Have you ever paused before ordering a white blood cell (WBC) count for a patient with a probable infection, knowing its inaccuracy for predicting bacterial infection, but then send it anyway? Physicians gain great comfort from tests, even tests that perform badly. We use WBC results to support our suspicions, and when they don't, we often ignore them, but occasionally we're led to wrong decisions. The WBC count doesn't even help you know who might get sicker: it isn't an independent predictor of mortality in the Pneumonia Severity Index (PSI).
But what if there were a test better than the WBC, better at predicting bacterial infection, antibiotic need, and prognosis? Some claim that procalcitonin (PCT) is just such a test. Increasingly, labs are considering this rapid turnaround test (about an hour), so PCT could be coming to your ED soon. But before we sign up for yet another test, let's see if there's evidence to support its use.
PCT is a calcitonin precursor released from the thyroid in response to hypercalcemia. Normally, it's barely detectable. In response to systemic bacterial infection, however, PCT is released within hours in large amounts from various tissues. Viral infections do not generally elicit PCT, which increases with infection severity and decreases as patients stabilize, as opposed to C-reactive protein (CRP), which stays elevated after clinical improvement.
In 2008, BRAHMS USA, Inc., received U.S. Food & Drug Administration approval of its PCT Kryptor test that serially assesses risk for progression to severe sepsis and septic shock for ICU patients over their first 24 hours. While interesting, unless your ICU admission process has major problems, this test feature would not help you much in the ED. PCT levels that increase to >2.0 ng/ml represent a high risk for progression to severe sepsis and septic shock, while levels <0.5 ng/ml represent low risk, and levels <0.3 ng/ml represent a healthy condition (www.brahms-usa.com/clinical_performance.htm.) Low levels do not exclude infection, and can occur in localized bacterial infections without systemic signs.
“PCT can improve critical decisions ICU physicians have to make every day, including whether or not to administer antibiotics,” said Eric H. Gluck, MD, the director of critical care services for Swedish Covenant Hospital in Chicago. (www.medicalnewstoday.com/articles/103359.php.) While PCT was approved to help predict progression to severe sepsis and septic shock, Dr. Gluck goes one step further, asserting that PCT can help decide which patients need antibiotics. We certainly make antibiotic decisions in the ED, so let's see if PCT could be useful to emergency physicians.
One potential use for PCT in the ED would be to assist in deciding if a patient has a viral infection or a bacterial infection requiring antibiotics. Studies have generally found PCT's performance characteristics — sensitivity and specificity — for detecting serious bacterial infections to be superior to those of the WBC count and as good or better than CRP at any given cutoff value. Tests are compared using a receiver operator curve (ROC) with greater area under this curve (AUC), generally corresponding to better test performance.
It's well-recognized, for example, that clinical evaluation of febrile infants under 90 days old is difficult. Various risk-stratifying strategies have been created to detect occult bacteremia and UTIs, such as the Philadelphia and Rochester criteria, which look to exclude bacterial infection in low-risk, well-appearing infants. These use WBC count or absolute band count thresholds. In a study of 234 young infants in which 32 (13%) had definite bacterial infection (bacteremia, UTI), the AUC of PCT was 0.82 compared with 0.74 for absolute neutrophil count and 0.66 for WBC count. (Pediatrics 2008;122:701.) At a 0.12 ng/ml cutoff, PCT had a sensitivity of 95%, specificity of 25%, and negative predictive value of 95%.
Other studies in young children, using various PCT cutoffs (i.e., 0.5 and 0.9 ng/ml), have found PCT and CRP to have similar performance but PCT superior in identifying invasive infections, including within the first eight hours of fever. The Philadelphia and Rochester criteria have high sensitivity and negative predictive values (∼97%), but it's unclear if these guidelines might perform better overall if PCT were substituted for the WBC components. Note that for this and other diagnostic scenarios, the optimal PCT cutoff has not been established. Recent reviews of PCT in adults concluded that PCT was superior to CRP to distinguish bacterial and viral infections (Clin Infect Dis 2004;39:206), but had only moderate diagnostic performance for bacteremia. (Ann Emerg Med 2007;50:34.)
