What term is used for the ability of a test or technique to detect a disease when present?

Sensitivity, Specificity, and Predictive Values

It is critical to understand the characteristics of both screening and diagnostic tests. Sensitivity and specificity are characteristics inherent in the test and are independent of the prevalence of the disease.58,61 Sensitivity is the probability, expressed as a percentage, that if the disease is present, the test is positive. The numerator is the number of patients with the disease who have a positive test, and the denominator is the total number of patients with the disease tested. Specificity is the probability, expressed as a percentage, that if the disease is absent, the test is negative. The numerator is the number of subjects without disease who have a negative test, and the denominator is the total number of subjects without disease tested.

Although the sensitivity and specificity of a test are important considerations when deciding whether or not to order a test, we become more interested in the predictive values when the test results have returned. Predictive values are also much more intuitive to patients. Predictive values, in contrast to sensitivity and specificity, depend on the prevalence of the outcome in the population tested.

A positive predictive value (PPV) is the probability that if the test is positive, the subject has the disease. The numerator is the number of subjects with the disease who have a positive test, and the denominator is the total number of subjects with a positive test. A negative predictive value (NPV) is the probability that if the test is negative, the subject does not have the disease. The numerator is the number of subjects without disease who have a negative test, and the denominator is the total number of subjects with negative tests. Given the same sensitivity and specificity, the PPV will increase and the NPV will decrease as the prevalence increases. Likewise, as the prevalence decreases, the PPV decreases and the NPV increases.

These abstract concepts are best demonstrated with a clinical example. Peaceman and colleagues performed a prospective cohort study at multiple centers to assess whether cervicovaginal fetal fibronectin could be used as a diagnostic test in women with symptoms of preterm labor62; fetal fibronectin has also been assessed in other studies as a screening test.59 In the Peaceman study, women with symptoms of early preterm labor were enrolled, and cervicovaginal swabs for fibronectin testing were obtained. Treating physicians and patients were blinded to the results of the fibronectin test, a strength of the study. The outcomes assessed were the occurrence of delivery within 7 days, within 2 weeks, and before 37 weeks' gestation. The results of the analysis of delivery within 7 days (Table 17.7) may be used as an example to illustrate sensitivity, specificity, PPV, and NPV.

Some would look at these results and the high NPV and suggest that fetal fibronectin testing is a useful tool in this setting to rule out an imminent delivery. Another way of looking at these same data would be to look closely at the low prevalence of delivery within 7 days (3%). After reading this article, the following questions emerge: Is it appropriate to use a diagnostic test in such a low-prevalence group? More importantly, what would be the impact of testing a higher-prevalence population (i.e., a population with a greater chance of preterm birth within 7 days)?

Definition

A test is useful only to the extent that clinicians can understand the implications of a positive or negative result. Few, if any, tests always correctly identify the presence or absence of disease in all patients. Clinicians can decide what to do with a “positive” test result only when they have a clear knowledge of the test's characteristics and its statistical predictive value when applied to a specific patient population. Such “medical decision analysis” directly affects the clinical care of patients.

Some tests provide a qualitative positive or negative result. Many test results, however, are quantitative and define a range of “normal” values around a mean. Therefore, a few members of any population will have an “abnormal” test result but not actually have a disease. This means that a test result may be misleading owing to variability in the patient population.

The test itself can give an incorrect result due to (1) inaccuracy, (2) imprecision, or (3) incorrect performance. The accuracy of a test is the difference between the mean value of test results and the true result, as measured by a gold standard test. The precision of a test is the reproducibility of results between instruments or persons performing the test. An incorrectly performed test can invalidate any result.

Test accuracy can be described in several ways. The sensitivity of a test measures the proportion of individuals who have a disease and are correctly identified as being positive for that disease, based on the test. Specificity measures the proportion of individuals who do not have a disease and are identified as being disease free, based on the test. False-positive results are more likely with tests that have a high sensitivity, low specificity, or both. Sensitivity and specificity are characteristics of the test and do not change with the prevalence (frequency) of disease in the population. Said another way: a test's sensitivity and specificity do not affect the probability of a patient having a disease.

