Day-to-day variability in euvolemic body mass

Abstract

Short-term variability in body mass is a common, everyday phenomenon; however, data on body mass variability are scarce. While the physiological variability of body mass is negligible in healthy individuals, it could have implications for therapy in patients with impaired volume homeostasis, for example, patients with kidney failure undergoing kidney replacement therapy. We analyzed a long-term dataset comprising 9521 days of standardized body mass measurements from one healthy male individual and assessed the variability in body mass as a positive or negative relative difference in body mass measured on subsequent days. The average and median relative differences were zero, with a standard deviation (SD) of 0.53% for the one-day interval, increasing to 0.69% for the 7-day interval, and this variability was constant throughout the observation period. A body mass variability of approximately 0.6% (±450 mL in a 75-kg patient) should be taken into consideration when weight-dependent treatment prescriptions, e.g. the ultrafiltration rates in patients on hemodialysis, are being set. Consequently, a “soft target weight”, considering the longitudinal variation of volume markers, such as body mass, might improve treatment quality.

Brief report

Measuring and correcting volume excess has remained a major challenge in individuals with chronic kidney disease (CKD) who require kidney replacement therapy since the early days of dialysis [1, 2]. At the beginning of the hemodialysis session, volume excess is usually determined by taking the difference of the actual body mass to an assumed reference mass at which the patient is expected to be at his or her “dry” or target weight. Most difficulties reside with an inadequate estimation of this target weight, which ideally reflects the status of euvolemia and depends on age, sex, body size, clinical condition, and body composition. In a conclusion document resulting from a Kidney Disease Improving Global Outcomes (KDIGO) controversies conference, the authors suggested in their passage referring to extracellular volume management, that a “physical examination paired with review of longitudinal weights […] should be performed at least once per month […]” [3]. Whatever the identified target weight is, it is usually assumed as being constant under stable clinical conditions and guides the removal of fluid by ultrafiltration in the typical intermittent, thrice-weekly treatment schedule. To the best of our knowledge, whether this weight is constant has never been shown. Ashby et al. recently proposed accounting for the variability in the prescription of ultrafiltration using a soft target weight rather than a fixed value [4]. The magnitude of the variability in euvolemic body mass cannot be examined in CKD patients where the main component of volume control is disrupted. Clinically, this question must be addressed in patients with normal kidney function. Interestingly, we have not been able to identify published information on intraindividual body mass variability under everyday conditions or for prolonged periods of time.

Therefore, the purpose of this brief report was to analyze and quantify the day-to-day variability in body mass for extended periods of time under everyday conditions, ultimately aiming to derive information that might be placed in the context of ultrafiltration requirements for patients on hemodialysis.

In a self-experiment, one of the experimentally experienced authors, a male, normally trained nonsmoker without apparent medical condition, recorded his body mass to the nearest 0.1 kg every morning, using a standard electronic weighing scale (Bosch PPW4200, Germany) with an accuracy of 100 g. Measurements were always taken after the overnight fast, undressed, voided, and before intake of food or fluid for a period of almost 30 years. The recording started on December 29, 1993, at an initial body mass of 75 kg (age 33 years), and as of April 21, 2023 (and a body mass of 83.1 kg) covered a period of 10707 days. The continuous recording was interrupted by short periods when measurements were not possible because of out-of-home holidays and business trips. The subject did not follow any specific diet and maintained constant eating behavior during the observation period. Body mass increased during the entire observation period from about 75 to 84 kg, with stable phases as well as increases and decreases in weight depending on lifestyle (e.g. changes in family or job situation) (Figure 1).

Figure 1.

Absolute body mass.

Time course of absolute body mass (n = 9211) during almost 30 years of observation.

Overall, measurements were done on 9521 days (26.1 years) and missed on 1185 days (3.2 years) for the above-mentioned reasons. The average length (mean ± standard deviation (SD); median and interquartile range (IQR)) of 619 uninterrupted measurement periods was 30.7 ± 33.3 (21, 12 to 40) days, and the average interruption (n = 618) lasted 3.8 ± 3.9 (2, 1–5) days (Figure 2).

Figure 2.

Observation phases and interruptions.