PCT also has been studied as a way to discri-minate between bacterial and aseptic meningitis. With current vaccinations, bacterial meningitis has become relatively rare. In one large U.S. study, only 3.7 per-cent of children having an LP and found to have pleocytosis had bacterial meningitis, but 83 percent were hospitalized. (Pediatrics 2008;122:726.) Not infrequently, patients who received antibiotics before an LP, with subsequently sterile CSF cultures, get a full course of empiric antibiotics even though probably most have viral disease.
A review of pediatric meningitis studies found that PCT had superior performance compared with CRP, CSF protein, and neutrophil count, with AUC levels of 0.98, 0.89, 0.88, and 0.87, respectively. (Arch Pediatr Adolesc Med 2008;162:1157.) While these results may reflect publication bias, they are nonetheless impressive. Like for febrile young infants, a risk-stratification system has been developed called the Bacterial Meningitis Score. (Pediatrics 2008;122:726.) It consists of:
- Positive CSF Gram stain.
- >1,000 CSF polys/mm3.
- CSF protein >80/uL.
- Peripheral WBC count >10,000/mm3.
- Associated seizure.
In a validation study of this score, only two of 1,714 (0.1%) children with a score of 0 had bacterial meningitis, that is, a negative predictive value of 99.9%. Studies of PCT in adults with meningitis have not been as consistent, and cases of bacterial meningitis with initially low PCT levels have been reported in all age groups. Whether PCT can better discriminate among the full range of patients and CSF findings remains to be seen.
PCT has been investigated in several other clinical scenarios. PCT was found to be better than CRP, ESR, and WBC count for predicting which children with UTI had pyelonephritis (Pediatrics 2004;114;e249), but had poor performance in identifying bacterial infection among cirrhotics with ascites (Hepatogastroenterology 2001;48:502), patients suspected of endocarditis (Am J Med 2010 Sep 24. [Epub ahead of print]), and for diagnosis of appendicitis. (Eur Surg Res 2009;43:291.)
PCT appears to be better than WBC count and other available biomarkers to identify or exclude serious and systemic bacterial infections. It's logical to presume PCT replacement of the WBC count could enhance some risk-stratification schemes, such as for febrile young infants and meningitis. At this point, however, PCT has not been shown to supplement our current clinical evaluation. Medicare reimbursement is approximately $27 for PCT and about $10 for a CBC, so if PCT were used as indiscriminately as WBC count, there is potential for substantial additional costs.
In addition to being used as a marker to determine if antibiotics should be started, PCT has also been proposed as a means to determine when antibiotics should be stopped. Serial evaluations of PCT appear to be prognostic. This characteristic has led to a provocative study in JAMA that suggested that PCT could significantly reduce the duration of antibiotics in adults treated for community-acquired pneumonia. (CAP; JAMA 2009;302: 1059.) Let's see what the buzz is all about.
In the ProHOSP study, Schuetz and colleagues compared outcomes and antibiotic use among 1,359 adults with lower respiratory tract infections, 68 percent with CAP (others mainly had bronchitis), who were randomized to standard care or care guided by PCT levewls. Doctors were “strongly encouraged” to stop antibiotics if the patient's PCT level was <0.1 ng/ml, and “encouraged” to stop them if the level was <0.25 ng/ml or dropped 80 percent. PCT measurements were done in real time every two days for hospitalized patients with initially high levels. If antibiotics were withheld, PCT levels were repeated after six to 24 hours to ensure they stayed low. Alternatively, antibiotics were recommended at higher PCT levels.
In the control group, antibiotics were recommended for five to 10 days in uncomplicated CAP and longer in complicated cases consistent with published guidelines at the time. The primary clinical outcome was the proportion of those experiencing death, ICU, and readmission, and CAP-related complications. The secondary outcome was antibiotic use.
