Volumetric capnography as a screening test for pulmonary embolism in the emergency department – clinical investigations

Franck Verschuren

Study objective: To compare the diagnostic performance of volumetric capnography (VCap), which is the plot of the expired C[O.sub.2] partial pressure against the expired volume during a single breath, with the PaC[O.sub.2] to end-tidal C[O.sub.2] (EtC[O.sub.2]) gradient, in the case of suspected pulmonary embolism (PE).

Design: Single-center, prospective study.

Setting: Emergency department of a teaching hospital.

Patients: A total of 45 outpatients with positive enzyme-linked immunosorbent assay d-dinner levels of > 500 ng/mL. The diagnosis of PE was confirmed in 18 outpatients according to a validated procedure based on the ventilation-perfusion lung scan and/or spiral CT scanning. Interventions: Curves of VCap were obtained from a compact monitor connected to a computer. A sequence of four to six stable breaths allowed the calculation of the following several variables: alveolar dead space fraction; the ratio of alveolar dead space (VDalv) to airway dead space (VDaw); the VDalv to physiologic dead space (VDphys) fraction; the slope of phase 3; and the late dead space fraction (Fdlate) corresponding to the extrapolation of the capnographic curve to a volume of 15% of the predicted total lung capacity.

Results: The mean ([+ or -] SD) PaC[O.sub.2]-EtC[O.sub.2] gradient was 5.3 [+ or -] 0.7 mm Hg in the PE-positive group and 2.8 [+ or -] 0.7 mm Hg in the PE-negative group (p = 0.019). Four variables of the VCap exhibited a statistical difference between both groups, as follows: the VDalv/VDaw fraction, the slope of phase 3; the VDalv/VDphys fraction; and the Fdlate, which was 8.2 [+ or -] 3.3% vs -7.7 [+ or -] 2.8%, respectively (p = 0.000011). The diagnostic performance expressed as the mean area under a receiver operating characteristic curve comparison was 75.9 [+ or -] 7.4% for the PaC[O.sub.2]-EtC[O.sub.2] gradient and 87.6 [+ or -] 4.9% for the Fdlate (p = 0.02).

Conclusion: Fdlate, a variable of VCap, had a statistically better diagnostic performance in suspected PE than the PaC[O.sub.2]-EtC[O.sub.2] gradient. VCap is a promising computer-assisted bedside application of pulmonary pathophysiology. Future research should define the place of this technique in the diagnostic workup of PE, especially in the presence of positive d-dimers.

Key words: C[O.sub.2] emergency department; pulmonary embolism; volumetric capnography

Abbreviations: CI = confidence interval; ELISA = enzyme-linked immunosorbent assav: EtC[O.sub.2] = end-tidal C[O.sub.2]; Fdlate = late dead space fraction; ExpC[O.sub.2] 15% TLC = extrapolated C[O.sub.2] partial pressure at an exhaled volume of 15% of the estimated total lung capacity; PE = pulmonary embolism; ROC = receiver operating characteristic; TLC = total lung capacity; VCap = volumetric capnography; VDalv = alveolar dead space; VDalv/VTalv = alveolar dead space fraction; VDaw = airway dead space; VDphys = physiologic dead space; V/Q = ventilation-perfusion ratio; VT = tidal volume


Pulmonary embolism (PE) has always been a challenging diagnosis. In 1959, Robin et al (1) were the first authors to suggest a physiologic approach to this difficult diagnosis by measuring the PaC[O.sub.2] to end-tidal C[O.sub.2] (EtC[O.sub.2]) gradient as a percentage of the ventilated, but not the perfused, lung (ie, alveolar dead space). This direct application of basic physiopathology represented an exciting challenge for clinicians and physiologists, who began to explore the role of capnography as a simple, noninvasive, rapid bedside method in the diagnostic workup of patients with PE. (2-4) Nevertheless, in 1966 Nutter and Massumi (2) emphasized the pitfalls and sources of errors when using the C[O.sub.2] gradient as an assessment of alveolar dead space, pointing out the relatively high incidence of false-positive and false-negative results of this approach. This weak diagnostic performance, the technical limitations or artifacts, and the lack of sufficient validation explained why this test was abandoned and was considered to be an “orphan” test for nearly 20 years. (5) The following three elements brought about its resurrection in the late 1990s: (1) the use of volumetric capnography (VCap); (2) the development of new analytic devices, acquisition systems, and microcomputers, providing quick and reliable information; and (3) the combined use of d-dimer assays with capnographic information to increase its sensitivity in ruling out the diagnosis of PE.

VCap, also known as the single breath test for C[O.sub.2] displays an expirogram, which is the plot of the expired C[O.sub.2] concentration against the expired volume during a single expiration. The technique has been extensively described by Fletcher and colleagues. (6,7)

Even if VCap constitutes a direct computerized application of the pulmonary pathophysiology at the bedside, it is seldom used in emergency departments. (8,9) However, the current trend toward less invasive procedures for the diagnosis of PE justifies a careful evaluation of its place in the diagnostic workup of patients with PE, especially in combination with d-dimer measurement.

The objectives of this study were as follows: (1) to validate the use of VCap in a subset of outpatients presenting to an emergency department with a clinical suspicion of PE with positive results of a rapid enzyme-linked immunosorbent assay (ELISA) for d-dimers; and (2) to compare the diagnostic performance of VCap with the PaC[O.sub.2]-EtC[O.sub.2] gradient.


