Introduction

In individuals with a spinal cord injury (SCI), a positive energy balance increases the risk of obesity and is associated with numerous adverse metabolic sequelae, including glucose intolerance and cardiovascular disease.1, 2 By contrast, reduced caloric intake could lead to further reduction in the level of physical activity and may affect activities of daily living.3 Basal metabolic rate (BMR) accounts for ~70% of the total daily energy expenditure in sedentary individuals4 and, therefore, a reliable estimation of BMR would be clinically important to inform and to appropriately adjust caloric intake in individuals with SCI.

Previous studies reported mean BMR to range from 1256 to 1855 kcal day−1 in the SCI subjects.5 Specifically, Yilmaz et al.6 and Gorgey and Gater7 reported a difference in BMR between paraplegic and tetraplegic subjects of 370 and 224 kcal day−1 respectively, while Collins et al.8 reported no difference between the two groups.

BMR is often estimated as a function of body weight and/or height, using predictive equations that have been validated in able-bodied subjects. When used for individuals with SCI, these equations have been reported to overestimate 4–92% of the measured values of BMR.5

Researchers suggested that lower than predicted BMR in individuals with SCI could be because of reduced fat free mass (FFM).9 After adjustment for FFM done by dividing BMR by FFM (kcal kg−1), the BMR in individuals with SCI was found to be comparable to that of able-bodied individuals.10 Therefore, it might be necessary to predict the BMR on the basis of FFM in individuals with SCI.5

However, available reports estimating BMR in individuals with SCI based on the assessment of FFM are scarce. In one of them, Sedlock and Laventure11 measured FFM using hydrodensitometry in four men with SCI and reported that, in three of these patients, the predicted BMR values were higher than the measured values (by 24%, 16%, 9%, respectively), whereas, a 3% lower value was observed in the remaining fourth subject.

In this study, we compared the BMR values measured by indirect calorimetry (IC) with those estimated from FFM, assessed by dual X-ray absorptiometry (DXA). Through this comparison, we aimed to validate the existing BMR predictive equation which includes FFM as an independent variable12 in the motor complete SCI population; our second objective was to develop a new SCI population-specific equation based on the FFM assessment.

Methods

Participants

A total of 50 individuals (38 males, 12 females) with motor complete SCI presenting to the outpatient clinic of the authors’ institute, a tertiary level general hospital, were recruited for the study. All participants were over 18 years of age (Table 1). Participants with psychologic problems, a history of infection, such as urinary tract and respiratory tract infections, within one month of the assessment were excluded. Participants were classified with regards to the level and severity of neurological impairment on the American Spinal Injury Association Impairment Scale (AIS), based on an assessment performed by an experienced rehabilitation physician during clinic visits. The methods used in the present study were approved by the institutional review board of the Seoul National University Hospital (IRB No. 1412-115-634), and all participants provided written informed consent.

Table 1 Characteristics of participants
Full size table

Indirect calorimetry

Participants underwent BMR measurements using IC between 0800 and 1000, following an overnight fast after dinner the previous day was provided, with no intake of food, caffeine, nicotine, or medications. All BMR measurements were completed at a room temperature of 25±1 °C. Before the BMR measurement, bladder was emptied by catheterization, as all participants were not able to void voluntarily. After catheterization, the participants rested in the supine position for 20 min. The Quark CPET system (Cosmed, Italy) was calibrated according to the manufacturer’s guidelines. After calibration, the canopy system was placed over the participants’ heads, with the subjects resting in the hood for an additional 5 min before the start of the data collection. Following this accommodation period, the gas exchange data were recorded for 30 min. However, only the data from the lowest 10 min were selected and then averaged as the final BMR data for each subject. Oxygen consumption (VO2) and carbon dioxide production (VCO2) were automatically recorded every 5 s, with a variability within 10% of the mean VO2 and VCO2, indicative of a steady state gas exchange.