The predictive value of tests, in contrast, depends on the prevalence of a disease in a population of patients. The predictive value of a positive test indicates the proportion of those with a positive test who actually have the disease. Often, the predictive value of tests is expressed as the probability, or odds, that a condition is present. Likelihood ratios express the amount that the odds change when the results of the test are available (Table 36-1). In this respect, an important concept is Bayes' theorem, which “relates the probability of an item (e.g., a patient) being a member of a particular group (e.g., clinical class), given the presence of an attribute (e.g., an abnormal test result), to the probability of known group members having the attribute and the probability of obtaining a group member when picking at random an item from the universe of items.”1 It allows the calculation of changes in the probability of disease as new information (e.g., test results) becomes available. The post-test probability is calculated with the Fagan nomogram, similar to that shown in Figure 36-1. A Web-based interactive nomogram can be accessed at http://www.cebm.net/nomogram.asp.

Predictive Value of Laboratory Tests

Predictive value (PV) theory deals with the usefulness of tests as defined by their clinicalsensitivity (ability to detect a disease) andspecificity (ability to define the absence of a disease).

Sensitivity=Number positive by testTotal number positive×100

Specificity=Number negative by testTotal number without disease×100

PV of a positive test result=True-positive resultsTotal positive results×100

PV of a negative test result=True-negative resultsTotal negative results×100

The problems addressed by PV theory arefalse-negative andfalse-positive test results. Both are major considerations in interpreting the results of screening tests in general and neonatal screening tests in particular.

Testing for human immunodeficiency virus (HIV) seroreactivity illustrates some of these considerations. If it is assumed that approximately 1,100,000 of 284,000,000 residents of the United States are infected with HIV (prevalence = 0.39%) and that 90% of those infected demonstrate antibodies to HIV, then we can consider the usefulness of a simple test with 99% sensitivity and 99.5% specificity (seeChapter 302). If the entire population of the United States were screened, it would be possible to identify most of those infected with HIV:

1,100,000× 0.9×0.99=980,100(89.1%)

Interpretation of specific IgE/skin tests

The sensitivity and specificity of diagnostic cut-points for clinical relevance for either skin testing and/or serum allergen-specific IgE will depend on a number of factors. These include the quality of extracts used and the experience of the operator (or laboratory). Geographic location and variation of the prevalence of environmental allergens will have an enormous impact. Similarly, the predictive value of tests will be influenced by the prevalence of the suspected allergy in the population studied. For example the prevalence of ‘true’ allergy in relation to positive IgE tests is likely to be higher in referrals to a specialist allergy clinic compared with those screened in primary care.

Whereas it is true that the negative predictive value of skin tests and specific IgE is far more robust, the phenomenon of local IgE synthesis and expression has been increasingly recognized. Again this illustrates the importance of re-evaluation of the history in the interpretation of discordant IgE/skin tests, whether positive or negative. The presence of a negative test in the face of a strong positive history on re-evaluation is an indication for provocation testing in the target organ, whether food challenge or nasal or bronchial inhalation challenge.

Sensitivity, Specificity, and Positive and Negative Predictive Value

The validity of a screening or diagnostic test (or one used for prediction) is initially measured by its ability to categorize individuals who have preclinical disease correctly as “test positive” and those without preclinical disease as “test negative.” A simple two-by-two table is typically used to summarize the results of a screening test by dividing those screened into four distinct groups (Table 9.2). In this context, sensitivity and specificity provide fundamental measures of the test's clinical validity.Sensitivity is the probability of testing positive when the disease is truly present and is defined mathematically as a/(a + c). As sensitivity increases, the number of individuals with disease who are missed by the test decreases, so a test with perfect sensitivity will detect all individuals with disease correctly. In practice, tests with ever-higher sensitivity tend to also classify as “diseased” many individuals who are not actually affected (false positives). Thus thespecificity of a test is the probability of screening negative if the disease is truly absent and is defined mathematically as d/(b + d). A test with high specificity will rarely be positive when disease is absent and will therefore lead to a lower proportion of individuals without disease being incorrectly classified as test positive (false positives). A simple way to remember these differences is that sensitivity is “positive in disease” whereas specificity is “negative in health.”

A perfect test has both very high sensitivity and specificity and thus low false-positive and false-negative classifications. Such test characteristics are rare, however, because there is a trade-off between sensitivity and specificity for almost every screening biomarker, diagnostic, or predictive test in common clinical use. For example, although high LDL-C levels usually serve as a biomarker for atherosclerotic risk, up to half of all incident cardiovascular events occur in those with LDL-C levels well within the normal range, and many events occur even when levels are low. If the diagnostic cutoff criterion for LDL-C is reduced so that more people who actually have high risk for disease will be test positive (i.e., increase sensitivity), an immediate consequence of this change will be an increase in the number of people without disease in whom the diagnosis is made incorrectly (i.e., reduced specificity). Conversely, if the criterion for diagnosis or prediction is made more stringent, a greater proportion of those who test negative will actually not have the disease (i.e., improved specificity), but a larger proportion of true cases will be missed (i.e., reduced sensitivity).

