Article Text

PDF

The additional use of end-tidal alveolar dead space fraction following D-dimer test to improve diagnostic accuracy for pulmonary embolism in the emergency department
  1. Young Hoon Yoon1,
  2. Sung Woo Lee1,
  3. Dong Min Jung1,
  4. Sung Woo Moon1,
  5. Jan Kurt Horn2,
  6. Yun-Sik Hong2
  1. 1Department of Emergency Medicine, College of Medicine, Korea University, Seoul, Korea
  2. 2Department of Surgery, San Francisco General Hospital, University of California San Francisco, San Francisco, CA, USA
  1. Correspondence to Professor Sung-Woo Lee, Korea University Ansan Hospital Emergency Department, 516 Gojan-dong, Danwon-gu, Ansan 425-707, Korea; kuedlee{at}korea.ac.kr

Abstract

Purpose To determine the diagnostic performance of bedside assessment of end-tidal alveolar dead space fraction (ADSF) for pulmonary embolism (PE) and whether the use of additional ADSF assessment following D-dimer assay can improve the diagnostic accuracy in suspected PE patients in the emergency department.

Methods A prospective observational study of 112 consecutive adult patients suspected of PE of whom 102 were eligible for analysis. ADSF was calculated using arterial carbon dioxide and end-tidal carbon dioxide. An ADSF less than 0.2 was considered normal.

Results PE was confirmed in 11 (10.8%) of 102 patients. D-dimer assay alone as a reference standard test for PE had a sensitivity of 100%, specificity of 38.5% and false negativity of 0%. Area under the receiver-operator characteristic curve for the diagnosis of PE using ADSF values alone was 0.894, Sensitivity, specificity and false negativity for the combined results of a positive D-dimer test and abnormal ADSF were 100%, 78.0% and 0% for the presence of PE, respectively. Of 65 patients with a low or intermediate clinical probability and a positive D-dimer assay, 36 (55.4%) patients displayed normal ADSF and had no PE.

Conclusions By itself ADSF assessment performed well in diagnosis of PE. The combined result of a positive D-dimer and abnormal ADSF increased the specificity for diagnosing PE compared with the D-dimer test alone. The use of additional bedside ADSF assessment following a positive D-dimer test may reduce the need for further imaging studies to detect PE in patients with a low or intermediate clinical probability.

  • Alveolar dead space fraction
  • clinical probability
  • D-dimer
  • diagnosis
  • end-tidal carbon dioxide
  • pulmonary embolism
  • respiratory
  • thrombo-embolic disease

Statistics from Altmetric.com

Diagnostic tests for pulmonary embolism (PE) are often based on clinical algorithms that include clinical probability, various laboratory tests and the findings of radiological studies.1–3 In many emergency departments (ED) including our institution, the confirmation of PE relies on chest computed tomography (CT) scan. High clinical probability or positivity of a D-dimer assay are considered indications for obtaining chest CT.4 The D-dimer test alone as an indicator of clot formation is perhaps the most promising laboratory assessment, because it can safely rule out PE.4 The sensitivity of the D-dimer assay depends on the methods of assay.1 5 6 ELISA-based methods for detecting D-dimer levels are recognised as the most sensitive tests available. A newer D-dimer assay using fluorescence immunoassay has demonstrated sensitivity profile similar to standard ELISA method.6 However, because D-dimer levels may also be elevated with cancer, following recent trauma and surgery and in many other medical conditions, it is a non-specific test. The low specificity of the D-dimer assay increases the use of chest CT in the ED.1 4 To improve clinical accuracy, additional bedside tests are required to exclude PE and avoid the need for costly imaging studies.

The ideal additional PE screening test for use in the ED in combination with the D-dimer assay must have high sensitivity and low false negativity for the diagnosis of PE7 because delay in the diagnosis of PE contributes to poor outcomes such as death or disability.8 Furthermore, the use of an additional test must improve the specificity for the exclusion of PE compared with the D-dimer test alone, thereby reducing the need for further imaging studies. In addition, it should be easily achieved at the bedside.