Data were collected in the emergency department of an urban teaching hospital with an annual census of 52,000 patients. The patients were prospectively included in the study by one investigator (FV) during a 10-month period from January to November 2001. Patients were eligible for the study when they presented with a clinical suspicion of PE and a plasma d-dimer level of > 500 ng/mL. The exclusion criteria were as follows: age < 18 years; hemodynamic instability (ie, BP, < 80 mm Hg); inability to cooperate: necessity for intubation and artificial ventilation; and a plasma d-dimer level of < 500 ng/mL. The physician in charge determined the clinical pretest probability of the patients according to a scoring system from Wells and coworkers, (10) All the eligible patients were asked to sign an informed consent from. The studies were carried out in accordance with the guidelines of the institutional review board and were approved by the local ethics committee.

D-Dimer Assay

Plasma d-dimer assays (rapid ELISA Vidas d-dimers: bioMerieux: Marcy l’Etoile, France) were performed on undiluted plasma samples and were analysed in the coagulation laboratory by a technician who was unaware of any clinical information. This d-dimer assay is a quantitative ELISA method that is automated on an immunoanalyzer (VIDAS; bioMerieux), combining the sandwich immunoenzymatic method in two steps with a final fluorescence detection. (11) The results were communicated by phone to the physician in charge of the patient within 1 h. A recent meta-analysis (12) showed that the use of the ELISA d-dimer test in the diagnosis of PE in the adult emergency department population was associated with a sensitivity of 94% (95% confidence interval [CI], 88 to 97%) and a specificity of 45% when using a cutoff level of 500 ng/mL. Moreover, a subgroup of the meta-analysis using the same rapid ELISA d-dimer test as in our study showed an outstanding sensitivity of 109% (95% CI, 97 to 100%). Thus, a normal value safely rules out PE, obviating the need for any capnographic measurement, with the rare exceptions being patients with a high pretest clinical probability of PE.

Volumetric Capnogram Measurement

The curves of VCap were obtained using experimental technology (Datex Ohmeda Division of Instrumentarium Corp; Helsinki, Finland). A commercially available compact monitor (CS/3; Datex-Ohmeda) with a gas analyzer (M-COVX; Datex-Ohmeda) simultaneously recorded the time-based capnography, tidal volume (VT) and minute volume, and the respiratory flow of the patient via a light sensor (D-lite: Datex-Ohmeda) [flow sensor and gas sampler had a dead space volume of 9 mL]. The C[O.sub.2] measurement was based on sidestream technology, with the gas being transported through a 2-m sampling line at a flow rate of 200 mL/min to an improved infrared sensor (with thermopile detector). Appropriate data acquisition software (S/5 Collect, version 3.0; Datex-Ohmeda) collected all the data and stored it on-line in a computer. The data then were converted off-line and displayed as a volumetric capnogram. Moreover, respiratory parameters like peak expiratory flow, respiratory rate, and minute ventilation were also available for analysis. All the equipment was mounted on a compact rolling stand that allowed easy access to the patient. The gas measurement unit was calibrated before each measurement with a 5% volume of C[O.sub.2] calibrating gas.

The quality of the VCap curve depends on the synchronization between the expired C[O.sub.2] concentration curve and the expired volume. Indeed, there is a lag time of about 1.6 s for the gas sample to travel from the sampling system to the C[O.sub.2] sensor, in contrast with the almost instantly measured expired volume. This delay is automatically determined in vivo, with pressure changes compensated for, and mar be confirmed off-line by checking the adequacy of the crossing between the EtC[O.sub.2] point and the end of the expiratory volume plateau phase.

Data Collection

Figure 1 shows an example of VCap during one single expiration in a healthy patient, with its dead space subdivisions derived from the work of Fletcher et al. (6) The shape of the curve is divided into the following three phases: phase 1, the C[O.sub.2]-free volume of the airway dead space (VDaw); phase 2, a transition phase; and phase 3, the sloping phase of the alveolar volume. The right vertical line joins the expired VT and the EtC[O.sub.2]. The PaC[O.sub.2] value must be entered manually, and it determines the superior horizontal line. The area under the right vertical line and the upper horizontal line delimits a rectangle, the surface of which corresponds to an ideal system in which all the C[O.sub.2] from the pulmonary perfusion would be eliminated without any dead space and any ventilation/perfusion ratio (V/Q) mismatch. (6) The slope of phase 3 is automatically, determined as the least square fit of the last 440 ms of the expiration and can be manually adapted. Finally, a middle vertical line is defined by an equal p and q area surrounding the curvature of phase 2, and it determines the VDaw. Area X, which is enclosed between the slope line, VDaw, and VT, is equal to the area under the C[O.sub.2] curve, and corresponds to the expired C[O.sub.2] volume of the effective tidal volume. Area Z is calculated by multiplying VDaw by PaC[O.sub.2], and it represents the wasted ventilation due to VDaw. Area Y is then calculated by subtracting areas Z and X from the total area (VT x PaC[O.sub.2]), and determining the alveolar dead space (VDalv). The physiologic dead space (VDphys) represents the sum of VDaw and VDalv. These areas can be used to express the following various dead space fractions:

* Vdphys/VT = Y + Z/X + Y + Z, which represents the VDphys fraction;

* VDalv/VTalv = Y/X + Y, which represents the percentage of the alveolar volume occupied by alveolar dead space per breath;

* VDalv/VDaw = Y/Z, which represents the relative contribution of the alveolar dead space to the VDaw; and

