Is energy intake underreported in hemodialysis patients ?

Estimating energy intake is not an easy task in any population. Food intake assessment is usually based on individual self-report and epidemiological studies, a process that can be distorted and result in underor over-reporting intake.1-2 When not identified, underreporting produces measurement biases and inconsistent results, leading to a misinterpretation of energy and nutrient intake,3 which can hamper nutritional diagnosis and generate inadequate dietary interventions. Despite studies showing an energy intake below recommendations for hemodialysis patients, a significant prevalence of excess weight and body fat,4-7 or maintenance of anthropometric parameters over time8 have been found in this population. This paradox may be related to food intake underreporting. Underreporting is a distortion of selfreported food intake1. Overall, it results from the individual’s failure to recall all consumed food or the underestimation of the quantities ingested. We also understand it as a gap between said measured energy intake and energy expenditure measured without weight change or weight maintenance over a period of follow up.9 Although it is well referenced in the literature for healthy individuals, underreporting is still not well studied


IntroductIon
Estimating energy intake is not an easy task in any population.2] When not identified, underreporting produces measurement biases and inconsistent results, leading to a misinterpretation of energy and nutrient intake, 3 which can hamper nutritional diagnosis and generate inadequate dietary interventions.
Despite studies showing an energy intake below recommendations for hemodialysis patients, a significant prevalence of excess weight and body fat, [4][5][6][7] or maintenance of anthropometric parameters over time 8 have been found in this population.This paradox may be related to food intake underreporting.
Underreporting is a distortion of selfreported food intake 1 .Overall, it results from the individual's failure to recall all consumed food or the underestimation of the quantities ingested.We also understand it as a gap between said measured energy intake and energy expenditure measured without weight change or weight maintenance over a period of follow up. 9lthough it is well referenced in the literature for healthy individuals, underreporting is still not well studied

Introduction:
Underreporting of energy intake is not much studied in hemodialysis population.Objective: To evaluate the underreporting of energy intake and associated factors in hemodialysis patients.Methods: A cross-sectional study, with 344 patients stable adults, of ten hemodialysis centers in Goiânia-GO.Energy intake was assessed by six 24-hour dietary recalls and basal metabolic rate (BMR) was estimated using the Harris Benedict equation.It was considered as underreporting when the ratio between the average energy intake and basal metabolic rate was lower than1.27.For analysis of factors associated with underreporting, the Poisson regression with robust variance estimation was applied.Results: The prevalence of underreporting was 65.7%, being more significant in individuals who are overweight and in the non dialysis days.The result of the multivariate analysis identified four factors independently associated with the underreporting: being a female (PR = 1.27,CI = 1.10 to 1.46), body mass index ≥ 25 kg/ m 2 (PR = 1.29,CI = 1.12 to 1.48), three meals or lower/day (PR = 1.31,CI = 1.14 to 1.51) and hemodialysis length lower than 5 years (PR = 1.19CI = 1.01 to 1.40).Conclusion: The population showed a high prevalence of underreporting of energy intake.Being a female, presenting overweight, lower number of meals/day and lower length time on hemodialysis were factors associated with underreporting.
in chronic kidney disease, [3][4][5]8 especially in the hemodialysis population. Klppenburg et al. 8 evaluated the under-reporting from hemodialysis individuals, identifying a prevalence higher than 60%.In Brazil, few studies have investigated underreporting among chronic kidney patients [4][5] and only one evaluated the population under hemodialysis, finding a prevalence of 65%.7 Thus, the lack of national data regarding underreporting of food intake among patients with chronic kidney disease fostered this study, aimed at establishing the prevalence of energy intake underreporting and associated factors in hemodialysis patients.