PE contributes to ventilation/perfusion mismatch by increasing functional dead space.4 9 The end-tidal alveolar dead space fraction (ADSF) can be easily calculated using the measurement of end-tidal carbon dioxide (Petco2) and arterial carbon dioxide (Paco2). The single use of ADSF alone may be limited for the diagnosis of PE. Hogg et al10 reported that when the Bohr-calculated respiratory dead space is combined with the latex agglutination D-dimer test, the joint test has high sensitivity and specificity for the diagnosis of PE. Several studies have demonstrated a high negative predictive value for ADSF measurement plus D-dimer negativity in ambulatory patients with stable vital signs.11–13 However, because a negative latex agglutinin D-dimer result is highly sensitive by itself in the exclusion of PE, additional negativity of ADSF may not improve the ability to exclude PE.

The aims of this study are to determine: (1) the diagnostic efficacy of ADSF alone for PE; (2) whether the combination of a positive D-dimer result and positivity for bedside assay of ADSF of 0.2 or greater can improve the diagnostic accuracy for PE in the ED; and (3) whether the use of additional ADSF following a positive D-dimer assay can reduce the need for further imaging in patients with a low or intermediate clinical probability.

Patients and methods

Selection of study patients

This was a prospective observational study conducted in the ED of a university teaching hospital. Patients over 18 years of age who presented to the ED up to 24 h after the onset of symptoms suspicious for PE were consecutively enrolled from January 2007 to March 2008. The symptoms suspicious for PE are shown in table 1. Exclusion criteria were: (1) late presentation more than 24 h after symptom onset; (2) anticoagulation treatment during the 48 h before study commencement; (3) terminal illness: (4) electrocardiography (ECG) findings consistent with acute myocardial infarction; or (5) contraindication to contrast administration. The Institutional Review Board of our hospital approved the study and the need for informed consent was waived because this study deploys clinical assessments deemed minimally invasive to patients. Patient identity was protected throughout.

Table 1

Clinical characteristics of patients with suspected PE

Study protocol

When patients presented to the ED, blood pressure, respiratory rates, heart rates, ECG and capnography were routinely measured and patient demographic information and chief complaints were recorded. If patients had respiratory or circulatory compromise elicited in the primary survey, they received emergency resuscitation including intravenous fluid infusion and mechanical ventilation as required. After the primary survey, we assessed the clinical probability of PE and performed a D-dimer test. The need for further diagnostic assessment with multidetector row chest CT angiography (MD-CTA) and clinical follow-up were decided upon based on the initial findings of the clinical probability assessment and the D-dimer result. The criteria for obtaining MD-CTA were: (1) a high clinical probability regardless of D-dimer positivity; (2) a low or intermediate clinical probability and a positive D-dimer test; or (3) an intermediate clinical probability, a negative D-dimer test and the absence of an alternative diagnosis more likely than PE.3 The clinical probability of PE was assessed using the revised Geneva scoring system.14 D-dimer levels were determined using a fluorescence immunoassay agglutinin method (Biosite, San Diego, USA). A D-dimer of 350 μg/l or greater was deemed a positive result.

Paco2 and Petco2 levels were assessed before MD-CTA in the ED. Although patients were informed of the results of the D-dimer assay and arterial blood gas analysis to decide on the need for CT, the results of ADSF were not considered in the decision to proceed with MD-CTA.

All patients, regardless of diagnosis, were followed up clinically at our outpatient clinic for 3 months. We reviewed their medical records after completion of the study.

Reference standard

The combination of clinical probability, latex agglutinin D-dimer test, and MD-CTA (Brilliance CT 64-channel scanner; Philips, Eindhoven, The Netherlands) served as reference standard tests and a 3-month follow-up review was instituted to exclude or confirm the diagnosis of PE. PE was excluded in patients with an MD-CTA negative finding for PE or in patients with a low clinical probability and a negative D-dimer test. PE was confirmed positive on MD-CTA by standard radiological MD-CTA criteria for PE. All MD-CTA were evaluated by two independent radiologists and findings were further confirmed by a thoracic radiology specialist. The results of ADSF levels were blinded to the radiologists.

