If Leonard Feldman, MD, FACP, got his wish, scores of hospitalists are now doing a lot less at their jobs than they were before attending his talk at Hospital Medicine 2012 in April.
Specifically, those hospitalists are doing fewer chest X-rays, nasogastric lavages, and blood transfusions and are calculating fewer fractional excretion of urea (FEUrea) and fractional excretion of sodium (FENa) scores. All of these practices were targeted in Dr. Feldman's popular session, “Non-Evidence-Based Medicine: Things We Do for No Reason.”
NG lavage overload
Nasogastric (NG) lavage is arguably the most egregious of the overused practices, Dr. Feldman said. A 1999 study in Annals of Emergency Medicine found that patients and physicians rated NG intubation as more painful than abscess incision, fracture reduction, urethral catheterization…and all other commonly performed procedures in the emergency department (ED), Dr. Feldman said.
“Patients think NG lavage and intubation really stink, so we need to have a good reason to do this,” said Dr. Feldman, who is director of the general medicine consult service and the med-peds program at Johns Hopkins Hospital in Baltimore. “This is information you might want to take to your ED.”
NG lavage is often done to look for bloody aspirate, which points to a greater likelihood of high-risk lesions on endoscopy. “So the question becomes, does it actually improve outcomes to know who is at higher risk?” Dr. Feldman said. “What we need is a randomized controlled trial on this question, and the best we have is a propensity-matched retrospective analysis.”
The analysis, published in 2011 in Gastrointestinal Endoscopy, looked at 632 patients who were admitted with gastrointestinal bleeding to Veterans Affairs (VA) hospitals. Patients with and without NG lavage were compared and matched on more than two dozen confounders. No significant differences were found between the patients in terms of mortality, mean length of stay (LOS), or rates of emergency surgery or blood transfusion.
Two differences did emerge, however: Patients who got an NG lavage were much more likely to get an endoscopy, and their time to endoscopy was faster. “Does that mean there was a confounder we couldn't actually measure—that the ER physicians who are ordering that NG lavage are just very aggressive?” Dr. Feldman said.
“The question is, is it really a good thing to get endoscopy more often or faster if your mortality, length of stay, or number of units of blood transfused are all similar?” he continued.
The editorialists for the VA study didn't seem to think so: They concluded that using NG lavage to manage patients with acute upper GI bleeding was “antiquated,” he said.
“We certainly know it's painful, and if it isn't giving us great results for management, then it probably is antiquated, as well,” Dr. Feldman said.
To transfuse or not to transfuse?
NG lavage may win the prize for most painful procedure, but blood transfusions aren't exactly fun, either. So it's good news that recent research suggests one can use a more restrictive transfusion strategy with cardiac patients.
A 2010 study in the Journal of the American Medical Association (JAMA) found no mortality difference in cardiac patients treated with a target hematocrit of 30% (liberal strategy) and those with a target of 24% (restrictive strategy) to trigger transfusion. This confirms the findings of the 1999 TRIC trial, a mixed-ICU study that suggested patients on restrictive strategies are more likely to survive—though the finding wasn't statistically significant, he said.
Likewise, a study published in late 2011 in the New England Journal of Medicine compared outcomes for hip fracture surgery patients with cardiovascular disease where the liberal transfusion cutoff was less than 10 g/dL and the restrictive was less than 8 g/dL or being symptomatic. The outcomes were death or the ability to walk across a room without human assistance at 60-day follow-up.
Patients in the liberal group got a lot more blood than those in the restrictive group: 1,866 units versus 652 units, respectively. At 30 and 60 days, there was no statistically significant difference in mortality or walking ability.
“So when the orthopedist says ‘We need to give that blood so this patient will do better in their rehab,’ you can say, ‘No, they really have the same chance of walking at 30 or 60 days whether you are liberal or restrictive in the blood use,’” Dr. Feldman said.