Methods
This is a cross-sectional study carried out involving ten hemodialysis centers in Goiânia-GO, between May 2009 and March 2010.The study was approved by the Ethics in Research Committee of the University Hospital -Federal University of Goiás (HC/UFG, Protocol No. 011/2009) and the Santa Casa de Misericórdia de Goiania (Protocol 046/2009), and each participant signed an informed consent.
The sample was estimated at 302 individuals, considering the total number of patients on hemodialysis in the municipality (n = 1400), margin of error and significance level at 5%, test power of 80% and prevalence of 50% underreporting for this population.We added 20% more individuals to the sample, because we forecasting losses and refusals, totaling 362 individuals.Of this total, 344 patients completed the study.For sample selection purpose, we used a simple random draw, proportional to the total number of patients in a hemodialysis center.
We included non-institutionalized clinically stable patients older than 18 years, of both genders, with stable weight and no clinical evidence of inflammation and/or infection, in the last three months and in hemodialysis for more than three months.The exclusion criteria were: cancer, tuberculosis, acquired immunodeficiency syndrome, chronic obstructive pulmonary disease and severe cardiovascular disease, uncontrolled severe diabetes mellitus, current pregnancy, dialysis through a venous catheter, situations that prevented the anthropometric evaluation or the investigation of food intake (advanced bone disease, stroke sequelae, disability or amputations).
Clinical (disease etiology, comorbidities, and duration of hemodialysis), demographic (gender and age), socioeconomic (education, occupation, marital status, income), anthropometric (weight, height), food intake and physical inactivity data was obtained from medical records.The patient was interviewed through nutritional assessment by four nutritionists, using standardized forms tested in a pilot study.Physical inactivity was defined according to criteria from the Pan American Health Organization. 10ccording to Lohman et al., 11 weight and height were measured after the intermediate dialysis session of the week.We obtained the body mass index (BMI) from the dry weight/square of height ratio, considering the World Health Organization classification. 12The ideal weight was obtained from the BMI 23 (kg/m 2 ). 13To calculate energy intake per kilogram we used the ideal weight.When their weight adequacy was less than 95% or greater than 115%, we used the adjusted weight, calculated as recommended by the NKF/KDOQI (2000). 14ood intake was evaluated using six 24-hour recalls, over three days in and three days out of dialysis, obtained in a three-week period.Of the six recalls, one was obtained on a Saturday.The food quantities reported by patients were recorded in household measures, with the aid of utensils such as glasses, cups and spoons and then converted into grams or milliliters.Energy intake was calculated by a software designed for the study (www.dbcheckout.com.br/nutri), which is mainly based on data from the Brazilian Table of Food Composition. 15e considered it to be underreporting when the ratio between the average energy intake and basal metabolic rate (EI/BMR) was less than 1.27.For sedentary individuals, it is estimated that the minimum daily energy needed to maintain body weight is 1.27 times the BMR, 12,16 below which weight maintenance is biologically and statistically unlikely. 1,12A value higher than 2.39 was considered to characterize the overreporting energy intake. 17BMR was calculated by the Harris Benedict equation, a method that provides an acceptable prediction of basal metabolism for chronic renal patients. 18he data was entered in duplicate using regularity check by the Epi-Info 6.0 software (Centers for Disease Control and Prevention, Atlanta).Analyses were performed in the STATA/SE 12.0 statistical software (Stata Corporation, Texas).Categorical variables were expressed as frequencies and percentages and compared by the Descartes Rule of Signs.Continuous variables had their normality checked by the Kolmogorov-Smirnov test (p ≥ 0.05) and expressed as mean or median and standard deviation and compared by the t-test (p < 0.05) or the Wilcoxon test.In the bivariate analysis among socioeconomic, demographic, clinical and anthropometric data and underreporting, we used the Pearson's chi-square test.To analyze the factors associated with underreporting, we initially used a simple Poisson regression.The variables with p ≤ 0.20 were tested in the multivariate analysis using Poisson regression with robust variance estimation.The ones with a p < 0.05 remained in the model.

results
Table 1 depicts the patients' characteristics.Table 2 shows a high prevalence of underreporting in the study sample (65.7%), being significantly higher for individuals with BMI ≥ 25 kg/m 2 (78.4%).None of the patients showed overreporting of energy intake.Energy intake per kilogram of body weight and the EI/BMR ratio were significantly lower in the group with BMI ≥ 25 kg/m 2 .
Table 3 depicts the intake of energy and macronutrients in dialysis days, days off dialysis, under-reporters and valid reporters.We found that underreporting was significantly higher (74.4%) on the dialysis-off day, and under-reporters had higher protein intake.
The prevalence of underreporting was significantly higher among females (75%), in individuals with up to eight years of education (70.18%), with up to five years on hemodialysis (69.91%), with BMI ≥ 25 kg/ m 2 (78.38%) and those who had up to three daily meals (76.56%), as shown on Table 4.  general health, comorbidities, etiology, marital status, smoking, alcohol consumption, albumin, pre and post-dialysis urea), however, these were not associated with underreporting among the patients studied.

