Article Text

Download PDFPDF

Measurement of carbon dioxide production in very low birth weight babies


BACKGROUND CO2production is most commonly measured by using indirect calorimetry to quantify elimination of CO2 in breath (Vco 2). An alternative is to measure the rate at which CO2 appears in the body pool (Raco 2) by infusing a 13C labelled bicarbonate tracer. Vco 2 and Raco 2 generally differ but are related byc, a factor that adjusts for the incomplete recovery of infused tracer in the breath. The literature relating to human studies cites a wide range of values forc but the only neonatal study to determinec empirically estimated a mean value of 0.77.

AIM To estimate fractional recovery rate, c, in very low birthweight babies, and assess the feasibility of using the isotopic technique to measure CO2 production during mechanical ventilation.

METHOD Eleven spontaneously breathing, continuously fed, very low birthweight infants (median birth weight 1060 g, median gestational age 29 weeks) were studied.

RESULTS Mean (SD) Vco 2 was 9.0 (2.0) ml/min (standard temperature and pressure dry, STPD) and mean (SD) Raco 2 was 9.6 (2.1) ml/min (STPD). The mean (SD) value ofc was estimated as 0.95 (0.13). The 95% confidence intervals of the mean were 0.87–1.03.

CONCLUSIONS The results emphasise the importance of measuringc for a given study population rather than assuming a value based on adult studies. The close approximation of Raco 2 and Vco 2 in this group of babies implies that the labelled bicarbonate infusion technique could be used to measure simply CO2 production during mechanical ventilation.

  • carbon dioxide
  • carbon isotopes
  • calorimetry
  • very low birthweight babies

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The measurement of CO2 production has important clinical applications in the management of very small babies. If O2 consumption is measured simultaneously, respiratory quotient can be calculated, allowing total energy expenditure and nature of fuel oxidised to be deduced. Potentially this could help in the choice of more appropriate nutritional interventions for infants who show constrained capacity to eliminate CO2 because of lung disease. Measuring CO2 production is especially problematical in ventilated very low birthweight (VLBW) infants.1 Conventionally, breath CO2elimination (Vco 2) is calculated by measuring gas flow and the CO2 content of inspired and expired breath. This technique, indirect calorimetry, is subject to both technical errors, such as leakage of expired gas, and confounding by changes in minute ventilation.2 An alternative approach is measurement of the rate at which CO2 produced by metabolism appears in the body bicarbonate pool (Raco 2, rate of appearance). This can be accomplished by measuring the isotopic enrichment (ratio of 13C to 12C) of CO2 in expired breath while continuously infusing a13C labelled sodium bicarbonate tracer at a known rate.3 ,4

Studies of animals and adult humans have consistently shown that Vco 2 and Raco 2differ.4 ,5 The discrepancy is principally attributable to incomplete recovery of 13C labelled CO2 in breath when there is insufficient time for infused label to equilibrate between body bicarbonate pools. A correction factor (c), given by the quotient of Vco 2 and Raco 2, can be used to adjust for the fractional recovery of tracer.6-8The literature cites mean values of cranging between 0.5 and 1.06 in different patient groups,4 ,5 and a value of 0.80 is often assumed in studies of substrate oxidation. A single published study of neonates yielded a mean (SD) estimate of 0.77 (0.05) but included only three babies weighing < 1500 g.7

In view of the potential importance of accurately quantifying CO2 production in VLBW babies and uncertainty about the magnitude of c in this group, we have simultaneously measured Vco 2 and Raco 2 in spontaneously breathing babies and demonstrated the feasibility of applying the isotopic technique during mechanical ventilation.



Eleven spontaneously breathing VLBW (birth weight < 1500 g) babies were studied. All had clinically indicated venous access and weighed < 1500g at the time of the study. Four were breathing ambient oxygen, seven were breathing air. Feeds were either expressed breast milk (n = 8) or a standard preterm formula (n = 1) administered by hourly intragastric bolus. One infant received total parenteral nutrition, and one both total parenteral nutrition and expressed breast milk. Table 1 summarises clinical details. A single ventilated baby was later studied to address problems of sampling from the endotracheal tube and to examine within patient variation in Raco 2. St George's Healthcare NHS Trust research ethics committee approved the study; informed written consent was obtained from the parents.

Table 1

Clinical characteristics of the patients studied


Figure 1 summarises the protocol. Intragastrically fed babies were changed from hourly bolus feeds to continuous feeding at least one hour before the start of the study. A bolus dose of tracer (NaH13CO3) was given intravenously, followed by continuous tracer infusion. Simultaneously a Deltatrac II metabolic monitor (Datex, Helsinki, Finland) was used to measure Vco 2 throughout the first hour, after which breath was sampled intermittently to measure isotopic enrichment. Babies were routinely nursed in incubators, and temperature, heart rate, respiratory rate, Sao 2 continuously monitored.


