Recent Evidence Challenges Four Inpatient Management Habits

Why did you order that test? Dr. Leonard S. Feldman wants you to turn off the autopilot and consider the evidence from the medical literature on the following four practices:

• Is the best target hematocrit for a cardiac patient 30% (or a hemoglobin level of 10 g/dL)?

• Should nasogastric lavage be routine in patients with suspected GI bleeds?

• Is it helpful to measure the fractional excretion of sodium or fractional excretion of urea nitrogen when evaluating acute kidney injury?

• Are daily chest x-rays a good idea in patients on mechanical ventilation in an ICU or step-down unit?

"Many of us do all of these things regularly on a reflexive basis," said Dr. Feldman, director of the general medicine comprehensive consult service at Johns Hopkins University, Baltimore.

Yet recent studies challenge the value of these approaches:

Transfusions

Three studies influenced the AABB (formerly the American Association of Blood Banks) to publish a clinical practice guideline this year recommending a restrictive strategy when considering blood transfusions in hospitalized patients with preexisting cardiovascular disease. The AABB says not to transfuse if the hemoglobin level is above 7-8 g/dL and to consider transfusing patients who are symptomatic or who have a hemoglobin level of less than 8 g/dL, the AABB says (Ann. Intern. Med. 2012;157:49-58).

One recent study randomized 502 patients undergoing coronary artery bypass grafting (CABG) or valve replacement who were on cardiac bypass support to either a liberal strategy aiming to maintain a hematocrit of at or above 30% from the start of surgery until discharge from the ICU or a restrictive strategy that aimed to keep the hematocrit at or above 24%.

Both groups were able to meet these goals. The two groups did not differ significantly, however, in a composite end point of 30-day all-cause mortality and severe morbidity occurring during hospitalization (JAMA 2011;304:1559-67).

There was a trend toward higher risk of death with the restrictive strategy, with a hazard ratio of 1.28, but the ratio ranged from 0.6 to 2.7, so the risk was not statistically significant. Analysis of a slew of secondary outcomes also found no significant differences between groups.

Another study randomized 2,016 patients aged 50 years or older who either had a history of cardiovascular disease or cardiac risk factors and who had a hemoglobin level below 10 g/dL after surgery for hip fracture. The liberal strategy triggered transfusion in all patients with a hemoglobin below 10 g/dL. The restrictive strategy transfused only if the patient showed symptoms of anemia or at the physician’s discretion in patients with a hemoglobin level less than 8 g/dL.

Patients received a lot more blood under the liberal strategy compared with the restrictive strategy – 1,866 vs. 652 units of blood, respectively – but the groups did not differ significantly in death rates or the ability to walk across a room without human assistance at 30 and 60 days after surgery (N. Engl. J. Med. 2011;365:2453-62). Sixty days after surgery, 35% of patients in each group had died or were unable to walk across a room unaided.

"This is a trial that you should be showing all of your orthopedists," Dr. Feldman said. "When the orthopedist says that we need to give a patient blood so the patient will be able to do better in rehab" after hip surgery, point to the data showing that this isn’t necessarily true, he suggested.

An earlier trial of 838 critically ill patients who had hemoglobin levels below 9 g/dL within 72 hours of admission to the ICU found 30-day mortality rates of 19% in those randomized to a restrictive transfusion strategy and 23% in patients randomized to a liberal transfusion strategy, a difference that was not statistically significant (N. Engl. J. Med. 1999;340:409-17).

The liberal strategy transfused when hemoglobin levels fell below 10 g/dL and maintained the hemoglobin at 10-12 g/dL. The restrictive strategy waited until hemoglobin levels fell below 7 g/dL before transfusing and maintained hemoglobin at 7-9 g/dL.

"So, we have three different studies here that really seem to indicate that for transfusion of those patients who are at highest risk – cardiac patients – it did not seem to make a difference in their outcomes, particularly in mortality, if you were restrictive or liberal" in setting thresholds for transfusion, Dr. Feldman said. "This is not the same, though, for patients who are having an acute coronary syndrome. We don’t have any good data for them," and the AABB doesn’t recommend for or against liberal or restrictive transfusion thresholds, he added.

The evidence behind the AABB recommendation is only of moderate quality because there are only the three studies. Although the literature generally favors a restrictive versus a liberal approach, there’s no robust evidence for any particular transfusion threshold, said Dr. Daniel S. VanderEnde, a hospitalist and member of the joint transfusion committee at Emory University, Atlanta.

Pointing to the limited amount of research, the AABB itself calls the recommendation’s quality of evidence "very low" and the strength of the recommendation "uncertain."

