Introduction

Until recently metabolic carts were the only means to obtain short-duration measurements of extrapolated 24-hour (24-h) resting energy expenditure (24-h REE) of ~30–45 min [1, 2]. The main disadvantage was the necessity for subject connection to the instrumentation using ventilated hoods or face masks. Furthermore, metabolic carts tend to be inherently inaccurate, with the possibility of underestimating extrapolated 24-h REE by up to 10% [3].

All indirect calorimetry methodologies have an inherent disadvantage. This is the inability to accurately correct for the presence of water vapor in the baseline and sample gases to allow expression of the metabolic results as Standard Temperature Pressure Dry [4, 5]. The utilization of new technologies [4] allows continuous direct measurement of water vapor pressure in both the baseline and sample gases [4] thus eliminating the need for its removal. This allows accurate extrapolations of 24-h REE from only a 1-h metabolic measurement utilizing whole-room indirect calorimetry (WRIC) [3].

WRIC may eliminate some of the subjects’ discomfort and anxieties associated with the use of metabolic carts thus possibly reducing errors in extrapolated 24-h REE [6, 7]. In the past, shorter-duration metabolic measurements were not possible with WRICs due to the inability of the instrumentation to obtain accurate measurements of oxygen and carbon dioxide in less than several hours. This was mainly due to inaccuracies caused by incomplete drying of the sample and baseline gases and the resolution of the oxygen and carbon dioxide sensors [4].

This report describes the derivation of extrapolated 24-h VO2, VCO2, RQ, and REE from recalculation of 60-min metabolic data from a previous study [3] to reflect a 30-min measurement duration thus being more on par with that of metabolic carts [8, 9].

Materials and methods

Instrumentation

The Sable Systems Promethion (Model GA3m2/FG250) integrated system (Sable Systems International, Las Vegas NV, USA) that was connected to a whole-room indirect calorimeter (WRIC), specific for the measurement of resting energy expenditure (REE), was utilized to obtain the data for this analysis [3]. This WRIC had an interior volume of 4,597 liters. This instrument contains two channels, each comprising of separate fuel cell oxygen, near inferred carbon dioxide, and thin film capacitive water vapor pressure sensors. A novel switching system is utilized to ensure continuous measurements of the sample gases from within the REE_WRIC. This involves switching between the two channels in a staggered time fashion. For example, 10-min after starting the metabolic test, channel #1 switches from measuring the sample gases to that of the baseline air stream for 2-min while channel #2 remains sampling the gases from within the REE_WRIC. After 2-min channel #1 returns to sampling the gases from the REE_WRIC. After an additional 10-min this same cycle occurs again but with Channel #2 switched to measuring the baseline air while Channel #1 remains sampling the gases from within the REE_WRIC. For both the 60-min propane burns and 1-h REE subject metabolic tests utilized for this analysis [3], baseline air was sampled by each sensor channel for a duration of 2-min, every 10-min. Upon completion of the metabolic 60-min tests the “gaps” that occurred in each channel sample gas data from the REE_WRIC were filled in mathematically by an Expedata software macro (version 1.9.27, Sable Systems International, North Las Vegas, NV) utilizing that derived from the opposite channel sampling the gases from within the REE_WRIC [4]. Upon completion of the metabolic test, continuous differences in oxygen and carbon dioxide between the baseline air and the sample gases were utilized for the final calculation of all the metabolic parameters utilizing the Expedata software.

Correction for the presence of water vapor

One unique feature of the Promethion integrated instrumentation is the continuous correction of the baseline air, sample gases, and airflow rates through the REE_WRIC for the presence of water vapor. This correction is achieved by direct measurements of water vapor pressure (kPa). The integrated water vapor pressure sensors utilized by Sable Systems International eliminate the need for separate measurements of temperature (C) and relative humidity (%) that would be necessary for the calculation of water vapor pressure [10].

Linearity of propane burn rates

Separate from the 60-min metabolic data utilized for this analysis, an additional 8-h propane combustion test was conducted to determine the linearity of burn rates according to procedures as described [11]. For this propane combustion test an analytical balance (Mettler Toledo Model MS1602S/03, Mettler Toledo LLC, Columbus, OH) was connected to the computer (Dell Optiplex 7070, Round Rock, TX) for continuous recording of the decrement of propane weight during this combustion test utilizing Lab X Direct Software (version. 2.5 Greifensee, Switzerland).

Recalculated 30-min propane combustion tests

One-hour propane combustion data, from a previous study [3] were recalculated utilizing Expedata software to reflect a 30-min duration for extrapolated 24-h (24-h) ventilation rates (V; l/d) of oxygen (O2), carbon dioxide (CO2), respiratory quotient (RQ; VCO2/VO2) and REE (MJ/d). The first and last 10-min of data were reserved to serve as a beginning and ending for Expedata’s smoothing algorithms, respectively, for the original 60-min duration. This was reduced to 5-min in terms of a 30-min testing duration. For the 60- and 30-min durations mean VO2 (l/min), VCO2 (l/min), and EE (MJ/min) across the middle 40- and 20-min data blocks, respectively, were then multiplied by 1440 (min in 24-h) to obtain the extrapolated 24-h metabolic parameters. The daily RQ was calculated as the mean across both the 40 and 20-min data blocks, respectively.

