ, Avram MM: Altered taste perception in uremia. Trans Am Soc Artif Intern Organs, 24 1978
P
ROTEIN AND ENERGY malnutrition is common in both continuous ambulatory peritoneal dialysis (CAPO) and hemodialysis patients, and contributes to increased mortality and morbidity.I-3 Malnutrition increases the risk for infections, surgical complications, and in-hospital stay. The dialysis treatment per se is associated with loss of amino acids and proteins to the dialysate. 4 •5 A strong inverse correlation has been found between the serum albumin concentration and mortality.3 Hunger and appetite are low in CAPO patients, 1,6 which can partly be explained by insufficient dialysis. 7,8 Another explanation for poor food intake in CAPO patients has been that the feeling of fullness from the dialysate in the abdomen reduces the drive to eat. The contribution of glucose from the dialysate as a caloric source could also explain the lack of hunger and appetite. 2 ,9 Renal failure patients are often nauseated and may experience altered senses of taste. 10 Furthermore, effects of medication, mental depression, or poor physical activity may all From the Department 0/ Medicine, Section o/Nephrology, and Health Behaviour Research, Karolinska Hospital, Stockholm, Sweden. Received April 23, 1992; accepted in revised/orm July 21, 1992. Supported by the Swedish National Foundation/or Kidney Patients. Address reprint requests to Britta Hylander MD, PhD, Section o/Nephrology, Division o/Medicine, Karolinska Hospital, Box 60500, S-J04 01 Stockholm, Sweden. © 1992 by the National Kidney Foundation, Inc. 0272-6386/92/2006-0011 $3.00/0
592
contribute to reduce appetite and hence food intake in both hemodialysis and CAPO patients. 10 We undertook the following study to compare food intake and eating motivation in a group of dialysis patients and matched healthy controls. PATIENTS AND METHODS Three matched groups of each 12 subjects (eight men, four women) were studied. Both dialysis groups were matched as regards their clinical condition, other concomitant diseases, and medication. No subject in any group was diabetic. Hypertension was diagnosed in all but one patient. No patient had chronic obstructive pulmonary disease. Patients in the CAPO group were studied after having been on CAPO for more than 3 months. The mean time on CAPO was 10 ± 5 (SO) months (range, 3 to 18). Their mean age was 60 ± 14 years (range, 37 to 79). Their initial diagnoses were glomerulonephritis (5), nephrosclerosis (3), pyelonephritis (I), congenital aplasia (I), and others (2). Mean body mass index (BMI) was 23.0 ± 2.8 kg/m 2 (reference value, 20 to 25 kg/m2). The 12 hemodialysis patients had been on hemodialysis for more than 3 months and their mean time on dialysis was 23 ± 28 months (range, 3 to 108). The mean age was 60 ± 13 years. Their initial diagnoses were glomerulonephritis (5), nephrosclerosis (3), cystic disease (2), and others (2). Mean BMI was 22.7 ± 3.4 kg/m2. All patients received hemodialysis treatment three times a week. Six patients were treated with erythropoietin. In the third group of healthy matched controls, the mean age was 60 ± 14 years and the mean BMI was 23.4 ± 1.8 kg/m2. No CAPO patient had experienced peritonitis within 5 months preceding the study. The dialysate volumes varied between 2.0 I and 2.5 1. All patients had four exchanges of dialysate per day. Body weights were noted at both "VIKTOR" test meals on an electric scale.
American Journal of Kidney Diseases, Vol XX, No 6 (December), 1992: pp 592-597
593
EATING BEHAVIOR IN DIALYSIS PATIENTS
The following blood tests were determined in the two patient groups: hemoglobin, serum urea, creatinine, standard bicarbonate, and albumin. In all CAPO patients, these values were obtained within 2 days of the test meal to reflect the average dialysis situation. In hemodialysis patients, the serum urea and creatinine values were obtained after dialysis as a posttreatment value, whereas the albumin concentration was calculated as the mean of the values postdialysis before the test and predialysis after the test. This value was taken to reflect an average estimate of the degree of hydration. Hemoglobin and bicarbonate values were obtained from the blood test before dialysis closest to the test meal.
