Use of the Body Composition Monitor for Fluid Status Measurements in Elderly Malnourished Subjects

Abstract

Most hemodialysis (HD) patients are able to finish dialysis at or below the normally hydrated weight determined using the body composition monitor (BCM). However, a minority become symptomatic when they are still fluid overloaded based on BCM-measured overhydration (OH). Malnourished patients frequently fall into this group, suggesting that they may have OH that is inaccessible to ultrafiltration. To isolate any effect of malnutrition on BCM-measured OH from those relating to renal failure, OH measurements for 20 elderly subjects with normal renal function who were classified as malnourished were compared with an age-matched cohort with no known nutritional issues. Body composition monitor measurements were also made on five malnourished HD patients. Mean OH for malnourished subjects with normal renal function was not significantly different from an age-matched cohort without known nutritional deficiencies (1.3 and 1.1 L, respectively; p = 0.5). Post-dialysis OH for HD patients ranged from −0.1 to +4.5 L. A slightly elevated BCM-measured OH appears to be common in elderly subjects and may be explained by changes in the composition of adipose tissue. The effect of malnutrition could not be isolated from sarcopenia, but this study supports the need for caution when reducing target weight in vulnerable patients.

Fluid management is one of the core elements of hemodialysis (HD). Chronic fluid overload is strongly linked to mortality,^1,2 whereas removing too much fluid from HD patients is associated with morbidity and loss of residual renal function.³ Taken together, this highlights the importance of accurate methods of determining how much fluid to remove during dialysis.

For many years, clinical assessment has been the cornerstone of fluid management in dialysis, but there is growing acceptance that the use of bioimpedance alongside clinical assessment can improve the assessment of fluid status.⁴ The body composition monitor (BCM; Fresenius Medical Care, Bad Homburg, Germany) uses bioimpedance spectroscopy (BIS) together with models to calculate fluid volumes⁵ and normally hydrated (NH) weight.⁶ Overhydration (OH) is simply the difference between the actual and NH weight. It is positive if the subject is overloaded and negative if they are fluid deficient.

Studies have shown that at a population level, some degree of normalization of OH can improve outcomes for HD patients.^7,8 However, in patients with little or no urine output, the ability to reduce time-averaged fluid overload depends on tolerance of fluid depletion. Body mass index (BMI) has been shown to be inversely associated with pre- and post-dialysis BCM-measured OH^2,9,10 and BCM users soon become aware that patients with high BMI can finish dialysis several liters below normal hydration. Local audit carried out when BCM commenced in our unit suggested that, before introducing BCM, the practice of probing for dry weight meant that patients often finished dialysis at the lowest weight they could tolerate.¹¹ Through a comparison with bariatric surgery candidates with normal kidney function, we have shown that the post-dialysis fluid depletion measured in obese HD patients is because of tolerance and not an artifact of the model.¹²

At the other extreme, the tendency for residual post-dialysis fluid overload observed in patients with low BMI may be because of poor tolerance of fluid depletion, to an alteration in fluid distribution related to malnutrition, or to a combination of factors. The presence of edema and expanded fluid volumes are relatively common in protein-deficient states of malnutrition¹³ and has been linked to hypoproteinemia and leaky capillaries.^14,15 Subjective global assessment (SGA), a validated tool to aid interpretation of malnutrition risk in patients on dialysis,¹⁶ has been shown to correlate with OH in both HD¹⁷ and peritoneal dialysis patients.¹⁸ Body composition monitor-measured OH has previously been shown to be associated with a mild degree of malnutrition as assessed by SGA in subjects with normal renal function.¹⁹ The BCM model has also been applied to a malnourished indigenous population from the Republic of Colombia, showing values of OH proportional to the degree of malnutrition (in some cases, with OH as high as 6.2 L).⁶ However, this was a reanalysis of data from 1978 which used dilution-based estimates of extracellular water (ECW) and intracellular water (ICW).²⁰

This study aimed to corroborate these observations using BCM-measured OH in malnourished subjects (with normal renal function) that are more representative of current HD populations. If BCM-measured OH is found to be usual in malnourished subjects with normal renal function, it is likely to be confined to the tissues with a relatively low impact on cardiovascular morbidity. However, if the BCM indicates normal hydration in these subjects, the OH commonly observed in malnourished HD patients may be “real” excess fluid that is accessible for removal by ultrafiltration (UF). In this case, the reasons for poor tolerance of fluid depletion will need to be investigated and addressed.

Subjects and Methods

Ethical approval was granted by a local committee and the work was undertaken in accordance with the Declaration of Helsinki.

Subjects

The study population consisted of three groups:

1. Group A: Subjects with normal renal function without known nutritional deficiencies (n = 30)
2. Group B: Subjects with normal renal function classified as malnourished (n = 20)
3. Group C: HD patients classified as malnourished (n = 5)

Group A, the subjects with normal renal function without known nutritional deficiencies, came from a large cohort with no history of renal failure or cardiovascular complications used for the definition of reference ranges for BCM parameters.²¹ For comparison with the subjects in group B, a subgroup matched for age was extracted from the dataset.