Human subjects’ data

Metabolic data involving 35 healthy, (37% female, Age: 38.6 ± 14.6 years, BMI 25.3 ± 4.9 kg/m2) from a previous study of 60-min metabolic measurements within the REE_WRIC [3] were recalculated based on a theoretical 30-min measurement duration as described for the propane combustion tests above. Furthermore, these subjects were weight stable, consumed a consistent diet, and were moderately active for three months prior to REE measurements [3].

Statistics

Statistical analyses utilized SPSS (version 29, Chicago, IL). Independent t-tests were utilized to determine differences between propane stoichiometry and combustion for extrapolated 24-h VO2, VCO2, RQ, and REE for both the original 1-h and 30-min recalculated propane data. Paired t-tests were utilized to determine if any differences existed in 24-h extrapolated VO2, VCO2, RQ, and REE when derived from 30-min recalculated versus that from the original 1-h subject data [3]. Pearson Correlations were utilized to determine the relationships between stoichiometrically derived 24-h extrapolated VO2, VCO2 RQ, and REE and actual results from propane combustion for both the 1-h and 30-min recalculated metabolic data. Similar correlations were utilized to determine the relationship between time and propane consumption during the 8-h propane combustion test. For the human subjects from the prior study [3] correlations were also performed between 30-min recalculated and the original 60-min extrapolated 24-h VO2, VCO2, RQ, and REE. Finally, both 30-min recalculated and the original 60-min 24-h extrapolated REE were correlated with that predicted by the Mifflin equation [3, 12].

The Bland-Altman [13] limit analysis was applied first to both the 1-h and 30-min recalculated propane combustion data to determine the agreement between related stoichiometries and actual combustion regarding extrapolated 24-h VO2, VCO2, and REE. This same analysis was applied to the human subject metabolic data to determine the magnitude of agreement between these same parameters including the RQ. Finally, proportional bias (p < 0.05) was determined by regressing mean differences for propane combustion for each duration for extrapolated 24-h VO2, VCO2, and REE on mean values obtained by both durations. A similar proportional bias was performed for the human subject data, but including the RQ, utilizing mean differences in the metabolic parameters obtained from each duration. All results are presented as mean ± standard deviation (p < 0.05) unless otherwise noted.

Results

Linearity of propane burn consumption

The mean burn rate across the 8-h combustion test was 0.1650 ± 0.0130 g/min. Furthermore, the consumption of propane remained linear over the course of this propane combustion test (Fig. 1).

Fig. 1
figure 1

The linear relationship between propane gas consumption and time for an 8-h combustion test.

Full size image

Recalculated 30-min propane burn combustion tests

All descriptive statistics for extrapolated 24-h VO2 (l/d), VCO2 (l/d), RQ (VCO2/VO2), and REE (MJ/d), when derived from 30-min recalculated propane stoichiometries and combustion data, are shown in Table 1.

Table 1 Descriptive statistics for 24-h extrapolated variables from recalculated 30 and previously published 60-min of REE metabolic data.
Full size table

For the 30-min duration, no differences existed between propane stoichiometry and combustion for any of the extrapolated 24-h metabolic parameters (Table 1). Furthermore, 30-min derived extrapolated VO2 (r = 0.91, p < 01), VCO2 (r = 0.99, p < 0.01), and 24-h REE (r = 0.93, p < 0.01), when calculated from propane stoichiometry, were correlated with that from combustion (data not shown).

The Bland-Altman limit analysis for the related comparisons between stoichiometry and combustion for the recalculated 30-min duration is presented in Table 2. There was good agreement between 30-min recalculated propane stoichiometry and actual combustion for extrapolated 24-h VO2, VCO2, and REE as reflected by the narrow confidence intervals and much tighter limits of agreement. Moreover, no significant proportional bias existed for these parameters, suggesting that larger average values do not lead to greater disagreements between propane stoichiometry and combustion.

Table 2 Bland-Altman limit analysis for comparison between 30 and 60-min extrapolated 24-h RMR metabolic data.
Full size table

Sixty-minutes propane combustion tests used as the standard of comparison

For the 60-min duration, no differences existed between propane stoichiometry and combustion for any of the extrapolated metabolic parameters (Table 1). Furthermore, 60-min derived extrapolated 24-h VO2 (r = 0.97, p < 01), VCO2 (r = 1.00, p < 0.01), and REE (r = 0.98, p < 0.01), calculated from propane stoichiometry, were correlated with that from combustion (data not shown).

The Bland-Altman limit analysis for the related comparisons between stoichiometry and combustion for the 60-min duration are presented in Table 2. There was good agreement between propane stoichiometry and actual combustion for extrapolated 24-h VO2, VCO2, and REE as reflected by the narrow confidence intervals and much tighter limits of agreement. Moreover, no significant proportional bias existed for any of these parameters.

