Use of infrared thermographic calorimetry to determine energy expenditure in preterm infants

¹ From the Department of Family Medicine, University of Wisconsin, Madison; the Department of Medical Research, The Carle Foundation, Urbana, IL; the Department of Internal Medicine, University of Illinois at Urbana-Champaign; the Department of Pediatrics, University of Iowa, Iowa City; and the University of Illinois College of Medicine, Chicago.

² Supported in part by Summer Medical Student Research Fellowship (to AKA) from the Society for Pediatric Research and NIH grants HD26945 and RR00059.

³ Address reprint requests to RA Nelson, Department of Medical Research, The Carle Foundation, 611 West Park Street, Urbana, IL 61801. E-mail: ralph.nelson{at}carle.com.

ABSTRACT  
Background: Measurement of infant energy expenditure in the clinical setting is difficult and is rarely done. Both indirect and direct calorimetry require long measurement periods and frequent calibration.

Objective: The objective of this study was to validate in infants a newly developed method of determining energy expenditure, infrared thermographic calorimetry (ITC), against an established method, respiratory indirect calorimetry (IC). ITC measures mean infant body surface temperature. ITC was used in conjunction with heat loss theory to calculate radiant, convective, evaporative, and conductive heat losses and thereby determine total energy expenditure.

Design: Ten healthy preterm infants were studied by obtaining concurrent ITC and IC measurements over a 3.5–5.5-h study period. Continuous IC measurements were compared with ITC measurements taken every 10 min during study periods. IC values were summed over 10-min intervals covering the 5 min before and 5 min after each ITC measurement, to allow comparisons between the 2 methods.

Results: Comparison of paired ITC and IC mean measurements for all 10 infants over the entire study period showed no significant difference between the 2 methods. However, individual paired IC and ITC values were significantly different for 7 of 10 infants. The overall mean difference between the 2 methods was 1.3%.

Conclusions: ITC is an accurate, noninvasive method for measurement of heat loss and energy expenditure in healthy preterm infants, and therefore it may be a useful clinical and research tool.

Key Words: Infrared thermographic calorimetry • energy expenditure • heat loss • indirect calorimetry • preterm infants • premature infants • metabolism • energy requirement • neonatology • neonatal nutrition • infant nutrition

INTRODUCTION  
For more than a century, investigators have been able to determine the energy expenditure of human subjects by direct and indirect calorimetry. Direct calorimetry is the determination of energy expenditure from measurement of heat loss by the body; indirect calorimetry (IC) is a method of indirectly calculating the heat produced by oxidative metabolism. Under appropriate conditions, these 2 methods have produced similar results, especially over longer periods (1, 2). Most studies of energy expenditure in infants have used indirect calorimetry because of the technical difficulties involved in using direct calorimetry. IC is easier to use than direct calorimetry technically, but IC is still rarely used for infants in the clinical setting because of the long measurement times and frequent calibration required.

Partitional calorimetry is the determination of the various forms of heat loss, which are summed to get overall energy expenditure; this method has been used previously in infants (3–7). Determinations of convective and radiant heat losses (3, 4) have been based on estimates of mean body surface temperature (MBST) derived from measurements of skin temperature at several sites. The use of thermistors and weighting of the resulting temperatures by proportional surface area to calculate estimates of nonevaporative heat loss requires assumptions about the distribution of body surface temperatures. The resulting estimates of mean surface temperature cannot be verified and may not reflect changes in vasomotor tone.

Infrared thermographic calorimetry (ITC) is an excellent method for determining MBST. When used in conjunction with heat transfer theory, MBST can be used to calculate heat losses from the body by radiation, convection, conduction, and evaporation. Total heat loss, and therefore energy expenditure, can be determined. Occasionally, corrections are necessary for changes in body temperature, which reflect heat storage. ITC is an innovative technique that enables the accurate study of heat loss, changes in energy expenditure, and metabolic heat production. ITC is a portable, fast, and noninvasive method, making it valuable for the study of human subjects in their natural environments. In adults, ITC has been used successfully to determine energy expenditure (8). ITC has been validated against IC in adults under a variety of conditions: during fasting, in patients receiving continuous parenteral nutrition, and during exercise at 30% of maximal work (8–10). The aim of this study was to validate ITC for determining the energy expenditure of infants by comparing it with the more established method of respiratory IC.

