This study was conducted in a twelve-bed mixed surgical/medical ICU of a tertiary referral hospital. It was approved by the regional ethical review board in Stockholm (2014/1778-31) and registered at Australia New Zealand Clinical Trials Registry (Trial ID ACTRN12615000205538). Patients and relatives were informed about the study orally and in writing before written informed consent was obtained. All mechanically ventilated patients ≥18 years of age were considered for recruitment. Exclusion criteria were: 1) gas leaks (chest tubes, pneumothorax, bronchoesophageal fistulas, etc; leaks registered by the ventilator <10 % of minute volume (MV) were accepted); 2) NO therapy or ECMO 3) fraction of inspired oxygen (FiO2) >0.60; 4) respiratory rate (RR) >35; and 5) absence of informed consent. Although continuous renal replacement therapy (CRRT)-induced CO2 diversion can interfere with the accurate determination of metabolic CO2 production, the effect on pulmonary gas exchange measurements should be identical for all instruments. Patients with CRRT were therefore eligible for inclusion providing flow rates and filtration remained constant during the study period. All patients were mechanically ventilated by Evita XL ventilators (Dräger, Germany) in continuous positive airway pressure-assisted spontaneous breathing (CPAP-ASB), biphasic positive airway pressure-assisted spontaneous breathing (BIPAP-ASB) or mandatory minute volume ventilation (MMV) modes. When present, active humidification was turned off at least 30 min prior to measurements.
All aspects of patient care were ultimately decided by the attending physician. To avoid excessive changes in metabolic rate, staff in charge of patients included in the study were encouraged to keep feeding rates, vasopressor infusions and sedation constant 1 h prior to and during measurements. Changes in ventilator settings, endotracheal suctioning, disconnections in the ventilator circuit or mobilization were avoided unless deemed urgently required by medical staff. In case of such events the ongoing measurement was discontinued and restarted after a 15-min resting period. If unexpected deterioration in a patient’s condition required more extensive interventions by clinical staff the measurement was aborted. Multiple measurement series in single patients were performed with a minimum interval of 24 h.
In order to minimize the influence of potential baseline drift in resting energy expenditure (REE), measurements with the study devices and reference method were conducted simultaneously as previously described by Graf et al. [12]. Daily calibrations were performed as recommended by manufacturers or technical manuals. All instruments were connected in parallel to the ventilator circuit. As this required a temporary disconnection of the ventilator and entrainment of room air into the tubing, a 15-min resting period was mandated before commencing measurements. This allowed sufficient time for the mixing chamber of the Deltatrac to equilibrate with alveolar gas from the patient. The flowmeters of the COVX and Quark were connected to the y-piece and expiratory port of the ventilator respectively. The collection tube for expiratory gas to the Deltatrac mixing chamber was then attached to the Quark turbine flowmeter with a plastic adapter. This connection was reinforced with duct tape to prevent gas leaks. Gas sampling lines for both study instruments were connected to the COVX flowmeter using a three-way stopcock, facilitating a switch between devices without disconnecting the ventilator. This setup was approved by the manufacturers of both study devices. An illustration of all connections can be seen in Fig. 1.
Simultaneous measurements with Deltatrac II and either Quark RMR or E-sCOVX were then undertaken for 20 min. After measuring with the first device, another simultaneous measurement was immediately performed with the second device and Deltatrac II. Simple computer randomization was used to determine the order of measurement.