* Late dead space fraction (Fdlate) = 1 – (extrapolated C[O.sub.2] partial pressure at an exhaled volume of 15% of the estimated total lung capacity [ExpC[O.sub.2] 15% TLC]/PaC[O.sub.2]). The concept of Fdlate is explained in Figure 2. Fdlate is defined according to Eriksson et al. (13) These authors showed that the extrapolation of phase 3 of the capnographic curve reached the PaC[O.sub.2] value at a volume corresponding to approximately 15% of the TLG in healthy subjects and in patients with obstructive lung diseases, but failed to reach this PaC[O.sub.2] value in cases of PE, creating a gradient between the ExpC[O.sub.2], 15% TLC and PaC[O.sub.2]. This ExpC[O.sub.2] 15% TLC may be regarded as a C[O.sub.2] value at a forced expiration where time is no longer a limiting factor for the diffusion of C[O.sub.2] between the pulmonary blood and the alveoli. Total lung capacity (TLC) has been estimated taking the sex and height of the patient into consideration. (14) The equation for the ExpC[O.sub.2] 15% TLC is: (EtC[O.sub.2] + [TLC x 0.15 – VT] x slope) x (PB – P[H.sub.2]O), where slope is expressed as the percentage per liter, and TLC is expressed in liters, PB is the barometric pressure, and P[H.sub.2]O is the water vapor pressure.


Steady State Status of the Patient

Since VCap was only performed on a spontaneously breathing patients, the steady-state status had to be confirmed. Nose-clips were placed, and the patients were breathing room air in a 45[degrees] semi-Fowler position through a mouthpiece connected to the flow sensor and gas sampler (D-lite; Datex-Ohmeda). Before starting the data collection, the subject was allowed to breathe quietly for 2 min in order to become accustomed to the technique, and to identify a stable respiratory rate and EtC[O.sub.2] values. On-line data were then registered for 2 min, and a blood gas sample was slowly collected during the last minute of registration (ABL 725; Radiometer; Copenhagen, Denmark). After checking for the adequacy of the synchronization between the volume and the C[O.sub.2] curves, the investigator determined off-line a sequence of four to six breaths among the 10 to 20 registered ones, in order to obtain an EtC[O.sub.2] value with a < 5% variations coefficient, and a VT variation coefficient of < 20%,

Diagnostic Strategy for PE

After determining the clinical pretest probability for PE, the physician in charge of the patient performed a rapid ELISA d-dimer determination. Patients with a d-dimer level of 500 ng/mL were investigated by the combined use of both V/Q lung scintigraphy (V/Q lung scan) and thin-collimation, multi-detector row spiral CT scanning (Mx 8000; Marconi; Lorain, OH). The V/Q lung scan classified the probability of PE as normal, low, intermediate, or high according to the classification of Hull et al. (15) PE was considered to be absent if the results of a normal spiral CT scan were consistent with those of a normal or low-probability V/Q lung scan. PE was confirmed if the positive results of a spiral CT scan were concordant with those of a high-probability V/Q lung scan. (16) Pulmonary angiography was performed only when the results of the V/Q lung scan and spiral CT scan were discordant or when the V/Q lung scan was of intermediate probability. Patients presenting with a contraindication for spiral CT scan due to a serum creatinine concentration of > 1.5 mg/dL, an allergy to iodinated contrast medium, or the intake of metformin were investigated using the concordant information of a low clinical pretest probability with a normal or low-probability V/Q lung scan, a high clinical pretest probability with a high-probability, V/Q lung scan, or an intermediate probability V/Q lung scan with a normal result of a lower limb venous ultrasonography.

6-Month Follow-up

Patients were contacted by phone at 6 months after their admission to the emergency department. Patients without a diagnosis of PE were asked to answer three questions regarding the occurrence of a PE or deep venous thrombosis and the requirement of anticoagulant therapy.

Statistical Analysis

Respiratory and blood gas variables from both groups with and without PE were compared according to the Student t test. The diagnostic performances of the parameters from VCap are expressed as the area under the receiver operating characteristic (ROC) curve. (17) Comparisons between parameters are expressed as the difference between their respective areas under the ROC curve, and the statistical significance has been computed according to the nonparametric approach based on the generalized two sample Wilcoxon statistics of Lee and Rosner. (18) The results are expressed with one-sided p values.


Capnographic measurements were performed in 51 patients. Six patients were excluded from the study for the following reasons: three patients had no steady-state EtC[O.sub.2] (variation coefficient, > 5%) or VT (variation coefficient, > 20%) during the off-line analysis; two patients had poor quality VCap with no plateau phase due to a high respiratory rate (29 and 35 breaths/min); and one patient was lost to follow-up. These six patients had clinical characteristics that were similar to those of the included patients, and two of them had a final diagnosis of PE.

A total of 4.5 nonconsecutive outpatients suspected to have PE with d-dimer levels of > 500 ng/mL were included for analysis. The clinical and anthropometric data are reported in Table 1. The clinical data were statistically comparable between the PE-positive and PE-negative groups. Men constituted only 20% of the patients studied (n = 9). PE was diagnosed in 18 patients (40%) and was ruled out in 27 patients (60%), resulting in a high prevalence of PE that differs from the usual 25% prevalence in an outpatient population with positive d-dimer levels. (19,20) This high prevalence may be explained by the nonconsecutive inclusion of patients in the study. Nine of the 11 patients with a high pretest probability had a PE event (82%), as well as 6 of the 17 patients with an intermediate pretest probability (35%) and 3 of 17 with a low pretest probability (18%). These numbers were consistent with the performance of the clinical signs in evaluating the presence of PE, (10) and the difference between each clinical probability was significant with the results of the [chi square] test ([chi square] = 11.711; p = 0.0029). The alternative diagnoses when PE was finally ruled out were the following: pleural or pericardial effusion (six patients); pulmonary infection (five patients); no diagnosis (five patients); COPD in exacerbation (four patients); heart failure (four patients); and hyperventilation (three patients).