VIKTOR The lunch intake was measured by VIKTOR, a universal eating monitor of the type first developed by Kissileff et alII and later modified by us. 12·14 The VIKTOR equipment consists of a hidden scale built into a table and connected to a microcomputer, which registers the amount of food eaten from the plate placed on top of the scale. The computer (LUXOR ABC 8(0) is equipped with an analog/digital converter to convert the readings from the scale to the computer. Before each experiment, the whole system is calibrated by an interactive calibrating process with known weights. Aforkholder provides signals to the computer to start or stop data readings when the fork is removed or returned to the holder. As soon as the person under study removes the fork from the fork-holder and starts to eat, a weight reading is stored three times per second for subsequent analysis. The readings are also shown graphically on-line as an eating curve on the computer monitor. All computer programs were written in BASIC. After each test meal, the computer analyzes the eating data and calculates the total intake (grams), duration of consumption (minutes), rate of consumption (g/min), and the relative rate of consumption (1m = the intake of food during the first half of the meal minus the intake of the second half of the meal divided by the total intake of food). 1m values greater than 0.1 indicate a decelerated eating pattern, that is, an eater who slows down the eating rate toward the end of the meal. II This indicates strong internal satiety signals. An 1m value less than 0.1 indicates that the subject is a nondecelerated eater with weaker internal satiety signals.
Eating Motivation Assessments of subjective feelings of hunger, appetite, fullness, prospective consumption, food choice, and food preferences were made by a model system developed by Blundell et ai, translated and adapted to Swedish conditions by US . 14 The assessments are described below. Visual analogue scales. Visual analogue scale (VAS) ratings are made on lOO-mm horizontal scales, by which subjects answer questions with a cross on each scale. The scales are anchored with the description in brackets: "How strong is your desire to eat?" (not strong at all to very, very strong), "How hungry do you feel?" (not hungry at all to very, very hungry), "How full do you feel?" (not full at all to very, very full), " How much food do you think you could eat?" (none to a very big portion). The last question is also referred to as measuring " prospective consumption." The pleasantness of
the meals was checked by using a further VAS at the end of each meal where the subjects rated the question "How pleasant have you found your food?" (not at all pleasant to very, very pleasant). A f ood preference checklist. Subjects tick off the food items out of a list of 32 items they would like to eat immediately. Tbe food items represent four food groups (eight from each group): high-protein, high-fat, high-carbohydrate, and lowenergy items. The lists are used to amplify measures of general motivation (V AS), that is, the more hungry a person is, the more food items he ticks of[ A f orced-choice (food preference) list. SUbjects are forced to chose between 32 pairs of high-carbohydrate and high-protein items. Half of the high-carbohydrate foods are sweet and half are savory. This procedure is designed to reveal a specific high preference for carbohydrates or proteins. Examples of food pairs are roast chicken breast or a baked potato with a small knob of butter; panfried codfiIIet or a slice of jam-filled sponge cake. The list gives a preference quota. The quota is used to determine whether a high (32:0 to 29:3), medium (28: 4 to 24:8), low (23:9 to 19: 13), or no (18: 14 to 14: 18) preference for either macronutrient is present.
Test Meal The meal served was a microwave-heated excess portion of an industrially produced Swedish hash with a standard composition and an energy content of 150 kcalflOO g consisting of diced meat, onions, and potatoes that were mixed and fried (Table 1).
Procedure The participants were informed before coming to the laboratory that ratings of subjective feelings of hunger, fullness, appetite, and food preferences in connection with the meals were to be recorded, but no information of how food intake would be measured was given before the meals. The patients were instructed to eat their ordinary breakfast in their homes between 7:00 and 8:00 AM. The interval between breakfast and lunch was 4 to 5 hours. Patients were instructed not to drink or eat between breakfast and the test lunch. The CAPO patients came for lunch twice to the eating laboratory after breakfast with the same food composition, once with dialysate in their abdomen and once without. This order was at random and unknown to the VIKTOR investigator. The test occasions were about 1 week apart. In CAPO patients, the last dialysate was drained approximately 30 minutes before one of the test meals. Ouring the alternate meal, all patients had standard lactate dialysate (2.0 to 2.5 L) in the abdominal cavity, instilled approximately 30 minutes before the meal. The hemodialysis patients and the healthy controls came once Table 1. Composition of "Swedish Hash Test Meal"
Per 100 9 Energy %
Energy (kcal)
Protein (g)
Fat (g)
150
6 16
43
7
Carbohydrates (g)
15 41
594
HYLANDER, BARKELlNG, AND ROSSNER
for lunch. When the patients arrived at the laboratory, they were given the test meal together with a glass of water (200 mL). They were asked to eat as much as they wanted of the food to pleasant fullness and to drink the water when they wanted throughout the meal. Immediately before and after lunch, repeated VAS, food preference checklists, and forced-choice food preference lists were filled in by the patients.