In addition to sensitivity and specificity, the performance or yield of a screening, diagnostic, or predictive test also varies depending on the characteristics of the population being evaluated. Positive and negative predictive values are terms used in epidemiology that refer to measurement of whether an individual actually has (or does not have) a disease, contingent on the result of the screening test itself.

Predictive Value of a Test Result

When one is testing mixed populations that contain subjects with and without coronary artery disease, the concept most often employed is the “predictive value” of a positive or negative test. The predictive value of a positive test is the proportional likelihood of the disease being present after a positive test result is found in a given individual. For instance, a predictive value of 0.7 (or 70 percent) would simply mean that a given subject who displays a positive test result would have a 70 percent likelihood of having disease. Conversely, the predictive value of a negative test expresses the likelihood that a subject is free of disease after a test is found to be negative. In contrast to sensitivity and specificity, which are derived from homogenous populations with or without disease, predictive value is strongly influenced by disease prevalence (pre-test probability) within the population to be tested, as well as by the sensitivity and specificity of the test itself. The relationship among these factors is expressed by the following formulas:

Predictive value of a positive test(PVPT):

PVPT=(Sensitivity×Probability of disease)(Sensitivity×Probability of disease)+{(1−Specificity)×(1−Probability of diseases)}

Where Probability of disease represents the proportion of the pre-test population harboring the disease.

From this formula, one sees that the predictive value is affected by both the sensitivity/specificity of the test itself and the pre-test probability. It is, however, most strongly influenced by the values in the latter part of the denominator (i.e., 1 - Specificity and 1 - Probability of disease in the pre-test population). The higher the values of specificity and prior probability of disease, the lower are these two numbers (and their product), and the closer the predictive value approaches 1 (or 100 percent), the highest value obtainable. Thus, both high test specificity and/or the pre-test disease prevalence play equal roles in allowing a given test—when positive—to predict with greater likelihood the presence of disease.

Predictive value of a negative test (PVNT):

PVNT=Specificity×(1−Probability of disease){Specificity×(1−Probability of disease)}{(1−Senstivity)×(Probability of diseases)}

In this instance, one sees that the PVNT is most strongly influenced by the test sensitivity as well as the pretest prevalence, i.e. the higher the sensitivity and the lower the pretest disease prevalence, the greater is the resulting PVNT value, therefore the greater is the opportunity to exclude the likelihood of disease, given a negative test result.

These concepts are frequently referred to as Bayes' theorem, which expresses the idea that the post-test likelihood of disease depends upon the qualities of both the test itself as well as the composition of the population subjected to this test. This is useful in understanding the meaning of test results in general, as well as in test selection. For instance, if one wishes to select the proper test to provide the greatest chance of excluding a disease (high predictive value of a negative test), one would best select a test with the highest possible sensitivity and apply it to a population with the lowest possible prevalence of disease. On the other hand, if one wishes to select a test to provide the greatest chance of positively identifying a disease while minimizing the number of false-positive responders (high predictive value of a positive test), one would select a test with the highest possible specificity and apply it to a population that harbors a high initial prevalence of the disease in question.

Diagnosis

Accurate diagnosis requires recognition of the clinical syndrome, a reasonable epidemiologic likelihood of exposure, and laboratory confirmation. Although much has been said in the past about the inaccuracy of testing for Lyme disease, it is comparable, overall, with other serologic testing. There remain technical arguments about the best antigen and methodology to use, but it is likely that more of the difficulty arises from misinterpretation of test results than from inaccurate ones.

Laboratory diagnosis relies primarily on serologic diagnosis—the demonstration of antibody in peripheral blood that adheres to B. burgdorferi. The sensitivity of microbiologic culture remains low (except in erythema migrans, where it is unnecessary) and probably reflects the small number of spirochetes present in readily accessible clinical specimens. The polymerase chain reaction assay has remained problematic, probably for the same reason (in addition to contamination-related issues leading to false-positive results). Antigen detection methods have not been found to be reproducible. Immune complexes containing Borrelia antigens have been described, but do not appear to be a reliable independent marker of infection. Hence, most laboratories rely on ELISAs to quantitate immunoreactivity.