"Subpar" evidence is one reason that Dr. VanderEnde’s institution leaves transfusion decisions to individual clinicians, he said in an interview. One ICU may transfuse at a hemoglobin threshold "in the mid-7s, and another ICU will have a transfusion threshold in the mid-8s."

Emory is starting a computer order entry protocol requiring physicians to tell why they’re transfusing blood, compared with no previous oversight. "It doesn’t stop them from transfusing for any reason. It is just trying to collect data, in the hopes that maybe they will be more restrictive in their use rather than liberal," he said.

Anecdotally, transfusion practices do seem to be shifting, at least among newly-trained physicians, Dr. VanderEnde added. Five years ago when he would ask medical students about transfusion thresholds, many hewed to "the old 10/30 rule," but far fewer do so today, he said. "The younger orthopedists tend to not transfuse as much as the older orthopedists."

Nasogastric Lavage

Few procedures performed in emergency departments are more painful for patients than nasogastric intubation, and there’s a study to prove that (Ann. Emerg. Med. 1999;33:652-8).

"Patients think nasogastric lavage and nasogastric intubation really stink, so we need to have a good reason to do it," Dr. Feldman said.

Sherry Boschert/IMNG Medical Media

And, like all invasive procedures, there are risks involved, Dr. Chad T. Whelan said in an interview. There are only modestly convincing data suggesting that nasogastric lavage can provide some prognostic or "localizing" information (such as differentiating upper vs. lower bleed). "Therefore, the risk/benefit ratio of routinely performing them for all patients has shifted with our increasing understanding of their risks and benefits," said Dr. Whelan, a hospitalist at the University of Chicago.

International consensus recommendations on the management of patients with nonvariceal upper GI bleeding suggest that physicians consider placing a nasogastric tube in selected patients because the findings may have prognostic value – not very helpful advice in decision making, Dr. Feldman said. (Ann. Intern. Med. 2010;152:101-13).

The rationale until now has been that patients with bloody aspirate on nasogastric lavage are significantly more likely to have high-risk GI lesions on endoscopy, compared with patients with clear or bilious aspirates on lavage. But does knowing this improve outcomes?

One review of the literature on how to determine if a patient has a severe upper GI bleed confirmed that a bloody aspirate on nasogastric lavage increases the likelihood of an upper GI bleed but there’s only a mildly increased likelihood of a severe bleed, "and the negative likelihood ratio is not unimpressive," Dr. Feldman said (JAMA 2012;307:1072-19).

Results of a separate propensity-matched retrospective analysis of data on 632 patients admitted with GI bleeding are "as good as we’re going to get on this topic," he said. The study found that getting or not getting nasogastric lavage did not change 30-day mortality, mean length of stay, transfusion requirements, or emergency surgery rates (Gastrointest. Endosc. 2011;74:971-80).

The only things that nasogastric lavage did change were an increase in the rate of patients undergoing endoscopy, a shorter interval to endoscopy, and a shorter length of stay among patients who had endoscopy.

That suggests that there was an individual-provider confounder that the study could not measure. Perhaps emergency physicians or gastroenterologists who order nasogastric lavage are simply more aggressive, Dr. Feldman said. "This is information that you might want to take to your emergency department," he said.

In a joint editorial accompanying the study, an emergency physician and an endoscopist concluded that the practice of nasogastric lavage in patients with acute upper GI bleeding is "antiquated."

Dr. Whelan said the role of nasogastric lavage "is in transition rather than antiquated." As upper GI bleeding epidemiology evolves and endoscopic interventions improve, "the widespread use of nasogastric lavage as a universal piece of the upper GI bleed protocol should decrease. Whether nasogastric lavage ultimately becomes a completely unnecessary procedure remains to be seen," he said.

Physicians at his institution no longer routinely perform nasogastric lavage when evaluating suspected upper GI bleeding, but "it has not completely disappeared from practice, either," he said. That’s less a factor of "aggressive" physicians and more a result of how practice changes and environmental factors, Dr. Whelan added. "Not all emergency rooms have access to full-service endoscopy on site, so emergency room physicians may have a different set of risk/benefit tradeoffs to consider."

Acute Kidney Injury

Can the fractional excretion of sodium (FE_Na) or fractional excretion of urea nitrogen (FE_UN) help narrow the differential diagnosis in acute kidney disease? Widespread use of these measures began after a 17-person study in 1976 suggested that patients with prerenal azotemia had a FE_Na of less than 1 and patients with acute tubular necrosis had a FE_Na greater than 3 (JAMA 1976;236:579-81).