SUBJECTS AND METHODS  
Subjects
Ten preterm infants were enrolled. All the infants had birth weights appropriate for their gestational ages, were gaining weight at the time of study, and were in stable condition. None of the infants required supplemental oxygen or ventilatory support and none were receiving intravenous amino acids or lipids. Infant 5 was receiving intravenous dextrose and electrolytes and was the only subject that had had a serious complication, a grade 4 intraventricular hemorrhage with hydrocephalus. Four infants were initially diagnosed with respiratory distress syndrome and 2 infants, infants 8 and 10, were receiving caffeine and theophylline, respectively. Seven of the infants were fed by orogastric tube during the study period, whereas 3 infants (infants 1, 2, and 7) were removed from the incubator and bottle-fed.

Written informed consent was obtained from the infants' parents, and all infants were judged by their physicians to be in stable condition before each study. The study was approved by the University of Illinois and Carle Foundation Institutional Review Boards and the University of Iowa Human Subjects Review Committee.

Equipment
To perform simultaneous IC and infrared thermography measurements, a specially designed incubator was used. The incubator (C-86 Isolette; Air-Shields, Hatboro, PA) had a rectangular opening (43 cm long x 21 cm wide) cut in the top, with a polytetrafluoroethylene (hard, clear plastic) lid that opened for infrared scans. All of the infants were studied in this incubator regardless of the type of bed usually used (ie, crib or incubator) to minimize differences among study conditions for different infants.

Before each study, the incubator was set overnight to an air temperature equal to the infant's usual incubator temperature, or to 30°C for cradled infants. Infants 1 and 2 were studied with the incubator in the skin-temperature-servocontrol mode (incubator temperature was controlled by using a skin probe on the infant's back), but because the air temperature fluctuations were large (>1°C) with these infants, infants 3–10 were studied with the incubator set to air temperature control.

Study protocol
On the morning of the study, infants were placed into the warm incubator CO₂) were measured continuously. The infant's respiratory and heart rates were continuously monitored and were recorded at 10-min intervals. Infant activity level was graded every 5 min by using the Brück scale, which ranges from -4 (eyes closed, no movement) to 5 (crying) ( Temperature and environmental measurements
Temperatures were measured with thermistor probes (Series 400 probes; Yellow Springs Instrument Co, Yellow Springs, OH) attached to a data logger (Data-Logging System, model 2280B; John Fluke Manufacturing Co, Everett, WA) that recorded measurements at 15-s intervals. Each probe had been tested against a certified mercury thermometer (US National Bureau of Standards, Washington, DC) in a water bath, and correction equations generated for each probe were programmed into the data logger. The temperatures of incubator air (10 cm below the center of the top), room air, the incubator wall (inner slanted surface of the front top wall), the inner wall of the polytetrafluoroethylene head hood, hood air, infant skin (midback, above the iliac crest), and the 3 black bodies (reference temperature standards for the infrared camera) were also recorded by the data logger every 15 s. Infant axillary temperature, used as a measure of infant core temperature (12), was measured several times throughout the experiment with a mercury-in-glass thermometer. Incubator temperature was adjusted if axillary temperatures were <36.5°C or >37.4°C (11).