Data Analysis Statistical analyses of the eating data and motivational ratings between groups were performed using one-way analysis of variance (ANOVA) and appropriate post hoc analyses. Student's t test was used for the within-group analysis. KruskalWallis' one-way ANOVA by ranks was used to evaluate forcedchoice food preference lists and food preference checklists between groups, and Wilcoxon's signed-rank test was used for analysis within groups. When CAPD patients were compared with the other groups, data from the eating occasion with the dialysate-filled abdomen were used. All results are expressed as the mean ± SD if not otherwise indicated. The study was approved by the Karolinska Ethics Committee.
RESULTS
Lunch Meal Intake
Fifteen CAPD patients were invited, but only 12 completed two lunch meals. The three patients who dropped out had been transplanted after the first test meal. All invited hemodialysis patients and healthy controls came for their lunches as scheduled. The total intake of food for CAPD patients with dialysate in the abdomen was 206 ± 70 g and without dialysate 190 ± 76 g (NS). The hemodialysis patients consumed 295 ± 105 g and the controls 357 ± 175 g. Comparison of the intake offood of the three groups with ANOVA showed a significant difference between the groups (P < 0.05). Further analysis indicated that these differences were between controls and CAPD patients (P < 0.05, Fisher). Other variables related to eating behavior, such as duration of consumption, rate of consumption, and relative rate of consumption, are shown in Table 2. There were no significant differences between dialysis groups in duration of consumption and relative rate of consumption. The rate of consumption was significantly different between groups (dialysis and controls) (P < 0.05, ANOVA). These differences were found between controls and CAPD patients (P < 0.05, Fisher) and between controls and hemodialysis patients (P < 0.05, Fisher), but not between CAPD and hemodialysis patients.
VAS
The motivation to eat as measured by VAS differed between groups before the meals (P < 0.05, ANOVA). CAPD patients had less hunger (P < 0.05, Fisher) and less desire to eat (P < 0.05, Fisher) than both controls and hemodialysis patients. Eating lunch caused a large and significant decrease in ratings of desire to eat, hunger, and prospective consumption, and a significant increase in rated fullness in all groups compared with ratings before lunch (paired t test, at least P < 0.01). After the lunch meals, there were no differences between groups in motivation to eat, fullness, etc. The lunch meal was also equally well liked by the groups (Table 3). There was a tendency for CAPD patients to be less hungry and have less desire to eat before lunch with dialysate in the abdomen compared with empty abdomen, but these differences were not statistically significant. After the lunch meals, there were no differences in motivation to eat in CAPD patients with or without dialysate in the abdomen. Food Preference Checklist
Eating lunch caused a significant decrease in the number of item chosen from the food preference list for all groups, compared with before lunch (Wilcoxon's rank test, at least P < 0.01). There were no differences in the number of items chosen between these groups, either before or after the meals. Dialysate in the abdomen did not affect the number of items chosen by CAPD patients. Forced-Choice (Food Preference) List
The forced-choice (food preference) list demonstrated relative preferences for high-protein food items in favor of high-carbohydrate foods before lunch in all groups. The median preference ratio (carbohydrate to protein items) before meals were similar in all groups: 6:26 for CAPD patients with both full and empty abdomen, 6:26 for hemodialysis patients, and 8:24 for controls. After lunch, the relative preferences changed significantly to preferences for high-carbohydrate food items for hemodialysis patients (median ratio, 18: 14) and controls (median ratio, 24:8) (Wilcoxon's rank test, P < 0.05 for both groups), but no
595
EATING BEHAVIOR IN DIALYSIS PATIENTS
Table 2. Characteristics of Food Intake in Dialysis Patients and in Controls CAPO (Full)
Intake of food (g) Duration of consumption (min) Rate of consumption (g/min) Relative rate of consumption (1m)
206 8.9 24 0.09
CAPO (Empty)
± 70 ± 3.4 ±8 ± 0.21
190 8.0 25 0.18
± ± ± ±
Hemodialysis
76 3.0 6 0.16
295 12.4 24 0.09
± ± ± ±
Controls
105 3.6 8 0.13
356 11.0 32 0.14
± ± ± ±
175 4.7 10 0.14
NOTE. Values are means ± SO. See text for statistical analysis.