Group B, the subjects with normal renal function classified as malnourished, were recruited from the wards for the care of the elderly within the local hospital, a tertiary referral hospital. This group is regularly assessed for malnutrition and renal function and is monitored to ensure that their fluid intake is sufficient and that they are not dehydrated. All subjects had no history of renal failure and were not taking diuretics. Renal function was assessed with an estimated glomerular filtration rate (eGFR), calculated by the local laboratory with the modification of diet in renal disease (MDRD) 4 parameter formula²² using the most recent creatinine measurement for the individual (not being more than 1 month from the date of BCM measurement). Normal renal function was defined as an eGFR of greater than 45 ml/min/1.73 m2 to account for the normal reduction in eGFR with aging.²³ The definition of malnutrition was based on the Malnutrition Universal Screening Tool (MUST).²⁴

The HD patients in group C were recruited from the hospital and satellite dialysis units under the care of a large Teaching Hospital. Subjects were greater than 18 years of age and had been treated with HD for more than 3 months to allow time for fluid imbalance on presentation to be corrected. BMI was calculated based on the patients recorded height and their NH weight as defined by BCM. Defining malnutrition using MUST is not validated for HD patients. A specialist dietician identified patients for this cohort based on the following criteria:

1. SGA score greater than 3;
2. Presence of two of the three core nutritional variables associated with malnutrition (unintentional weight loss, BMI less than 20 kg/m² and reduced dietary intake);
3. Scoring greater than 2 on a novel malnutrition screening tool (Leeds Screening Tool) developed locally²⁵

Procedure

Data for group A came from a previously reported study.²¹

Group B had a BCM measurement according to the manufacturer’s instructions. Measurements were checked visually for artifacts and accepted where repeat measurements of OH were no more than 0.2 L different.

The malnourished HD patients in group C had a BCM measurement made pre- and post-dialysis. Pre-dialysis measurements were made after the patient had been supine for at least 2 minutes (from manufacturer’s guidance) and post-dialysis measurements made after the patients’ vascular access had been sealed and the patients had been weighed and returned to a supine position for at least 2 minutes, allowing sufficient time for redistribution of fluids. UF volume, blood pressure, and any symptoms experienced were recorded. Hemodialysis treatments were taken as the midweek session from standard 3 × 4 hour per week regimes, with a dialysate temperature of 36°C, where target weights were defined on the basis of clinical examination and BCM on indication.

Data Analysis

Continuous data were assessed for normality using Shapiro–Wilk tests. Body composition monitor-measured OH was described by mean (standard deviation) for normally distributed data and median (range) for non-normal data. The small number of HD patients in group C meant that the data could not be described as normal and median (range) was used.

The primary objective was to assess the difference in OH between a cohort of subjects with normal renal function who were classified as malnourished (group B) and an age-matched cohort without known nutritional deficiency (group A), which was assessed using a two-sided Student’s t-test.

The data analysis was carried out using the statistical software package “R,” version 3.0.2 (R Foundation for Statistical Computing, Vienna, Austria).

Sample Size

The dataset used for validation of the BCM consists of 1,296 subjects. For group A, all subjects above 80 years of age, to match a cohort recruited from the elderly care unit, were selected. This gave 30 subjects with a standard deviation in measured OH of 1.0 L. Assuming a similar standard deviation for BCM-measured OH in subjects with normal renal function classified as malnourished, recruiting 20 subjects to group B allowed a difference of approximately 0.8 L to be detected at the level of 5% type I error with 80% power.

An association between BCM-measured OH and nutritional state in HD has already established.⁹ A convenience sample of five HD patients classified as malnourished was based on the estimated prevalence of eligible patients in the study time-frame and used for descriptive purposes only.

Results

The characteristics of the patient groups can be seen in Table 1. There was no significant difference in age or gender between groups A and B, but as expected, subjects without known nutritional deficiencies had greater BMI, lean tissue mass, and adipose tissue mass. Group B subjects had a median (interquartile range) eGFR of 78.5 (59 to 90) ml/min/1.73 m² and 15 of the 20 subjects had an eGFR of greater than 60 ml/min/1.73 m².

Table 1.: Subject Characteristics

The mean BCM-measured OH in group B was 1.3 L. This was not significantly different from the mean BCM-measured OH of 1.1 L for group A, the subjects without known nutritional deficiency (mean difference: −0.2 l; p = 0.5; 95% CI: −0.8 to 0.4).

The inclusion criteria for malnourished HD patients in group C did not include age and the five patients recruited were aged between 68 and 82 years. Three of the patients had no recorded comorbidities, one had heart failure alone and one had acute coronary syndrome, heart failure, smoking, and peripheral vascular disease as comorbidities. The post-dialysis median BCM-measured OH was slightly higher than the OH observed in groups A and B at 1.8 L, but ranged widely from −0.1 L fluid depleted to 4.5 L fluid overloaded (Figure 1).

