1. What decision rules do you use to determine if a patient is low or very low risk for PE?
The most commonly used decision instruments for risk stratification are the Wells’ score, Geneva score, and PERC rule. Additional tools include the Kline rule and the Pisa model, but these are less utilized.
The Wells score is a scoring system consisting of 7 criteria, to be used in conjunction with d-dimer. It assigns patients a risk of low, moderate, or high based on the total points. In low-risk patients, a negative d-dimer is sufficient to rule out PE without further testing. One of the initial validation studies found a negative predictive value of 99.5% for this group . One major weakness of Wells’ is that one of the highest weighted criteria, “an alternative diagnosis is less likely than PE,” is a very subjective measure; however, use of Wells’ has repeatedly been validated (Wolf, 2004).
The original Geneva score was a similar scoring system comprised of 8 variables that was well-validated but limited by its reliance on lab and imaging interpretation (blood gas and chest x-ray). Therefore it was revised to only include risk factors, signs, and symptoms. The Revised Geneva score (RGS) still consists of 8 variables, and like the Wells’ score, stratifies patients into low, moderate, or high risk. There is also a Simplified Revised Geneva score (sRGS) that assigns points uniformly for each variable rather than assigning different weights (Fesmire, 2011).
Both Wells’ and the sRGS can alternatively be subject to a dichotomous interpretation of PE likely or PE unlikely. Wells’ was used in this way by the Christopher Study Investigators when they demonstrated the safety of a simple algorithm involving d-dimer for patients deemed PE unlikely versus CT for patients deemed PE likely. Of those who were PE unlikely and had a negative d-dimer, 0.5% ultimately had a venous thromboembolism (VTE) at 3-months (Christopher, 2006).
Any discussion of decision rules warrants mention of clinical gestalt and how it compares to validated criteria in assessing the pretest probability of PE. The PIOPED study was the first to corroborate the use of gestalt. Since then, a number of comparative studies (Sanson, 2000; Runyon, 2005) have found that gestalt and objective criteria such as Wells’ perform equally well. This premise was also supported by a recent meta-analysis of 52 studies estimating sensitivity and specificity for gestalt and/or a decision rule. The results are summarized here:
|Decision Tool||Sensitivity (%)||Specificity (%)|
Among the 52 studies included there was significant heterogeneity of results, which was related to differing prevalences of PE (Lucassen, 2011).
The Pulmonary Embolism Rule-out Criteria (PERC) can be used to exclude the diagnosis of PE in patients who are deemed low risk. It is critically important to recognize that the clinician must first decide that the patient is low risk for PE based on clinical gestalt before applying PERC. This criteria was derived by Kline, et al., in 2004. Using a multi-center cohort of over 3,000 patients, the investigators collected data on 21 variables related to the diagnosis of PE. Then, by logistic regression and stepwise backwards elimination, they narrowed down the criteria to 8 variables that could be used in a block rule to justify not ordering a d-dimer if all questions are answered “no.” They internally validated the rule in a low-risk group and found a sensitivity of 96%. Of the 362 patients who were PERC negative, 5 (1.4%) were later diagnosed with VTE. This miss rate fell under the predetermined test threshold of 1.8%, a point of equipoise below which there is a greater chance of harm than benefit by proceeding with further testing.
Kline, et al., performed a large validation study in 2008, in which they sought to validate the premise that being both low-risk by gestalt (estimated risk <15%) and PERC negative renders patients very low-risk. This categorization would signify that they should have a pretest probability lower than the test threshold and therefore not benefit from further testing. Their study identified 1666 patients who were gestalt low-risk and PERC negative. In this group, 16 (1%) had VTE or death within 45 days. The sensitivity of this very low-risk classification was 97.4% (Kline, 2008).
When the authors applied PERC to the entire study population without gestalt interpretation, they found that the false negative rate would be unacceptably high if the population’s pretest probability was >6%. As most populations have a prevalence higher than this, the authors comment that this highlights the importance of risk stratification prior to application of PERC so that the pretest probability is sufficiently low.
PERC has been repeatedly validated. A 2012 systematic review and meta-analysis by Singh, which included 12 studies and nearly 15,000 patients in 6 countries, concluded that “because of the high sensitivity and low negative likelihood ratio, PERC rule can be used confidently in clinically low probability population settings” (Singh, 2012).
2. Are there any situations in which you modify the d-dimer threshold for ruling out PE? If so, when?
D-dimer testing has become standard practice for ruling out PE in low-risk patients due to its high sensitivity. But it is also known to lack specificity and lead to a high number of false positives. The result is an obligation to pursue the diagnosis of PE with CT pulmonary angiogram (CTPA), which poses a risk of contrast-induced nephropathy (CIN) and radiation-induced malignancy. Green and Yealy claim that 1 contrast-induced renal failure will result from every 200 CTs and that 1 radiation-induced cancer will result from every 2,000 CTs. They also assert that half of both types of adverse effects will ultimately be fatal (Green, 2012).