Raw data for all measured parameters were extracted by software applications from each instrument at the highest possible sampling rate. As all three systems provide a different amount of data points over time and treat artifact suppression differently, we used a standardized approach that was communicated to the manufacturers involved prior to the study. When a patient inhales or exhales sharply (as in the case of coughing) gas concentrations and spirometry cannot be properly synchronized, resulting in very low VO2/VCO2 values. Even though these values are not representative of the patient’s metabolic state, the contribution of single artifact data points potentially offsets the average value of the measurement. The user interface of the E-sCOVX omits these values when displaying the average over time, but when raw data is extracted using the manufacturer’s software these data points are given a value of zero. The Quark attempts to calculate VO2/VCO2 from available parameters, resulting in artifacts with variable low values. To solve this issue, all data points with the value zero were omitted in calculations of averages from the E-sCOVX module. For the Quark RMR all values which did not fulfil a set of pre-specified criteria (RR >3/<60, tidal volume >0.2/<3 L, fractional content of expired CO2 > 0.5/<8 %) were excluded as artifacts. As the Deltatrac is not affected by small variations in breathing patterns, no method of artifact suppression was used.
Using the processed data, average values of VO2, VCO2, MV and respiratory quotient (RQ) were obtained. REE was calculated using the modified Weir equation, not accounting for nitrogen excretion. Measurements where the reference method registered a mean RQ of <0.6 or >1.2 were discarded.
Materials
Deltatrac II Metabolic Monitor
The Deltatrac Metabolic Monitor uses a mixing chamber technology where all exhaled gas from the patient is collected from the expiratory port of the ventilator. It does not measure flow directly. Gas from the mixing chamber is diluted with room air at a constant flow rate and VCO2 is calculated as the product of flow and the concentration of CO2 post-dilution (FCO2). VO2 is then calculated using the Haldane transformation, which assumes that N2 is biologically inert and present in the same concentrations in inhaled and exhaled gas. As the denominator in this equation is (1 – FiO2), VO2 measurements will become increasingly inaccurate as FiO2 approaches 1. It is therefore recommended that measurements are restricted to patients with a maximum FiO2 of 0.6. It uses a paramagnetic O2 analyzer and an infrared CO2 analyzer.
Breath-by-breath instruments
The common feature of breath-by-breath systems is that both flow and gas concentrations are measured over the respiratory cycle. As there is a delay in gas concentration measurements due to the transport time in the sampling lines, flow and gas concentration curves need to be synchronized by software algorithms. VO2 and VCO2 are then calculated as the product of volumes and concentrations. Due to the difficulties of accurately measuring the small differences in inspired and expired volumes, the assumptions of the Haldane transformation are used to allow for unidirectional flow measurement. This imposes the same limitations on accuracy at higher FiO2 levels as with the Deltatrac. As synchronization of gas and flow measurements become increasingly difficult at very high respiratory rates, a maximum RR of 35 was allowed for the purposes of this study.
Quark RMR
The Quark RMR is a breath-by-breath system for gas exchange measurements. It measures CO2 and O2 concentrations over the respiratory cycle through a sampling line connected close to the endotracheal tube. Flow is measured using a turbine flowmeter attached to the expiratory port of the ventilator. There is a software application to compensate for bias flow in the ventilator circuit, which has to be set by the user before the start of every measurement. It uses a paramagnetic O2 analyzer and an infrared CO2 analyzer.
E-sCOVX
The E-sCOVX module also measures gas exchange breath-by-breath. Gas samples are drawn from a sampling line connected to the flowmeter and analyzed by paramagnetic and infrared methods for O2 and CO2, respectively. Flow rates are measured by a pneumotach flowmeter connected directly to the Y-piece of the ventilator circuit.
Statistical analysis
Limits of agreement and bias between VO2/VCO2 as measured by study instruments (E-sCOVX/Quark RMR) and the reference method (Deltatrac II) were compared using Bland-Altman plots [14]. Agreement of REE values as measured by study devices and the reference method were also analyzed with the Bland-Altman method, although it is a dependent variable calculated from VO2 and VCO2. A sample size of 50 measurements in at least 20 patients was considered sufficient to determine limits of agreement within ±2.0 standard deviations (SD). Continuous variables with parametric distribution were analyzed for statistical significance using a two-tailed Student’s t-test for paired samples and Mann-Whitney U-test for non-parametric data, and an α level of ≤0.05 was considered as statistically significant.