Think pulmonary embolism and you may find it….

This is part of a recurring series examining landmark articles in Emergency Medicine, in the style of ALiEM’s 52 Articles.

Discussing:  “Prospective Multicenter Evaluation of the Pulmonary Embolism Rule-out Criteria.” (J Thromb Haemost 2008;  Kline JA et al.)

Screen Shot 2015-09-07 at 1.58.00 PM

Main Points

  1. Pulmonary Embolism Rule-out Criteria (PERC):
  • age < 50,
  • heart rate < 100 bpm
  • SaO2 < 95%
  • no hemoptysis
  • no estrogen use
  • no surgery/trauma requiring hospitalization within 4 weeks
  • no unilateral leg swelling
  • no prior venous thromboembolism

2) PERC in combination with clinical gestalt for low suspicion of pulmonary embolism (PE) reduces the probability of venous thromboembolism (VTE) to below 2%.


Chest pain is the most common emergency department chief complaint.  The differential diagnosis is vast, and includes high acuity conditions (such as myocardial infarction, aortic aneurysm, and pulmonary embolism) down to low acuity conditions (such as reflux, muscle strain and anxiety.)  Pulmonary embolism is one of the high risk clinical conditions that should not be missed. However, determining which patient should or should not be worked up for pulmonary embolism can be difficult.  

The authors of this study stratified patients across multiple hospitals into risk categories based upon clinical gestalt and PERC in order to determine post-test probability for VTE or death.  They successfully determined that patients with low suspicion of PE who were PERC (-) had <2% probability of PE.  This study is incredibly useful to the emergency physician in order to prevent unnecessary testing, thus preventing the overutilization of resources and decreasing patients’ exposure to radiation.  


This was a prospective, multicenter study (13 Emergency Departments) that enrolled 8138 patients.  The main outcome was image proven VTE or death within 45 days.  85% of all patients in the study presented with the chief complaint of chest pain (53%) or dyspnea (33%).  Of the 8138 patients, 1666 (20%) were classified as low suspicion and PERC (-).  Within this group, 15 were VTE (+) and one patient died, which equals 1% of all patients studied.  PERC (-) as an independent diagnostic test has a sensitivity of 95.7% (95% CI 93.6%-97.2%) and a specificity of 25.4% (95% CI 24.4%-26.4%.  PERC (-) in combination with low suspicion (low clinical gestalt for PE) yielded a diagnostic sensitivity equal to 97.4% (95% CI 95.8%-98.5%) and a specificity of 21.9% (95% CI 21.0%-22.9%).  Of all 8138 patients who were tested for PE in the ED, 6.9% were VTE (+) within 45 days.  

Level of Evidence

According to the ACEP grading template for grading of evidence for diagnostic questions, this study receives a grade 1.  It was a prospective study that enrolled 8138 patients over 13 institutions.  It focused on a direct and highly applicable clinical question.  The only question remains is: is there an even more effective way of decreasing testing for the low suspicion patient? 


What was most surprising about this study was that so much of the decision making is based upon clinical gestalt.  Clinical gestalt and PERC, vs PERC alone, increases sensitivity by almost 2%.  Other surprises were that only 6.9% of all patients ultimately have VTE,  two-thirds of physicians when ordering a test for PE have a low-suspicion for it, and 80% of clinicians, when ordering a test for PE, believe in an alternative diagnosis.   

Source Article

Kline JA, Courtney DM, Kabrhel C, Moore CL, Smithline HA, Plewa MC, Richman PB, O’Neil BJ, Nordenholz K. Prospective Multicenter Evaluation of the Pulmonary Embolism Rule-out Criteria. J Thromb Haemost 2008; 6: 772-80


2 thoughts on “Think pulmonary embolism and you may find it….

  1. Great review. Nice job. I know you mentioned it but just want to further highlight that you should only apply PERC to those patients in which you have low pretest probability (about 15%) in order to reduce to an acceptable 2% post test probability. If you apply to someone at a higher pretest probability then you will miss an unacceptable amount. Also PERC is different than most other rules in which is it unidirectional. If a patient has a positive feature, you don’t necessarily need to continue your evaluation as it was not studied in this fashion. This is unlike the NEXUS criteria which is bidirectional. Thanks Dr. Blackwood for your post.

Leave a Reply