More things you shouldn't do based on the lack of evidence

Leonard Feldman, MD, FACP, has one word for physicians who order folic acid tests for patients with altered mental status: “Don't.”

Dr. Feldman, an associate professor of internal medicine and pediatrics at Johns Hopkins University in Baltimore, said that word a lot during his session “Non-Evidence Based Medicine: Things We Do for No Reason,” held at Hospital Medicine 2014 in Las Vegas in March.

The talk, which Dr. Feldman updates with new topics every year for the meeting, also targeted creatine kinase muscle-brain (CK-MB) testing and 24-hour patient observation after switching from IV to oral antibiotics as generally useless practices.

Such unwarranted testing and use of resources make up the biggest chunk of unnecessary health care costs, estimated at between $250 and $325 billion per year, Dr. Feldman noted.

“If you look at the growth in volume of physician services from 2000 to 2009, imaging and other diagnostic tests have gone up most rapidly, faster than [evaluation and management] services,” Dr. Feldman said, and with his tongue firmly implanted in his cheek, added, “I mean, I know at Hopkins, no patient gets through the door without going through the doughnut of truth [CT].”

Doctors may order unnecessary tests because they don't understand the costs, they are bending to patient expectations, they haven't gotten the data they need (such as test results) from other health care facilities, they don't actually understand the results of the tests they have, or they are practicing defensive medicine, Dr. Feldman said. But he suggested that inertia may be the biggest reason.

Forget about folate

Before the late 1990s, there was good reason to check patients for suspected folate deficiency, Dr. Feldman said, since historically about 6% of the population had folate-deficiency anemia. But in 1996, the U.S. and Canada implemented mandatory folic acid fortification of enriched cereal grains, and soon everything changed.

A 1999 Lancet study of Kaiser Permanente endocrinology outpatients looked at folate deficiency levels between 1994 and 1998, during the period of cereal fortification, and found that serum folate deficiency rates declined yearly from 1.3% in 1994 to 0.3% in 1998.

At Baystate Health System in western Massachusetts in 2002, researchers retrospectively reviewed the charts of all in- and outpatients who had folic acid tests ordered in a 4-month period in 2002. Of the 1,007 tests ordered on 980 patients, 60% were for anemia and 7.2% were for anemia with macrocytosis. Thirty percent of the tests were ordered because the patients had dementia or altered mental status. Of those 1,007 folic acid tests ordered, only 4—or 0.4%—showed a low serum folate level (less than 3 ng/mL).

Curiously, the Lancet study also found that, as the rate of folate deficiency declined, the rate of folate testing actually increased, from 7.3 per 1,000 patients in 1994 to 10.2 per 1,000 patients in 1998. Meanwhile, a 2001 study at the Cleveland Clinic found that, of the small number of patients who were found to have low serum folate levels, only half actually received folate replacement, “either because the physicians didn't think it was important, or they forgot to give it to them,” he said.

If low serum folate levels are so rare, and if physicians are only going to do anything about them half the time they are found to exist, why do clinicians still order them? Habit, said Dr. Feldman.

“A big reason for unnecessary test ordering is learned behavior; we see this especially in residency programs. Senior residents do something, then pass it along to the interns watching them, who then become residents and pass it along further,” Dr. Feldman said.

The practice of ordering folic acid tests for dementia, delirium, or altered mental status was never based in hard evidence but emerged as the result of published case reports suggesting a link between these conditions and low folate, he added. The same is true for peripheral neuropathy and malnutrition.

Another big reason physicians order unnecessary tests is a lack of awareness of cost. Folic acid tests do indeed add up. Beth Israel Deaconess Medical Center (BIDMC) in Boston did a retrospective review of serum folate level tests of admitted and ED patients in 2011. Clinicians had ordered 2,093 tests for 1,900 patients; only 2 of those patients were found to have a low serum folate level. With a charge of $151 per test at that institution, BIDMC spent $160,000 per test that yielded a deficiency, Dr. Feldman noted.

The real kicker? In both cases, there was no change in management.

CK-MB? Not OK

Another questionable test that is still ordered often is the CK-MB for diagnosis of acute myocardial infarction or acute coronary syndrome, Dr. Feldman said. “No evidence exists that CK-MB adds to the accuracy at this point,” beyond what already can be found with troponin assays, he said.

In the 1970s, CK-MB tests were useful for detecting acute myocardial infarction and reinfarction and for estimating infarct size, he said. However, the specificity of the test has never been great—there are many false positives—so when troponin testing came along in 1997, it should have replaced CK-MB. Troponin I and T tests have excellent specificity and are more sensitive and yield results earlier than CK-MB, he said. Still, many clinicians often order both.