Mean incubator wall temperature was estimated from the upper front wall temperature by using the following equation:

RESULTS  
For the 10 infants, mean (±SD) values were as follows: birth weight, 1.28 ± 0.47 kg; gestational age, 31 ± 3 wk; study weight, 1.74 ± 0.23 kg; and age at the time of the study, 34 ± 23 d (Table 1). Mean CO₂ was 9.69 ± 0.98 mL•kg^-1•min^-1 and mean respiratory quotient was 1.08 ± 0.04 (Table 2
View this table:
TABLE 1.. Clinical characteristics of subjects  

View this table:
TABLE 2 .. Results of indirect calorimetry¹  
Mean air temperature in the study incubator was 31.7 ± 1.9°C (range: 29.7–34.7°C) for all infants (Table 3). The first 2 infants experienced the highest mean incubator air temperatures (34.7 and 34.4°C); these infants were studied with the incubator in the skin temperature control mode. Incubator air temperature was also more variable for these infants, with a range of > 1.0°C. The rest of the infants were studied with the incubator in the air temperature control mode, and the variability in air temperature decreased, with a range of <0.6°C. Mean incubator wall temperature was 30.2 ± 0.9°C for all infants and varied less than air temperature. Relative humidity was similar for the infants, with a mean of 29.6%. Activity levels were similar among the infants, with most of the activity scores being <-2 (mean: -2.4 ± 0.5).

View this table:
TABLE 3 .. Temperature and activity measurements  
MBST did not differ between scans done with and without head hoods (Table 3). MBST was significantly correlated (r = 0.90, P < 0.05) with incubator air and wall temperatures but not with skin (back) temperature as measured by thermistor probe (r = 0.31). Six infants showed increases in surface temperature over time (MBST, final - initial) over time. The mean difference between beginning and ending temperatures for infants 3–10 was sufficiently small to justify neglecting heat storage in the calculation of energy expenditure, with the largest increases in infants 1 and 2 (Table 3). Four infants showed slight decreases in MBST over time.

Variability of energy expenditure was greater for values determined by IC than for those determined by ITC. Use of mean No significant differences in energy expenditure were found among paired ITC and IC results (within-infant comparisons) for 3 of 10 infants, but significant differences were found for the other 7 infants (Table 4). The direction of the differences was not consistent. For the purpose of discussion, infants were divided into 3 groups: group 1 (infants 1, 2, and 10), where ITC < IC; group 2 (infants 5, 8, and 9), where ITC = IC; and group 3 (infants 3, 4, 6, and 7), where ITC > IC based on individual results of paired ITC and IC values (Table 4).

View this table:
TABLE 4.. Summary of heat loss and heat production data used for within-infant comparisons¹  
The results for group 1 may be explained by a heat-gaining environment for infants 1 and 2, causing considerable heat storage that was not dissipated during the study period. Infant 10 was receiving theophylline, which may explain some of the increased When ITC and IC means were compared for all 10 infants based on 19 paired measurements for each infant (among-infant comparison), no significant difference between the 2 methods was found (Table 5); the mean difference between the 2 methods was 0.15 ± 1.17 kJ•kg^-1•h^-1. Mean IC and ITC values based on these 19 time points (Table 5) were not significantly different from means based on all time points (Table 4). The mean difference between IC and ITC based on all time points was –0.24 ± 1.22 kJ•kg^-1•h^-1 (NS; Table 4). The mean energy expenditure (IC) was 11.62 ± 1.24 kJ•kg^-1•h^-1 (Table 5). In Figure 2, the mean (±SD) heat loss and heat production for the 10 infants over the entire time period are shown. The mean difference between the 2 methods for each infant is also presented. These numbers were derived from the summary data presented in Table 5. The minimum difference between the 2 methods was 0.004 kJ•kg^-1•h^-1, which is only 0.10 kJ•kg^-1•d^-1 (0.04% of the IC value). The maximum difference between the 2 methods was -2.14 kJ•kg^-1•h^-1, or 51.31 kJ•kg^-1•d^-1 (18% of the IC value). The 95% CI for the differ-ence between the 2 methods was -0.26 to -0.04 kJ•kg^-1•h^-1, or -10.51% to 4.79%.

