Pulmonary
Prospective validation of wells criteria in the evaluation of patients with suspected pulmonary embolism

https://doi.org/10.1016/j.annemergmed.2004.04.002Get rights and content

Study objective

The literature suggests that the d-dimer is useful in patients suspected of having pulmonary embolism and who have a low pretest probability of disease. A previously defined clinical decision rule, the Wells Criteria, may provide a reliable and reproducible means of determining this pretest probability. We evaluate the interrater agreement and external validity of Wells Criteria in determining pretest probability in patients suspected of having pulmonary embolism.

Methods

This was a prospective observational study. Trained research assistants enrolled patients during 120 random 8-hour shifts. Patients who underwent imaging for pulmonary embolism after a medical history, physical examination, and chest radiograph were enrolled. Treating providers and research assistants determined pretest probability according to Wells Criteria in a blinded fashion. Two d-dimer assays were run. Three-month follow-up for the diagnosis of pulmonary embolism was performed. Interrater agreement tables were created. κ Values, sensitivities, and specificities were determined.

Results

Of the 153 eligible patients, 3 patients were missed, 16 patients declined, and 134 (88%) patients were enrolled. Sixteen (12%) patients were diagnosed with pulmonary embolism. The κ values for Wells Criteria were 0.54 and 0.72 for the trichotomized and dichotomized scorings, respectively. When Wells Criteria were trichotomized into low pretest probability (n=59, 44%), moderate pretest probability (n=61, 46%), or high pretest probability (n=14, 10%), the pulmonary embolism prevalence was 2%, 15%, and 43%, respectively. When Wells Criteria were dichotomized into pulmonary embolism–unlikely (n=88, 66%) or pulmonary embolism–likely (n=46, 34%), the prevalence was 3% and 28%, respectively. The immunoturbidimetric and rapid enzyme-linked immunosorbent assay d-dimer assays had similar sensitivities (94%) and specificities (45% versus 46%).

Conclusion

Wells Criteria have a moderate to substantial interrater agreement and reliably risk stratify pretest probability in patients with suspected pulmonary embolism.

Introduction

The role of the d-dimer assay in evaluating patients with suspected pulmonary embolism remains unclear to many physicians. Many clinicians are uncertain about the differences between the various types of d-dimer assays and the appropriate clinical setting in which to use them. Recent reviews1., 2., 3., 4., 5., 6. and published studies7., 8., 9., 10., 11., 12., 13. have shown that the sensitivities and negative likelihood ratios of the rapid enzyme-linked immunosorbent assay and the immunoturbidimetric d-dimer assays are the most promising for the emergency department (ED) setting. However, use of these assays alone is insufficient to exclude the diagnosis of pulmonary embolism in all patients.2., 4., 5., 6., 7. Several authors suggest that the d-dimer assay has the greatest utility in the patient with a low pretest probability.2., 3., 5., 6., 7., 8.

A renewed interest in pretest probability has developed in recent years, leading investigators to develop more reliable means of determining pretest probability using clinical decision rules. Several such rules have been recently published.14., 15., 16., 17., 18., 19. Unfortunately, many of these rules are complicated and difficult to apply, especially in a busy ED environment.14., 15., 17., 18. In recent years, Wells et al14., 19., 20. published, refined, and then internally validated a clinical decision rule for pretest probability of pulmonary embolism.

Wells Criteria for pretest probability of pulmonary embolism consist of the 7 following weighted criteria: (1) clinical signs and symptoms of deep venous thrombosis (+3.0); (2) an alternative diagnosis that is less likely than pulmonary embolism (+3.0); (3) pulse rate greater than 100 beats/min (+1.5); (4) immobilization or surgery in the previous 4 weeks (+1.5); (5) previous deep venous thrombosis/pulmonary embolism (+1.5); (6) hemoptysis (+1.0); and (7) malignancy (on treatment, treated in the past 6 months, or palliative; +1.0). Summation of these point values can be trichotomized into low (<2), moderate (2 to 6), or high (>6) pretest probability with prevalences for pulmonary embolism of 2% to 4%, 19% to 21%, and 50% to 67%, respectively. Alternatively, the total score can be dichotomized into a pulmonary embolism–unlikely (≤4) or a pulmonary embolism–likely (>4) pretest probability with prevalences for pulmonary embolism of 5% to 8% and 39% to 41%, respectively.14., 19.