dIscussIon
When analyzing the energy intake report of the patients evaluated in this study, we noticed a considerable percentage of underreporting, while there was no overreporting.[5]8 In different populations, overreporting seems to be far less common 1 than underreporting, with prevalence rates between 1 and 39%, 2,19,20 and it is usually found among females, 2,9 eutrophic and young patients. 2][5] Although well explored in healthy subjects, with prevalence ratios between 2 and 85%, 3 in hemodialysis patients, underreporting is not much studied.There is only one Dutch 8 and one Brazilian 7 studies which investigated underreporting in this population, finding prevalence rates of 61% and 65%, respectively, similar to this study (65.7%).This data indicates that the prevalence of underreporting in hemodialysis individuals is high, similar to that found in other populations.
Another interesting data found in this study was the higher prevalence of underreporting in the dialysis-off day and the increased protein intake by under-reporters compared to valid reporters.The difference in underreporting on days with and without dialysis, [3][4][5]7,8 was not addressed by other authors. As or macronutrient intake between valid reporters and under-reporters, findings are not uniform in the literature.While some studies have identified a lower intake of carbohydrates 2,21 and/or lipids 2,21,22 between under-reporters, others found only higher protein intake 2,8,23 in the same group.
For the patients evaluated in this study, the factors associated with underreporting were BMI starting at 25 kg/m 2 , female gender, having up to three meals a day and being in hemodialysis for up to five years.There are a number of associated factors or predictors of underreporting when healthy individuals are concerned.The most frequently cited are females, 19,20,24 age, lower educational level 2 and income, 9 overweight (by IMC) [3][4][5]20,23,24 and high body fat. 7,19,20 n the study by Bazanelli et al., 4 BMI was the only determinant of underreporting in patients undergoing peritoneal dialysis. Other sudies carried out with patients with chronic kidney disease did not evaluate predictors of underreporting.3,5,8 Mafra et al. 7 did not investigate determinants of underreporting; however, they found a negative correlation between the Goldberg index and BMI, and between the reported energy intake and BMI for hemodialysis patients.
Excess weight has been pointed out as an important factor associated with underreporting. 9,20,22,25,26][9]21,24 In general, there is a mismatch between the energy intake reported by those who are overweight, especially the obese, 9,21 and the energy expenditure assessed by the double marked water method. 27[5]8 Monitoring studies of stable patients on hemodialysis have shown that for part of them, BMI and weight are maintained or increased with time, despite an energy intake reported to be lower than the recommended 30 to 35 kcal/kg/day. 8,28,29n this study, we found a higher prevalence of underreporting in the overweight group, with BMI starting at 25 kg/m 2 a factor associated with underreporting for the individuals evaluated.Although we did not follow the patients evaluated in order to monitor anthropometric parameters, we found that the high prevalence of underreporting is not consistent with the mean EI/BMR ratio, nor with the high percentage of overweight patients according to BMI.These results are in agreement with those from Avesani et al. 5 and Bazanelli et al., 4 who observed greater underreporting, with prevalence rates of 80 and 83.3%, respectively, in chronic renal failure patients with a BMI ≥ 25 kg/m 2 .Interestingly, in general, overweight individuals undergo frequent dietary restrictions in order to reduce weight. 9Thus, they become proficient connoisseurs of food with higher energy content 30 and develop a selective underreporting for food rich in carbohydrates and lipids, 2 which may arise from greater access to information on healthy eating, whether due to health policies or clarifications published by means of comunication. 2eing female is another factor commonly associated with underreporting. 3,7,9,22,25,26,28Novotny et al. 19 found a prevalence of underreporting of 61% versus 85% for men and women, while Gomes & Leão 2 reported 61% versus 79.5%, respectively.Among some female characteristics that may explain the greater underreporting in women, we list dissatisfaction with weight and body image (leading to repeated dietary restrictions), guilt 19 , fear of a negative evaluation 9 and desire for social approval.Dissatisfaction with weight, body image, guilt 18 , fear of a negative evaluation 9 and repeated dietary restrictions in an attempt to lose weight, 23 are female characteristics identified as causes of underreporting among women.Another explanation is the desire to adjust or a lust for social approval, defined as the tendency to answer what is socially accepted. 20hus, at the time of answering the dietary diary, 3 women tend to report a consumption considered appropriate rather than declaring their actual intake.