Dual energy X-ray absorptiometry

Body composition was measured using a whole-body DXA scanner (GE Lunar Prodigy, WI, USA). The participants were transferred from the wheelchair to an adjustable height stretcher car and then subsequently moved to the scanner bed using a transfer board. All scans were performed and analyzed by a trained technician using the Lunar software (version 13.31) and whole body fat mass and FFM (lean mass and bone mineral content) were calculated. The BMR value was estimated, based on the FFM values using the Cunningham equations; BMR=21.6 × FFM+370.12

Statistical analysis

To develop a new BMR predictive equation for the persons with motor complete SCI, a multiple linear regression model with backward elimination was used. Age, sex, level of injury (paraplegia or tetraplegia), duration, fat mass, and FFM were included in the model as possible predictors of BMR. The residual was examined to assess modeling assumptions. The differences between the measured values of BMR and those estimated from the new equation, as well as the existing equation, were calculated. The limits of agreement (mean difference±2 s.d.) were calculated and denoted on Bland–Altman plots.

The percent difference between the estimated and the measured values of BMR was calculated using equation (1):

Intra-class correlation coefficients (ICCs) were also computed to evaluate the agreement between the measured and the estimated BMR values.

All analyses were performed using the Statistical Package for the Social Science (SPSS, version 20.0, IBM, NY, USA); a P-value<0.05 was considered to be statistically significant.

Results

Of the 50 participants included in the final analysis, 27 had a cervical spine lesion and 23 had a thoracic spine lesion, with 29 participants categorized as class ‘A’ on the AIS and 21 as class ‘B’. The characteristics of the participants are summarized in Table 1. The mean value of BMR measured by IC was 1274.8 (range: 857–1757, s.d.=235.2) kcal day−1.

Validation of the existing equation

The mean value of BMR predicted by the Cunningham equation was 1280.1 (range: 933.6–1676.8, s.d.=181.0) kcal day−1 and the ICC between values measured by IC and estimated from the equation was 0.845.

The mean of differences was 5.4 (s.d.=118.0) kcal day−1 and the limits of agreement ranged from −230.6 to 241.3 kcal day−1. The mean of % difference was 1.62 (s.d.=10.1) % with the limits of agreement ranged from −18.5 to 21.8%.

Development of the prediction equation

Using the previously described multiple linear regression model, equation (2) was developed to estimate BMR in motor complete SCI patients. Predictors other than FFM were eliminated and the equation included only FFM and a constant with the R-square value of 0.76.

The mean value of BMR predicted by the newly developed equation in this study was 1276.8 (range: 883.8–1726.8, s.d.=205.3) kcal day−1 and the ICC between values measured by IC and estimated from the equation was 0.866 (Figure 1).

Figure 1
figure 1

Basal metabolic rate (BMR) predicted by the developed equation (BMR-Equ) and BMR measured by indirect calorimetry (BMR-IC) with the line of equality shown.

Full size image

The mean value of differences was 2.1 (s.d.=115.5) kcal day−1 and the limits of agreement ranged from –229.0 to 233.1 kcal day−1 (Figure 2a). The mean of % difference was 1.0 (s.d.=9.76) % with the limits of agreement ranged from –18.5 to 20.6% (Figure 2b).

Figure 2
figure 2

Bland–Altman plot. The differences between predicted and measured values of the basal metabolic rate (BMR) when the developed equation was applied were plotted. Horizontal solid line represents the mean value and horizontal dashed lines the 95% confidence interval; (a) difference in BMR (kcal day−1); (b) difference in percent ratio.

Full size image

Discussion

When the newly developed equation and the Cunningham equation were applied, the mean values of predicted BMR were 1276.8 kcal day−1 and 1280.1 kcal day−1 respectively for 50 subjects with SCI. These mean values were very close to the measured BMR value of 1274.8 kcal day−1. However, the limits of agreement where 95% of individual differences lie between these limits ranged from −229.0 to 233.1 kcal day−1 in the newly developed equation and from −230.6 to 241.3 kcal day−1 in the Cunningham equation. These results suggest that at individual level, there is a considerable amount of bias from the BMR values measured by IC when applying FFM-based BMR prediction equations to persons with SCI.

