Published in the October 2013 issue of Today’s Hospitalist
OLD HABITS DIE HARD. Although there are many reasons why doctors order tests and perform procedures there is no evidence for “or even evidence against “the rationale that this is how things have always been done is a powerful one to overcome.
Take the common practice of ordering carotid ultrasounds for patients who come to emergency departments after a fainting spell, despite no indication that such scans influence how patients will be treated.
“I dont know how we pass this down from generation to generation, but we need to stop,” said Leonard Feldman, MD, assistant professor of medicine and pediatrics at Johns Hopkins and director of its general internal medicine comprehensive consultation service. Dr. Feldman gave a session on nonevidence-based medicine at this year’s Society of Hospital Medicine conference.
Given how much experience paramedics, EDs and other medical personnel have with “syncope Sundays” “in his hospital, Dr. Feldman said he regularly reads the notation “DFOIC,” for “done fell out in church,” on the chart of elderly people he is called to see “it is mind-boggling how much over-testing occurs.
Syncope and carotid ultrasound
Syncope, which happens in 3% of the population every year and to one-third of all people in their lifetimes, represents 5% of ED visits every year, Dr. Feldman said. More than one-third of these patients injure themselves when they faint, and their treatment costs the U.S. more than $2 billion.
Most commonly, the etiology of these episodes is vasovagal, followed by orthostatic hypotension. But most of the time, there is no known cause. A July 27, 2009, study in JAMA Internal Medicine found that many tests done on these patients neither helped determine the cause nor affected treatment. Carotid ultrasound “affected management 2% of the time,” Dr. Feldman noted, and helped “determine the etiology less than 1% of the time.” Meanwhile, the cost per test to affect diagnosis and management totaled nearly $20,000.
But that same study found that the cheapest test “taking a postural blood pressure “affected management more than any other: between 25% and 30% of the time. Further, its cost per test affecting diagnosis or management was $17. But according to the study, recording patients’ postural blood pressure took place only 38% of the time.
“Why are we not doing the one test that can be done right away and easily and that actually affects management?” Dr. Feldman asked. The only other study on the topic, published in Mayo Clinic Proceedings in April 2005, found that although patients presenting with syncope and sent for neurovascular imaging may have cerebrovascular disease, the disease probably did not cause the syncope.
“Neurovascular ultrasonography should be reserved for the subset” of patients who have “focal signs or symptoms or carotid bruits,” the paper concluded.
“If you look in high-risk patients, you do find disease,” Dr. Feldman said, but cerebrovascular causes of syncope are uncommon. “You are more likely to find cerebrovascular disease than find the actual cause of the syncope. People with cardiac causes of syncope, not surprisingly, also often have cerebrovascular disease.”
The takeaway message: “Don’t order carotid ultrasound for patients without focal neuro signs or bruits,” said Dr. Feldman. “It’s very unlikely that you will find the cause of the syncopal episode.”
Prophylactic antiepileptic drugs
Another common practice based on little evidence is prescribing antiepileptic medicine “usually phenytoin, but also now levetiracetam (Keppra) “for patients just diagnosed with brain tumors who have never had a seizure. While only one-third of patients with brain tumors experience seizures, more than 80% of neurosurgeons report prescribing antiepilectics to brain tumor patients.
As Dr. Feldman pointed out, that practice may be more harmful than helpful. A study published this April in the Journal of Neurosurgery found that the medicine failed to stop seizures. (The number of seizures among patients randomized to receive phenytoin exceeded those in the group just being observed.) Moreover, the study found that the group taking phenytoin ended up with “a lot of adverse events,” including several “major” gastrointestinal and neurological issues.
“If you are not treated with phenytoin, you don’t get adverse events from phenytoin,” Dr. Feldman pointed out. The study also found that the number of seizures that actually occurred in the observation group “those who did not get the antiseizure drug prophylactically “was much lower than expected: 8%, not 30%.
“Maybe our expectations of how many of these patients have seizures are overblown,” he said.