Diagnosis of PE

PE was diagnosed in 18 patients (40%) and ruled out in 27 (60%). Table 2 shows the diagnostic criteria used to exclude of confirm the diagnosis. Thirty-one of the 45 patients included those who were investigated according to the association of a V/Q lung scan and a spiral CT scan. Six patients could not undergo a spiral CT scan because of the following contraindications: a serum creatinine concentration of > 1.5 mg/dL (one patient); an allergy to iodinated contrast medium (three patients); and the intake of metformin (two patients). An additional eight patients were investigated according to concordant clinical pretest probabilities with the V/Q lung scan of the spiral CT scan alone.

C[O.sub.2] Gradient

The mean ([+ or -] SD) PaC[O.sub.2]-EtC[O.sub.2] gradient was 5.3 [+ or -] 0.7 mm Hg in the PE-positive group and 2.8 [+ or -] 0.7 mm Hg in the PE-negative group (p = 0.019). Arbitrarily considering a pathologic PaC[O.sub.2]-EtC[O.sub.2] gradient to be > 3 mm Hg, we saw that four patients with a normal gradient had a PE event (ie, false-negative results). On the other hand, eight patients presented with a gradient of > 3 mm Hg in the absence of PE (ie, false-positive results). With an arbitrary cutoff gradient value such as 3 mm Hg, the sensitivity of the test would be 77% and the specificity would be 70%.


The respiratory, blood gas, and ventilatory data computed by VCap are reported in Table 3. Four variables from the VCap shared with the C[O.sub.2] gradient a statistical difference between the PE-positive and the PE-negative groups, as follows: VDalv/ VDaw; VDalv/VDphys; the slope of phase 3; and Fdlate. The Fdlate achieved the statistical difference (p = 0.000011). The alveolar dead space ratio (area Y of the VCap curve) was not statistically different between the two groups. The values of the PaC[O.sub.2]-EtC[O.sub.2] gradient were significantly correlated with VDalv/VDaw (r = 0.537; p = 0.0002), VDalv/ VDphys (r = 0.531; p = 0.0002), VDalv/VTalv (r = 0.853; p = 0.000001), the slope of phase 3 (r = 0.331; p = 0.0.3), and Fdlate (r = 0.753; p = 0.000001).

The areas under the ROC curve are reported in Figure 3. The mean areas were 75.9 [+ or -] 7.4% for the PaC[O.sub.2]-EtC[O.sub.2] gradient and 87.6 [+ or -] 4.9% for the Fdlate, which represents a mean gain of 11.7 [+ or -] 5.3% for the Fdlate (p = 0.02).



Late dead space fractions are reported in Figure 4. In 18 patients with PE, Fdlate ranged from -3.2 to 40.6%, with a mean value of 8.2 [+ or -] 3.3%. In 27 patients without PE, Fdlate ranged from -20.4 to 9.4% with a mean value of -7.7 [+ or -] 2.8%. If we consider a cutoff point of 12% in separating positive PE from negative PE, as suggested by Erikkson et al, (13) all of the six patients with an Fdlate value of [greater than or equal to] 12% had PEs. Moreover, all of the eight patients with false-positive results for the PaC[O.sub.2-]EtC[O.sub.2] gradient had Fdlate values of < 12%. Three of these eight patients were considered to be COPD patients in acute exacerbations.



Our study shows that VCap can be used as a bedside technique for spontaneously breathing outpatients who are suspected of having PE. The measurement of the Fdlate obtained from VCap is more powerful than the traditional PaC[O.sub.2]-EtC[O.sub.2] gradient in separating a positive from a negative PE event. Moreover, the population that we studied was pre-selected with a very sensitive d-dimer assay in order to avoid a capnographic measurement in patients in whom a negative d-dimer value could easily exclude the diagnosis of PE without recourse to any other additional test. Nevertheless, the small number of patients as well as the nonconsecutive inclusion defines our study as a preliminary study.

About 14 human clinical studies (1-4,8,9,13,21-28) have been reported in the literature concerning the use of C[O.sub.2] monitoring in the setting of the clinical suspicion of PE. Unfortunately, all of these studies differ in terms of the population’s origin (ie, inpatients or outpatients), patient status (ie, spontaneously breathing or mechanically ventilated), the reference tests used for PE diagnosis, the dead space calculation, the type of capnograph, or the era of performance (ie, 1959 to 2001). These differences explain to some extent the lack of sufficient validation of the C[O.sub.2] measurement as a diagnostic tool in patients with PE. In this way, the PaC[O.sub.2]-EtC[O.sub.2] gradient is probably not specific enough to replace the radiologic and nuclear imaging, and is not sensitive enough to definitely rule out PE when the gradient is normal.