View this table:
TABLE 5.. Mean heat loss measured by infrared thermographic calorimetry (ITC) and heat production measured by indirect calorimetry (IC): data used for among-infant comparisons  

View larger version (14K):
FIGURE 2. . Mean (±SD) heat loss, as measured by infrared thermographic calorimetry (ITC; ), and heat production, as measured by indirect calorimetry (IC; ), for all subjects (n = 10). The mean difference () between ITC and IC was -0.15 kJ•kg^-1•h^-1. Paired differences were not significant.

 
Results of the repeated-measures analysis of variance showed no significant difference between IC and ITC for period A, the initial postfeeding period (85–135 min), but a significant interaction between time and method occurred (P < 0.001). This may have been the result of slight decreases in IC values over this time period. For period B, the second postfeeding period (10–120 min), differences between the 2 methods were significant (P < 0.001). During this period, there was a small increase in IC results over the first 30 min postfeeding.

DISCUSSION  
Few studies have simultaneously compared heat loss and heat production in infants. In this study, we described and validated ITC, a new method of determining infant heat loss and energy expenditure. ITC is an accurate method for the measurement of MBST and interferes less with the infant's environment than does IC. In conjunction with heat loss equations adapted for this purpose, ITC enables the determination of radiant, convective, evaporative, and conductive heat losses from infants. Ten preterm infants were studied with concurrent ITC and IC during study periods lasting a mean of 281 min. No significant difference was found between the 2 methods overall, with a mean difference (ITC - IC) of -0.15 kJ•kg^-1•h^-1, or 1.3%.

Similar results were obtained by other investigators who compared heat loss and heat production. Howland (1) performed simultaneous direct and indirect calorimetry in 2 infants and found a difference of <3% in 13 comparisons. In a study of 25 preterm infants, Day (3) used concurrent direct and indirect calorimetry and found no significant difference between the mean values for energy expenditure. Bell (27) used concurrent direct and indirect calorimetry to determine energy expenditure in 20 growing preterm infants during 2 consecutive postprandial periods; the mean difference between the 2 methods (2.4%) was not significant.

In our study, no significant difference was found between the 2 methods of determining energy expenditure overall, although the degree of concordance varied among infants. In 7 of 10 infants, within-infant paired t tests showed a significant difference between the 2 methods. Significant differences were not found for the remaining 3 infants. The range of differences found was 0.04–18.51%, and the larger difference was greater than the error of either method. However, because there was no consistent relation of ITC to IC, further adjustments of coefficients in the heat loss equations would not alter the individual differences. Most of the infants had differences of <10% between the 2 methods, which would result in minimal differences in energy expenditure per day if estimated by ITC alone compared with IC.

Studies by Benedict (28) showed that on only 8 occasions in 288 periods of 2 h each were direct and indirect calorimetry in agreement; the differences were in the range of ±10% of total energy expenditure. However, direct and indirect calorimetry were in nearly perfect agreement when summed over 24 h. Webb (29) also noted that heat production and heat loss rarely were the same over short periods of time. In infants, similar differences were found between direct and indirect calorimetry. In the group of 25 infants studied by Day (3), 17 of 50 cases had measurements that differed by >10%. In a study of concurrent direct and indirect calorimetry in 14 growing preterm infants, Sauer et al (30) found that energy expenditure determined by direct calorimetry was lower by an average of 7.4%.

Physiologic factors are most likely the primary cause of differences between the 2 methods at individual time points. Heat production and heat loss are in constant flux and are affected by changes in environment, feeding, handling, activity, and medical condition. These influences on energy expenditure affect heat production and heat loss differently and there may be a time lag before the effects appear. Heat loss will exceed heat production if body core temperature rises and heat is stored, which in turn stimulates heat loss and suppresses metabolism. In this case, ITC would exceed IC. The opposite occurs if body temperature decreases; heat conservation is stimulated and metabolic rate increases. In that case, IC exceeds ITC. This negative feedback mechanism for maintaining normal body temperature is probably responsible for differences in heat production and heat loss over short periods of time (8).