Likewise, the AAAB (formerly the American Association of Blood Banks) recommends a restrictive strategy for patients with cardiovascular disease—meaning transfuse only when the patient is symptomatic or has a hemoglobin of less than 8 g/dL. “So now we have an organization that specializes in this area lending credence to [a restrictive strategy]. They said the evidence was moderate quality, but that's because there just isn't much,” he said.
A word of caution, he added: The jury is still out on transfusion strategies for patients who have an acute coronary syndrome. “There really is no evidence on this,” Dr. Feldman said.
FENa or FEUrea…or neither?
The evidence on whether to use FENa or FEUrea to evaluate acute kidney injury is a bit complicated. The first suggestion that FENa might be helpful in this area came from a 17-person study published in JAMA in 1976, Dr. Feldman noted. Patients with prerenal azotemia appeared to have a low FENa and those that didn't had a high FENa, he said.
“Now we know that FENa isn't perfect, since lots of intrinsic kidney disorders can cause it, and it can be elevated in patients with prerenal states that are contributed to by diuretic use,” Dr. Feldman said. “So then we started looking at FEUrea—thinking maybe it would fix the problems with those on diuretics.”
One study, published in 2002 in Kidney International, examined this hypothesis. It comprised 102 ICU patients, 27 who had prerenal azotemia while on diuretics, 50 who had prerenal azotemia without diuretics, and 25 who had acute tubular necrosis. The results suggested FENa, FEUrea and the urine-to-plasma-creatinine ratio all looked “pretty good” for deciding who was prerenal in the absence of diuretics, he noted.
“It also looked like FEUrea really kicked butt for those patients who were on diuretics,” Dr. Feldman added. “But this was only 102 ICU patients, and they excluded patients with acute glomerulonephritis [AGN] and obstructive nephropathy, so you have to make sure you have excluded those patients if you are going to use FEUrea.”
On the other hand, a 99-patient study, published in 2007 in the American Journal of Kidney Disease, found the sensitivity and specificity of FEUrea and FENa “weren't very impressive” whether the patients were on diuretics or not. What's more, the study excluded patients with rhabdomyolysis, adrenal insufficiency, obstructive nephropathy, AGN, nephrotoxic acute kidney injury, and chronic kidney disease.
In February 2012, an article in the Cleveland Clinic Journal of Medicine looked at all existing research on the two indices and gave an evaluation of the specificities and sensitivities in a best- and worst-case scenario. “The best case for a patient using FENa is actually pretty good. A positive FENa gives you a likelihood ratio of 24, and if your FENa is greater than 3%, it's a likelihood ratio of 75, and all these look really good,” Dr. Feldman said.
The problem, he added, is that the worst-case scenario “really stinks. In the worst case, these tests didn't help whatsoever. So you are left not knowing if it's the best- or worst-case scenario,” he said.
The authors themselves cautioned that a single index calculated at a specific time is often insufficient to characterize the pathogenesis of acute kidney injury. As well, they noted that collecting urine samples after acute changes in volume or osmolarity can compromise their usefulness for diagnosis, Dr. Feldman noted.
In the end, he said, FENa and FEUrea don't really help very much in determining who has acute kidney injury, or for what reason. “You should spend a lot of time doing a really great history and physical, and trying to figure out why your patient has acute renal failure, rather than trying to use these indices,” he said.
Reining in X-rays
High-quality evidence suggests the number of chest X-rays being done in ICUs and stepdown units is too high as well, Dr. Feldman said.
A trial of mechanically ventilated patients in 21 French ICUs, published in Lancet in 2009, found no difference in outcomes when chest X-rays were done routinely every day or only on demand. Days on mechanical ventilation, length of stay in the ICU and mortality were the same, but the number of chest X-rays was reduced by 32% in on-demand patients, and these patients had their ventilators changed more often as well.
The authors of a 2010 meta-analysis of eight studies, published in Radiology, also concluded that daily routine chest radiography can be eliminated without increasing adverse events in adult ICU patients, Dr. Feldman noted.
“So, on-demand chest X-rays really seem to make a lot of sense for these patients,” he said. “And I would certainly have you all think about that as you go back to your hospitals.”