IF YOU TEND to order stress tests or DVT prophylaxis for low-risk patients just to be safe, you were the target of this years “Things We Do For No Reason” lecture at the annual Society of Hospital Medicine meeting.
The presenter—Anthony C. Breu, MD, director of medical resident education at the VA Boston Healthcare System—took aim at both practices, hoping to change the minds of audience members who think such orders have value.
But Dr. Breu said he also wanted to reach those who are already non-believers, who think stress tests are low value and that hospitalists overdo chemoprophylaxis for VTE. “I want to give you ammunition so you can go back to your home institutions and make the case,” he said.
Stress tests for low-risk patients
Why do doctors order stress tests for low-risk patients? For one, said Dr. Breu, the ACC/AHA recommends that you get a stress test for these patients either during admission or within 72 hours of discharge. “And if you want to be guideline concordant, you better get it during the hospitalization because more than 90% of these patients don’t get a stress test within 72 hours of discharge,” he said. In fact, ordering stress tests for low-risk patients remains the standard of care at many hospitals.
“A stress test doesn’t rule out short-term bad outcomes.”
~ Anthony C. Breu, MD
VA Boston Healthcare System
That was borne out in an informal poll that Dr. Breu did on Twitter. He described a patient with low-risk chest pain and asked followers if they’d suggest a stress test after the patient’s second negative troponin. Among respondents, 24% said they’d get one during admission, 37% said they’d want one done within 72 hours and 21% thought the patient should have one within a few months. Only 18% said “no.”
Dr. Breu then asked those who responded “yes” what they were trying to achieve with a stress test. Among replies, 23% wanted to diagnose unstable angina, while 21% wanted to predict patients’ short-term risk for acute MI or death. The majority—55%—wanted to diagnose or rule out coronary artery disease (CAD). Dr. Breu then proceeded to pick apart each of those rationales.
Unstable angina: testing the wrong population
Even with conventional troponin assays—and hospitals are increasingly using high-sensitivity tests—the diagnosis of unstable angina is increasingly rare, Dr. Breu said.
“You really have only two things: stable angina or an MI,” he said. Five years ago, in fact, experts started recommending doing away with the term altogether.
But Dr. Breu said he believes unstable angina does exist and that hospitalists do see these patients. “But this is a clinical diagnosis” made on the basis of clinical history, ECGs and troponins, he noted, not with a positive stress test.
“I don’t think this first justification is sound. If you’re making this diagnosis with a stress test, you’re applying the test to the wrong population,” he said. “We really need to be diagnosing unstable angina upstream of a stress test.”
As for using stress tests to identify patients at short-term risk of bad outcomes, Dr. Breu said that four studies published between 1997 and 2016 all looked at patients with chest pain who had serial negative troponins and nonischemic ECGs.
“Consistently, patients across all these studies had a less than 1% risk of acute MI or death within 30 days,” he noted. As to whether or not a stress test could identify those rare cases, a study published in the December 2013 issue of the Journal of Nuclear Cardiology is the only randomized trial of stress tests in patients with low-risk chest pain.
After a nonischemic ECG and negative serial troponins ED patients with chest pain were randomized to either a nuclear stress test or clinical assessment, then followed with telemetry, troponins and ECGs. “A total of eight patients had an acute MI within 30 days, four in each arm,” Dr. Breu pointed out. Importantly, two of the four patients in the stress test arm who had MIs had had a negative stress test.
“A stress test doesn’t rule out short-term bad outcomes,” he concluded. “Even with a negative test, you can’t give patients a clean bill of health.”
Any role with CAD?
How about goal No. 3, identifying or ruling out CAD? Dr. Breu presented two studies, the first published in the June 24, 2013, issue of JAMA Internal Medicine. It looked at ED patients with negative ECGs and troponins who were stress tested, with 11% of those tests coming back positive.
But when a sizeable percentage of those patients were given angiograms, only half the angiograms were positive as well. “Best case scenario: We’re talking about a false positive stress-test rate of 50% for obstructive CAD.”
In the second study, which was published in the November 2014 Circulation: Cardiovascular Imaging, 23% of patients had known CAD.
“Yet only 9% of the stress tests were positive, and the false positive rate was again around 50%,” said Dr. Breu. “When you apply these tests—which have low specificity to patients who don’t have high pre-test probabilities, you get a lot of false positives.”
Plus, in their efforts to rule out CAD with stress testing, doctors are creating other problems. Research in the April 2014 JAMA Internal Medicine looked at more than 200 hospitals evaluating patients with suspected ischemia who didn’t have MI. Researchers divided hospitals into quartiles based on their stress-test rates, with rates ranging from 6% to 35%.
“Unsurprisingly,” Dr. Breu noted, “as you go from quartile 1 to quartile 4, you get not only more stress tests, but also more angiograms and revascularizations.” Despite the higher rates of testing and procedures, hospitals across all four quartiles had the exact same rate of short-term acute MIs: 0.3%.
While that kind of overuse is problematic for individual patients, it’s estimated that upwards of $3 billion a year is spent on tests and procedures in this patient population. Dr. Breu’s recommendations: Determine patients’ risk of acute MI or death with either a HEART or TIMI score, and make use of troponins and ECGs.
And not only should you not order a stress test in the hospital for low-risk patients, “I don’t think we should be recommending one in our discharge summary.” For doctors who are nervous about not following guidelines, he noted that he expects the ACC/ AHA to update its guidelines soon.
Turning to DVT chemoprophylaxis, Dr. Breu made it clear he wasn’t talking about postop surgical patients or those in the ICU, nor patients hospitalized with acute MI or stroke.