Infants 1 and 2 showed the greatest differences between the 2 methods, with IC values being greater than ITC values. Mean skin temperature increased in these infants throughout the study period. Calculation of heat storage based on change in MBST showed values of 0.95 and 0.62 kJ•kg^-1•h^-1 for infants 1 and 2, respectively. Addition of these heat storage values to ITC values would significantly decrease the differences between the 2 methods for these infants. Calculation of heat storage by using this method for the remaining infants yielded values of <0.17 kJ•kg^-1•h^-1 and did not explain differences between the 2 methods.

Infants 3–7 (group 3), for whom ITC-derived values were greater than those of IC, had an expected increase in ^• VO₂ in the half hour after feeding. As seen in studies of adults, for the remainder of the study periods, heat loss exceeded heat produced and the difference between ITC and IC increased. Infants in group 3 may have been showing an earlier or enhanced diet-induced thermogenesis compared with those in group 2. This may have been a result of differences in body composition, maturation, or ability to conserve heat. No trends in birth weight, age (in days), mode of feeding, activity level, or diagnosis were noted to influence differences between the 2 methods. Given that determinations of energy expenditure by IC are most accurate when summed over periods of hours (31) because of the variability of energy expenditure determined over shorter periods, it seems that the differences seen in this study were small and within the range of experimental and biological variation.

The CO₂ data in this study were similar to results found in previous research with similar preterm infants (7). In our group of growing preterm infants, all respiratory quotients determined by IC were >1.0, which indicated a state of net lipogenesis (31). The variability of energy expenditure as determined by IC was greater than that determined by ITC, and was partly a result of the short (10-min) intervals of time used in the analyses. Bell et al ( ITC is an accurate method for the determination of energy expenditure in groups of healthy preterm infants over periods of several hours. Paired differences between ITC and IC over time were similar to previous observations (3, 8, 28, 29) and appeared to mainly reflect true differences between metabolic heat production and total heat loss. ITC may be less reliable for ascertaining the energy expenditure of individual infants; however, it should correctly assess changes in energy expenditure resulting from interventions, such as changes in the thermal environment, even in individual cases. Recent use of ITC in infants under radiant warmers was valuable for monitoring infant responses to changes in the thermal environment (33). In addition, because of the decreased intra-infant variability in ITC measurements compared with IC, estimation of energy expenditure by ITC may require a shorter period of measurement than IC. Further study will be necessary to determine the minimum number of ITC measurements needed to accurately quantify 24-h energy expenditure in infants. ITC is a portable and noninvasive method that holds promise for measuring the energy expenditure of infants in research and clinical settings. It makes possible the measurement of heat loss by each route—radiant, convective, evaporative, and conductive. Use of heat loss equations also allows prediction of changes in energy expenditure resulting from changes in environmental conditions.

ACKNOWLEDGMENTS  
We thank Karen J Johnson, who provided valuable assistance by helping to monitor the subjects during the studies.