Several particular problems confront serologic testing for Lyme disease. First, the disease is focally endemic. In highly endemic areas, such as areas of Long Island, 10 percent or more of the population may have been exposed. In contrast, in nearby New York City, there were about as many cases of malaria in 2010 as of confirmed Lyme disease. This highly varying background prevalence has a major effect on the positive and negative predictive values of test results, which depend on the background prevalence. In areas with Lyme disease prevalence of 1 in 100,000 and a statistically defined negative cut-off of 3 standard deviations, 1 sample in 1,000 will be a false positive, but only 1 in 100,000 will be a true positive. Second, as with any antibody-based testing, it must be remembered that it takes several weeks or more to develop a detectable antibody response in blood. Thus, negative results are common at the time of the erythema migrans, but, as with the syphilitic chancre, the rash by itself is sufficiently diagnostic to mandate immediate antibiotic treatment. Third, unlike other serologic tests, most clinicians rely on a single Lyme disease ELISA value rather than comparing acute and convalescent titers. This may be reasonable in syphilis, where screening reaginic tests are typically positive only during active infection, and is understandable insofar as there is a desire to treat Lyme infection as soon as possible. However, it makes test interpretation difficult in patients who have had prior infection and have a persistently elevated titer. As in most infections, the immune system usually continues to generate detectable antibodies for an extended period after the infection has resolved, sometimes making the relevance of the positive serology to current symptoms difficult to judge.

Finally, as with most serologic tests, some cross-reactions occur. To address this a two-step test procedure is used, but unfortunately this too has produced confusion. Western blots are routinely used to confirm positive or borderline ELISAs. The criteria for interpreting these (Table 39-3) were defined in individuals with positive or borderline ELISA results; therefore, they can only be applied in individuals with negative ELISA results with extreme caution. Second, the criteria were derived statistically. Interpretation is based not on the uniqueness of any of the bands but on the statistical probability of accurate diagnosis. Importantly, just as in syphilis, testing algorithms are evolving. While the standard approach for over two decades has been to screen with an ELISA and confirm positive or borderline results with the more specific Western blot, the CDC now also endorses an alternative approach, testing with two independent ELISAs (i.e., testing for reactivity to different antigens, much like the European algorithm for syphilis testing) provides better test accuracy and eliminates much of the confusion associated with Western blot interpretation.10

Table 39-3. Western Blot Criteria for Confirmation of Positive Serology in Lyme Disease

As in syphilis, spinal fluid studies can be extremely useful, but they also are the source of considerable confusion. Most, if not all, patients with active CNS infection will have elevations in the CSF white blood cell count, protein concentration, or both. Chronic infection is often accompanied by increased local production of antibodies, resulting in an increased relative CSF IgG concentration and even oligoclonal bands. In most acute cases (about 90%) and in many patients with more longstanding CNS Lyme disease (at least 50% and possibly more), synthesis of specific anti–B. burgdorferi antibodies in the CNS can be demonstrated by appropriately comparing concentrations of specific antibody in CSF and serum. Simple measurement of antibody concentration in the CSF alone can be very misleading in individuals with blood–brain barrier breakdown, CNS inflammation from other causes, or positive peripheral blood Lyme disease serologies, as discussed previously in considering CSF serologies in syphilis. Mounting evidence suggests that measurement of the B cell–attracting cytokine, CXCL13, in CSF may be useful in clarifying the diagnosis.11,12

COMPARATIVE VALUE OF PRICK AND INTRADERMAL TESTS

The value of routine intradermal skin tests in asthmatic individuals is controversial. Skin tests by any technique are not simple “yes/no” predictors for the occurrence of clinically significant reactions when the lower airway (or any other tissue) is exposed to the allergen. Skin tests obviously challenge mast cells in the skin rather than mast cells in any other particular organ. Not surprisingly, the predictive value of tests in the skin is variable, and in some settings quite poor. Determination of the sensitivity, specificity, and predictive values for skin tests requires a gold standard test for comparison (and for predictive values, knowledge also of the disease prevalence). In the setting of food allergy, the gold standard is the placebo-controlled, double-blind, oral food challenge, a practical test that is reflective of real world exposures associated with disease. For nasal allergy, challenge by nasal instillation of pollen or pollen extract can be used, but such a nasal challenge is limited by uncertainty over the amount of allergen that would reflect real world exposure and the time frame over which that exposure should occur. For asthma, the analogous test is bronchial challenge with a nebulized solution of the allergen. Even more so than with nasal challenge, there is uncertainty as to the maximum amount of allergen that should be inhaled and tolerated before the allergen is judged irrelevant to the patient's disease.