In another poll he’d posted on Twitter, he asked his followers what their choice would be if they had to choose between prophylaxing all their patients or none. The results: Sixty-eight percent would give it to everyone, while 32% would give it to no one.
In a follow-up question on Twitter, Dr. Breu described a patient with hypertension, hyperlipidemia, substernal chest pressure, a normal ECG and one negative troponin, who was admitted to rule out MI—in other words, a patient at low risk of VTE. Did his followers want to order DVT prophylaxis? Fifty-seven percent would order enoxaparin 40 mg QD, 24% wanted heparin 5000 U BID or TID, and 8% voted for ordering “something else.” Only 11% would order nothing.
A study published in the June 2018 Journal of Hospital Medicine revealed similar results. Researchers found that 71% of high-risk patients were treated with chemoprophylaxis, “which seems reasonable,” Dr. Breu said.
Unfortunately, 74% of low-risk patients were likewise prophylaxed. “We’re just giving it to everyone and not bothering to risk-stratify.”
To tamp down such unnecessary chemoprophylaxis, Dr. Breu said he’d make three arguments. First, evidence for prophylaxis is weak. Two, rates of VTE in hospitalized medical patients “are lower than you think.” And three, “there are harms to prophylaxis, even in high-risk patients, that we don’t recognize or give enough weight to.”
To illustrate how weak the evidence is for chemoprophylaxis, he discussed three randomized trials that, he said, “form the backbone for why we give DVT chemoprophylaxis.” The first was the MEDENOX trial, published in the September 1999 New England Journal of Medicine. That trial randomized patients at relatively high risk to either enoxaparin (two doses, either 20 or 40 mg) or placebo, and it followed patients for 14 days.
The authors found no difference between the two arms in the rates of symptomatic DVT or PE. Event rates in both were under 1%.
The second study—the PREVENT trial published in Circulation in August 2004—randomized patients to either dalteparin 5000 U or placebo and followed them for 21 days. Again, the authors found no difference in symptomatic DVT or PE rates, nor in mortality.
But the ARTEMIS trial published in the February 2006 British Medical Journal, which randomized patients to either fondaparinux or placebo, did find a difference in fatal PE rates at 15 days with chemoprophylaxis, said Dr. Breu. But here’s why: In this industry-funded trial, “any sudden or unexplained death in the absence of autopsy was classified as being by PE.” That result was so sketchy that when the Cochrane Review did its analysis in 2014 (more on that later), “they actually excluded those patients.”
So if all three trials found no difference in symptomatic PE or DVT with prophylaxis, why are these studies cited to support its use? Because all three, said Dr. Breu, also looked at rates of asymptomatic DVT.
“No doubt, you reduce the risk of asymptomatic DVT with prophylaxis,” he pointed out. Further, the researchers in the MEDENOX and ARTEMIS trials screened all enrollees with venography looking for asymptomatic DVT, while the PREVENT trial used ultrasound at 21 days or sooner if patients had symptoms.
“Who screens patients to see if they’re asymptomatic?” Dr. Breu asked. “That isn’t the standard of care.” The composite endpoint of all three trials “mirrors the rates of asymptomatic DVT quite closely.”
In looking for a reason to track asymptomatic DVT, Dr. Breu noted that upwards of 80% of patients who die with a PE are found on autopsy to have asymptomatic DVT.
“But only 10% or 15% of asymptomatic DVTs progress,” he said, and the number of both fatal and non-fatal PEs is small. The 2014 Cochrane Review, for instance, looked at 14,000 patients given heparin vs. a similar number on placebo. The event rate in the heparin arm was 0.2% vs. 0.3% for placebo.
“The number needed to treat with prophylaxis to prevent one PE is 769,” Dr. Breu pointed out. “How disheartening is that? Some of us work an entire year and don’t prevent even one PE.”
A study published by Mayo Clinic Proceedings this February revealed another wrinkle with chemoprophylaxis. The authors screened patients on admission for asymptomatic DVT, then excluded the 2% who had one. They then looked to see how many patients developed proximal DVT in the hospital.
Out of 1,170 patients, only three experienced a proximal DVT, with only one being symptomatic. But here’s the rub: Two of those three patients, Dr. Breu said, were on DVT prophylaxis with low molecular weight heparin.
“There’s often a failure of prophylaxis,” he noted, “not a failure to prophylax.”
Risk of harm
There’s also, Dr. Breu pointed out, the risk of harm with prophylaxis, primarily bleeding. The Cochrane Review found that the bleeding risk among patients prophylaxed with heparin was 0.65% vs. 0.34% with placebo, and that the number needed to harm for major bleeding was 323.
“If you include rates of minor bleeds, the number needed to harm is around 100,” he said. “Don’t discount that risk.”
There’s also the issue of patient experience. If you need to prophylax about 770 patients to avoid one PE and each patient is in the hospital for around five days, it takes more than 3,800 injections of low molecular weight heparin or more than 11,500 jabs of unfractionated heparin to prevent a PE.
Further, all the data in all the studies he’d just presented were exclusively for patients at high risk for VTE, Dr. Breu explained. No data support prophylaxing low-risk patients at all.
“If you find the case wanting for high-risk patients, and I do, how could you possibly justify prophylaxing low-risk ones?” he asked.
His recommendations: Don’t prophylax low-risk patients. And “pause a bit even giving prophylaxis to high-risk patients,” he said. “The prophylaxis can fail.” Finally, don’t underestimate the harms.
Phyllis Maguire is Executive Editor of Today’s Hospitalist.
Published in the October 2019 issue of Today’s Hospitalist.