The FE_Na is not perfect, because many intrinsic kidney disorders can cause low FE_Na and the FE_Na can be elevated when diuretic use contributes to prerenal states, so a few studies looked at adding the FE_UN to the diagnostic tools. Their results were contradictory.

One study of 102 patients in the ICU found that incorporating FE_UN was 85% sensitive and 92% specific in detecting prerenal injury, but the study excluded patients with acute glomerulonephritis and obstructive nephropathy, "so you have to make sure that you exclude those patients if you’re going to use FE_UN," Dr. Feldman said (Kidney Int. 2002;62:2223-9).

In a separate study of 99 patients, however, the FE_Na and FE_UN were much less impressive in patients with or without diuretics. In patients on diuretics, FE_UN had a sensitivity for distinguishing transient from persistent acute kidney injury of 79% and a specificity of 33%, and in patients not on diuretics the sensitivity was 48% and the specificity was 75% (Am. J. Kidney Dis. 2007;50:568-73).

A recent analysis reviewed the literature to provide some guidance for clinicians, but the end result is confusing, Dr. Feldman said. Under best-case scenarios, these two measures would be likely to make a difference in diagnosing the cause of acute kidney injury, but under worst-case scenarios, "they really stink," he said (Cleve. Clin. J. Med. 2012;79:121-6).

The authors cautioned that a single index calculated at a specific time often is insufficient to properly characterize the pathogenesis of acute kidney injury. "In the end, probably FE_Na and FE_UN really don’t help you very much to decide" the reason behind acute kidney injury, Dr. Feldman said.

Chest X-Rays

Routine chest x-rays in mechanically ventilated patients in ICUs provide, well, too many unneeded x-rays, recent data show.

A crossover study that randomized 21 French ICUs to either routine daily chest x-rays for these patients or x-rays on demand found that the on-demand strategy reduced the number of x-rays by 32% without affecting the number of days on ventilation, length of ICU stay, or mortality. With the daily x-ray strategy, 424 patients got 4,607 x-rays, compared with 3,148 x-rays in 425 patients under the on-demand strategy (Lancet 2009;374:1687-93).

Patients had their ventilators changed more often under the on-demand strategy, probably as clinicians were troubleshooting potential problems, but the number of interventions did not differ significantly by x-ray strategy, Dr. Feldman noted.

A meta-analysis this year of eight trials including 7,078 adult ICU patients concluded that routine daily x-rays can be eliminated without increasing adverse outcomes (Radiology 2012;255:386-95).

Dr. Feldman suggested specific goals for these four scenarios, which he presented at the annual meeting of the Society of Hospitalist Medicine.

"Do not reflexively transfuse cardiac patients to hematocrits of 30%. Do not do routine daily chest x-rays. Do not reflexively NG [nasogastric] lavage our patients. And spend more time doing a really great history and physical and thinking about why your patient has acute renal failure than trying to use indices that don’t actually help us very much," he said.

He added a personal goal: "If I can make the residents at Johns Hopkins change, that will be a real feat, because they love to order tests on everybody."

Dr. Feldman, Dr. Whelan, and Dr. VanderEnde reported having no financial disclosures.

Why did you order that test? Dr. Leonard S. Feldman wants you to turn off the autopilot and consider the evidence from the medical literature on the following four practices:

• Is the best target hematocrit for a cardiac patient 30% (or a hemoglobin level of 10 g/dL)?

• Should nasogastric lavage be routine in patients with suspected GI bleeds?

• Is it helpful to measure the fractional excretion of sodium or fractional excretion of urea nitrogen when evaluating acute kidney injury?

• Are daily chest x-rays a good idea in patients on mechanical ventilation in an ICU or step-down unit?

"Many of us do all of these things regularly on a reflexive basis," said Dr. Feldman, director of the general medicine comprehensive consult service at Johns Hopkins University, Baltimore.

Yet recent studies challenge the value of these approaches:

Transfusions

Three studies influenced the AABB (formerly the American Association of Blood Banks) to publish a clinical practice guideline this year recommending a restrictive strategy when considering blood transfusions in hospitalized patients with preexisting cardiovascular disease. The AABB says not to transfuse if the hemoglobin level is above 7-8 g/dL and to consider transfusing patients who are symptomatic or who have a hemoglobin level of less than 8 g/dL, the AABB says (Ann. Intern. Med. 2012;157:49-58).

One recent study randomized 502 patients undergoing coronary artery bypass grafting (CABG) or valve replacement who were on cardiac bypass support to either a liberal strategy aiming to maintain a hematocrit of at or above 30% from the start of surgery until discharge from the ICU or a restrictive strategy that aimed to keep the hematocrit at or above 24%.

