Measurement of Serum Procalcitonin: Title and subTitle BreakA Step Closer to Tailored Care for Respiratory Infections?

The advent of antibiotics led to dramatic reductions in mortality and medical complications from bacterial respiratory tract infections.¹ Although the armamentarium of diagnostic and therapeutic approaches has increased exponentially since that time, the effect of these advances on patient outcomes is less dramatic. Nevertheless, the goal of care remains constant: to match the patient's illness with a treatment approach that optimizes recovery while causing no harm. Harm exists in at least 2 forms: patient-specific, from an adverse event due to a diagnostic test or a treatment,² and population-based, from the development of bacterial resistance promoted by the misuse or overuse of antibiotics.^{1 ,3}

Lower respiratory tract infections (LRTIs) are a pervasive public health problem and cause more disease and death in the United States than any other infection.⁴ Patients and their physicians share a goal of improving symptoms from LRTIs as quickly as possible, often viewing antibiotics as the most expeditious intervention to achieve this goal. This one-size-fits-all approach for patients with illnesses ranging from acute bronchitis to community-acquired pneumonia (CAP) fails to consider the basic questions of who would benefit (or be harmed) from antibiotic therapy, and if treated, what is the optimal duration.

Answering these questions requires an accurate assessment of which patients have bacterial infections and how their illness might progress with or without treatment. Many conventional bacterial detection techniques used in LRTIs, such as sputum Gram stains and cultures, lack sufficient diagnostic accuracy and turnaround time to inform real-time decisions.^{5 - 6} In the setting of a severe LRTI with evidence of organ derangement, assessing the benefits and risks of treatment often favors empirical antibiotic therapy. However, such clinical scenarios are uncommon; the vast majority of patients have mild to moderate illness, rendering this assessment of cause and response more challenging. Clinical decision rules based on bedside clinical and laboratory data could help physicians improve antibiotic management for such patients and optimize the duration of therapy. The ultimate clinical value of such rules is a function of diagnostic accuracy, effectiveness in changing clinician behavior, safety for patients, and both individual and societal cost.⁷

In this issue of JAMA, Schuetz and colleagues⁸ build on prior work^{9 - 12} by studying the effectiveness and safety of using a novel decision rule to guide antibiotic therapy of patients diagnosed with LRTIs ranging from acute bronchitis to exacerbation of chronic obstructive pulmonary disease (COPD) to CAP. In their current study, the authors evaluated 1359 patients presenting to emergency departments in 6 tertiary care hospitals who were randomized to receive care guided by evidence-based LRTI guidelines (control group) or by an experimental algorithm that used serum procalcitonin (PCT) to quantify the likelihood of a bacterial infection and gauge the response to therapy. PCT is released in response to bacterial infection and correlates with illness burden and severity; furthermore, it is rarely elevated in patients with viral infection.¹³ The algorithm specified 1 of 4 antibiotic treatment recommendations, ranging from strongly discourage to strongly recommend, based on the measured PCT levels. For hospitalized patients, repeat PCT measurements guided continuation of antibiotic therapy.

In contrast to many effectiveness trials that seek to change clinician behavior through the implementation of decision rules,¹⁴ compliance with treatment recommendations in this trial were commendably high (79.4% in the control group and 90.8% in the PCT group). Overall, the PCT-guided strategy led to reduced antibiotic exposure by an average of 3 days (8.7 vs 5.7 days) and led to 8.2% less antibiotic-related adverse effects (28.1% vs 19.8%), with slightly lower overall adverse event rates in the PCT group (15.4% vs 18.9%). Although similar outcomes were observed for patients with acute bronchitis, COPD exacerbation, and CAP, the paths to reducing antibiotic exposure differed in a predictable fashion. For acute bronchitis, decreased antibiotic exposure was principally from nontreatment, whereas for CAP it was principally from reduction in duration of therapy.

Even though this PCT-guided decision rule holds promise for the management of patients with LRTI, several issues must be carefully considered before broadly translating this research into clinical practice. First, the study had a high proportion of patients with pneumonia (68.1%), more than half of whom were considered high risk (pneumonia severity index risk classes IV or V).¹⁵ In such patients, the high likelihood of bacterial disease and high disease acuity render PCT guidance unlikely to alter the decision to initiate antibiotic therapy and unlikely to augment the standard severity assessment of the patient.¹⁶ Decision rules, like diagnostic tests, are best studied and most useful for patients with intermediate pretest probabilities of having the disease of interest. For example, application of a PCT-guided approach to antibiotic therapy is likely to have the greatest effect in the subset of patients with an intermediate risk of bacterial LRTI, such as those with acute exacerbations of COPD or patients with underlying cardiopulmonary disease who present with symptoms compatible with CAP.

Second, the trial by Schuetz et al⁸ was conducted exclusively at Swiss tertiary care hospitals with teaching programs. More generalizable data on the effectiveness and safety of the intervention in less intensive, nonteaching facilities and in differing locations (including North America) are required prior to broad adoption of this management approach.

Third, the safety assessment of PCT-guided management was based on a composite end point of adverse events within 30 days of enrollment, ranging from recurrent LRTI to death. This approach may not completely allay the concerns of patients or physicians regarding widespread use of this decision rule. While the noninferiority margin of the PCT-guided approach based on the study findings for this composite outcome is reassuring, the slightly higher mortality in the PCT group (5.1% vs 4.8% in controls) is consistent with an absolute mortality difference of up to 2.5%. Moreover, although outcomes assessment at 30 days is justified for the study of an acute illness such as LRTI, a longer surveillance interval would assuage concerns about late complications or death associated with less antibiotic use.

Fourth, other pragmatic questions regarding implementation of a PCT-guided decision rule to manage LRTI require consideration. The current trial did not deploy clinician behavior-change techniques of proven effectiveness, such as reminders or audit and feedback, in either study group.^{17 - 18} The investigators did not use aggressive time-to-clinical-stability instruments shown to reduce duration of parenteral antibiotic therapy to guide antibiotic discontinuation in the control group.¹⁹ It is unclear how incorporation of these state-of-the-art components might have affected the study findings.

Fifth, the investigators did not assess the costs of PCT testing, implementation of the study interventions, or overall medical care. The issue of whether the likely cost-savings attributed to reduced antibiotic use and reduced antibiotic-related adverse effects would outweigh the accrued costs of PCT testing and implementation of the decision rule remains undefined. Another issue is how the cost-effectiveness of the PCT-guided rule compares with other methods to reduce the use of antibiotic therapy for LRTIs.^{20 - 22} Future work should address these important questions.

In summary, Schuetz and colleagues⁸ have charted the waters for more tailored management of LRTIs by demonstrating that a PCT-guided decision rule safely diminishes antibiotic use and adverse effects in patients with such illness. In the future, an increasing number of such “theragnostic” approaches are likely to be possible in which blood samples or tissue specimens can be used to quickly measure microbial fragments, circulating markers of organ stress and system responses, and genetic patterns that predict clinical outcomes, drug effectiveness, or both. PCT-guided care is an initial step toward such a tailored approach that could lead to more appropriate antibiotic therapy for patients with LRTI, while promoting antibiotic stewardship for the entire population.