The PaC[O.sub.2]-EtC[O.sub.2] gradient in our study showed a mean ROC curve area of 75.9 [+ or -] 7.4%, with a sensitivity of 77% and a specificity of 70% when using a cutoff value of 3 mm Hg. These figures correlate well with the usual diagnostic performance that has been reported in previous studies. (2-4) When taking the numerous pitfalls and sources of error of the PaC[O.sub.2]-EtC[O.sub.2] gradient as the assessment of the VDalv into consideration, we confirm that this technique is probably not useful enough to help the clinician in his diagnostic workup for PE. (2)

VCap presents at least three theoretical advantages over the PaC[O.sub.2]-EtC[O.sub.2] gradient, as follows: (1) the calculation of the three components of the expired volume (VT, VDalv, and VDaw) from area ratios, which cannot be done with time-based capnography (6); (2) a higher sensitivity in estimating VDalv, since the area Y of the VCap curve takes not only the PaC[O.sub.2]-EtC[O.sub.2] gradient into consideration, but also the slope of phase 3 (6); and (3) a higher specificity in separating PE flora COPD thanks to the Fdlate.

In our series, four patients with a normal PaC[O.sub.2]-EtC[O.sub.2] gradient had a PE event. Unfortunately, VCap could not improve these false-negative results, since Vdalv/VTalv, Fdlate, and the slope of phase 3 remained within normal ranges. All of these patients showed only segmental or subsegmental branches obstructed on spiral CT scans, and one of them had a pulmonary infarction. Therefore, it seems reasonable to assert that dead space determination, determined by whatever capnographic technique, will never be sensitive enough in eases of peripheral PE or in cases of the adaptation of the V/Q ratio mismatches due to pulmonary infarction, atelectasis, or a potential hypocarbic bronchoconstrictive reflex. (29) These results differ from a study by Patel and coworker, (27) who, in 53 outpatients, applied a derived neural network model with six variables derived from VCap to detect a PE with a sensitivity of 100% (95% CI, 89 to 100%). The expression of our results according to ROC curves seems to be more accurate than the arbitrary use of a cutoff point to achieve 100% sensitivity, (30) since a proven PE may be accompanied by normal capnographic measurements in numerous clinical conditions. (2)

The theoretically better specificity of VCap over the traditional PaC[O.sub.2]-EtC[O.sub.2] gradient is due to the measurement of tire Fdlate, which takes the EtC[O.sub.2], the arterial C[O.sub.2], and the slope of phase 3 into consideration. The concept of Fdlate has the following three theoretical advantages: first, it should improve the specificity, of the dead space measurement by separating patients with obstructive lung diseases from those with PE, both conditions being frequently associated with an increased PaC[O.sub.2]-EtC[O.sub.2] gradient; second, it respects the steady-state status, since it does not necessitate a forced expiration of the patient; and third, it might correct a false-positive C[O.sub.2] gradient due to an incomplete diffusion time in cases of high respiratory rates or low VT values. (6)

Three series (13,22,23) have shown that VCap coupled with Fdlate measurement might give more accurate information than the PaC[O.sub.2]-EtC[O.sub.2] gradient alone. Eriksson et al (13) showed that an Fdlate cutoff of 12% correlated significantly with the angiographic findings in 38 patients with suspected PE. Moreover, VCap with Fdlate appeared to be superior to measurements of the traditional PaC[O.sub.2]-EtC[O.sub.2] gradient in differentiating PE patients from patients with normal, obstructive, or restrictive lung function. Nine years later, Olsson et al (22) validated the same technique in comparison with the V/Q lung scan in a larger number of inpatients and outpatients (total number of patients, 223) and showed that the Fdlate cutoff of 12% suggested by Eriksson et al (13) led to the diagnosis of PE with a sensitivity of 85% and a specificity of 93%. In their conclusion, the authors suggested starting anticoagulation therapy when a very high Fdlate was associated with a high clinical probability of PE, and avoiding further investigation for a very low Fdlate in combination with a low clinical probability. Finally, a study by Anderson et al (23) explored the use of VCap with Fdlate in a small group of 12 surgical patients including 10 mechanically ventilated subjects and 4 ARDS patients. The authors pointed out the promising role of a noninvasive bedside screening test for PE in a subset of critically ill patients who were difficult to transport.

Our study showed that all eight patients with a false-positive PaC[O.sub.2]-EtC[O.sub.2] gradient of > 3 mm Hg in the absence of PE had Fdlate values of < 12%, confirming the results of the three previous series that focused on the Fdlate measurement, but in a more specific population comprising outpatients with positive d-dimer levels.

The second advantage of using the Fdlate measurement lies in the steady-state situation of the patient. Hatle and Rokseth (4) had patients perform a deep expiration to reduce the false-positive PaC[O.sub.2]-EtC[O.sub.2] gradients, which were due to the slow emptying of the C[O.sub.2] in COPD patients, with an increased time constant. Even if they had noted a normalization of the PaC[O.sub.2]-EtC[O.sub.2] gradient in 19 of the 21 COPD patients after a forced exhalation, it may be argued that the use of the maximal end-expired C[O.sub.2] value instead of EtC[O.sub.2] for alveolar dead space determination has three disadvantages, as follows: it is dependent on patient cooperation; it suppresses the steady-state condition; and it is difficult to perform in severely ill patients. (13)

The last advantage of Fdlate measurement concerns its capacity to correct a falsely positive C[O.sub.2] gradient due to an incomplete diffusion time in patients with high respiratory rates or low VT values. If we had arbitrary excluded from the study patients with VT values 28 breaths/min, we could have decreased the rate of false-positive results of the C[O.sub.2] gradient, but at the cost of the exclusion of three of the eight patients. Since Fdlate might be regarded as a forced expiration in which time is no longer a limiting factor for C[O.sub.2] elimination, its use allows the interpretation of capnographic curves in more tachypneic patients.