Both groups were able to meet these goals. The two groups did not differ significantly, however, in a composite end point of 30-day all-cause mortality and severe morbidity occurring during hospitalization (JAMA 2011;304:1559-67).

There was a trend toward higher risk of death with the restrictive strategy, with a hazard ratio of 1.28, but the ratio ranged from 0.6 to 2.7, so the risk was not statistically significant. Analysis of a slew of secondary outcomes also found no significant differences between groups.

Another study randomized 2,016 patients aged 50 years or older who either had a history of cardiovascular disease or cardiac risk factors and who had a hemoglobin level below 10 g/dL after surgery for hip fracture. The liberal strategy triggered transfusion in all patients with a hemoglobin below 10 g/dL. The restrictive strategy transfused only if the patient showed symptoms of anemia or at the physician’s discretion in patients with a hemoglobin level less than 8 g/dL.

Patients received a lot more blood under the liberal strategy compared with the restrictive strategy – 1,866 vs. 652 units of blood, respectively – but the groups did not differ significantly in death rates or the ability to walk across a room without human assistance at 30 and 60 days after surgery (N. Engl. J. Med. 2011;365:2453-62). Sixty days after surgery, 35% of patients in each group had died or were unable to walk across a room unaided.

"This is a trial that you should be showing all of your orthopedists," Dr. Feldman said. "When the orthopedist says that we need to give a patient blood so the patient will be able to do better in rehab" after hip surgery, point to the data showing that this isn’t necessarily true, he suggested.

An earlier trial of 838 critically ill patients who had hemoglobin levels below 9 g/dL within 72 hours of admission to the ICU found 30-day mortality rates of 19% in those randomized to a restrictive transfusion strategy and 23% in patients randomized to a liberal transfusion strategy, a difference that was not statistically significant (N. Engl. J. Med. 1999;340:409-17).

The liberal strategy transfused when hemoglobin levels fell below 10 g/dL and maintained the hemoglobin at 10-12 g/dL. The restrictive strategy waited until hemoglobin levels fell below 7 g/dL before transfusing and maintained hemoglobin at 7-9 g/dL.

"So, we have three different studies here that really seem to indicate that for transfusion of those patients who are at highest risk – cardiac patients – it did not seem to make a difference in their outcomes, particularly in mortality, if you were restrictive or liberal" in setting thresholds for transfusion, Dr. Feldman said. "This is not the same, though, for patients who are having an acute coronary syndrome. We don’t have any good data for them," and the AABB doesn’t recommend for or against liberal or restrictive transfusion thresholds, he added.

The evidence behind the AABB recommendation is only of moderate quality because there are only the three studies. Although the literature generally favors a restrictive versus a liberal approach, there’s no robust evidence for any particular transfusion threshold, said Dr. Daniel S. VanderEnde, a hospitalist and member of the joint transfusion committee at Emory University, Atlanta.

Pointing to the limited amount of research, the AABB itself calls the recommendation’s quality of evidence "very low" and the strength of the recommendation "uncertain."

"Subpar" evidence is one reason that Dr. VanderEnde’s institution leaves transfusion decisions to individual clinicians, he said in an interview. One ICU may transfuse at a hemoglobin threshold "in the mid-7s, and another ICU will have a transfusion threshold in the mid-8s."

Emory is starting a computer order entry protocol requiring physicians to tell why they’re transfusing blood, compared with no previous oversight. "It doesn’t stop them from transfusing for any reason. It is just trying to collect data, in the hopes that maybe they will be more restrictive in their use rather than liberal," he said.

Anecdotally, transfusion practices do seem to be shifting, at least among newly-trained physicians, Dr. VanderEnde added. Five years ago when he would ask medical students about transfusion thresholds, many hewed to "the old 10/30 rule," but far fewer do so today, he said. "The younger orthopedists tend to not transfuse as much as the older orthopedists."

Nasogastric Lavage

Few procedures performed in emergency departments are more painful for patients than nasogastric intubation, and there’s a study to prove that (Ann. Emerg. Med. 1999;33:652-8).

"Patients think nasogastric lavage and nasogastric intubation really stink, so we need to have a good reason to do it," Dr. Feldman said.

Sherry Boschert/IMNG Medical Media

And, like all invasive procedures, there are risks involved, Dr. Chad T. Whelan said in an interview. There are only modestly convincing data suggesting that nasogastric lavage can provide some prognostic or "localizing" information (such as differentiating upper vs. lower bleed). "Therefore, the risk/benefit ratio of routinely performing them for all patients has shifted with our increasing understanding of their risks and benefits," said Dr. Whelan, a hospitalist at the University of Chicago.