Study design

For the purposes of this study, we designated the sensitivity, specificity and false negativity of the D-dimer assay as reference standards for PE. We examined the receiver-operator characteristic (ROC) curve for the diagnosis of PE versus ADSF alone. We calculated the sensitivities, specificities and false negativity for the combined test of a positive D-dimer result and ADSF of 0.2 or greater for a diagnosis of PE. If both the D-dimer assay and ADSF assessment were positive, the combined test result was considered positive. If either or both tests were negative, the combined test result was designated as negative.

Further subgroup analysis was performed to assess the sensitivity, specificity and false negativity of ADSF measurement alone to exclude PE in patients screened with a low or intermediate clinical probability and positive D-dimer test. As the high clinical probability subgroup (two patients in our study) will always require further imaging studies, the utility of additional screening tests was not analysed for this group, and we excluded the patients with high clinical probability from this subgroup analysis.15

Data collection

Data collection for clinical probability and assessment of D-dimer levels and ADSF were performed in the ED by emergency physicians and nurses all of whom received previous training by the investigators. Petco2 was measured using side stream capnography (Ordion, Jerusalem, Israel) with continuous display, because it is suitable in the ED setting for non-intubated patients and requires minimal patient cooperation.16 The capnograph was calibrated according to the manufacturer's instructions before use. Exhaled gas was sampled for Petco2 measurements with the endotracheal tubes when deployed or by using a nasal catheter (Ordion) in the majority of cases. To avoid sampling errors, we established steady state conditions for respiration, blood pressure and heart rate after ventilation had stabilised (respiratory variation of plus or minus two breaths per minute over 1 min). Once the respiratory rate was invariable, Petco2 was recorded and arterial blood gas was determined simultaneously. At the end of the study, an independent researcher who was blinded to the results of clinical status, D-dimer assay and MD-CTA then calculated the ADSF using the values of Petco2 and Paco2. ADSF was calculated by dividing the difference between Paco2 and Petco2 by Paco2.11–13 ADSF values of 0.2 or greater were considered abnormal and are referred to as positive results.12 If the difference between Paco2 and Petco2 fell into the range of 0 to −3 mm Hg, this was considered negative for ADSF.

Statistical analysis

We compared the sensitivity, specificity and false negativity for a diagnosis of PE between the D-dimer test alone versus the combined test of a positive D-dimer assay and ADSF of 0.2 or greater. All statistical analyses were carried out using SPSS version 11.0, and data are expressed as means±SD. Sensitivity and specificity for diagnosing PE are expressed as values (95% CI). The 95% CI of these values were calculated using the Wilson ‘score’ method.17

Results

The characteristics of selected patient and reference standard outcome

One hundred and twelve patients were enrolled in this study, and 10 patients were excluded from the analysis because of renal failure in three patients, ECG findings indicative of acute myocardial infraction in six patients and late presentation in one patient (figure 1).

Figure 1

Reference standard and outcome. The incidence of pulmonary embolism (PE) was assessed with a multidetector row chest computed tomography angiography (MD-CTA) and a 3-month follow-up review was instituted to confirm initial impressions.

The clinical characteristics and demographics of 102 patients appropriate for analysis are displayed in table 1. MD-CTA was obtained in 71 patients and PE was confirmed in 11 (10.8%) of 102 patients (figure 1). The prevalence of PE stratified by the clinical probability was 100% for a high clinical probability, 8.3% for an intermediate clinical probability and 10.3% for a low clinical probability. Fourteen (13.7%) patients died in the hospital, and two of these mortalities were attributed to PE (table 1).

Eighty-seven patients were followed up in the outpatient clinic of our hospital and one patient was contacted by telephone to achieve 100% follow-up in 88 surviving patients. No cases of new-onset PE were encountered within 3 months of follow-up.

D-dimer assay as a reference standard test for PE

The sensitivity, specificity and false negativity of the D-dimer assay as a reference standard test for PE were 100% (74.1–100%), 38.5% (29.1–48.7%), and 0% (0–9.9%), respectively, in 102 study patients (figure 1).