For the purpose of this discussion, I'll focus on CAP patients. Subjects were older (median, 73 years), 93 percent were hospitalized, and about 50 percent were PSI IV-V. Compliance with PCT guidance was approximately 90 percent. Clinical outcome was statistically similar between the PCT and control groups, 16.1 percent vs. 20.2 percent (95% CI of difference, -9.1% to 0.9%). Median antibiotic days were less in the PCT-guided vs. control groups, seven vs.10 (95% CI of mean % change, -37.6 to -26.9), with reduction in antibiotic side effects of about 10 percent. The investigators stated, “PCT guidance will have substantial clinical and public health implications to reduce antibiotic exposure.”
A closer look, however, raises doubts about this conclusion. There's good reason to believe that five days of antibiotics is sufficient for most CAP patients regardless of PCT testing, which is shorter than median durations in both study groups. In 2003, levofloxacin (off patent in June 2011) was FDA-approved as a five-day, 750 mg daily regimen for CAP, and a large study found similar outcomes with five and 10 days of ceftriaxone. (Med Mal Infect 2002;32:369.) Most CAP patients stabilize within 72 hours and current ATS/IDSA CAP guidelines also recommend antibiotic discontinuation for these patients after five days of treatment. (Clin Infect Dis 2007;44[Suppl 2]:S27.) The study relied on older guidelines that did not address antibiotic discontinuation. One wonders how different antibiotic use would have been had study clinicians been better informed or required to treat according to shortest duration consistent with current guidelines. Although emergency physicians aren't usually involved in discontinuing antibiotics, we can facilitate less antibiotic exposure by prescribing five-day regimens and close follow-up.
While it's reassuring that the serious outcomes studied — death and ICU admission, for instance — were similar between groups, probably most patients who experienced these were on antibiotics regardless of their care assignment. And because neither clinicians nor patients were blinded to the treatment assignment, assessment of outcomes may have been biased as well.
I suspect that serial clinical evaluation is as good as or better than PCT in predicting antibiotic need, but doctors are much more likely to discontinue antibiotics if they have the excuse of a normal test result. I have a theory that antibiotics can be stopped when a patient says his appetite is back. Call it Dave's Cheeseburger Test (DCT), which is only a few bucks at McDonald's. I have even heard proponents advocate for doctors to use PCT to justify not prescribing antibiotics for acute bronchitis, an infection for which multiple trials have already conclusively shown that antibiotics are of no value. Shame on them.
Even for PCT's approved indication in the ICU, no one knows whether survival of critically ill patients can be improved by using PCT-guided care and if a decreasing PCT level identifies someone who no longer needs antibiotics or one particularly sensitive to their salutatory effects in whom antibiotics should be continued. The value of PCT in the ED currently remains speculative, and the potential for abuse, if the WBC count is any indication, great.
PCT must now be viewed in the context of simultaneous introduction of rapid and accurate molecular diagnostic testing for viral infections and soon bacteria. In one study, 64 percent of 1,142 children with meningitis had a positive enteroviral test, and none was found to have bacterial meningitis. (Clin Infect Dis 2010;51: 1221.) I suspect that most clinicians would rely on a specific positive enteroviral test (two are FDA-approved with turnaround times of three to four hours) over a low PCT to feel comfortable discharging a child with meningitis from the ED. The effectiveness of this approach has already been demonstrated. (Arch Pediatr Adolesc Med 2000;154:817.) Rapid multiplex PCR of nasal specimens for respiratory viruses already are available, and broad-based PCR assays of blood and CSF for pathogenic bacteria are around the corner.
Just like PCT, we'll need to see if these tests lead to true benefits for patients and not just false courage for doctors. For now, while PCT looks promising and many studies are ongoing, it does not have convincingly demonstrated utility. PCT is a test still looking for a home.
Comments about this article? Write to EMN at firstname.lastname@example.org.
Click and Connect! Access the links in this article by reading it on www.EM-News.com.