International consensus recommendations on the management of patients with nonvariceal upper GI bleeding suggest that physicians consider placing a nasogastric tube in selected patients because the findings may have prognostic value – not very helpful advice in decision making, Dr. Feldman said. (Ann. Intern. Med. 2010;152:101-13).

The rationale until now has been that patients with bloody aspirate on nasogastric lavage are significantly more likely to have high-risk GI lesions on endoscopy, compared with patients with clear or bilious aspirates on lavage. But does knowing this improve outcomes?

One review of the literature on how to determine if a patient has a severe upper GI bleed confirmed that a bloody aspirate on nasogastric lavage increases the likelihood of an upper GI bleed but there’s only a mildly increased likelihood of a severe bleed, "and the negative likelihood ratio is not unimpressive," Dr. Feldman said (JAMA 2012;307:1072-19).

Results of a separate propensity-matched retrospective analysis of data on 632 patients admitted with GI bleeding are "as good as we’re going to get on this topic," he said. The study found that getting or not getting nasogastric lavage did not change 30-day mortality, mean length of stay, transfusion requirements, or emergency surgery rates (Gastrointest. Endosc. 2011;74:971-80).

The only things that nasogastric lavage did change were an increase in the rate of patients undergoing endoscopy, a shorter interval to endoscopy, and a shorter length of stay among patients who had endoscopy.

That suggests that there was an individual-provider confounder that the study could not measure. Perhaps emergency physicians or gastroenterologists who order nasogastric lavage are simply more aggressive, Dr. Feldman said. "This is information that you might want to take to your emergency department," he said.

In a joint editorial accompanying the study, an emergency physician and an endoscopist concluded that the practice of nasogastric lavage in patients with acute upper GI bleeding is "antiquated."

Dr. Whelan said the role of nasogastric lavage "is in transition rather than antiquated." As upper GI bleeding epidemiology evolves and endoscopic interventions improve, "the widespread use of nasogastric lavage as a universal piece of the upper GI bleed protocol should decrease. Whether nasogastric lavage ultimately becomes a completely unnecessary procedure remains to be seen," he said.

Physicians at his institution no longer routinely perform nasogastric lavage when evaluating suspected upper GI bleeding, but "it has not completely disappeared from practice, either," he said. That’s less a factor of "aggressive" physicians and more a result of how practice changes and environmental factors, Dr. Whelan added. "Not all emergency rooms have access to full-service endoscopy on site, so emergency room physicians may have a different set of risk/benefit tradeoffs to consider."

Acute Kidney Injury

Can the fractional excretion of sodium (FE_Na) or fractional excretion of urea nitrogen (FE_UN) help narrow the differential diagnosis in acute kidney disease? Widespread use of these measures began after a 17-person study in 1976 suggested that patients with prerenal azotemia had a FE_Na of less than 1 and patients with acute tubular necrosis had a FE_Na greater than 3 (JAMA 1976;236:579-81).

The FE_Na is not perfect, because many intrinsic kidney disorders can cause low FE_Na and the FE_Na can be elevated when diuretic use contributes to prerenal states, so a few studies looked at adding the FE_UN to the diagnostic tools. Their results were contradictory.

One study of 102 patients in the ICU found that incorporating FE_UN was 85% sensitive and 92% specific in detecting prerenal injury, but the study excluded patients with acute glomerulonephritis and obstructive nephropathy, "so you have to make sure that you exclude those patients if you’re going to use FE_UN," Dr. Feldman said (Kidney Int. 2002;62:2223-9).

In a separate study of 99 patients, however, the FE_Na and FE_UN were much less impressive in patients with or without diuretics. In patients on diuretics, FE_UN had a sensitivity for distinguishing transient from persistent acute kidney injury of 79% and a specificity of 33%, and in patients not on diuretics the sensitivity was 48% and the specificity was 75% (Am. J. Kidney Dis. 2007;50:568-73).

A recent analysis reviewed the literature to provide some guidance for clinicians, but the end result is confusing, Dr. Feldman said. Under best-case scenarios, these two measures would be likely to make a difference in diagnosing the cause of acute kidney injury, but under worst-case scenarios, "they really stink," he said (Cleve. Clin. J. Med. 2012;79:121-6).

The authors cautioned that a single index calculated at a specific time often is insufficient to properly characterize the pathogenesis of acute kidney injury. "In the end, probably FE_Na and FE_UN really don’t help you very much to decide" the reason behind acute kidney injury, Dr. Feldman said.

