Testing the tests

An expert gives advice on which tests may be most helpful in diagnosing intensive care patients.

During an Internal Medicine 2011 session on diagnostic testing in the ICU, Darryl Y. Sue, FACP, revealed his conclusion before he'd even begun his lecture.

“Everything's going to turn out to be helpful; nothing's going to turn out to be perfect,” said Dr. Sue, who is a professor of medicine at the University of California, Los Angeles.

Darryl Y Sue FACP Photo by Kevin Berne
Darryl Y. Sue, FACP. Photo by Kevin Berne.

Despite that warning, Dr. Sue went on to provide attendees with evidence and education about which of the imperfect available tests may be most helpful in diagnosing intensive care patients.

Volume depletion

The first diagnostic dilemma under discussion was whether a patient is hypovolemic. “It sounds like something that might be relatively easy,” Dr. Sue said.

Yet research has shown that the easiest method for assessing volume depletion—the physical exam—is not very effective. A 1999 study published in the Journal of the American Medical Association (JAMA) found that exam findings such as dry axillae, sunken eyeballs, longitudinal furrows on the tongue and dry mucous membranes were not strongly predictive of hypovolemia. “They are not very powerful tools and we all know that patients in the ICU have many other reasons that they could have these abnormalities,” said Dr. Sue.

Presumably the standard against which those study authors evaluated the effectiveness of physical exam is a much better tool. However, that would be the volume challenge and it isn't always possible, according to Dr. Sue. He reported on a study, published in Critical Care Medicine in February, which found that septic patients who got the most fluid in the ICU had the worst survival and patients who got the least fluid had the best survival.

“You could say that's because the sickest patients needed more fluid, but in fact it's entirely possible that how much fluid they got was unrelated to how much fluid they needed,” Dr. Sue said.

Although fluid replacement based on central venous pressure (CVP) monitoring seems to be effective in early septic shock, the evidence may not support longer continuation of this treatment, according to Dr. Sue, who noted that the same study found no relationship between CVP and fluid balance by the fourth day of hospitalization.

“Yet, at least in our ICU, we continue to give these people fluid every time they drop their pressure or they drop their urine output,” he said. That practice may need to be rethought in light of the new evidence indicating that more fluid may actually harm such patients.

Instead, physicians may want to consider both old and new methods of measuring dynamic change. An old strategy is passive leg raising, in which a patient is moved from a legs-horizontal, torso-elevated position to torso horizontal, legs elevated. “I remember doing this when I was a house officer,” said Dr. Sue. “It's an old idea that may be coming back.”

Patients who had a 9% or more change in stroke volume from the leg raising were likely to respond to fluid in a March 2010 study from Critical Care Medicine. Measuring stroke volume might be impractical, but changes in pulse pressure can also be used, Dr. Sue noted. Variations in pulse pressure (of more than 12.5%) with respiration can also predict response to fluid, he said.

The new method for measuring volume depletion requires ultrasound. Patients who have a greater than 12% change in the diameter of the inferior vena cava with each breath are more likely to respond to fluid, researchers have found. “As the ventilator kicks in and gives the positive pressure breath, the inferior vena cava diameter increases. If the change is fairly large, it suggests that the patient is volume depleted,” said Dr. Sue.

This technique will work in both ventilated and spontaneously breathing patients (although the changes in pressure are opposite), but may not be effective with lower tidal volumes or obesity. “I did this the other day and it worked extremely well in one patient and it worked terribly in another,” said Dr. Sue.

Finally, there's one very old-fashioned method that may be a good measure of volume depletion in some patients. “If you ask people who can tell you if they're thirsty, that works pretty well,” Dr. Sue said, “as long as they are not water-depleted (hypernatremic).”


Even the most communicative patients, however, are unlikely to be able to tell you whether they have an infection. And your clinical judgment alone may be ineffective as well. “I thought in my practice I'm pretty good at telling whether a patient is infected. It turns out I'm not that good,” said Dr. Sue.

Better identification of infected patients could allow physicians to start appropriate antibiotics earlier (improving outcomes) and discontinue unneeded antibiotics sooner (reducing antibiotic resistance and saving money).

The biomarker procalcitonin is showing some potential to help with this task, Dr. Sue reported. It is a peptide precursor of calcitonin that's produced by thyroid parafollicular cells and neuroendocrine cells of lungs and intestines, and it can be measured relatively inexpensively with a point-of-care test.

A study of pneumonia patients published in JAMA in September 2009 found that the marker could be used to decrease the duration of antibiotic therapy with no negative impact on outcomes. “This is a study showing that we can get away a little bit with adjusting antibiotics based on our clinical judgment plus a measurement of procalcitonin,” said Dr. Sue.

The test may also help with identifying patients with systemic inflammatory response syndrome (SIRS) resulting from an infection. “Very, very many patients in the ICU have SIRS yet few of them, relatively speaking, have infections,” Dr. Sue said.

Research has shown that daily procalcitonin tests may pinpoint which patients are at risk for sepsis, he reported. The test appears to help with this even for trauma patients. This group has been shown to have high procalcitonin levels at the start of hospitalization that drop but then go up again in those who develop septic complications—a day before the sepsis is identified. “Maybe there procalcitonin would be able to help us make a decision about starting antibiotics or just watching and seeing what happens,” said Dr. Sue.

Of course, procalcitonin measurements would need to be used in conjunction with clinical judgment, he cautioned. “The idea that there is a magic test that will identify everything so we don't have to see the patient isn't likely. It's going to be in combination,” Dr. Sue said.

Troponin and BNP

Clinical judgment should come into play before a physician even orders testing of troponin and B-type natriuretic peptide (BNP). Dr. Sue described the common scenario where a patient who was hospitalized for a non-cardiac cause is found to have elevated levels of the biomarkers. “Now what do you do? Do you cath him? Do you do an echo? Do you get a cardiology consult?” asked Dr. Sue. “Aren't you kicking yourself for ordering it?”

Recent research has shown that elevated troponin levels are associated with worse outcomes. The tricky part is that many of the outcomes are long-term. Studies of patients with gastrointestinal bleeds and respiratory disease have found that patients with elevated troponin have higher mortality years after hospitalization.

“I'm not exactly 100% sure what I'm going to tell this patient who has this elevated troponin that doesn't seem to be perfectly related to why he's in the ICU,” Dr. Sue said. “That you have a worse five-year survival? Maybe.”

Research on patients with acute pulmonary embolism has shown a need to respond more urgently: Elevated troponin, as well as BNP, predicts higher in-hospital mortality in these patients. The sensitivity of these tests for predicting poor outcome is high, so if neither is elevated, you usually don't need to do further studies to evaluate risk, according to Dr. Sue.

There are a number of caveats to consider when evaluating BNP levels, however. “Atrial fibrillation increases BNP, so you might have to set a different cutoff value. Renal failure ditto. And obesity decreases BNP so the BNP has to be adjusted,” said Dr. Sue.

While its main application is in heart failure, BNP has been shown to predict mortality in septic shock patients. To some extent, findings like this, of elevated biomarkers in sicker patients, should be obvious, Dr. Sue noted. “I'm shocked to hear that patients who have more abnormal tests may do more poorly,” he joked.

While the troponin and BNP may have good prognostic value, that doesn't necessarily mean that they will be helpful for diagnosis, especially in patients with multiple disorders.

“The usual advice is don't order if you don't know what to do with the result,” Dr. Sue concluded. “It's still going to be useful to take the history, do the physical exam, and then decide how to use lab tests.”