Diagnosis of Thoracic Aorta Dissection
Abstract & Commentary
By Michael H. Crawford, MD, Editor
Source: Rogers AM, et al. Sensitivity of the aortic dissection detection risk score, a novel guideline-based tool for identification of acute aortic dissection at initial presentation. Circulation 2011;123:2213-2218.
Thoracic aortic dissection is notoriously difficult to diagnose. Since the presenting symptoms are protean, it is not feasible to image everyone with symptoms that could be due to dissection. Thus, a risk assessment tool was devised by an expert committee, but it has never been validated clinically. These investigators applied the risk score to the International Registry of Acute Aortic Dissection database to test its utility for diagnosing aortic dissection. More than 2500 patients in the registry were categorized by 12 clinical markers: five predisposing conditions, three pain features, and four exam features. Those with no risk markers were scored 0; those with markers in at least one of the three categories were scored 1; and markers in two or three categories were scored 2 or 3, respectively. Score 0 was considered low risk; score 1 was intermediate risk; and 2 or 3 was high risk. A score of 0 was found in 4%; 1 in 37%; and 2 or 3 in 59%. Among the 108 low-risk score 0 patients, 72 had chest x-rays and 49% had a widened mediastinum. Using an algorithm based upon score and chest x-ray when appropriate, the overall sensitivity for the detection of aortic dissection was 96%. The most common of the 12 individual risk markers were abrupt onset of pain (79%); severe pain (73%); ripping or tearing pain (22%); new murmur of aortic regurgitation with pain (24%); and a pulse deficit or upper extremity blood pressure differences (20%). The authors concluded that this clinical risk marker score was highly sensitive for detecting aortic dissection.
This is an interesting study because clinical factors believed to be helpful in the diagnosis of aortic dissection were collated into a proposed risk score by a group of experts without any clinical testing. Of course this happens all the time and we often never know exactly how useful these scores will be. In this case, a large database was used to test the scores utility in retrospect. Although it did well (sensitivity 96%), a prospective study would give us more confidence in its utility. However, it is difficult to study a low incidence event like aortic dissection prospectively.
Inspection of the 12 individual markers shows that some were much more useful than others: Abrupt onset of pain and severe pain occurred in more than 70%. A new murmur of aortic regurgitation in conjunction with pain, ripping or tearing pain, and a pulse deficit or systolic blood pressure difference between limbs occurred in 20%-24%. These three features of the pain history and two physical examination findings seem more specific for aortic dissection than other less common findings such as known thoracic aortic aneurysm, known aortic valve disease, focal neurologic deficit, and hypotension or shock, which occurred in 11%-16%. The other three markers (Marfan Syndrome, family history of aortic disease, and recent aortic manipulation) occurred less than 5% of the time.
Unfortunately, this study cannot assess specificity because all the patients had aortic dissection. It is likely that specificity and hence positive-predictive valve will be lower than the sensitivity. This puts the aortic dissection score in the category of other highly sensitive tests with high negative predictive values such as d-dimer, troponin, and BNP. How much use of such a score will cut down on excessive imaging in the emergency department remains to be seen.