NaH13CO3 (99% 13C) was obtained from Promochem Ltd, Welwyn Garden City, Herts, UK. A 2 mg portion was diluted in 1 ml unlabelled 2.74% NaHCO3 and packaged in 5 ml sterile pyrogen-free ampoules by the Northwick Park and St Mark's NHS Trust Pharmacy.


A priming dose of NaH13CO3 (0.25 mg/kg) was given after collection of two baseline samples to measure background 13C enrichment. NaH13CO3(0.2 mg/kg/h) was then infused continuously using an IVAC syringe driver (Alaris, Basingstoke, Hants, UK). Isotope dosage was measured by weighing on a Sartorius top loading balance with a resolution of 1 mg. A loose fitting face mask and 10 ml syringe were used to sample breath at 30 minute intervals. Breath was then expelled into evacuated tubes (Exetainers; Labco Ltd, High Wickham, Bucks, UK) for storage at room temperature. 13C enrichment was measured by isotope ratio mass spectrometry at the Bureau of Stable Isotope Analysis, Brentford, Essex, UK. (The Bureau provides a postal service at commercial rates.) δ (del) values were converted into atom percent excess (APE) using the formula8:Embedded Image The rate of appearance of CO2(Raco 2) was then calculated using the standard single pool model equation6 ,8:Embedded Image in which F represents the rate of infusion of NaH13CO3 (μmol/kg /min), Ei the enrichment of infusate (99% APE), Eb the 13C enrichment of expired breath at steady state (in APE). Raco 2 was then converted into ml/kg/min, standard temperature and pressure dry (STPD) by applying Avogadro's constant (1 mole of gas ≡ 22.4 litres, STPD). Breath isotopic enrichment was plotted against time, and the plateau defined according to the convention of taking four or more consecutive points with a coefficient of variation of < 5%.


The single pool model assumes that all bicarbonate administered and CO2 produced by metabolism enters and leaves the body from a single, rapid turnover pool (fig 2). In practice, this probably communicates with slow turnover pools, representing relatively inert tissues—for example, bone.3 During short studies, insufficient time may elapse for tracer equilibration between pools, causing incomplete recovery of tracer in the breath. This will cause overestimation of Raco 2, as breath isotopic enrichment (Eb) forms the denominator in the calculation (equation 2). To compensate for this, a correction factor (termedc, also known as the fractional recovery rate) can be derived as follows.6-8 It is first assumed that all CO2 administered as tracer and produced by metabolism is eliminated only in breath, that body CO2/bicarbonate pool size is unchanged, and that full equilibration occurs. Under such conditions: Vco 2 = Raco 2 (3)

Figure 2

Model of CO2/HCO3 kinetics. All infused tracer and HCO3 /CO2 generated by metabolism enters and leaves the body from a central rapid turnover pool. This eventually attains equilibrium with slow turnover pools. The factor c corrects for overestimation of breath13C/12C enrichment when a study is too short for this state to be attained.

The factor c, correcting for incomplete recovery and consequent overestimation of Raco 2is then given by: Vco 2 = cRaco 2 (4)

which can be rearranged as: c = Vco 2/Raco 2 (5)


Respiratory elimination of CO2(Vco 2) was measured using a commercially available open circuit indirect calorimeter (Deltatrac II metabolic monitor). This device offers a choice of four preset canopy flow rates: “baby” (3.1 litres/min), “child” (10.3 litres/min), “adult”, and “obese adult”. In accordance with manufacturer's recommendations, the instrument was warmed up for at least 30 minutes before two point calibration with room air (CO2 content assumed to be 0.04%) and Datex calibration gas (4.99% CO2, later verified by BOC Analytical Division, Crawley, Sussex, UK). The baby's head and shoulders were placed under the transparent perspex canopy, and a partial seal created by tucking the integral flexible skirt beneath the body and mattress. The child flow range was chosen because Bauer et al 9 have concluded that a flow rate of at least 4.5 litres/min is required for accurate measurement of Vco 2 using the Deltatrac in canopy mode. We also formally compared child and baby range canopy flow rates by studying four VLBW continuously fed babies over four consecutive one hour periods in a randomised 4 × 4 latin square design (see Results).


The manufacturer recommends that the canopy flow is calibrated by burning alcohol at rates approximating adult O2 consumption and CO2 production. It must then be assumed that canopy flow changes proportionately when the device is switched between baby, child, and adult ranges. We chose in addition to calibrate the instrument directly in child and baby settings by infusing medical grade CO2 (confirmed as 100% CO2 by BOC Analytical Division) at rates similar to those we encountered clinically.