Chest X-Rays

Routine chest x-rays in mechanically ventilated patients in ICUs provide, well, too many unneeded x-rays, recent data show.

A crossover study that randomized 21 French ICUs to either routine daily chest x-rays for these patients or x-rays on demand found that the on-demand strategy reduced the number of x-rays by 32% without affecting the number of days on ventilation, length of ICU stay, or mortality. With the daily x-ray strategy, 424 patients got 4,607 x-rays, compared with 3,148 x-rays in 425 patients under the on-demand strategy (Lancet 2009;374:1687-93).

Patients had their ventilators changed more often under the on-demand strategy, probably as clinicians were troubleshooting potential problems, but the number of interventions did not differ significantly by x-ray strategy, Dr. Feldman noted.

A meta-analysis this year of eight trials including 7,078 adult ICU patients concluded that routine daily x-rays can be eliminated without increasing adverse outcomes (Radiology 2012;255:386-95).

Dr. Feldman suggested specific goals for these four scenarios, which he presented at the annual meeting of the Society of Hospitalist Medicine.

"Do not reflexively transfuse cardiac patients to hematocrits of 30%. Do not do routine daily chest x-rays. Do not reflexively NG [nasogastric] lavage our patients. And spend more time doing a really great history and physical and thinking about why your patient has acute renal failure than trying to use indices that don’t actually help us very much," he said.

He added a personal goal: "If I can make the residents at Johns Hopkins change, that will be a real feat, because they love to order tests on everybody."

Dr. Feldman, Dr. Whelan, and Dr. VanderEnde reported having no financial disclosures.

Why did you order that test? Dr. Leonard S. Feldman wants you to turn off the autopilot and consider the evidence from the medical literature on the following four practices:

• Is the best target hematocrit for a cardiac patient 30% (or a hemoglobin level of 10 g/dL)?

• Should nasogastric lavage be routine in patients with suspected GI bleeds?

• Is it helpful to measure the fractional excretion of sodium or fractional excretion of urea nitrogen when evaluating acute kidney injury?

• Are daily chest x-rays a good idea in patients on mechanical ventilation in an ICU or step-down unit?

"Many of us do all of these things regularly on a reflexive basis," said Dr. Feldman, director of the general medicine comprehensive consult service at Johns Hopkins University, Baltimore.

Yet recent studies challenge the value of these approaches:

Transfusions

Three studies influenced the AABB (formerly the American Association of Blood Banks) to publish a clinical practice guideline this year recommending a restrictive strategy when considering blood transfusions in hospitalized patients with preexisting cardiovascular disease. The AABB says not to transfuse if the hemoglobin level is above 7-8 g/dL and to consider transfusing patients who are symptomatic or who have a hemoglobin level of less than 8 g/dL, the AABB says (Ann. Intern. Med. 2012;157:49-58).

One recent study randomized 502 patients undergoing coronary artery bypass grafting (CABG) or valve replacement who were on cardiac bypass support to either a liberal strategy aiming to maintain a hematocrit of at or above 30% from the start of surgery until discharge from the ICU or a restrictive strategy that aimed to keep the hematocrit at or above 24%.

Both groups were able to meet these goals. The two groups did not differ significantly, however, in a composite end point of 30-day all-cause mortality and severe morbidity occurring during hospitalization (JAMA 2011;304:1559-67).

There was a trend toward higher risk of death with the restrictive strategy, with a hazard ratio of 1.28, but the ratio ranged from 0.6 to 2.7, so the risk was not statistically significant. Analysis of a slew of secondary outcomes also found no significant differences between groups.

Another study randomized 2,016 patients aged 50 years or older who either had a history of cardiovascular disease or cardiac risk factors and who had a hemoglobin level below 10 g/dL after surgery for hip fracture. The liberal strategy triggered transfusion in all patients with a hemoglobin below 10 g/dL. The restrictive strategy transfused only if the patient showed symptoms of anemia or at the physician’s discretion in patients with a hemoglobin level less than 8 g/dL.

Patients received a lot more blood under the liberal strategy compared with the restrictive strategy – 1,866 vs. 652 units of blood, respectively – but the groups did not differ significantly in death rates or the ability to walk across a room without human assistance at 30 and 60 days after surgery (N. Engl. J. Med. 2011;365:2453-62). Sixty days after surgery, 35% of patients in each group had died or were unable to walk across a room unaided.

"This is a trial that you should be showing all of your orthopedists," Dr. Feldman said. "When the orthopedist says that we need to give a patient blood so the patient will be able to do better in rehab" after hip surgery, point to the data showing that this isn’t necessarily true, he suggested.

