Published in the January 2007 issue of Today’s Hospitalist.
Consider this scenario: A 50-year-old woman with no past medical history arrives in the emergency room with pleuritic chest pain and shortness of breath. Her oxygen saturation is 95%, her heart rate is 105 beats per minute and her physical exam shows nothing remarkable. Her chest X-ray, EKG and other vital signs appear to be normal, while a troponin test rules out heart attack.
And a negative CT angiogram seems to rule out another big threat: pulmonary embolism (PE). Do you need to order more tests to be sure you don’t have to start her on anticoagulation?
That was the question posed by hospitalist Tracy Minichiello, MD, during her update on venous thromboembolism at the Fall 2006 Hospitalist CME Series meeting in Tampa, Fla. Judging by the show of hands at the presentation-some indicating "yes" to ordering more tests, but just as many "no’s" and "maybe’s"-hospitalists are still unclear about what tests they need to comfortably rule out PE.
"It is generally agreed that CT scans have decreased sensitivity for peripheral pulmonary emboli," said Dr. Minichiello, who is director of the anticoagulation clinic at University of California, San Francisco, "so there is some debate about the need for additional testing to rule out PE in patients with negative CT scans."
But instead of piling on more tests for all patients with suspected PE, she urged hospitalists to take a step-by-step, evidence-based approach, starting with calculating patients’ pre-test probability of the disease. She also highlighted validated risk scoring systems that can help physicians make that calculation and determine the other factors they need to make the right call.
A question of probability
Dr. Minichiello briefly reviewed findings from a meta-analysis published in the April 27, 2005, issue of Journal of the American Medical Association. That study looked at the clinical validity of negative CT results for patients with suspected PE.
Researchers included studies that used CT scanning to rule out PE and that had at least three months of follow-up data available to assess recurrence rates. Based on those data, they found that CT’s negative predictive value was 99%.
"That’s an incredibly high value," said Dr. Minichiello, pointing out that the resulting negative likelihood ratio was a "very low" 0.07. "These results would lead you to believe that CT scanning was enough in all cases."
Looking more closely at the analysis, however, Dr. Minichiello noted that many of the participating patients had a lower pre-test probability-a key factor in calculating patients’ post-test PE probability. (Post-test probability equals pre-test probability times the likelihood ratio.)
That means that these results can be applied only to patients with similar pre-test probability. How do you make that essential calculation?
A better way to score risk
While physicians have traditionally relied on the "gestalt" approach to pre-test probability, many are now applying widely validated PE risk-scoring systems such as the Wells criteria.
The Wells criteria assign "risk" points to patients for certain clinical features, including deep-vein thrombosis symptoms, tachycardia, hemoptysis and cancer, among others. Patients’ total points correlate to a low, moderate or high pre-test PE probability of 7%, 30% or 75% risk-of-disease prevalence.
But one challenge with the Wells criteria is the category of "other diagnosis less likely" than PE, said Dr. Minichiello. "This can be a grey area."
She reviewed another risk scoring system that was published in the Feb. 7, 2006, issue of Annals of Internal Medicine: the revised Geneva risk score, which has a similar post-test prevalence as the Wells criteria.
It has "eliminated this grey area," she said, by assigning points only for history, signs and symptoms. (See "Calculating pre-test risk: the revised Geneva scoring system," left.)
Another piece of the puzzle
Returning to the patient in the ER, say you score her risk, find out her pre-test probability is low, then use the negative likelihood ratio to calculate that she has a post-test PE probability of only 1%.
"Everybody would be comfortable holding back on anticoagulation," said Dr. Minichiello.
But what if the patient’s pre-test probability is in the higher end of the intermediate range? Using post-test probability calculators to demonstrate this point, Dr Minichiello showed that the patient’s post-test probability would then be too high to withhold anticoagulation based simply on a negative CT scan.
"This demonstrates the importance of knowing the negative likelihood ratio or negative predictive value for all the tests we use," she said. "Only then can we accurately interpret test results in the context of pre-test probability."
Fortunately, Dr. Minichiello pointed out, hospitalists have established evidence on likelihood ratios that they can turn to. In a review published in the July 30, 2005, British Medical Journal (BMJ), researchers calculated the likelihood ratios of different tests commonly used to diagnose PE. They determined the range of pre-test probabilities over which each test could confirm or exclude PE.
What were key findings? Researchers identified the two imaging modalities that physicians can rely on for intermediate risk patients. "CT plus ultrasound and normal V/Q [ventilation perfusion scan] appear to be the most helpful to rule out PE," Dr. Minichiello explained.
Another very useful component of the review: Researchers calculated negative likelihood ratios associated with different types of D-dimer tests. "Not all D-dimers are created equal," Dr. Minichiello warned. According to the BMJ study, the qualitative ELISA is the most sensitive and the best test for ruling out disease.
Putting it all together
What does all this mean for your ER patient? "If you decide she has a low pre-test probability, you could use a CT scan, a D-dimer assay or a low probability V/Q scan to rule out disease," Dr. Minichiello said.
But when you get into intermediate probability, "you need to be more discerning," she pointed out. To rule out disease using CT, you’d have to add ultrasound. Or you could use a qualitative ELISA D-dimer or a V/Q scan that is normal or near normal, she added.