A mannequin was placed under the canopy to simulate a baby, and CO2 injected at constant rate using a Harvard rotating screw syringe driver with four parallel mounted gas tight 50 ml polypropylene syringes. Connections were made with gas tight three way taps and PVC lines (Datex). Syringes were flushed four times with CO2 to remove air before use. Timed collections of water delivered in the range 4–12 ml/min showed that 2.777 ml was reproducibly displaced by each revolution of the syringe driver screw. The CO2 injection rate was therefore established by counting rotations of the screw using a vane and slotted optical switch connected to a microcomputer. Gas temperature was measured using an ELAB type CTD thermocouple (ELAB, Copenhagen, Denmark) and volumes corrected to STPD.



Analysis of variability in background 13C enrichment in the 11 pairs of breath samples collected at baseline of each study showed that there was a small but statistically significant variation between studies (one way analysis of variance, F  =  766; 10, 11 df; p < 0.001). This is allowed for in the calculation of APE and δ. Isotopic steady state was achieved in eight subjects by 120 minutes and in all 11 by 200 minutes (fig 3). Mean (SD) Raco 2 was 9.6 (2.1) ml/min STPD (table 2), equivalent to 8.5 (1.7) ml/kg body weight/min.

Figure 3

Individual plots of breath enrichment (atom percent excess (APE)) v time. Patient identification numbers correspond to those in tables 1 and 2. Weight at time of study is given.

Table 2

Raco 2 and Vco 2 measurements by patient with individual estimates of c


In vitro calibration showed that the Deltatrac consistently underestimated true CO2 injection rates (fig 4), showing greatest discrepancy at highest rates. The relation between CO2 injection rate (y) and simultaneous Deltatrac measurement (x) using the child canopy flow setting was given by y  =  1.12x (n = 16, r = 0.99; SD residuals = 0.36). As there was no intercept term (a), the regression coefficient (slope, b  =  1.12) was subsequently used to correct values of Vco 2 measured in vivo. A similar calibration constant was obtained in the baby flow range (y = 1.13x (n = 15, r  = 0.99; SD residuals = 0.16)).

Figure 4

Prediction of true Vco 2 from Deltatrac measurement.

Within patient measurements of Vco 2 made using the baby and child flow rates were not significantly different (baby range median 7.4 ml/min; child range median 7.9 ml/min; p = 0.7, Kruskal-Wallis; four babies × four one hour studies).

Mean (SD) Vco 2 measured with the Deltatrac in the 11 babies studied was 9.0 (2.0) ml/min, equivalent to 7.9 (1.3) ml/kg body weight /min. Raco 2 exceeded Vco 2 in seven.


Mean (SD) c was 0.95 (0.13). The 95% confidence intervals of the mean (0.87 to 1.03) included unity. No statistically significant correlation betweenc and body weight, energy intake, or Vco 2 was apparent.


To assess the feasibility of measuring Raco 2 in small ventilated babies, we studied a 20 day old 658 g infant with chronic lung disease (gestational age 25 weeks, birth weight 587 g). We and others2 have found it impossible to measure Vco 2 using the Deltatrac in such circumstances. She was continuously fed with human milk through a nasogastric tube and ventilated with a Babylog 8000 plus ventilator (Dräger, Hemel Hempstead, Herts, UK) at constant pressure, rate, and inspired time settings throughout the 24 hour study period. Fio 2 varied between 0.4 and 0.7, and Pco 2 between 5.0 and 6.5 kPa. A primed continuous intravenous infusion of NaH13CO3 was administered at a constant rate for 24 hours using doses given above (see Methods). Breath samples were collected using a syringe connected to a side port (designed for surfactant administration) on the endotracheal tube connector. Figure 5 shows the results. Raco 2 (STPD) was measured over three periods: plateau 1 (260—410 minutes, n = 4, Raco 2 = 5.5 ml/min); plateau 2 (630—695 minutes, n = 5, Raco 2 = 4.9 ml/min); plateau 3 (1175—1348 minutes, n = 5, Raco 2 = 5.4 ml/min).

Figure 5

Study of ventilated 658 g infant. Plot of breath enrichment (atom percent excess (APE)) against time. See text for estimates of Raco 2.


Arterial Pco 2 reflects a balance between the rate at which CO2 is eliminated through the lungs and the rate at which it is produced by metabolism. Much attention in neonatal intensive care has been focused on controlling elimination by mechanical ventilation, but less has been paid to minimising CO2 production by varying the amount and type of dietary fuel supplied.1 Inherent difficulties in measuring the rate of CO2 production in small ventilated babies2 may be one reason why nutritional interventions have been largely overlooked. We have shown that the labelled bicarbonate infusion technique can be used to measure simply the rate of CO2 production in VLBW babies, whether spontaneously breathing or mechanically ventilated. We found moreover that estimates of Vco 2 and Raco 2 in this group are comparable even in a short term study lasting two to four hours. In this respect, VLBW infants differ significantly from older children and adults.