Distribution of the duration of uninterrupted observation phases (a) and interruptions (b). Note the differences in scale on both x-axes.

Variability in body mass was assessed as positive or negative relative differences in body mass measured on subsequent days (1-day interval, ΔM1%) and for 2-, 3-, and 7-day intervals (ΔM2%, ΔM3%, ΔM7%), respectively. For each of the intervals, average and median relative differences were (close to) zero, with a SD of 0.53% for the 1-day interval, which increased to 0.69% for the 7-day interval (Table 1).

Table 1.

Relative body mass differences.

	ΔM1%	ΔM2%	ΔM3%	ΔM7%
n	9211	8999	8858	8579
Average	0.0002	−0.0018	−0.0017	−0.0078
SD	0.5279	0.6109	0.6474	0.6927
Median	0.0000	0.0000	0.0000	0.0000
Q25	−0.3571	−0.3831	−0.3937	−0.4706
Q75	0.3559	0.3802	0.3920	0.4728

ΔM1% to ΔM7%, relative difference in body mass (in %) measured at 1- to 7-day intervals; SD, standard deviation; Q25 and Q75, 25% and 75% percentiles.

The time series and distribution of the relative body mass differences for the 1-day interval are shown in Figure 3.

Figure 3.

Relative body mass differences.

Day-to-day differences in relative body mass (n = 9211) are shown as serial data (a) and frequency histogram with a non-Gaussian distribution (b). Full and broken lines indicate the mean ± standard deviation.

Despite of temporal changes and an overall increase in body weight, day-to-day body mass was constant on average (the relative difference in body mass was zero), albeit with a SD of approximately 0.53%. The SD in relative differences increased with increasing observation interval and reached 0.69% in the 7-day interval. This variability was maintained throughout the observation period of almost 30 years, covering approximately half of an adult’s lifetime with phases of long-term weight gain and weight loss, owing to variations in work and lifestyle determined by professional requirements and personal preferences. However, day-to-day and weekly variability remained constant on average throughout the three decades.

The stability of average body mass changes is close to the small day-to-day difference of only −0.058 kg in large weight control studies (9768 participants) using automatic smart scales [5]. Only a few studies have examined individual day-to-day variability in body mass, and these studies cover much shorter observation phases. For example, day-to-day variability in body mass was 0.4 kg in 86 male students (21.4 ± 1.0 years, body mass 74.7 ± 7.8 kg, BMI 22.7 ± 1.7 kg/m²) measured on five consecutive days within the same week [6]. This result corresponds to a relative variability of 0.54%, which is identical to the variability identified in the present study. In an earlier study, the coefficient of variation (CV, i.e. SD divided by the mean) of body mass in 65 healthy men (18 to 42 years, body mass 55.40 to 107.90 kg) representing a cross-section of U.S. Army personnel and participating in an exercise program was 0.66 ± 0.24% [7]. Among ten healthy individuals (38 ± 11 years, 3 males), within-individual CV of daily weight was 0.71% and larger compared to the other studies [4].

Variability in body mass must not be confused with the typical measurement error in the range of 0.1 kg or about 0.12% of initial body mass, identified in this study. This error is expected to decrease with higher resolution of weighing scales as shown elsewhere (0.06 kg, 0.05 to 0.07 kg 95% confidence interval with 0.05 kg resolution) [6].

A variation in fat, protein, carbohydrates, or minerals in the magnitude of 0.5% of body mass (58.3 ± 6.7% of which is water in adult males) [8] within 1 day is physiologically improbable, if not impossible. The physiologic day-to-day fluctuation in the magnitude of 0.4 kg must be due to variation in water, which is linked to the variability in osmotically active sodium and other moieties. This variability is probably not only due to variable sodium intake (and delayed excretion) but also to variable sequestration of sodium in the interstitial tissue matrix, thereby varying the amount of osmotically active sodium, and probably also to the variation in glycogen, binding, or releasing free water in the liver and skeletal muscle [9–12]. Intestinal volume is also mostly water, which is reabsorbed in the colon. Normal stool consists of 60–80% water [13]. To which extent day-to-day fluctuations in body mass relate to variable colonic content and mobility remains speculative. Loose stool with a higher water content and faster colonic transit is likely to increase variability in body mass. In any case and whatever the background, a variability in body mass of 0.5% is to be expected in presence of physiological volume control.