The place of VCap in the diagnostic arsenal for PE needs to be evaluated. Indeed, the limited availability or the time-consuming use of radiologic and nuclear testing, as well as tire current trend toward less invasive diagnostic procedures for PE, justify a careful evaluation of the role of VCap in combination with d-dimer measurement. Most authors have studied the combination of the capnographic measurement with d-dimer assays to achieve the best sensitivity in ruling out PE without recourse to additional radiologic or nuclear imaging. (8-10,24) In two multi-center studies by Kline et al (8) and Rodger et al, (9) totaling > 600 patients, the combination of a normal alveolar dead space fraction (VDalv/VTalv) with a normal d-dimer value excluded PE with 98.4% and 97.8% sensitivity, respectively (95% CI, 91.6% and 88.5 to 100%, respectively). Such a normal combination was observed in 43% and 28% of the enrolled patients, respectively, avoiding the recourse to vascular imaging in this subset of patients regardless of the pretest clinical probability. Nevertheless, these encouraging results must be compared with the diagnostic performances of the new-generation ELISA d-dimer assays that have been validated as screening tests for the exclusion of PE with a sensitivity of 94% (95% CI, 88 to 97%) and with a similar exclusion proportion (31% of the patients). (12,19) Consequently, the benefit of associating agglutination d-dimer tests and VDalv/VTalv measurements seems similar to the performance of a rapid ELISA d-dimer test alone. Finally, Wells and coworkers (10) showed in a series of 930 outpatients, in whom capnographic measurements were not performed, that the association of a well-scored clinical probability with the results of a whole-blood agglutination d-dimer assay could safely rule out PE in 47% of the population. Consequently, the relative importance of VCap, d-dimer levels, and clinical probability, as first-line diagnostic tools for patients with suspected PE, needs further evaluation. In our study, in which all of the patients had positive ELISA d-dimer levels, 3 of 17 patients with a low clinical probability had a final diagnosis of PE. Only one of these three patients had a normal PaC[O.sub.2]-EtC[O.sub.2] gradient of 2.4 mm Hg as well as normal parameters from VCap (slope, 1.3%/L; VDalv/VTalv, 13%; Fdlate, -3.2%). This patient had a subsegmental PE shown on the CT scan, but his findings would have constituted a false-negative result if a normal VCap was used in combination with a low clinical probability and a positive d-dimer level to exclude PE.

This study had several limitations. First, we did not use a diagnostic “gold standard” procedure. Nevertheless, we considered that the systematic recourse to pulmonary angiography for all patients could be advantageously replaced by a less invasive diagnostic strategy because our clinical 6-month follow-up was meticulous. Moreover, the diagnostic strategy that we used could be less sensitive or specific, since spiral CT scanning is less accurate than pulmonary angiography in cases of subsegmental defects. Nevertheless, the use of a thin-collimation, multi-detector row, spiral CT scanner improves the detection of peripheral PE. Second, our rapid ELISA d-dimer assay is not the one that is most commonly used throughout emergency departments, because it is more expensive, less rapid, and less specific than the traditional whole-blood agglutination d-dimer assays. However, we considered that the d-dimer levels we used, thanks to their outstanding sensitivity, were the cornerstone of a first step in the diagnostic workup for PE. Third, we have arbitrarily excluded from the study protocol all of the patients with negative d-dimer levels. The diagnostic performance of VCap in our study will therefore not necessarily apply to a general ambulatory population with suspected PE.

In conclusion, this preliminary study suggests that the use of parameters from VCap could improve the relatively low performance of the traditional PaC[O.sub.2]-EtC[O.sub.2] gradient for the diagnosis of PE. A future clinical management protocol should answer the question of the place of this technique in combination with the clinical information and the measurement of d-dimer levels. Nevertheless, we must keep in mind that VCap in spontaneously breathing patients remains strongly dependent on patient cooperation and on the careful verification of a steady-state measurement, limiting the use of this technique to patients with a relatively stable respiratory status.

Table 1–Clinical and Demographic Data in 45

Patients Suspected for PE *

PE-Positive PE-Negative

Variables (n = 18) (n = 27)

Risk factor for PE

Surgery within 3 mo 2 0

Recent immobilization 2 0

Previous VTE 4 3

Family history of VTE 3 3

Cancer in treatment 1 2

Smoking and oral contraception 3 0

Clinical signs of PE

Dyspnea 13 21

Pleuritic chest pain 4 13

Nonretrosternal chest pain 6 18

Hemoptysis 3 0

Pleural rub 0 1

Heart rate > 90 beats/min 10 10

Leg symptoms 3 1

Low-grade fever 1 5

Chest radiograph compatible with PE 0 4

Syncope 2 1

Electrocardiographic signs of PE 3 2

Clinical probability of PE ([dagger])

Low 3 14

Intermediate 6 11

High 9 2

Demographic data

Age, ([double dagger]) yr 65 [+ or -] 4 62 [+ or -] 4


Male 2 7

Female 16 20

Weight, ([double dagger]) kg 67 [+ or -] 3 72 [+ or -] 3

Height, ([double dagger]) cm 164 [+ or -] 1 165 [+ or -] 1

* VTE = venous thromboembolism.

([dagger]) According to Wells et al. (10)

([double dagger]) Values given as clean [+ or -] SD.

Table 2–Criteria Used To Exclude or Confirm the Diagnosis of PE

Diagnostic Criteria Patients, No.

