Don't Order a Test Just Because You Can

Benjamin T. Galen, MD; Robert L. Fogerty, MD, MPH; Manisha Juthani-Mehta, MD

Disclosures

South Med J. 2015;108(5):251-253. 

Community-acquired pneumonia (CAP) is a clinical diagnosis based on typical features, including fever, cough with sputum production, and pleuritic chest pain, supported by physical examination findings of lung consolidation, laboratory evidence of leukocytosis or bandemia, and new infiltrates on chest imaging.[1] Urinary antigen tests for Legionella pneumophila serotype 1 and Streptococcus pneumoniae are used frequently in the inpatient evaluation of patients with suspected CAP.[2,3] These urinary antigen assays have the advantage of being obtained noninvasively and with rapid results; however, given their widespread availability and ease of acquisition, these molecular tests run the risk of being overused at a cost that potentially outweighs their benefit. As with all diagnostic tests, providers should be aware of disease prevalence and test characteristics when ordering and interpreting urinary pneumonia antigen tests. Careful consideration of the limited impact that these assays have in the management of routine cases of CAP also may prevent overuse.

The Centers for Disease Control and Prevention estimates that there are 900,000 cases of S. pneumoniae pneumonia annually in the United States. L. pneumophila, on the contrary, is the predominant cause of at most 18,000 cases annually of Legionnaires disease in the United States, the syndrome of pulmonary involvement from this organism.[4] A sporadic and epidemic phenomenon, Legionnaires disease comprises only 3% of pneumonia cases in adults older than age 50, whereas S. pneumoniae is the most common cause of CAP.[1,3,5]

The sensitivity of the urinary L. pneumophila serotype 1 assay is 74% with a specificity of 99%.[3] The similarly high specificity of S. pneumoniae urinary antigen testing (96%) yields few false-positives but sensitivity also is limited to approximately 70%.[5] Depending on an individual's pretest probability of infection by these two microbes, these relatively low sensitivities may not clinically exclude CAP caused by them, given the intrinsic rate of false-negative test results. Furthermore, true-negative urinary antigen test results do not rule out bacterial pneumonia because more than 60% of CAP is caused by etiologies other than S. pneumoniae and L. pneumophila.[5] As a test for Legionnaires disease, the commercially available urinary antigen test is limited further by its failure to detect species or serotypes other than L. pneumophila serotype 1.[3]

The latest Infectious Diseases Society of America (IDSA)/American Thoracic Society guidelines recommend (level II, moderate) sending urinary antigen assays for L. pneumophila and S. pneumoniae in critically ill patients with severe CAP, defined primarily as criteria for admission to the intensive care unit. These guidelines define severe CAP as either of two major criteria (a requirement for mechanical ventilation or shock with vasopressors) or three of the following minor criteria: respiratory rate >30 breaths per minute, arterial oxygen pressure/fraction of inspired oxygen (PaO2/FiO2) ratio >250, multilobar infiltrates, confusion, blood urea nitrogen level >20 mg/dL, leukopenia resulting from infection, thrombocytopenia, hypothermia, or hypotension requiring aggressive fluid resuscitation.[6] We agree with these recommendations, but believe that clinicians are not adhering to these guidelines, either because of a lack of awareness of the recommendations or because of the allure of noninvasive, easily acquired tests. Performing both of these tests on every routine case of clinically diagnosed CAP is unnecessary, especially when the patient is neither severely ill nor failing empiric therapy. The IDSA/American Thoracic Society guidelines do not recommend routine use of these tests outside the subset of patients with severe CAP, although a benefit in nonresponders is speculated.[6]

Within one large urban health system with 1491 beds, the L. pneumophila or S. pneumoniae urinary antigen assays were each sent more than 4000 times in 2013.[7] Specifically for S. pneumoniae, positive results were obtained approximately 5% of the time in this cohort and in another retrospective cohort by Radigan et al.[7] We suspect that this low positivity rate results from the assay's poor sensitivity (a substantial number of false-negative results) and misuse of the assay in patients with a low likelihood of CAP. For example, a single institution in France studied the inpatient use of the S. pneumoniae urinary antigen test and found that more than 50% of the time the ordering provider did not believe the patient had pneumonia when the test was ordered.[8]

The relatively low sensitivity of the S. pneumoniae urinary antigen testing was confirmed in a prospective study of patients in Spain who were admitted with CAP and a blood culture positive for S. pneumoniae and subsequently underwent urinary antigen testing.[9] As expected for patients with invasive pneumococcal disease, many of the patients in this study would have met IDSA recommendations for urinary antigen testing, yet only 75% tested positive. This cohort of patients had high disease severity and 25% still had false-negative urinary antigen testing.