change was seen for CAPD patients. Their median ratios were 8:24 with full abdomen and 10: 22 with empty abdomen (Table 4). Laboratory Values Mean values for hemoglobin, serum creatinine and serum urea, albumin, and standard bicarbonate are shown in Table 5. DISCUSSION
Uremia is known to be associated with poor appetite and malaise. A study of the eating behavior of these patients may disclose possible dietary mechanisms for the clinically observed low eating drive and the body weight loss during dialysis therapy, and can thus help to design adequate dietary advice. 4 ,15 In our study, there was a significantly lower food consumption in CAPD patients compared with controls, which was also lower than in hemodialysis patients, although of borderline sigTable 3. Motivation to Eat Measured by VAS in Humans
Oesire to eat Before meal After meal Hunger Before meal After meal Fullness Before meal After meal Prospective consumption Before meal After meal Palatability
CAPO (Full)
CAPO (empty)
Hemodialysis
Controls
27 ± 26 4±2
36 ±30 5±3
53 ±26 4±3
53 ± 18 3±2
33± 28 3±2
38± 30 4±2
54 ± 22 4±3
57 ± 23 4±3
36 ± 28 82 ± 15
36 ± 36 84 ± 15
27 ± 22 83± 14
23 ±20 82 ± 21
39 ± 18 6±6 66 ± 24
40 ± 20 8±7 77 ± 24
51 ± 18 11 ± 9 69 ± 29
56 ± 18 15 ± 26 74 ± 22
NOTE. Values are means ± SO. See text for statistical analysis and description of scale anchors.
nificance. Theoretically, an increased intraabdominal pressure by the dialysate could explain this reduced food intake. However, we have previously measured intragastric pressures in CAPD patients with and without dialysate and found no differences in intragastric pressure. 16 Delayed gastric emptying in uremia, in particular in the CAPD patient group, is another alternative that might explain the differences between dialysis patients and controls. We cannot rule out this mechanism as a causal factor. The smaller food intake in CAPD patients might further be explained by the continuous glucose load suppressing hunger and appetite. 9 Normally, food preferences switch from protein to carbohydrates during the VIKTOR test meal. 12 In contrast to controls and hemodialysis patients, this did not happen in CAPD patients, which could reflect an effect of the high continuous carbohydrate load from the dialysate. Low standard bicarbonate levels are associated with poor appetite, but standard bicarbonate levels were 23 ± 1.5 mmol/L in CAPD patients and 22 ± 3.0 mmol/L in hemodialysis patients (normal range, 21 to 26). Poor intake could thus not be explained by acidosis. The differences between CAPD and hemodialysis patients in serum albumin concentrations seem to reflect the fact that sampling conditions were not identical, rather than clinically significant differences between the two paTable 4. Changes in Food Preferences Before and After Meal
Before meal After meal
CAPO (Full)
CAPO (Emtpy)
Hemodialysis
Controls
6:26 8:24
6:26 10:22
6:26 18:14
8:24 24:8
NOTE. Forced-choice (food preferences) between (carbohydrate:protein) items. See text for statistical analysis.
HYLANDER, BARKELlNG, AND ROSSNER
596
Table 5. Laboratory Parameters in CAPO and Hemodialysis Patients
CAPO patients Hemodialysis patients
Hemoglobin (gIL)
Creatinine ("moI/L)
Urea (mmoI/L)
Albumin (gIL)
Standard Bicarbonate (mmol/L)
94 ± 20.4 100.0 ± 16.3
814 ± 279 505 ± 127
20.9 ± 4.4 11.4 ± 2.0
33.5 ± 5.4 41.2 ± 4.3
23 ± 1.5 22 ± 3.0
tient groups. The higher hemoglobin levels in hemodialysis patients could partly be attributed to erythropoietin treatment in six of these patients, but the minor nonsignificant difference could probably not explain their higher food intake and more pronounced appetite. The slow consumption rate in association with low motivation to eat in CAPD patients remains difficult to explain, but might relate to the subjective feeling of fullness either due to delayed emptying from the stomach or to the dialysate despite unchanged intragastric pressures. The finding of an equally slow rate of consumption in hemodialysis patients implies that the uremic status per se contributes to the low motivation to eat. The loss of hunger and appetite and the development of anorexia during dialysis have numerous possible explanations such as inadequate dialysis, unpalatable or inadequate diets, medications, psychosocial factors, acidosis, depression, low physical activity, etc. 17- 19 At the
time of this study, Kt/V analyses were not regularly included in the clinical evaluation of these patients. We therefore do not know whether the food consumption data, in particular from the CAPD patients, reflect an inadequate dialysis. Despite all these possible factors relating eating behavior, food preferences, or eating motivation to the clinical state of the uremic patients, these relationships have not been much studied. Disturbances in taste threshold for sour and sweet in predialysis patients and improvement after dialysis have been demonstrated. 2o Low serum zinc levels could also be a contributing factor in loss of taste, but this was not measured in our patients. 21 Further study on factors affecting eating motivation is warranted. Even if adequate nutritional advice, based on theoretical calculations of the needs of these patients, is designed, this is oflittle help unless we understand what motivates the uremic patient to actually consume such a recommended diet.