PE confirmed 18

Positive spiral CT scan concordant with a 12

high-probability V/Q lung scan

High clinical pretest probability with a 5

high-probability V/Q lung scan

Positive spiral CT scan with high clinical pretest 1


PE excluded 27

Normal spiral CT scan consistent with a normal or 19

low-probability V/Q lung scan

Low clinical pretest probability with a normal or 6

low-probability V/Q lung scan

Pulmonary angiography for discordant results between 2

V/Q lung scan and spiral CT scan, or for

intermediate-probability V/Q lung scan

Table 3–Volumetric Capnography Data *

Variables PE Confirmed PE Excluded

PaC[O.sub.2], mm Hg 32.7 [+ or -] 1.1 32.3 [+ or -] 1.2

EtC[O.sub.2], mm Hg 27.3 [+ or -] 1.4 29.5 [+ or -] 1.0

PaC[O.sub.2]-EtC[O.sub.2] 5.3 [+ or -] 0.7 2.8 [+ or -] 0.7

gradient, mm Hg

Fdlate ratio, % 8.2 [+ or -] 0.7 -7.7 [+ or -] 2.8

VDalv/VDphys, % 41.9 [+ or -] 3.2 32.6 [+ or -] 2.7

VDalv/VDaw, % 82.9 [+ or -] 11.5 56.0 [+ or -] 7.2

Slope of phase 3, %/L 1.06 [+ or -] 0.14 1.77 [+ or -] 0.22

VDalv/VTalv, % 21.9 [+ or -] 2.5 16.9 [+ or -] 1.9

VDaw/VT, % 22.0 [+ or -] 1.0 25.1 [+ or -] 1.3

VDalv/VT, % 17.2 [+ or -] 2.1 12.6 [+ or -] 1.4

VDphys/VT, % 39.2 [+ or -] 2.1 37.7 [+ or -] 1.8

Peak expiratory flow, 30.84 [+ or -] 2.86 35.56 [+ or -] 3.42


Mean expiratory volume, 8.9 [+ or -] 0.6 9.4 [+ or -] 0.7


Predicted TLC, mL 5183 [+ or -] 292 5435 [+ or -] 248

Pa[O.sub.2], mm Hg 85.8 [+ or -] 4.9 84.4 [+ or -] 3.2

VT, mL 618.2 [+ or -] 49.2 570.9 [+ or -] 64.3

Variables p Value

PaC[O.sub.2], mm Hg 0.85 (NS)

EtC[O.sub.2], mm Hg 0.2 (NS)

PaC[O.sub.2]-EtC[O.sub.2] 0.019

gradient, mm Hg

Fdlate ratio, % 0.000011

VDalv/VDphys, % 0.031

VDalv/VDaw, % 0.043

Slope of phase 3, %/L 0.019

VDalv/VTalv, % 0.12 (NS)

VDaw/VT, % 0.095 (NS)

VDalv/VT, % 0.062 (NS)

VDphys/VT, % 0.6 (NS)

Peak expiratory flow, 0.34 (NS)


Mean expiratory volume, 0.61 (NS)


Predicted TLC, mL 0.31 (NS)

Pa[O.sub.2], mm Hg 0.81 (NS)

VT, mL 0.6 (NS)

* NS = not significant.

ACKNOWLEDGMENT: The authors thank P. Sweeney and E. Heinonen from Datex-Ohmeda Division of Instrumentarium Corporation, and L. Schmidt, I. Buschman, and A. Morias from Meda-Belgium, for their intellectual and technical support in the development of the expirogram curves. The authors thank V. Hubin for the data managing and C. Veriter for his constant dedication to this work.


(1) Robin ED, Julian DG, Travis DM, et al. A physiological approach to the diagnosis of acute pulmonary embolism. N Engl J Med 1959; 260:586-591

(2) Nutter DO, Massumi RA. The arterial-alveolar carbon dioxide tension gradient in diagnosis of pulmonary embolus. Dis Chest 1966; 50:380-387

(3) Vereerstraeten J, Schoutens A, Tombroff M, et al. Value of measurement of alveolo-arterial gradient of PC[O.sub.2] compared to pulmonary scan in diagnosis of thromboembolic pulmonary disease. Thorax 1973; 28:306-312

(4) Hatle L, Rokseth R. The arterial to end-expiratory carbon dioxide tension gradient in acute pulmonary embolism and other cardiopulmonary diseases. Chest 1974; 66:352-357

(5) Colp C, Stein M. Re-emergence of an “orphan” test for pulmonary embolism. Chest 2001; 120:115-119

(6) Fletcher R, Jonson B, Cumming G, et al. The concept of deadspace with special reference to the single breath test for carbon dioxide. Br J Anaesth 1981; 53:77-78

(7) Fletcher R. Deadspace, invasive and non-invasive. Br J Anaesth 1985; 57:245-249

(8) Kline JA, Israel EC, Michelson EA, et al. Diagnostic accuracy of a bedside D-dimer assay and alveolar dead-space measurement for rapid exclusion of pulmonary embolism: a multi-center study. JAMA 2001; 285:761-768

(9) Rodger MA, Jones G, Rasuli P, et al. Steady-state end-tidal alveolar dead space fraction and D-dimer: bedside tests to exclude pulmonary embolism. Chest 2001; 120:115-119

(10) Wells P, Ginsberg JS, Anderson DR, et al. Use of a clinical model for sale management of patients with suspected pulmonary, embolism. Ann Intern Med 1998; 129:997-1005