Colloquially, these tests can be mistakenly referred to as "urine pneumonia screens," which only serves to perpetuate their overuse, particularly in patients for whom the diagnosis of CAP is equivocal. Diagnostic uncertainty in cases of potentially multifactorial dyspnea is a common clinical scenario. Acute decompensated heart failure, acute exacerbations of chronic obstructive pulmonary disease/asthma, and bacterial pneumonia or some combination thereof may coexist in patients at risk for all three diagnoses. For comparison, B-type natriuretic peptide and N-terminal-pro-B-type natriuretic peptide tests with sensitivities up to 96% are clinically useful in excluding the diagnosis of heart failure in patients presenting to the hospital with dyspnea.[10] Congestive heart failure is a common problem, but Legionnaires disease is uncommon. The pretest probability for L. pneumophila pneumonia in a patient with CAP is typically low. Using Bayesian reasoning, the probability of Legionnaires disease in a patient with dyspnea (for whom the diagnosis of pneumonia is equivocal) is likely low enough not to warrant further testing. As stated above, expert guidelines recommend the use of these tests in a subset of patients with CAP, not as generic tests for all patients with respiratory distress.[6]

How clinicians should change testing practices during a known or potential L. pneumophila outbreak is a different question. We believe that if the features of the clinical presentation of CAP (eg, hyponatremia, multilobar infiltrates, absence of sputum) or a known local outbreak raise the pretest probability for Legionnaires disease, then a diagnostic workup for this uncommon condition may be indicated. Clinical prediction rules have been developed to help clinicians recognize potential cases.[3] The high specificity of the urinary L. pneumophila assay for the most common serotype makes this an attractive test to rapidly help diagnose Legionnaires disease in conjunction with the gold standard of a sputum or pleural fluid culture.[3]

Although the diagnosis of Legionnaires disease is of epidemiologic importance, the treatment is similar to empiric regimens for CAP, including fluoroquinolone or macrolide therapy. Neither urinary antigen assay provides antimicrobial sensitivities. For hospitalized cases of CAP in immunocompetent adults, the result of an S. pneumoniae urinary antigen assay does not provide a clear basis either for narrowing antibiotic coverage from empiric CAP regimens or reducing the duration of therapy. Empiric local regimens for CAP already take into account known S. pneumoniae β-lactam resistance in each community. In fact, a prospective randomized study of patients in Spain hospitalized with CAP failed to show either an outcome-related benefit or economic benefit to targeted therapy (based on these two urinary antigen tests) compared with empiric treatment.[11] The results of the L. pneumophila or S. pneumoniae urinary antigen assays do not appear to change management in routine cases.

Urinary pneumonia antigen testing may be beneficial in additional subsets of patients who are yet to be studied. For instance, patients with healthcare-associated pneumonia also are at risk for S. pneumoniae, but the test has not proven to be effective for antibiotic stewardship in this population.[7] The benefit of urinary pneumonia antigen testing in antibiotic-allergic patients or patients who do not respond to initial therapy remains to be clarified. Furthermore, evidence-based recommendations for L. pneumophila urinary antigen testing in conjunction with validated clinical prediction rules would be of great use to clinicians.

In routine cases of CAP, judiciously omitting these tests will likely not affect the clinical outcome; however, overuse of these molecular assays comes at a cost to the healthcare system. In the ongoing healthcare fiscal crisis, sending a test because we can is a procedure that should be more widely examined, and urinary antigen testing is a prime example. Although these tests are relatively inexpensive (Medicare reimbursement is $16.49 per test), they represent an opportunity to change physician test ordering practices so as to reduce waste.[12] Advocating for the prudent use of diagnostic tests is in keeping with the mission of the American Board of Internal Medicine Foundation's Choosing Wisely campaign.[13]

When caring for patients with CAP, clinicians should carefully consider whether sending both urinary pneumonia antigen tests is indicated and if the result will change their management. Clinicians also should keep in mind that the current guidelines recommend use only in cases of severe CAP. Further studies are required to demonstrate clinical scenarios in which L. pneumophila or S. pneumoniae urinary antigen assays are cost-effective for patients with CAP.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....