An earlier trial of 838 critically ill patients who had hemoglobin levels below 9 g/dL within 72 hours of admission to the ICU found 30-day mortality rates of 19% in those randomized to a restrictive transfusion strategy and 23% in patients randomized to a liberal transfusion strategy, a difference that was not statistically significant (N. Engl. J. Med. 1999;340:409-17).

The liberal strategy transfused when hemoglobin levels fell below 10 g/dL and maintained the hemoglobin at 10-12 g/dL. The restrictive strategy waited until hemoglobin levels fell below 7 g/dL before transfusing and maintained hemoglobin at 7-9 g/dL.

"So, we have three different studies here that really seem to indicate that for transfusion of those patients who are at highest risk – cardiac patients – it did not seem to make a difference in their outcomes, particularly in mortality, if you were restrictive or liberal" in setting thresholds for transfusion, Dr. Feldman said. "This is not the same, though, for patients who are having an acute coronary syndrome. We don’t have any good data for them," and the AABB doesn’t recommend for or against liberal or restrictive transfusion thresholds, he added.

The evidence behind the AABB recommendation is only of moderate quality because there are only the three studies. Although the literature generally favors a restrictive versus a liberal approach, there’s no robust evidence for any particular transfusion threshold, said Dr. Daniel S. VanderEnde, a hospitalist and member of the joint transfusion committee at Emory University, Atlanta.

Pointing to the limited amount of research, the AABB itself calls the recommendation’s quality of evidence "very low" and the strength of the recommendation "uncertain."

"Subpar" evidence is one reason that Dr. VanderEnde’s institution leaves transfusion decisions to individual clinicians, he said in an interview. One ICU may transfuse at a hemoglobin threshold "in the mid-7s, and another ICU will have a transfusion threshold in the mid-8s."

Emory is starting a computer order entry protocol requiring physicians to tell why they’re transfusing blood, compared with no previous oversight. "It doesn’t stop them from transfusing for any reason. It is just trying to collect data, in the hopes that maybe they will be more restrictive in their use rather than liberal," he said.

Anecdotally, transfusion practices do seem to be shifting, at least among newly-trained physicians, Dr. VanderEnde added. Five years ago when he would ask medical students about transfusion thresholds, many hewed to "the old 10/30 rule," but far fewer do so today, he said. "The younger orthopedists tend to not transfuse as much as the older orthopedists."

Nasogastric Lavage

Few procedures performed in emergency departments are more painful for patients than nasogastric intubation, and there’s a study to prove that (Ann. Emerg. Med. 1999;33:652-8).

"Patients think nasogastric lavage and nasogastric intubation really stink, so we need to have a good reason to do it," Dr. Feldman said.

Sherry Boschert/IMNG Medical Media

And, like all invasive procedures, there are risks involved, Dr. Chad T. Whelan said in an interview. There are only modestly convincing data suggesting that nasogastric lavage can provide some prognostic or "localizing" information (such as differentiating upper vs. lower bleed). "Therefore, the risk/benefit ratio of routinely performing them for all patients has shifted with our increasing understanding of their risks and benefits," said Dr. Whelan, a hospitalist at the University of Chicago.

International consensus recommendations on the management of patients with nonvariceal upper GI bleeding suggest that physicians consider placing a nasogastric tube in selected patients because the findings may have prognostic value – not very helpful advice in decision making, Dr. Feldman said. (Ann. Intern. Med. 2010;152:101-13).

The rationale until now has been that patients with bloody aspirate on nasogastric lavage are significantly more likely to have high-risk GI lesions on endoscopy, compared with patients with clear or bilious aspirates on lavage. But does knowing this improve outcomes?

One review of the literature on how to determine if a patient has a severe upper GI bleed confirmed that a bloody aspirate on nasogastric lavage increases the likelihood of an upper GI bleed but there’s only a mildly increased likelihood of a severe bleed, "and the negative likelihood ratio is not unimpressive," Dr. Feldman said (JAMA 2012;307:1072-19).

Results of a separate propensity-matched retrospective analysis of data on 632 patients admitted with GI bleeding are "as good as we’re going to get on this topic," he said. The study found that getting or not getting nasogastric lavage did not change 30-day mortality, mean length of stay, transfusion requirements, or emergency surgery rates (Gastrointest. Endosc. 2011;74:971-80).

The only things that nasogastric lavage did change were an increase in the rate of patients undergoing endoscopy, a shorter interval to endoscopy, and a shorter length of stay among patients who had endoscopy.