REFERENCES  

1. Howland J. Direct calorimetry of infants with a comparison of the results obtained by this and other methods. Trans 15th Int Congress Hygiene Dermatol 1912;2:438–51.
2. Benedict FG, Talbot FB. Metabolism of growth from birth to puberty. Washington, DC: Carnegie Institute of Washington, 1921:1–213. (Publication 302.)
3. Day R. Respiratory metabolism in infancy and childhood. XXVII. Regulation of body temperature of premature infants. Am J Dis Child 1943;65:376–98.
4. Hey E, Katz G, O'Connell B. The total thermal insulation of the newborn baby. J Physiol 1970;207:683–98.
5. Hammarlund K, Nilsson GE, Oberg PA, Sedin G. Transepidermal water loss in newborn infants. V. Evaporation from the skin and heat exchange during the first hours of life. Acta Paediatr Scand 1980;69:385–92.
6. Hammarlund K, Stromberg B, Sedin G. Heat loss from the skin of preterm and fullterm newborn infants during the first weeks after birth. Biol Neonate 1986;50:1–10.
7. Bell EF, Rios GR. A double-walled incubator alters the partition of body heat loss of premature infants. Pediatr Res 1983;17:135–40.
8. Shuran M, Nelson RA. Quantitation of energy expenditure by infrared thermography. Am J Clin Nutr 1991;53:1361–7.
9. Pogue NJ, Nelson RA. Quantifying mean skin temperature and sources of heat losses, before, during and after moderate steady state exercise. Fed Proc 1988;47:1313 (abstr).
10. Pogue NJ, Nelson PA, Boileau RA. Non-contact measurement of energy expenditure during steady-state aerobic exercise. Physiologist 1992;35:187 (abstr).
11. Brück K. Neutral temperature range and range of "thermal comfort" in premature infants. Biol Neonate 1962;4:32–51.
12. Mayfield SR, Bhatia J, Nakamura KT, Riss GR, Bell EF. Temperature measurement in term and preterm neonates. J Pediatr 1984; 104:271–5.
13. Lusk G. The elements of science and nutrition. 4th ed. Philadelphia: WB Saunders Co, 1928:64–8.
14. Eckburg J. Energy balance of a reclining patient. Master's thesis. University of Illinois, Urbana, 1990.
15. Adams AK, Gill AS, Ek JR, Nelson RA. Contribution of conduction to total infant energy expenditure. J Therm Biol 1998;23:267–73.
16. DuBois. Basal metabolism in health and disease. 3rd ed. Philadelphia: Lea & Febiger, 1936.
17. Gagge AP, Nishi Y. Heat exchange between human skin surface and thermal environment. In: Lee DHK, ed. Handbook of physiology—reactions to environmental agents. Bethesda, MD: American Physiological Society, 1977:69–92.
18. Fanger PO. Thermal comfort. Copenhagen: Danish Technical Press, 1970.
19. Nishi Y, Gonzalez R, Nevins R, Gagge A. Field measurement of clothing thermal insulation. ASHRAE Trans 1976;82:248–59.
20. Stolwick J, Hardy JD. Control of body temperature. In: Lee DHR, ed. Handbook of physiology—reactions to environmental agents. Bethesda, MD: American Physiological Society, 1977:45–68.
21. Hanson R de G. Respiratory heat loss at increased-core-temperature. J Appl Physiol 1974;37:103–7.
22. Nishi Y, Gagge AP. Moisture permeation factor for clothing—a factor governing thermal equilibrium and comfort. ASHRAE Trans 1970;75:137–45.
23. Nishi Y, Ibamoto KL. Model skin temperature—an index of thermal sensation in cold, warm and humid environments. ASHRAE Trans 1969;75:94–107.
24. Gagge AP. A new physiological variable associated with sensible and insensible perspiration. Am J Physiol 1937;120:277–87.
25. Smithsonian Physical Tables. 1918:183–4.
26. SAS Institute Inc. SAS user's guide: statistics, version 5.18. Cary, NC: SAS Institute Inc, 1988.
27. Bell EF. Direct calorimetry of preterm infants. Dev Physiopathol Clin 1991;2:83–6.
28. Benedict FG. Vital energetics. Washington, DC: Carnegie Institution of Washington, 1938:178–94.
29. Webb P. The measurement of energy exchange in man: an analysis. Am J Clin Nutr 1980;33:1299–310.
30. Sauer PJ, Dane HJ, Visser HKA. Longitudinal studies on metabolic heat loss, and energy cost of growth in low birth weight infants. Pediatr Res 1984;18:254–9.
31. Elia M, Livesey G. Theory and validity of indirect calorimetry during net lipid synthesis. Am J Clin Nutr 1988;47:591–607.
32. Bell EF, Rios GR, Wilmoth PK. Estimation of 24-hour energy expenditure from shorter measurement periods in premature infants. Pediatr Res 1984;18:254–9.
33. Ek J, Bell EF, Nelson RA, Radhi MA. Infrared thermographic calorimetry applied to preterm infants under radiant warmers. J Therm Biol 1999;23:97–103.