Gorgey et al.13 suggested that body weight can be primarily used to predict whole body FFM in men with chronic complete SCI and proposed the following prediction equation: whole body FFM=0.288 × body weight (kg)+26.3. When this equation was applied to our 38 male participants, the mean value of the estimated FFM amounted to 45.6±4.0 (mean±s.d.) kg and no statistical difference between the estimated and the DXA measured values of FFM in this study (mean value of 45.3 and s.d. of 6.8 kg) was observed. Furthermore, the ICC of the agreement between the predicted and the DXA measured values of FFM was 0.833. Therefore, the combined use of both the FFM prediction equation proposed by Gorgey et al.13 and the BMR prediction equation based on the FFM developed in this study could be considered in clinical environment where the body composition analysis techniques, such as DXA, are not available for male persons with chronic motor complete SCI.

In the present study, the mean FFM of 50 subjects was 42.1 (s.d.=8.4) kg, which was lower than the values reported in previous studies. For example, Gorgey et al.13 reported an FFM of 49.9 (s.d.=7.3) kg in 21 subjects with motor complete SCI, while Spungen et al.14 reported a lean mass of 45.9 (s.e.=0.1) kg in 67 subjects with paraplegic SCI. These differences might be due to ethnic differences. All participants recruited in the present study were East Asian and their average body mass index (BMI) was lower (22.0 kg m−2), as compared with the average BMI values in Gorgey et al.13 and Spungen et al.14 (26 kg m−2 and 25.8 kg m−2, respectively). On the other hand, in their analysis of 44 Japanese SCI patients (that is, ethnically closer to the subjects of the present study), Maruyama et al.15 reported the lean mass of 38.5±6.38 (mean±s.d.) kg, which was also lower than those reported in previous studies.13, 14

The limitations of this study need to be acknowledged. First, spasticity, which is commonly observed in individuals with SCI, might influence the BMR values measured by indirect calorimetry. However, in our study, we did not assess the distribution and severity of spasticity, nor did we control for pharmacological anti-spasticity treatment. Moreover, as the severity of spasticity can vary according to a patient’s physical and emotional condition,16 a single measurement of BMR by IC might be inappropriate for the evaluation of BMR in individuals with SCI having severe spasticity.17 Second, although DXA is becoming more accessible and widely accepted as a technique to measure body composition, the variabilities among manufacturers of the DXA instruments, different beam technologies within the same manufacturer and different software versions still remain issues.18 For example, significant mean differences of the total body fat proportion of 2.6–6.3% in able-bodied volunteers were observed between pairs of the DXA instruments.19 Therefore, when comparing the results of the present study with past or future studies on the BMR estimations using body composition measurement techniques in SCI populations, these variabilities in the FFM measurements should be taken into account.

Although there is a high level of correlation between BMR and FFM, as demonstrated by Gorgey et al.,17 the BMR–FFM models might be too general and thus unsuitable for individual BMR prediction, as gender- and age-related, as well as ethnic, hormonal, and other sources of variations in the BMR–FFM relationship should be considered on the individual level.20 Alternative methods for the measurement of BMR in individuals with SCI, such as a hand held IC might be considered useful in future research. As compared with traditional IC systems, hand-held IC systems only measure VO2, with the reserve quotient assumed to be 0.85, and require less expertize for use.21 Although the data from hand held devices are limited in terms of validity and clinical utility, these systems might provide more accurate measures of BMR than the predictive equations.22, 23

Conclusions

A considerable amount of bias from the BMR values measured by IC was still observed, which necessitates clinical consideration when applying FFM-based BMR prediction equations to individuals with SCI.

Data Archiving

There were no data to deposit.