The act of eating less often is a feature of underreporters. 20Among our sample population, the lowest number of daily meals was associated with underreporting.It is supposed that individuals who have fewer meals, consume a higher volume, which can hamper their memory about the quantities they actually ate 9 and cause underreporting, since the 24-hour recall depends on memory.Not reporting snacks 20 and drinks consumed between meals in the form of tweaks, which are not included in the calculation of daily energy consumption can also contribute to food intake underreporting.
]7,8 It is possible that having a nutritionist at each dialysis center, constantly monitoring and questioning patients about food intake may have conditioned those patients on longer time in dialysis to be more aware of their intake and this had not yet occurred in individuals under shorter treatment time, resulting in greater underreporting among the latter.
Underreporting is a major limitation of food consumption studies, 7,25 leading to inaccuracies in dietary intake monitoring.Among hemodialysis patients, an accurate assessment of energy consumption is crucial for assessing their clinical status, nutritional interventions and overall health outcomes. 7,21When not detected, it can negatively interfere in the nutrition and monitoring of hemodialysis patients.
Thus, the nutritionist monitoring these patients must analyze critically 1 the intake data.This will be useful to prevent misconduct in a nutritional intervention plan, 21 as, for example, supplying more calories which, over time, may contribute to overweight or unnecessary body fat build up, or even encouraging the consumption of certain foods, to correct certain nutrient that "may be considered in deficit," which can cause imbalance of serum levels of phosphorus, potassium, glucose, and others.The EI/BMR ratio can be a useful, simple and feasible tool to detect underreporting in clinical practice, which would help the healthcare professional interpret food data in a safer and more rigorous way for hemodialysis patients, obtained by traditional methods of dietary assessment.Removal of underreported data enables a more accurate assessment of food intake 24 regarding the evaluated group and provides for safer advice regarding nutritional counseling. 7 limitation of this study is the fact that the patients' BMR was not measured, but rather estimated by a predictive equation of energy expenditure.Although it is more appropriate to calculate the BMR of individuals with chronic kidney disease, the Harris Benedict equation may overestimate this calculation by about 5.8%, 18 which would result in a higher prevalence of underreporting in this population.
However, the underreporting prevalence we found was not different from that of other studies, 4,5 which used indirect calorimetry to find the energy expenditure and classify the self-reported energy levels of intake.
Another point to be considered is the use of 24-hour recalling to obtain dietary intake information.In general, more precise methods such as direct weighing are not feasible for studies with larger samples.Recalling is one method of choice because it is practical, easy to perform, inexpensive and can be applied in individuals with different levels of education. 21Despite these advantages, this method has limitations, such as having to depend on memory recall, and its application in just one day may not represent normal intake. 21In this study, we tried to compensate for these disadvantages using household measures to assist in a more adequate identifying of how much food was eaten. 9In addition to all these steps, we employed six reminders, comprising the days on dialysis and the interdialytic period, making it more representative of the normal dietary intake for valid responders.
]7,8 A multivariate analysis identified the variables that remained associated with underreporting, given the interdependence between the factors evaluated.

conclusIon
This population had a high prevalence of energy intake underreporting, especially among individuals who are overweight by BMI and on days without dialysis.The under-reporters had higher protein consumption than valid reporters and we did not detect the over-reporting of energy intake.Being female, overweight, having fewer meals and being in hemodialysis for a shorter time were factors associated with underreporting.
Our findings draw attention to the fact that the analysis of dietary data, regardless of underreporting, may result in a false interpretation of intake and even inadequacies in carrying out dietary treatment of the patients.Thus, it is suggested that food intake of female patients, those in overweight, with less daily meals and less time in hemodialysis be investigated more carefully and interpreted by the nutritionist of dialysis centers.

tAble 2
energy intake, estimated basal metabOlism rate and underrepOrting Of patients in hemOdialysis, accOrding tO the bOdy mass index.gOiânia-gO, 2012 Standard deviation; ** Independent t-test or Person's chi-square test; *** EI: Energy Intake in kcal/day; # EI/kg: energy intake by kilogram of ideal or adjusted weight; ¶ BMR: Basal Metabolic Rate in kcal/day; § EI/BMR: energy intake and basal metabolic rate ratio. *