In physiology and everyday life, a variability of 0.5% is negligible and body mass can be safely assumed to be constant. However, this is not the case in the current prescription practice for ultrafiltration volume in hemodialysis, which is usually based on actual interdialytic weight gain. The SD of body mass differences corresponding to the short (and long) dialysis intervals was roughly in the range of 0.6 and 0.65% of body mass. This SD amounts to ± 450 (and ± 488) mL SD in a 75 kg patient and, if not accounted for, leads to a discrepancy in prescribed ultrafiltration volume in the range of ± 11.3 to 15.0% for the short interval (and 12.2% to 16.3% for the long interval) of the typical 3 to 4 L of ultrafiltration volume. Considering the possibility that the optimal dry weight might follow the observed variation pattern of healthy individuals and considering that in a Gaussian normal distribution, 32% of values are beyond the SD (i.e. possibly more than half a liter of physiological variation since the last treatment can be expected in one out of three sessions on average), the variability in body mass could be a nontrivial source of intradialytic complications.

One approach to account for the inherent variability in dry weight and to adjust ultrafiltration requirements is to average the weight gains of previous treatments and prescribe the actual ultrafiltration volume as the average of the actual and average previous interdialytic weight gains [4]. This approach has become known as “soft target weight” prescription. It cuts the peaks and troughs of both extremely high and low ultrafiltration volumes expected with physiologic variation of euvolemic body mass, individual fluid intake, and unevenly spaced ultrafiltration intervals in the typical thrice weekly dialysis schedule and is thereby expected to improve ultrafiltration tolerance. In blunting ultrafiltration peaks, it helps to remain below the limits of critical specific ultrafiltration rates [14]. In engineering, this strategy resembles a proportional-integral-control (PI-control) that accounts for the history of the system and provides improved system stability compared to pure P-control based on actual deviation only [15]. So far, the soft target weight approach has not been formally tested in the hemodialysis setting.

The variability in body mass would be less important if a distinction could be made between euvolemic tissue and excess fluid volume for any given measurement of body mass, as claimed by bioimpedance analysis. This technology, however, is increasingly scrutinized as well [16–19]. Dry weight assessment is therefore more likely to be based on a combination of body mass and some additional diagnostic procedures in the foreseeable future [2].

In conclusion, as body mass is suggested to vary by about 0.5% from day-to-day assuming euvolemia, the reference mass to adjust volume excess in CKD patients could be expected to vary by a comparable magnitude as well. One approach to account for this variability in the prescription of ultrafiltration volume is to include information on average weight gains of preceding treatments rather than relying on actual weight gain alone.

Funding Statement

This work was supported by the Vienna Science and Technology (WWTF) Grant LS20-079 Precision Medicine.

Author contributions

Conceptualization, D.S.; methodology, P.H.; validation, D.S. and P.H.; formal analysis, D.S.; investigation, P.H.; resources, P.H.; data curation, D.S. and P.H.; writing—original draft preparation, D.S.; writing—review and editing, D.S., P.H., S.K., M.W., S.M., and M.H.; visualization, D.S.; supervision, D.S. and M.H.; project administration, M.H.; funding acquisition, M.H. All authors read and agreed to the submitted version of the manuscript.

Institutional review board statement

Not applicable.

Informed consent statement

Informed consent was obtained from the study participant.

Conflicts of interest

D.S. is a coinventor of patents in the field of blood volume and bioimpedance applications in hemodialysis and is a member of the American Renal Associates research board. M.H. served as a speaker and/or consultant for Astellas Pharma, AstraZeneca, Eli Lilly, Fresenius Medical Care, Janssen-Cilag, Siemens Healthcare, and Vifor, and had previously received academic study support from Astellas Pharma, Boehringer Ingelheim, Eli Lilly, Nikkiso, and Siemens Healthcare (not related to the present work). P.H., S.K., M.W., S.M., and S.K. declare no conflict of interest.

Data availability statement

The data supporting the findings of this study are available from the corresponding author, D.S., upon reasonable request.