A wide range of data has been reported regarding the incidence rates of these sequelae, but regardless of how heavily one weighs the data, it is evident that unnecessary CTs do carry the potential for adverse consequences. In order to reduce harm, some have suggested that d-dimer not be regarded as a “one size fits all” test.
A 2009 study examined the use of d-dimer at different thresholds that varied according to pretest probability. The authors proposed doubling the threshold for low-risk patients and cutting the threshold in half for high-risk patients. Overall they found that doubling the threshold in low-risk patients increased d-dimer specificity from 58% to 75%, but decreased sensitivity from 94% to 88%. They estimated that this would result in 15% fewer imaging studies but 3-4 additional missed PEs per 1000 patients (Kabrhel, 2009). Their data did not support use of d-dimer at any threshold in high-risk patients.
Kline has also suggested doubling the d-dimer threshold in certain groups of patients in an effort to increase the test’s specificity. He and his colleagues recently published a study investigating the effect of doubling the threshold on exclusion and miss rates. This was a multicenter prospective trial. All enrolled patients had a pretest probability calculated using Wells’ or RGS, a d-dimer measured, and a confirmatory CTPA performed upon enrollment as well as at 30-day follow-up.
They found two groups in which doubling the d-dimer threshold yielded a significantly increased exclusion rate with only a slight increase in the rate of missed PE. The first group was those deemed to have a pre-test probability of “PE unlikely” as defined by Wells’ less than or equal to 4 or RGS less than or equal to 6. In these patients, the exclusion rate doubled from about 16% to 32% when the d-dimer threshold was doubled. The PE miss rate increased from about 3.7% to 5.4%. Almost all of these missed PEs were isolated, sub-segmental PEs and none of these patients had concomitant DVT. (The implication that sub-segmental PEs are of less clinical consequence than segmental PEs is still highly controversial).
The other group of patients for which d-dimer could safely be doubled was those over 70 years of age. D-dimer is known to increase with age and this phenomenon was seen when d-dimer was plotted against age for the 552 patients in this study without PE. The slope of the curve increases most sharply around age 70. When the investigators systematically examined 8 different d-dimer thresholds for 4 different age groups, they found that doubling the threshold after age 70 resulted in an increased rule out rate without an increase in post-test probability of PE.
Looking more closely at the implications of this change in d-dimer threshold, the authors found that out of the 224 patients who had a creatinine measured before and after CTPA, 37 developed CIN. Of these 37 cases, 10 could have been prevented by using the doubled d-dimer threshold to rule out PE (Kline, 2012).
The 2012 study did not provide any results about d-dimer in pregnancy, possibly due to the protocol’s nondiscriminatory use of CT, but pregnancy is known to increase d-dimer and confound interpretation. In a small study in 2005, Kline, et al., sought to define the magnitude of increase in d-dimer during pregnancy by employing women seeking to conceive and following them through pregnancy. They found that the percent of women with a d-dimer within normal limits was 79% pre-conception, 50% in the 1st trimester, 23% during the 2nd trimester, 0% during the 3rd trimester, and 69% 4 weeks post-partum. They concluded that normal pregnancy will cause an elevated d-dimer and that the threshold should therefore be adjusted according to trimester (Kline, 2005).
3. To which patients do you give thrombolytics for PE?
Compared to indications for thrombolytics in MI or ischemic stroke, indications remain less established and more controversial in PE. Quite recently, however, ACEP, the AHA and Chest have all published updated guidelines which are relatively in agreement. At this point it is fair to say that all patients with massive PE should receive thrombolytics (unless they have a contraindication or the risks outweigh the potential benefits), and that thrombolytics should be considered on a case-by-case basis in patients with sub-massive PE.
What do the terms massive and submassive PE mean? Massive PE was previously defined by anatomical criteria: >50% obstruction of pulmonary vasculature or occlusion of 2 or more lobar arteries. It is now more commonly defined by hemodynamic instability, which is a function of both embolus size and underlying cardiopulmonary status. As such, a sub-massive embolus in a patient with poor cardiopulmonary reserve could produce similar outcomes to a massive embolus in a patient without prior cardiopulmonary disease. Therefore, any combination of angiographic obstruction and cardiopulmonary function that causes hemodynamic decompensation qualifies as massive PE (Wood, 2002).