The studies by Wells et al14., 19., 20. suggest that for a patient with a low pretest probability, as defined by Wells Criteria, and a negative whole-blood d-dimer assay, the diagnosis of pulmonary embolism can be reliably excluded. External, retrospective validation of Wells Criteria's ability to reliably risk stratify patients according to pulmonary embolism prevalence has also recently been published.21 However, the interrater agreement and validity of Wells Criteria have not been reported independently and prospectively.

We therefore conducted a study with 2 objectives: (1) to evaluate the interrater agreement of the Wells Criteria; and (2) to externally validate the ability of the Wells Criteria to determine pretest probability prospectively.

Section snippets

Study design

This was a prospective, observational study of patients presenting to the ED with suspected pulmonary embolism and who underwent diagnostic imaging for pulmonary embolism. The study hospital's institutional review board approved the study protocol, and written informed consent was obtained from all study participants.

Setting

The study was conducted from August 2001 through June 2002 at an emergency medicine residency–affiliated, community-based ED, serving a predominantly managed-care patient

Results

One hundred seventy-six patients who presented to the ED during the 120 study shifts met our enrollment criteria. Demographic data are listed in Table 1. Of these patients, 23 (23/176, 13%) were excluded for the following reasons: non–English speaking (2); pregnant at the time of the study (1); pregnant within the previous 6 months (3); morbidly obese (3); age older than 85 years (5); age younger than 18 years (2); unable to consent or critically ill (3); and known to have a recently elevated

Limitations

Our study has several important limitations. First, our study is limited in that 17 patients did not undergo an initial imaging study for pulmonary embolism because of negative d-dimer results but were considered to be ruled out for a pulmonary embolism according to a negative telephone follow-up. In all of these 17 follow-up instances, direct contact was made with the patients, and a proxy follow-up was unnecessary. One patient with hemolyzed d-dimers and a negative workup and follow-up was

Discussion

To our knowledge, this is the first study to examine the interrater agreement and prospective external validity of Wells Criteria. This clinical decision rule appears to reproducibly and reliably risk stratify patients with suspected pulmonary embolism into either a low, moderate, and high classification or a pulmonary embolism–unlikely and pulmonary embolism–likely classification. Furthermore, the 2 d-dimer assays used in our study appeared to perform with equal sensitivities and

References (34)

  • I. Chagnon et al.

    Comparison of two clinical prediction rules and implicit assessment among patients with suspected pulmonary embolism

    Am J Med

    (2002)
  • A. Perrier et al.

    Non-invasive diagnosis of venous thromboembolism in outpatients

    Lancet

    (1999)
  • S.J. Swensen et al.

    Outcomes after withholding anticoagulation from patients with suspected acute pulmonary embolism and negative computed tomographic findings: a cohort study

    Mayo Clin Proc

    (2002)
  • P.D. Stein et al.

    Clinical, laboratory, roentgenographic, and electrocardiographic findings in patients with acute pulmonary embolism and no pre-existing cardiac or pulmonary disease

    Chest

    (1991)
  • S.Z. Goldhaber

    Pulmonary embolism

    N Engl J Med

    (1998)
  • J.A. Kline et al.

    New diagnostic tests for pulmonary embolism

    Ann Emerg Med

    (2000)
  • H. Bounameaux et al.

    Plasma measurement of d-dimer as diagnostic aid in suspected venous thromboembolism: an overview

    Thromb Haemost

    (1994)
  • Cited by (177)

    • Echocardiographic Evaluation of Pulmonary Embolism: A Review

      2023, Journal of the American Society of Echocardiography
    • Venous Thromboembolism Management in Pregnant Patients

      2023, Techniques in Vascular and Interventional Radiology
    • 38-Year-Old Woman With Chest Pain

      2023, Mayo Clinic Proceedings
    View all citing articles on Scopus

    Author contributions: TRM and DMA conceived the study and obtained funding. SJW, TRM, and KMF supervised the conduct of the study. All authors were involved in data collection, and SJW managed the data. SJW, TRM, and KMF were responsible for quality control. Statistical analysis was performed by an independent statistician and in part by SJW. SJW drafted the manuscript, and SJW, TRM, RMF, JPF, and DMA contributed substantially to its revision. SJW takes responsibility for the paper as a whole.

    Presented at the Society for Academic Emergency Medicine annual meeting, St. Louis, MO, May 2002.

    Supported in part by a Medicare grant (Medicare 2001 Kaiser Permanente/Exempla Saint Joseph Hospital Research Fund).

    Reprints not available from the authors.

    View full text