Other challenges to the airway have been used. Some investigators have simultaneously exposed the upper and lower airways either in an exposure chamber or with a facial mask covering the nose, mouth, and possibly the eyes. These methods have potential to better replicate typical routes and levels of exposure but still may only be useful for determining immediate response to relatively large amounts of allergen, as opposed to ongoing low-level exposure. It is unclear whether low-level, chronic exposures to aeroallergen may be involved in the maintenance of airway inflammation in asthma. A further difficulty of all the challenge techniques (except food challenge) is that they are limited to investigational settings. The multiple issues cited above prevent investigators from assuming that a bronchial allergen challenge is the gold standard test with which to compare skin tests.

The models of food allergy and nasal challenge may offer some guidance in understanding the predictive value of skin tests for lower airway symptoms. For food allergy there is good evidence that the prick skin test is clinically useful to discriminate patients at high versus low risk of reaction when the tested food is eaten. Intradermal testing in those who are prick-test negative does not identify patients who will react on food challenge and is thus not indicated.4 A study of nasal challenge with timothy grass pollen has discriminated between the value of prick and intradermal tests. Patients with symptoms of seasonal rhinitis and a positive prick test reacted to relatively small amounts of intranasally administered pollen.5 Among subjects with a history of seasonal rhinitis but a negative prick skin test there was no difference in reactivity on nasal challenge between those who were positive or negative on an intradermal skin test. Both groups required far more pollen to induce nasal symptoms on challenge than did the prick-positive group (Fig. 10-5). Subjects with no history of nasal symptoms and no positive skin tests tolerated more pollen than the other three groups. Intradermal testing did not demonstrate a group of patients with increased nasal sensitivity that had been negative on prick testing; therefore, for this pollen, there was no additional information obtained by intradermal skin testing. The study also suggests that there are people with nonallergic nasal symptoms who have nonspecific (possibly irritant) reactions to pollen instillation at high doses of pollen. Nasal sensitivity to irritant effects could result from underlying inflammation of a nonallergic nature (such as in the syndrome of nonallergic rhinitis with eosinophilia). Another study of allergic rhinitis used nasal provocation to identify allergic and nonallergic patients before skin tests. There was no improvement in recognition of true positives when carefully performed intradermal end-point titration testing was compared to simple prick testing.6

Data for bronchial allergen challenge is more complicated. All the data regarding bronchoprovocation are limited by the uncertainty of the significance when increasing amounts of allergen are administered. Some authors have attempted to control for this issue by determining doses of inhaled allergen that are tolerated by nonallergic, nonasthmatic subjects.7 With this caveat, several observations on bronchoprovocation and skin tests are well supported.

The probability of bronchial sensitivity (by bronchoprovocation) increases with increasing size of the wheal produced by prick testing. The probability of bronchial sensitivity varies inversely with the concentration of antigen required to induce a positive intradermal skin test (Table 10-1). Prick skin tests with a wheal greater than 9 mm were associated with a 50% probability of positive bronchoprovocation. Intradermal skin tests at a 1:1000 dilution yielded 27% positive on bronchoprovocation, increasing to 61% positive among those with a positive test at 1:10000 dilution, a rate similar to prick test positives with a large wheal.

Given the similar predictive values of very dilute intradermal tests and strongly positive prick tests, it would be reasonable to prefer the simpler prick test unless the intradermal test identified challenge-positive individuals who would have been labeled as negative by the prick test. As seen above, such patients do not appear to be identified in studies of food allergy or rhinitis. The published bronchoprovocation studies do not adequately address whether intradermal tests identify such patients; however, a strong case against routine intradermal testing comes from a study using a cat allergen exposure chamber.8 Subjects in the study were exposed to cats in a small room that housed two cats. Subjects remained in the exposure chamber for up to several hours while nasal and eye symptoms, pulmonary symptoms, and lung function were monitored. No subjects who were prick-negative but intradermal-positive developed significant symptoms or a decrease in forced expiratory volume in 1 second (FEV1). The authors conclude that the intradermal test added no information about clinical cat sensitivity.

The data above lead to the conclusion that intradermal testing for aeroallergens in asthmatic individuals is not routinely necessary. The value of intradermal testing to aeroallergens remains hypothetical at best, since a group of subjects with provable allergic disease who are identified only by this technique has not been found. For a patient with a very suggestive history for asthma symptoms related to a particular allergen, but a negative prick skin test, intradermal testing might be justified, but only at a relatively dilute concentration of the specific allergen in question.

Level C recommendations

In adult patients with suspected nontraumatic thoracic aortic dissection, do not rely on D-dimer alone to exclude the diagnosis of aortic dissection.