Human subjects from a previous REE study

All descriptive statistics for extrapolated 24-h VO2 (l/d), VCO2 (l/d), RQ (VCO2/VO2), and REE (MJ/d) for the human subjects, as recalculated from 30, or derived from 60-min of metabolic data, are shown in Table 1. According to paired t-tests, no differences were found in any of the 24-h extrapolated metabolic parameters, whether recalculated from 30-min or derived from 1-h of metabolic data. The correlation between 60- and 30-min derived extrapolated 24-h REE is shown in Fig. 2A. Furthermore, 30-min derived extrapolated 24-h VO2 (r = 0.98, p < 01), VCO2 (r = 0.98, p < 01) and the RQ (r = 0.84, p < 0.01) were also correlated with that from the original 60-min metabolic measurements (data not shown). Finally, both the 30-min recalculated (r = 0.87, r < 0.01) and original 60-min 24-h extrapolated REE (r = 0.82, r = 0.01) were correlated with that derived by the Mifflin equation (data not shown).

Fig. 2: The validity of recalculated 30-minute extrapolated 24-h REE in human subjects.
figure 2

Relationships between recalculated 30 and 60-min derived extrapolated (Exp) 24-h REE (MJ/d) in 35 healthy adult non-smoking human subjects (plot A). The Bland-Altman limits of agreement analysis between recalculated 30 and original 60-min extrapolated 24-h REE (MJ/d) for 35 healthy adult non-smoking human subjects (plot B).

Full size image

The related comparisons in terms of the Bland-Altman limit analysis are presented in Table 2. There was good agreement between extrapolated 24-h VO2, VCO2, and REE from both recalculated 30-min data and that derived from 1-h subject metabolic measurements. This is reflected by the close agreement between recalculated 30-min and 1-h derived extrapolated 24-h REE, as shown by the Bland-Altman plot in Fig. 2B, where the delta extrapolated 24-h extrapolated REE was within 0.07 ± 0.33 MJ/d of zero. These close agreements are further reflected by the narrow confidence intervals and much tighter limits of agreement for each of these parameters. Moreover, no significant proportional bias existed for these parameters. However, the magnitude of the differences was greater (p < 0.05) between recalculated 30-min and that derived from 1-h metabolic measurements for the RQ, as reflected by the greater limits of agreement. This was also reflected in a much greater bias and 95% confidence interval, suggesting that greater disagreements will occur with larger average RQ values.

Discussion

This analysis suggests that only 30-min is necessary for accurate assessment of extrapolated 24-h VO2, VCO2, RQ, and REE utilizing the REE_WRIC. This is based on the recalculation of 60-min metabolic data from both propane combustion tests and human subjects from a prior study [3]. Furthermore, the percentage differences between stoichiometry and combustion for either the 1-h or recalculated 30-min extrapolated 24-h metabolic parameters were less than 2%. Similarly, these same percentage differences were maintained in the comparison of subject metabolic data between 1-h and that recalculated for 30-min. This agrees with the established accuracy criteria established for metabolic carts [2, 8, 9].

Linearity of propane burn consumption

Recalculation of 60-min propane burn data to reflect a 30-min duration depends on the linearity of propane consumption throughout a 60-min combustion test. This was verified by the linear decline in propane weight throughout an additional 8-h propane combustion test. Furthermore, the duration of this propane combustion test was necessary to determine if depressurization of the propane bottle would cause the burn rates to be non-linear towards the end of the test.

Propane combustion tests

This report demonstrated that all extrapolated 24-h metabolic parameters can be achieved from recalculated 30-min measurements derived originally from 1-h propane combustion tests [3]. This was shown by the lack of any differences between stoichiometry and propane combustion for recalculated extrapolated 24-h, VO2, VCO2, RQ, and REE. This is further substantiated by the resultant high correlations between recalculated 30-min propane stoichiometry and combustion for each of these metabolic parameters, except the RQ. Finally, stoichiometry versus propane combustion showed close agreement, narrow confidence intervals, and low bias for all the metabolic parameters according to the Bland-Altman limit analysis [13].

Human subjects from a previous REE study

The next phase of this analysis involved utilizing only the first 30-min of 60-min subject metabolic data from a previous study [3] to obtain extrapolated 24-h VO2, VCO2, RQ, and REE. This has been shown to be the case and further confirmed by the high correlations of the recalculated 30-min derived extrapolated metabolic parameters with that from 1-h [3], as well as good agreements according to the Bland-Altman limit analysis [13]. The lack of claustrophobic ventilated hoods or masks that might cause discomfort or anxiety [6, 7] was eliminated, possibly contributing to the stable results obtained. Furthermore, utilizing prediction equations, such as that by Mifflin, produced a predicted 24-h REE that was 17.5% lower than that obtained for the subjects. Our results suggest that 30-min measurement durations within the REE_WRIC are now on par with that of metabolic carts [2, 8, 9] and far more accurate than that derived by the Mifflin equation [12]. Finally, with all the above considered, using the REE_WRIC may be more appealing to individuals participating in various clinical or wellness programs where the measurement of 24-h REE is required. This is especially true for people with obesity who may not be able to have REE measured by a metabolic cart.