Diagnostic performance of single ADSF test for PE

Eleven patients who displayed the difference between Paco2 and Petco2 in the range of 0 to −3 mm Hg were considered to be negative for ADSF. Of 42 patients with ADSF of 0.2 or greater, 11 patients were diagnosed with PE. The area under the ROC curve for the diagnosis of PE using ADSF values alone was 0.894 (95% CI 0.866 to 0.973; figure 2). ADSF of 0.2 of greater had a sensitivity of 100% (95% CI 74.1% to 100%), specificity of 65.9% (95% CI 55.7% to 74.9%) and false negativity of 0% (95% CI 0% to 6.0%) for the diagnosis of PE.

Figure 2

Receiver-operator characteristic curve of alveolar dead space fraction (ADSF) alone for diagnosing pulmonary embolism. The area under the curve (AUC) was 0.894 (n=102).

Combined results of a positive D-dimer and positive ADSF for diagnosing PE

Table 2 shows the diagnostic characteristics when the reference standard latex agglutinin D-dimer test was combined with the positivity of ADSF.

Table 2

The combined result of D-dimer and ADSF for diagnosing PE (n=102)

In subgroup analysis of 65 patients who had a low or intermediate clinical probability and positive D-dimer test, 36 (55.4%) patients showed normal ADSF and did not have PE (table 3). The additional use of ADSF in patients with a positive D-dimer assay might reduce the need for MD-CTA by 55.4% (table 3).

Table 3

Additional use of alveolar dead space ventilation fraction for diagnosing PE in patients with a low or intermediate the clinical probability and positive D-dimer (n=65)

Discussion

We explored whether ADSF levels provide additional value for the diagnosis of PE and reduce the need for MD-CTA. Our study suggests that the combined results of a positive D-dimer assay and ADSF of 0.2 or greater increased the specificity from 38.5% (95% CI 29.1% to 48.7%) to 78.0% (95% CI 68.5% to 85.3%) for diagnosing PE with the same sensitivity when it was compared with the D-dimer test alone in suspected patients. In addition, a normal ADSF may reduce the need for MD-CTA to diagnose PE in the cohort of patients with a low or intermediate clinical probability and a positive D-dimer assay.

Kline et al7 reported that the use of a screening test for PE must be restricted to a population with a sufficiently low pretest probability. In our study, patients showed a prevalence of 10.8% for PE and they benefitted significantly from the use of non-invasive bedside screening for PE.

We showed that ADSF alone displayed a significant area under the ROC curve indicating reliability for diagnosing PE along with a high sensitivity and low false negativity for the diagnosis of PE. The single use of ADSF for the diagnosis of PE has also been investigated by others.10 11 For example, Hogg et al10 reported that ADSF measurements that were calculated from the Bohr, Enghoff and capillary sample Enghoff methods are not clinically reliable as stand-alone tests for the diagnosis of PE. Verschuren et al11 reported that the late dead space fraction (the extrapolation of the capnographic curve to a volume of 15% of the predicted total lung capacity) performed well for diagnosis in PE. In their study, the mean area under the ROC curve of the late dead space for diagnosing PE was 87.6 ± 4.9%11 and compares favourably with 0.894 (95% CI 0.866 to 0.973) as we demonstrated. In comparison, it should be noted that we explored alternative conditions in our study. We excluded patients with late presentation of over 24 h after symptom onset, calculated ADSF using Paco2 and Petco2, and used a cut-off value of 0.2 for abnormality of ADSF. In addition, the range of 0 to −3 mm Hg in the difference of Paco2 and Petco2 was considered negative for ADSF in our study. As carbon dioxide accumulates in air sampling catheters this might increase Petco2 values measured by sidestream capnometry and result in negative values for the difference of Paco2 and Petco2 despite frequent calibration of capnometry.18 The differences in the selection of study patients; method of calculation for ADSF and cut-off value of ADSF may affect the sensitivity of ADSF alone for the diagnosis of PE.