The AHA’s 2011 guidelines give more concrete definitions. Massive PE requires one of the following:
- sustained hypotension (systolic <90 for >15min or the need for inotropes)
- profound bradycardia
Sub-massive PE is characterized by normal BP but evidence of either:
- RV dysfunction (RV dilation, elevated BNP, suggestive EKG changes)
- myocardial necrosis (elevated troponin)
What is the evidence for thrombolytics? What we know about thrombolytics in PE is that at 24 hours they achieve reduced pulmonary artery pressures, increased pulmonary perfusion, and decreased RV dysfunction (Jaff, 2011). How this more rapid angiographic resolution of PE translates to overall outcomes is less certain. There is a paucity of high quality studies specifically looking at thrombolytics compared to heparin alone in patients with massive PE. A 2004 meta-analysis showed a 4% vs. 7% rate of recurrent PE and a 6% vs. 13% rate of death, respectively, for thrombolytics vs. heparin alone in massive PE (Wan, 2004). In cases of sub-massive PE, the jury is still out. No study has shown lower mortality in these patients who receive thrombolytics, but small studies, including one from the MOPETT trial, published in January 2013, have suggested that thrombolytics in some of these patients may decrease the risk of future CHF and pulmonary hypertension. Whether this is true, the effects on overall mortality, and the appropriate dose for sub-masive PE are still unclear.
However, the risks of thrombolytic therapy should not be underestimated. In the ICOPER registry, intracranial bleeding occurred in 3% of patients who received thrombolytics versus 0.3% of those who did not. The incidence of any major bleeding was 22% versus 9% in these groups (Goldhaber 1999). Absolute and relative contraindications to thrombolysis are the same as for MI. For patients in whom thrombolytics are contraindicated, or who remain hemodynamically unstable despite medical treatment, surgical or catheter embolectomy may be indicated.
Just as in MI and ischemic stroke, there is a “golden hour” in severe PE. In fatal cases of PE, two-thirds of patients die within 1 hour of presentation. Furthermore, thrombolytics are thought to be more effective in achieving reperfusion when administered early (Wood, 2002).
4. When, if ever, do you discharge a patient diagnosed with PE?
Standard treatment for PE has traditionally been inpatient-based and centered upon unfractionated or low molecular weight heparin (LMWH) bridged to a vitamin K antagonist. The down sides to this protocol include the resources and cost of hospital admission, frequent injections, multiple medications, and the need for extended lab monitoring. Therefore, there has been much research into whether outpatient treatment, and possibly even outpatient treatment with oral medication, is a viable option.
In 2010, one prospective and two retrospective studies published in the Journal of Thombosis and Haemostasis looked at outpatient treatment with LMWH for hemodynamically stable patients without significant comorbidities. All three found that outpatient treatment was safe and effective in this population (Agterof, 2010; Erkens, 2010; Kovacs, 2010).
In 2011, Aujesky, et al., conducted the first multicenter, prospective, randomized, non-inferiority trial comparing outpatient and inpatient treatment, both with LMWH. All enrolled patients were deemed to have a low risk of death (PESI class I or II; see below for more on this index). For outpatient and inpatient groups, respectively, rates of 90-day recurrent VTE were 0.6% and 0%; rates of 90-day major bleeding were 1.8% and 0%; rates of mortality were 0.6% in both groups. 14% of outpatients would have preferred to stay longer in the hospital. 29% of inpatients would have preferred early treatment at home. The conclusion made by the authors is that for select low-risk patients who prefer early discharge, outpatient treatment is a safe and effective option.
Are there any decision tools that can help determine which patients may be safe for outpatient treatment? The Aujesky study used the pulmonary embolism severity index, or PESI. The PESI is a prognostic model designed to risk stratify patients with acute PE according to estimated 30-day all cause mortality. It is the most accurate and easiest to use PE prognostic score. It consists of 11 variables that are given differing weight. The total score places patients into one of five severity categories, each with a discrete mortality risk. The more recently derived Simplified PESI (sPESI) consists of 6 variables. If all are negative, the patient is low-risk. If any are positive, the patient is high risk.
Erkens, et al., sought to determine if the original and sPESI could be used to accurately characterize which low-risk patients are appropriate for outpatient treatment. From their study they concluded that both models were accurate in identifying these patients; however, many high-risk patients were also treated as outpatients, without adverse outcomes. More research is clearly needed to further elucidate outpatient eligibility and outcomes (Erkens, 2012).
Troponin and brain natriuretic peptide (BNP) have also been investigated as markers of increased disease severity. The latter has shown a positive correlation, but lacks specificity. Church and Tichauer, in their review of ED management of PE, recommend using troponin, in conjunction with sPESI and focused echocardiography, to further risk stratify patients with PE (Church, 2012).