That suggests that there was an individual-provider confounder that the study could not measure. Perhaps emergency physicians or gastroenterologists who order nasogastric lavage are simply more aggressive, Dr. Feldman said. "This is information that you might want to take to your emergency department," he said.

In a joint editorial accompanying the study, an emergency physician and an endoscopist concluded that the practice of nasogastric lavage in patients with acute upper GI bleeding is "antiquated."

Dr. Whelan said the role of nasogastric lavage "is in transition rather than antiquated." As upper GI bleeding epidemiology evolves and endoscopic interventions improve, "the widespread use of nasogastric lavage as a universal piece of the upper GI bleed protocol should decrease. Whether nasogastric lavage ultimately becomes a completely unnecessary procedure remains to be seen," he said.

Physicians at his institution no longer routinely perform nasogastric lavage when evaluating suspected upper GI bleeding, but "it has not completely disappeared from practice, either," he said. That’s less a factor of "aggressive" physicians and more a result of how practice changes and environmental factors, Dr. Whelan added. "Not all emergency rooms have access to full-service endoscopy on site, so emergency room physicians may have a different set of risk/benefit tradeoffs to consider."

Acute Kidney Injury

Can the fractional excretion of sodium (FE_Na) or fractional excretion of urea nitrogen (FE_UN) help narrow the differential diagnosis in acute kidney disease? Widespread use of these measures began after a 17-person study in 1976 suggested that patients with prerenal azotemia had a FE_Na of less than 1 and patients with acute tubular necrosis had a FE_Na greater than 3 (JAMA 1976;236:579-81).

The FE_Na is not perfect, because many intrinsic kidney disorders can cause low FE_Na and the FE_Na can be elevated when diuretic use contributes to prerenal states, so a few studies looked at adding the FE_UN to the diagnostic tools. Their results were contradictory.

One study of 102 patients in the ICU found that incorporating FE_UN was 85% sensitive and 92% specific in detecting prerenal injury, but the study excluded patients with acute glomerulonephritis and obstructive nephropathy, "so you have to make sure that you exclude those patients if you’re going to use FE_UN," Dr. Feldman said (Kidney Int. 2002;62:2223-9).

In a separate study of 99 patients, however, the FE_Na and FE_UN were much less impressive in patients with or without diuretics. In patients on diuretics, FE_UN had a sensitivity for distinguishing transient from persistent acute kidney injury of 79% and a specificity of 33%, and in patients not on diuretics the sensitivity was 48% and the specificity was 75% (Am. J. Kidney Dis. 2007;50:568-73).

A recent analysis reviewed the literature to provide some guidance for clinicians, but the end result is confusing, Dr. Feldman said. Under best-case scenarios, these two measures would be likely to make a difference in diagnosing the cause of acute kidney injury, but under worst-case scenarios, "they really stink," he said (Cleve. Clin. J. Med. 2012;79:121-6).

The authors cautioned that a single index calculated at a specific time often is insufficient to properly characterize the pathogenesis of acute kidney injury. "In the end, probably FE_Na and FE_UN really don’t help you very much to decide" the reason behind acute kidney injury, Dr. Feldman said.

Chest X-Rays

Routine chest x-rays in mechanically ventilated patients in ICUs provide, well, too many unneeded x-rays, recent data show.

A crossover study that randomized 21 French ICUs to either routine daily chest x-rays for these patients or x-rays on demand found that the on-demand strategy reduced the number of x-rays by 32% without affecting the number of days on ventilation, length of ICU stay, or mortality. With the daily x-ray strategy, 424 patients got 4,607 x-rays, compared with 3,148 x-rays in 425 patients under the on-demand strategy (Lancet 2009;374:1687-93).

Patients had their ventilators changed more often under the on-demand strategy, probably as clinicians were troubleshooting potential problems, but the number of interventions did not differ significantly by x-ray strategy, Dr. Feldman noted.

A meta-analysis this year of eight trials including 7,078 adult ICU patients concluded that routine daily x-rays can be eliminated without increasing adverse outcomes (Radiology 2012;255:386-95).

Dr. Feldman suggested specific goals for these four scenarios, which he presented at the annual meeting of the Society of Hospitalist Medicine.

"Do not reflexively transfuse cardiac patients to hematocrits of 30%. Do not do routine daily chest x-rays. Do not reflexively NG [nasogastric] lavage our patients. And spend more time doing a really great history and physical and thinking about why your patient has acute renal failure than trying to use indices that don’t actually help us very much," he said.

He added a personal goal: "If I can make the residents at Johns Hopkins change, that will be a real feat, because they love to order tests on everybody."

Dr. Feldman, Dr. Whelan, and Dr. VanderEnde reported having no financial disclosures.