The trusted source for
healthcare information and
Source: Wicki J, et al. Assessing clinical probability of pulmonary embolism in the emergency ward: A simple score. Arch Intern Med 2001;161:92-97.
This five-year study from Switzerland examined the clinical characteristics of patients who presented to the emergency department (ED) with complaints of possible pulmonary embolism (PE). Data were obtained from two prior reports that studied plasma D-dimer in PE. The objective was to develop a simple standardized clinical score to stratify ED patients with clinically suspected pulmonary embolism into groups with a high, intermediate, or low probability of PE to refine the diagnostic approach to be less invasive. A database of 1090 consecutive patients was subjected to a standard algorithm that established diagnostic criteria for PE. Logistic regression was used to predict clinical parameters associated with PE. Two hundred ninety-six patients (27%) were found to have PE. The optimal estimate of clinical probability was based on eight variables: recent surgery, previous thromboembolic event, older age, hypocapnia, hypoxemia, tachycardia, and band atelectasis or hemidiaphragm elevation on chest x-ray. A probability score was calculated by adding points assigned to these variables. The most heavily-weighted variables were recent surgery and room-air hypoxia. The surgery had to have occurred during the last month, and included the following procedures: orthopedic, hip, knee, or extensive pelvic or abdominal surgery. The scores have been converted from European units (see Table).
Table: Probability Score Conversion
|Age > 80||2|
|Previous PE or DVT||2|
|Pulse rate > 100 bpm||1|
|PaCO2 < 36 mmHg||2|
|PaCO2 = 36-38.9 mmHg||2|
|PaO2 < 48.7 mmHg||4|
|PaO2 = 48.7-59.9 mmHg||3|
|PaO2 = 60-71.1 mmHg||2|
|PaO2 = 71.2-82.3 mmHg||1|
A cutoff score of 4 best identified patients with low probability of PE. A total of 486 patients (49%) had a low clinical probability of PE (score £ 4). This group had a 10% prevalence of PE. Patients with intermediate scores (5-8) represented 44% of the patients and had a 38% prevalence of PE; 6% of the patients had a score of 9-12 and an 81% prevalence of PE. When compared to the clinician’s pretest suspicions (second- and third-year internal medicine residents), the scoring system did better at predicting high and intermediate probabilities, but both predicted low probability patients with similar frequency. The authors conclude that less invasive testing would be warranted in patients with low clinical probability, and that this system identifies those patients.
A good friend of mine and I once decided that we’d rather have a successful gambler than a genius as our personal physician. The genius might be correct with a high probability—but one fatal mistake out of thousands is considered poor odds for the successful gambler, who would rather be wrong safely many times than fatally wrong once. Emergency medicine does force us to play the odds because we often must choose treatment and dispositions without complete answers. This dilemma plays out in PE in very concrete ways. We know from the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study that patients judged to have a low pretest clinical probability of PE and a low probability lung scan have a 4% incidence of PE.1 Other studies have determined that the addition of a normal lower limb venous ultrasound drives these odds even lower.2 Even though we may be wrong one time out of 100, at those odds, applying a standard like pulmonary angiography to all low probability patients might increase the morbidity and mortality from procedural complications beyond that native to disease prevalence. Is that a gamble worth taking to make the diagnosis? The answer is not clear. The good news is that this clinical scoring system gives us a concrete method of measuring the odds before we decide how far to go to make this diagnosis or when to begin empiric therapy. I might guess that a particular patient has a particular probability, but the scores could change my suspicions and prompt me to be more or less aggressive, as appropriate.
One limitation is that this study assumes that physicians have performed an arterial blood gas on every patient. It may not be valid to assume that normal pulse oximetry and respiratory rates always mean a normal room air blood gas. While studies like PIOPED suggested that the PaO2 alone was not helpful in predicting or excluding the diagnosis, this study suggests that it may provide useful information in playing the odds.
Bayes’ rule also reminds us that since this study was done in a population with a PE prevalence of 27%, the outcomes are most reproducible in similar populations. In general, if the variables in this study are weighted properly, these findings will be more robust in populations with higher prevalence of PE, and less so when prevalence is lower.3 As the bottle of hair tonic says: "Your results may vary."
1. The PIOPED Investigators. Value of the ventilation/ perfusion scan in acute pulmonary embolism. JAMA 1990;263:2753-2759.
2. Perrier A, et al. Noninvasive diagnosis of venous thromboembolism in outpatients. Lancet 1999;353: 190-195.
3. Sackett DL, et al. Clinical Epidemiology: A Basic Science for Clinical Medicine. 2nd ed. London, England: Little Brown & Co; 1991.