Our estimate of fractional recovery rate, c, in this sample of babies was 0.95, with 95% confidence intervals 0.73 to 1.2 including unity. This is substantially higher than values previously reported in short term studies. Consideration of the model would suggest that this reflects rapid equilibration of infused isotope between body pools, with minimal trapping in slow turnover pools such as bone.3 This seems physiologically plausible, as VLBW babies have a low bone and fat mass, high extracellular fluid volume, and high resting metabolic expenditure relative to older infants. The only other neonatal study we have identified7 estimated the mean value for c as 0.77 but recruited more mature babies (mean birth weight 2120 g, mean study weight 2100 g), only three of whom weighed < 1500 g. Estimates in older children and adults have varied from 0.5 to 1.06,4 ,5 although a value of 0.8 is commonly assumed in substrate oxidation studies. In a comparative study,10 Raco 2 was found to be much faster in children than in adults, which might further support our hypothesis.

The derivation of equation 4 requires an assumption that CO2 is eliminated only through the lungs. This may be questionable, particularly in immature babies. In adults, about 1% of CO2 is lost across the skin4 and less than 5% is excreted in the urine, although the exact amount is dependent on urine pH.3 Although we have found no estimates of skin CO2 loss in VLBW infants, a study11 of infants weighing < 1000 g (gestational age 23–29 weeks) confirmed that cumulative bicarbonate loss over the first four days of life was only 1.9 (0.5) mmol/kg (mean (SD)). This is equivalent to an equimolar non-respiratory loss of 7.4 μl CO2/min/kg body weight. These errors can therefore be considered insignificant.

A further potential source of error in the measurement of Raco 2 is fluctuating background 13C enrichment during the course of a study. Control studies have been undertaken in which subjects underwent the experimental protocol without bicarbonate administration. Most, however, have regarded this as an extremely small potential error and assumed that the initial background enrichment does not change.4

Some consider indirect calorimetry the ideal method for measuring CO2 production, but substantial errors can arise when studying small babies.2 Our in vitro experiment indicated that particular care is required to calibrate the Deltatrac for use with small infants. Other preliminary experiments (see Results) showed no significant difference between values obtained in vivo using baby and child settings. We took great care to simulate the in vivo situation when calibrating the Deltatrac, to the extent that we used physiologically appropriate rates of CO2 injection, validated the purity of gas used, and placed a baby sized mannequin within the canopy and flexible skirt. If calibration had not been performed, a considerable underestimate of Vco 2would have been made.

As we have shown, potential errors of different sorts apply to the measurement of both Raco 2 and Vco 2. This is not surprising, as they measure different aspects of CO2 metabolism. One (Vco 2) measures the elimination of CO2; the other (Raco 2) measures the turnover (or flux, Q) of bicarbonate within the body pool. In steady state, the latter is equivalent to both the rate of appearance (Ra) and rate of disappearance (Rd). Measurement of arterial blood gas status at the beginning and end of each study may have helped to confirm that the bicarbonate/CO2 pool was in steady state during each study, but we did not feel blood sampling ethically justifiable. Participants showed stable cardiorespiratory measurements throughout and most were breathing air. Moreover the satisfactory enrichment plateaux (fig 3) observed in the 11 studies themselves constitute evidence that the bicarbonate pool was in steady state.

The 11 studies we describe were conducted only for two to four hours, and we have not implied measurements to be representative of longer periods. In deriving c, we have made comparisons only between quantities simultaneously measured in the same baby. We performed only one 24 hour study, principally to assess the feasibility of applying the technique in a ventilated baby (fig 5). Although we attempted to measure Vco 2 by connecting the Deltatrac monitor to the expiratory port of the ventilator (Dräger Babylog 8000 plus), we obtained no satisfactory measurements, probably because a blow off mechanism in the ventilator allows inspiratory and expiratory gases to mix under certain circumstances. The many problems associated with the measurement of Vco 2 during mechanical ventilation have been well described previously. In contrast, measurement of Raco 2 was easily accomplished using the labelled bicarbonate technique and yielded comparable values at the three plateau periods studied.

In summary, we have shown that the labelled bicarbonate infusion technique is easily applicable to the measurement of CO2production in VLBW babies. In contrast with experience with older children and adults, the adjustment required for retention of infused isotope in short term studies is negligible in this group of patients. This confirms the value of determining the fractional recovery rate,c, for individual patient groups. The simplicity of this technique offers new opportunities to study the interaction between fuel metabolism and respiratory function even in the smallest ventilated babies.


We acknowledge the collaboration of Professor David Halliday who undertook breath analyses at the Bureau of Stable Isotope Analysis, Brentford. The study was funded by the St George's Hospital Special Trustees Research Fund.