It is well known that ADSF increases in many other medical conditions such as shock, sepsis, acute respiratory distress syndrome, and is not as specific for PE as the D-dimer assay.18 Therefore, the combined test of D-dimer assay and ADSF assessment may compensate for the weaknesses of individual tests by themselves and increase the clinical efficacy of PE diagnosis in suspected PE patients.

The use of combined bedside tests to screen for PE has been reported by others.10 13 In contrast to our study, Kline et al13 used a combined result of negative D-dimer and normal dead space to exclude PE safely. Our study demonstrated the additional value of ADSF following a positive D-dimer assay, recognising alternatively that a negative latex agglutinin D-dimer result is by itself highly sensitive for the exclusion of PE. Hogg et al10 reported that the combined use of Bohr-calculated respiratory dead space greater than 0.37 and a positive latex agglutination D-dimer test has a high sensitivity of 90.5% (71.1–97.3%) and specificity of 72.3% (67.5–76.7%) for the diagnosis of PE. Their result was similar to our findings for the combined use of a positive latex agglutinin D-dimer result and ADSF of 0.2 or greater. This combined test can easily be performed in the ED by an emergency physician and a nurse, and improves the accuracy of either test used individually.

We also showed that a normal ADSF result before imaging excluded PE in the cohort of patients with a low or intermediate clinical probability and positive D-dimer test; 55.4% of our patients with a low or intermediate clinical probability and positive D-dimer displayed negative ADSF and had no PE. We conclude that the additional finding of a normal ADSF result following a positive D-dimer assay may reduce the need for MD-CTA to rule out PE in the cohort of patients displaying a low or intermediate clinical probability.

Our findings add to a growing number of clinical studies reporting on the use of reliable bedside clinical assessments to reduce the need for expensive imaging studies. However, this study has several limitations. First, we did not perform a power analysis to estimate the sample size before the study. The low cohort number of this study prevented the combined use of a positive D-dimer result and ADSF of 0.2 or greater from achieving statistical significance as a screening test. Despite showing that negative results for the combined test (negative D-dimer or ADSF <0.2) displayed false negativity of 0% (95% CI 0% to 5.1%) for the exclusion of PE, approximately 400 patients would have been required to estimate a false negative rate less than 1%, with an upper limit of the 95% CI less than 2% in a cohort with a prevalence of 10%.7 10 Despite this limitation of a lower cohort number the combined test had statistically better diagnostic efficacy for PE than the D-dimer test alone in this study. Second, this study enrolled patients based upon the emergency physician's suspicion of PE. However, patients who had a low risk of PE, essentially a close to no-risk group, were not enrolled and released based upon the clinical decision of emergency physicians involved in the study. Hogg et al10 suggested the importance of objective, well-defined inclusion criteria for the selection of patients for the study of PE. In their study, the majority of patients with pleuritic chest pain (86.6%) scored a low clinical probability.10 In our study, 28 (25.7%) of 102 patients had low clinical probability. When we retrospectively analysed the 112 enrolled patients relying on the PE rule-out criteria without a D-dimer test,19 only seven low-risk patients would have been captured in this study. This low frequency of low-risk patients further contributed to our low cohort number in the study. Third, the result of D-dimer assays was used as a reference standard and not a research variable for the diagnosis of PE. Therefore, the results of the combined test would potentially be subject to incorporation bias.

Despite these limitations, we believe that our findings indicate that the use of additional ADSF can improve the diagnostic accuracy for PE compared with the D-dimer test alone, and may reduce the need for further imaging. To overcome these limitations and significantly prove that the combined testing of D-dimer and ADSF are efficacious, a multicentre study would be required utilising a large cohort and rigidly controlled inclusion criteria.

Conclusions

ADSF assessment showed good diagnostic performance for PE. The combined result of a positive D-dimer test and ADSF of 0.2 or greater improved the specificity for diagnosing PE compared with the D-dimer test alone, and the additional use of ADSF following a positive D-dimer test may reduce the need for further imaging for diagnosing PE in patients with a low or intermediate clinical probability.

References

View Abstract

Footnotes

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Korea University Ansan Hospital.

  • Provenance and peer review Not commissioned; externally peer reviewed.

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.