Head-to-head trials of the 2 tests indicate that CK-MB provides no extra useful information on infarct size or reinfarction or for evaluation of patients with chronic troponin elevation. Furthermore, a joint guideline from the American College of Cardiology, the American Heart Association, and the European Society of Cardiology recommends troponin testing over CK-MB in diagnosing MI, he noted.

“The guidelines say that if a troponin assay is not available, use CK-MB,” Dr. Feldman said. “To me that says, use troponin! And only use CK-MB if you don't have troponin.”

However, 2 major studies that compared CK-MB and troponin assays did find an interesting, albeit nonsignificant, trend whereby patients with negative troponin and positive CK-MB results had slightly higher inpatient mortality rates than those with double negative results, he noted.

“It's possible there may be some prognostic implications for these CK-MB-positive, troponin-negative patients,” Dr. Feldman said. “Or, it could be just false positives, and all you are seeing is that these patients were high risk to begin with because they came in with chest pain and a bunch of other comorbidities.”

Still, given the cost of adding CK-MB to troponin testing, eliminating CK-MB testing “is the right way to go,” he concluded. “Is CK-MB a high-value test? No way. We could save a lot of money for our health care system if we get rid of it.”

Transitioning to oral antibiotics

Another common, unnecessary practice is observing patients for a day after transitioning them from IV to oral antibiotics to make sure they don't crash, Dr. Feldman said.

Back in the 1980s and 1990s, this standard was motivated by a peer review agency (Island Peer Review Organization, now IPRO) that said it was a best practice to watch patients for 24 hours before you send them home, he noted.

Also, in many of the earlier trials of community-acquired pneumonia patients switched from IV to oral antibiotics—especially those involving fluoroquinolone—there was a built-in period of observation. “And when you start doing that, it becomes tradition,” Dr. Feldman said.

So why is observation a bad idea? First, physicians are already trying to get patients to see their primary care physicians within a week of discharge, so presumably the PCPs can monitor for transition difficulties, he said. More important, the half-lives of antibiotics are often long enough that therapeutic levels persist for hours. “So yeah, [patients] aren't on their IV anymore, but they are really still on their IV antibiotic,” Dr. Feldman said.

Also, the bioavailability of many of the oral antibiotics is excellent. “We know they are going to work,” Dr. Feldman said.

On the flip side, 24 hours may not even be enough to make a determination as to whether a patient should be discharged, he said. “Probably, the patient is going to do worse at 48 or 72 [hours], if at all. You might get by at 24, even if the oral medication isn't working well, because they still have that IV antibiotic onboard,” he said. “So hey, maybe we should keep them another 48 or 72 hours?”

There's not a great deal of data on whether, or how long, patients should be observed after switching from IV to oral antibiotics, he said. A 1995 study of 468 patients with pneumonia, urinary tract infection, or cellulitis at Mt. Sinai Hospital in New York City found no difference in the 2-week readmission rates between patients who were observed for 24 hours after transition and patients who were unobserved. Observed patients tended to be older and have more comorbidities; even so, only 1% had an adverse event during the observation period, and only 2% had any sort of event.

A study at the University of Pennsylvania in the 1990s found only 5 of 94 antibiotic-taking patients had adverse events. Just 2 of these events—rash and nausea—were attributable to the drug, and only one of them resulted in a change of the antibiotic, Dr. Feldman noted.

The most recent evidence, published in 2006 in the American Journal of Medicine, comes from the National Pneumonia Project. The 6,138 patients in the “observed” group—who were sicker—had similar readmission and 30-day mortality rates as the 3,182 patients who were discharged home after transitioning to oral antibiotics.

Another factor to consider is that there is a fair amount of data to suggest physicians are prescribing IV antibiotics in cases where they could be using oral antibiotics, he added. A 2007 Cochrane Review found no evidence that oral antibiotics are less effective for treating severe urinary tract infection than parenteral or initial parenteral therapy.

A 2012 Spanish study tested the use of a 3-step bundle to reduce the duration of IV antibiotic therapy and length of stay in inpatients with community-acquired pneumonia, compared to usual care. The intervention included early mobilization, use of objective criteria to switch to oral from IV antibiotics, and predefined criteria to discharge patients.

The results were that bundled patients had more than 2 days knocked off their lengths of stay, from 6 days in usual care to 3.9 days with the bundle, and went from 4 days on IV antibiotics to 2 days. They had fewer adverse drug reactions to boot. There were no differences between the groups in 30-day readmission rates, mortality at 30 days, or patient satisfaction.

“So the take-home is, systematic approaches like bundles are important, and we don't have any data that we should observe patients for a day after their transition,” Dr. Feldman said. “Granted, the evidence base is scant, but it certainly does not favor observation.”

At this point, there is little agreement on how long to treat many of the infections targeted by antibiotics, Dr. Feldman added. “That may be one of the things we look at next year: Why do we treat some of our infections for so darn long when there is evidence we don't need to do that?”