Key words/phrases for literature searches: thoracic aortic dissection, dissecting, aneurysm, D-dimer, diagnosis, predictive value of tests, sensitivity and specificity, likelihood functions, and variations and combinations of the key words/phrases.

Study Selection: Eighty-two articles were identified in the search. Twenty-four articles were selected from the search results for further review. One additional article was identified and added at the review stage, with 11 studies included for this critical question recommendation.

Traditionally, the diagnosis of acute nontraumatic thoracic aortic dissection has been based on diagnostic imaging. The use of a laboratory test to exclude the diagnosis of acute thoracic aortic dissection, similar to the use of D-dimer for ruling out acute pulmonary embolism, is appealing and could potentially save time and money.

Eleven Class III studies, including 2 meta-analyses, have evaluated the performance of D-dimer in diagnosing acute thoracic aortic dissection.19-29 These studies suffer from selection bias and vary widely in the assays used to measure D-dimer. Even though the cutoff value for a positive test result, as well as the type of assays used to measure D-dimer values, varied in the studies reviewed, D-dimer was highly sensitive for diagnosing acute thoracic aortic dissection, with sensitivities ranging from 91% to 100%. However, given the low quality of these Class III studies, strong recommendations about the routine use of D-dimer testing alone cannot be made. One Class III article evaluated the diagnostic accuracy of a negative D-dimer test result in conjunction with a risk-stratification score of 0.29 In those patients, none had an aortic dissection. In nonhigh-risk patients, the LR- was 0.04 (95% CI 0.01 to 0.15).29 This approach, however, needs prospective validation because of methodologic limitations of this study.

The following conditions, however, may result in a low or false-negative D-dimer value in patients with proven thoracic aortic dissection: chronicity, time from symptom onset, presence of thrombosis or intramural hematoma, short length of dissection, and young age of patient. Eggebrecht et al20 found a significant negative correlation between the absolute D-dimer values and time from onset of symptoms. D-dimer levels were higher in patients with acute versus chronic thoracic aortic dissections.20 Eggebrecht et al20 also noted that D-dimer levels were higher in patients with thoracic aortic dissection who died early, underwent emergency endovascular or surgical procedure, or had complications. Thrombosed false lumens or intramural hematomas may affect D-dimer levels. In multiple studies, D-dimer levels were lower in patients with thoracic aortic dissection and a thrombosed false lumen than in patients without a thrombosed false lumen.22,25,28 Hazui et al,23 in a 2006 study, found that patients with thoracic aortic dissection who were younger or had short dissection lengths and thrombosed false lumens without ulcerlike projections may have false-negative D-dimer results. Ohlmann et al25 identified 1 of 94 patients with a false-negative D-dimer test with a localized intramural hematoma without an intimal flap.

If a patient has a positive D-dimer result, the diagnosis of thoracic aortic dissection cannot be made definitively without imaging. D-dimer elevations are not specific for thoracic aortic dissection. Elevated D-dimer measurements can be found in patients presenting to the ED with many conditions, including but not limited to acute thoracic aortic dissection, pulmonary embolism, acute myocardial infarction, and inflammatory conditions. Based on the clinical presentation, a positive D-dimer result may prompt the physician to order an imaging study to further investigate the diagnosis. Sakamoto et al30 reported a sensitivity of diagnosing acute thoracic aortic dissection of 68.4%. According to Sakamoto et al,30 D-dimer levels were higher in patients with acute thoracic aortic dissection and pulmonary embolism compared with levels in patients with acute myocardial infarction. D-dimer was not able to reliably differentiate an acute thoracic aortic dissection from a pulmonary embolism with D-dimer values of 32.9 μg/mL (SD 66.7 μg/mL) for acute thoracic dissection and 28.5 μg/mL (SD 23.6 μg/mL) for pulmonary embolism. Because D-dimer is nonspecific, routinely obtaining this test in a large population of patients with symptoms suspicious for aortic dissection can result in harm, most notably, exposure to radiation and cost associated with advanced imaging.

Future Research

A prospective study evaluating D-dimer levels on undifferentiated ED patients who present with signs and symptoms concerning for thoracic aortic dissection is warranted. Studies clarifying the best way to integrate D-dimer testing into clinical algorithms that include risk stratification are needed.

3. In adult patients with suspected acute nontraumatic thoracic aortic dissection, is the diagnostic accuracy of CTA at least equivalent to TEE or MRA to exclude the diagnosis of thoracic aortic dissection?