And for patients with a high pre-test probability? "Ultimately," said Dr. Minichiello, "the recommendations are to go forward with pulmonary angiography."
COPD: the need for more suspicion
Dr. Minichiello also encouraged hospitalists to bring a heightened suspicion of pulmonary emboli to patients with COPD.
All too often when treating COPD patients, doctors attribute shortness of breath to just another unexplained exacerbation, she said. Yet studies have found that post-mortem incidence of PE in COPD patients is between 25% and 51%.
"In many of these patients," Dr. Minichiello grimly noted, "the diagnosis wasn’t made while they were alive."
We know that diagnostic tests perform equally well in COPD patients, she continued. "What’s missing is our ability to assess their risk and our threshold for pursuing a work-up."
A study published in the March 21, 2006, issue of Annals of Internal Medicine recruited nearly 200 COPD patients with unexplained exacerbations defined as shortness of breath without an increase in sputum, purulence, fever or urinary infection. While all received both ultrasound and CT, nearly 25% were found to have PE.
Even more troubling was the fact that researchers did not find that common indications of PE-like chest pain, hemoptysis, tachycardia, lower extremity edema, hypoxemia, trauma or surgery-were linked to PE. "There goes your risk-scoring system," she said.
Instead, PE was most strongly associated with previous thromboembolic disease, cancer and a drop in carbon dioxide of at least 5 mm Hg from baseline.
The bottom line? "All these other prediction rules may be less reliable in COPD patients," she concluded. "We need to maintain a very low threshold of suspicion for PE in these patients."
Making the case for evidence
Finally, Dr. Minichiello reviewed study findings that can help guide hospitalists’ quality improvement efforts and diagnostic approach: the value of using a written diagnostic algorithm to score patients’ clinical probability of PE.
The researchers in the Feb. 7, 2006, Annals study looked at three-month outcomes after ER-based evaluation of patients for suspected PE. They wanted to find out whether ER providers were applying appropriate evidence-based criteria to both rule in and rule out the disease. They also wanted to identify predictors for adherence to evidence-based diagnostic algorithms and see if the use of appropriate algorithms was associated with patient outcome.
Interestingly, they found that physicians use appropriate algorithms 92% of the time when they ruled in disease, "but only 40% of the time when they ruled it out," said Dr. Minichiello. The different use of algorithms had a big impact on outcomes: Patients who received appropriate care based on a diagnostic algorithm had a 1% recurrence rate, while 7% of those who did not had a recurrence. That’s a six-fold increase.
Researchers also identified several risk factors associated with the use of inappropriate criteria to rule out PE. A significant one was pre-existing congestive heart failure or COPD, with physicians probably attributing symptoms to those disorders and not pursuing further testing, said Dr. Minichiello. Other factors were advanced age and pregnancy, which often reflect concerns about contrast or radiation exposure.
But another very strong predictor of inappropriate management, she said, was the lack of a written diagnostic algorithm and of clinical probability scoring in the ER.
The findings are a stark reminder of the need to take an evidence-based approach. "You can influence outcomes on an individual level," Dr. Minichiello said, "and on a system level as well."
Michael S. Krivda is a freelance writer specializing in health care. He is based in Perkasie, Pa.
Low-molecular-weight vs. unfractionated heparin: which one is better?
Most physicians are comfortable with the efficacy of low-molecular-weight heparin to treat deep-vein thrombosis (DVT).
"For the most part, this disease is treated in the outpatient arena," said hospitalist Tracy Minichiello, MD, during her update on venous thromboembolism at the Fall 2006 Hospitalist CME Series meeting in Tampa, Fla.
"Low-molecular-weight heparin is easier to administer, does not require monitoring and therefore is ideal for outpatient therapy," she said. She did note, however, that many outpatients can’t afford low-molecular-weight heparin (LMWH) and have to resort to unfractionated heparin (UFH) with all the hassles of monitoring.
Several trials have compared LMWH to UFH. The studies have found that the drugs are clinically equivalent in treating DVT and non-massive PE and in preventing recurrence for up to three months, said Dr. Minichiello, who is director of the anticoagulation clinic at University of California, San Francisco.
Hospitalists still run into many situations where UFH is the anticoagulant of choice. "In patients at high risk for bleeding, in whom you want to be able to easily reverse anticoagulant effect, UFH is the way to go," she said, pointing to UFH’s shorter half-life and the ability to reverse it 100% with protamine.
Other inpatients routinely treated with UFH include those with renal insufficiency and those at both the high and low extremes of weight. Still, at discharge and in the outpatient arena, LMWH seems to be the clear winner.
But a study that Dr. Minichiello highlighted may change all that. Published in the Aug. 23/30, 2006, issue of Journal of the American Medical Association, findings were based on patients with acute DVT or pulmonary embolism who were randomized to either usual care with LMWH bridging to warfarin or subcutaneous, weight-based and dose-adjusted unfractionated heparin-without PTT monitoring to adjust UFH dosing. Researchers found no difference in thrombosis recurrence at 10 days or at three months, and no difference in bleeding.
"These are surprising results for those of us who are accustomed to using PTT to guide dosing of full-intensity unfractionated heparin," Dr. Minichiello said.
And results could signal a potential change in therapeutic approach. "The current standard is still LMWH," she told attendees, "but stay tuned for further studies and cost-effectiveness analyses on using UFH without monitoring
in the outpatient setting."