REFERENCES I. Lindholm B, Bergstrom J: Protein and amino acid metabolism in patients undergoing continuous ambulatory peritoneal dialysis (CAPD). Clin Nephrol 30:S59-S63, 1988 (suppl I) 2. Kopple JD, Blumenkrantz MJ: Nutritional requirements for patients undergoing continuous ambulatory peritoneal dialysis. Kidney Int 24:S-295-S-302, 1983 (suppl 16) 3. Lowrie EG, Lew NL: Death risk in hemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 15:458-482, 1990 4. Blumenkrantz MJ, Gahl GM, Kopple JD, et al: Protein losses during peritoneal dialysis. Kidney Int 19:593-602, 1981 5. Lindholm B, Bergstrom J: Nutritional aspects ofCAPD, in Gokal R (ed): Continuous Ambulatory Peritoneal Dialysis. London, England, Churchill Livingstone, 1986, pp 228-264 6. Popovich RP, Moncrief JW, Nolph KD, et al: Continuous ambulatory peritoneal dialysis. Ann Intern Med 88:444456, 1978 7. Lindsay RM, Spanner E: A hypothesis: The protein catabolic rate is dependent upon the type and amount of treatment in dialyzed uremic patients. Am J Kidney Dis 13:382389, 1989
8. Lysaght MJ, Pollock CA, Hallet MD, et al: The relevance of urea kinetic modelling to CAPD. ASAIO Trans 35:784788, 1989 9. Bouma SF, Dwyer JT: Glucose absorption and weight change in 18 months of continuous ambulatory peritoneal dialysis. J Am Diet Assoc 84:194-197, 1984 10. Barany P, Pettersson E, Ahlberg M, et al: Nutritional assessment in anemic hemodialysis patients treated with recombinant human erythropoietin. Clin Nephrol 35:270-279, 1991 II. KissileffHR, KIingsbergG, Van Itallie TB: A universal eating monitor for continuous recording of solid or liquid consumption in man. Am J PhysioI238:RI4-R22, 1988 12. Barkeling B, Rossner S, Bjorwell H: Effects of a highprotein meal (meat) and a high-carbohydrate meal (vegetarian) on satiety measured by automated computorized monitoring of subsequent food intake, motivation to eat and food preferences. Int J Obesity 14:743-751, 1990 13. Blundell JE, Rogers PJ, Hill AJ: Evaluating the satiating power offoods: Implications for acceptance and consumption, in Solms J (ed): Chemical Composition and Sensory Properties of Food and Their Influence on Nutrition. London, England, Academic, 1988, pp 1-15
EATING BEHAVIOR IN DIALYSIS PATIENTS
14. Hill AJ: Investigation of some short-term influences on hunger, satiety and food consumption in man. Thesis, Department of Physiology, University of Leeds, England, 1986 15. Blagg C: Importance of nutrition in dialysis patients. Am J Kidney Dis 17:458-461,1991 16. Hylander B, Burkart J, Dalton C, et al: Effect of intraperitoneal fluid volume changes on esophageal pressures. Studies in patients on continuous ambulatory peritoneal dialysis. Am J Kidney Dis 17:307-310,1991 17. Marckmann P: Nutritional status of patients with hemodialysis and peritoneal dialysis. Clin Nephrol 29:75-78, 1988
597 18. Young GA, Kopple JD, Lindholm B, et al: Nutritional assessment of CAPD patients. An international study. Am J Kidney Dis 17:462-471,1991 19. Bergstrom J: Anorexia and malnutrition in hemodialysis patients. J Blood Purif (in press) 20. Burge JC, Park HS, Whitlock CP, et al: Taste acuity in patients undergoing long-term hemodialysis. Kidney Int 15:49-53, 1979 21. Fornari AJ, Avram MM: Altered taste perception in uremia. Trans Am Soc Artif Intern Organs, 24 1978