Figure 1.: OH measurements by cohort. The region highlighted represents the 10th–90th percentiles of the population used to generate the reference ranges for the BCM. BCM, body composition monitor; OH, overhydration.

Discussion

The whole body model⁶ used in the BCM relies on the intra- and extracellular water content of NH lean and adipose tissue remaining essentially constant, although there is an age-related adjustment applied to the OH value generated from the model.²⁶ Extensive validation and successful clinical use of the device suggests that this model holds for most of the population, including patients with very high BMI.¹² However, if the normal ECW content of lean or adipose tissue is higher than the modeled value in specific conditions, the BCM will tend to indicate an elevated OH in patients with these conditions with normal fluid status and attempting to remove this fluid may result in hypovolaemia.

This study showed a systematically elevated BCM-measured OH in both groups of elderly subjects with normal renal function, who are expected to be euvolaemic. There was no difference between the groups with and without clinical diagnosis of malnutrition, which suggests that the condition leading to an elevated OH is primarily age-related sarcopenia. Inspection of the database described earlier suggests a tendency toward increased BCM-measured OH above 70 years of age.²¹ Dual energy x ray absorptiometry (DEXA) measurements of fat-free mass—which includes the ECW from adipose tissue—and body cell mass using total body potassium counting have suggested that there is a normal increase in the hydration of fat-free mass with age.²⁷ A possible explanation for the increase in the ECW content of NH tissue is an increase in the fat located within muscle. Increased levels of intermuscular and intramuscular adipose tissue (IMAT) are associated with age, disease, injury, inactivity, and obesity.²⁸ If IMAT has a higher ECW content than subcutaneous or other visceral fat, a proportion of adipose tissue as IMAT above the level expected by the BCM model will lead to systematic elevation of BCM-measured OH, as we observed in groups A and B.

If this hypothesis is correct, our results suggest that in the elderly, malnutrition has little impact on IMAT levels that are already increased because of aging. To isolate the effect of malnutrition and obtain a conclusive answer to the original research question would require the collection of BCM-measured OH in a younger cohort with a diagnosis of malnutrition, and ideally without any concomitant disease, injury, or restriction on activity. We did not have access to such a cohort and it is likely that this would be difficult to do; however, it is also true that such a cohort would be unrepresentative of the HD population.

The five HD patients classified as malnourished that formed group C achieved a post-dialysis BCM-measured OH that varied widely as shown in Figure 1. The four patients with OH that is comparable with that of patients in groups A and B may have reached a normal fluid status. If the OH is confined within the tissues where it does not present the cardiovascular risks associated with hypervolaemia, attempts to reduce target weights in these patients to reach a BCM-measured OH closer to zero may be unnecessary and could lead to intradialyic hypotension.

The fifth HD patient provides a case study that shows how rapidly tolerance of fluid removal can change. Three months before the study measurement, he transferred from peritoneal dialysis caused by UF failure and achieved normal hydration based on BCM, with a serum albumin of 37 g/L, within a week. He then began losing flesh weight rapidly and was hospitalized with an infection related to his PD catheter for 3 weeks. After discharge, the patient remained in poor health and was unable to cope at home. As is common in such patients, he was no longer able to tolerate adequate fluid removal. At the time of the study, the patient’s post-dialysis OH was +4.5 L and his serum albumin was diluted to 27 g/L. He continued to accumulate fluid until he died 19 days later. In this case, while part of the BCM-measured OH may have been associated with his worsening condition, much of it would have been real excess fluid that contributed to hypervolaemia.

In summary, although the study did not answer the original question, the data suggest that a slightly positive BCM-measured OH can be expected in NH elderly subjects. This could be explained by changes in the composition of adipose tissue because of intramuscular lipid accumulation that may also be found in individuals who are immobile through injury or persistent illness. Malnutrition per se may only alter the quantity of adipose tissue, but malnourished HD patients usually have at least one of the conditions that do lead to an increase in IMAT. The effect of low serum albumin to elevated BCM-measured OH is currently unclear.

Our data support a cautious approach to setting target weight in HD patients who are elderly or have limited mobility. Reductions should be made gradually allowing both intradialytic symptoms and increased post-dialysis recovery time to be reviewed. For those with residual function, interdialytic fluid gain (IDFG) should be monitored after a reduction in target weight as a decrease in IDFG is a good indication of excessive fluid removal.

In the absence of experimental quantification of the changes in fluid content, absolute blood volume measurements could help establish the degree of BCM-measured OH that can be expected in HD patients with altered adipose tissue composition. Continuous relative blood volume measurements, ideally extending beyond the UF time, could be used to establish if the measured OH in these patients is excess fluid that can transfer into the circulation. Such investigations could provide the means to optimize target weight in future. These techniques may also help in the development of appropriate fluid management strategies for patients with poor tolerance of UF.