The one systematic review, published in 2009 in the Journal of Neuro-Oncology, also concluded that patients with brain metastases who have not experienced a seizure due to their metastatic disease should not routinely be given prophylactic anticonvulsants.
While there isn’t much evidence behind that conclusion, Dr. Feldman said there is even less on using Keppra or identifying subsets of brain tumor patients who could benefit from antiseizure prophylaxis. “There are all sorts of noncancer related reasons that you might think about using prophylaxis like trauma, bleeds and stroke,” he said. “The problem is that there is not a lot of evidence in these areas.”
Optimal potassium levels
“We were all taught that potassium should be greater than four and magnesium greater than two,” Dr. Feldman pointed out. But is that correct? When you see a patient after a successful percutaneous transluminal coronary intervention (PTCI), for instance, are you “going to buff the lytes?”
Although experts have long recommended that potassium levels be between 4 and 5 mEq/L or even between 4.5 and 5.5, those guidelines are based on studies done years ago “before the routine use of beta-blockers, reperfusion therapy and early invasive management,” he said.
More recent studies suggest that lower may be much better. Last year, the Journal of the American Medical Association (JAMA) published a large multicenter study looking at the average potassium during a patient’s entire hospital stay, not the potassium level just on admission, which had been the focus of most previous research. It concluded that a range between 3.5 and 4.4 mEq/L was associated with the lowest rates of both mortality and arrhythmias.
“This is retrospective cohort data, but this is as good as we have, and this is different than what we have been told for decades,” Dr. Feldman said. “Maintaining serum potassium levels between 3.5 and 4.5 mEq/L may be more advisable than 4.0 to 5.0 mEq/L.”
Although evidence indicates that hospitalized patients on acid-suppressing medication experience less nosocomial gastrointestinal bleeding than patients who aren’t, the benefits are tiny and probably not worthwhile, Dr. Feldman said.
“While you are preventing one GI bleed by treating 700 or 800 patients, you have caused seven or eight patients to have a pneumonia and one or two of them to get C. diff,” he said. Those were the results of a large cohort study published in JAMA Internal Medicine in 2011.
“The number needed to harm is much less than the number needed to treat,” he said. “I would recommend against prophylactic acid-suppressing medications in patients outside the ICU.”
There may well be a subset of patients who benefit from PPIs, he added “perhaps patients on platelet inhibitors, steroids, or platelet inhibitors and steroids together. But “we don’t have those data,” said Dr. Feldman. “We need to find that subset.”
Deborah Gesensway is a freelance writer who covers U.S. health care from Toronto.
IN RESPONSE TO a call from the ABIM Foundation to come up with a list of tests and treatments that physicians and patients should always question, the Society of Hospital Medicine (SHM) last year culled through a list of 150 recommendations submitted by hospitalists around the country. It came up with a top five “Never Dos or Don’t Dos” for adult hospitalists and a separate list for pediatric hospitalists.
The following items made SHM’s lists:
Adult hospital medicine
1-Do not place or leave in place urinary catheters for incontinence, convenience or monitoring output for non-critically ill patients.
2-Do not prescribe medications for stress ulcer prophylaxis to medical inpatients unless at high risk for GI complications.
3-Avoid transfusions of red blood cells for arbitrary hemoglobin or hematocrit thresholds and in the absence of symptoms of active coronary disease, heart failure or stroke.
4-Do not order continuous telemetry monitoring outside the ICU without a protocol to govern continuation.
5-Do not perform repetitive CBCs and chemistry testing in the face of clinical and lab stability.
Pediatric hospital medicine
1-Don’t order chest X-rays in children with uncomplicated asthma or bronchiolitis.
2-Don’t routinely use bronchodilators in children with bronchiolitis.
3-Don’t use systemic corticosteroids in children under 2 with an uncomplicated lower respiratory tract infection.
4-Don’t treat gastroesophageal reflux in infants routinely with acid suppression therapy.
5-Don’t use continuous pulse oximetry routinely in children with acute respiratory illness unless they are on supplemental oxygen.