(11) de Moerloose P, Desmarais S, Bounameaux H, et al. Contribution of a new, rapid, individual and quantitative automated D-dimer ELISA to exclude pulmonary embolism. Thromb Haemost 1996; 75:11-13

(12) Brown MD, Rowe BH, Reeves MJ, et al. The accuracy of the enzyme-linked immunosorbent assay D-dimer test in the diagnosis of pulmonary embolism: a meta-analysis. Ann Emerg Med 2002; 40:13:3-144

(13) Eriksson L, Wollmer P, Olsson CG, et al. Diagnosis of pulmonary embolism based upon alveolar dead space analysis. Chest 1989; 96:357-362

(14) Standardized lung function testing: report working party. Bull Eur Physiopathol Respir 1983; 19(suppl):1-95

(15) Hull RD, Hirsh J, Carter CJ, et al. Diagnostic value of ventilation-perfusion lung scanning in patients with suspected pulmonary embolism. Chest 1985; 88:819-828

(16) Mayo JR, Remy-Jardin M, Muller NL, et al. Pulmonary embolism: prospective comparison of spiral CT with ventilation-perfusion scintigraphy. Radiology 1997; 205:447-452

(17) Hanley JA. Receiver operating characteristic (ROC) methodology: the state of the art. Crit Rev Diagn Imag 1989; 29:307-335

(18) Lee M-LT, Rosner BA. The average area under correlated receiver operating characteristic curves: a non parametric approach based on generalized two-sample Wilcoxon statistics. Appl Stat 2001; 50:337-344

(19) Perrier A. Desmarais S, Miron MJ, et al. Non-invasive diagnosis of venous thromboembolism in outpatients. Lancet 1999; 353:190-195

(20) Perrier A, Desmarais S. Goehring C, et al. D-dimer testing for suspected pulmonary embolism in outpatients. Am J Respir Crit Care Med 1997; 156:492-496

(21) Burki NK. The dead space to tidal volume ratio in the diagnosis of pulmonary embolism. Am Rev Respir Dis 1986; 133:679-685

(22) Olsson K, Jonson B, Olsson CG, et al. Diagnosis of pulmonary embolism by measurement of alveolar dead space. J Intern Med 1998; 244:199-207

(23) Anderson JT, Owings JT, Goodnight JE. Bedside noninvasive detection of acute pulmonary embolism in critically ill surgical patients. Arch Surg 1999; 134:869-874

(24) Kline JA, Meek S, Boudrow D, et al. Use of the alveolar dead space fraction (Vd/Vt) and plasma D-dimers to exclude acute pulmonary embolism in ambulatory patients. Acad Emerg Med 1997; 4:856-863

(25) Kline JA, Arunachlam M. Preliminary study of the capnogram waveform area to screen for pulmonary embolism. Ann Emerg Med 1998; 32:289-296

(26) Johanning JM, Veverka TJ, Bays RA, et al. Evaluation of suspected pulmonary embolism utilizing end-tidal C[O.sub.2] and D-dimer. Am J Surg 1999; 178:98-102

(27) Patel MM, Rayburn DB, Browning JA, et al. Neural network analysis of the volumetric capnogram to detect pulmonary embolism. Chest 1999; 116:1325-1332

(28) Kline JA, Kubin AK, Patel MM, et al. Alveolar dead space as a predictor of severity of pulmonary embolism. Acad Emerg Med 2000; 7:611-617

(29) Elliott CG, Pulmonary physiology during pulmonary embolism. Chest 1992; 101(suppl):163S-171S

(30) Grunkemeier GL, Ruyun J. Receiver operating characteristic curve analysis of clinical risk models. Ann Thorac Surg 2001; 72:323-326

* From the Departments of Emergency and Intensive Care (Drs. Verschuren, Thys, Zech, and Reynaert, and Mr. Roeseler), and Pneumology (Dr. Liistro), Cliniques Universitaires Saint-Luc, Universite Catholique de Louvain, Bruxelles, Belgium; and Datex-Ohmeda Division of Instrumentarium Corp (Mr. Coffeng), Helsinki, Finland

This work was supported by a technical and intellectual collaborative agreement with Datex-Ohmeda Division of Instrumentarium Corp, Helsinki, Finland.

Manuscript received November 18, 2002; revision accepted August 13, 2003.

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail: permissions@chestnet.org).

Correspondence to: Franck Verschuren, MD, Service des Urgences, Cliniques Universitaires Saint-Luc, Universite Catholique e-mail: Frank.verchuren@clin.ucl.ac.be

COPYRIGHT 2004 American College of Chest Physicians

COPYRIGHT 2004 Gale Group

You May Also Like

Premalignant evolution of lung cancer : Gilles F. Filley lecture

Premalignant evolution of lung cancer : Gilles F. Filley lecture Wilbur A. Franklin Abbreviations: ASD = angiogenic squamous dyspla…

Critical Interactions

Critical Interactions Kenneth D. Chinsky Man and Machine Positive pressure ventilation has been used for hundreds of years s…

Catching up on sleep: the National Sleep Disorders Research Plan

Catching up on sleep: the National Sleep Disorders Research Plan – Editorial Barbara A. Phillips Sleep disorders are remarkably com…

The Clinical and Economic Impact of Deep Chest Surgical Site Infections Following Coronary Artery Bypass Graft Surgery

The Clinical and Economic Impact of Deep Chest Surgical Site Infections Following Coronary Artery Bypass Graft Surgery – ) Christopher S. …