Procalcitonin Algorithms for Antibiotic Therapy Decisions: A Systematic Review of Randomized Controlled Trials and Recommendations for Clinical Algorithms | Clinical Pharmacy and Pharmacology | JAMA Internal Medicine | JAMA Network
[Skip to Navigation]
Figure 1. Flowchart depicting the process of article selection. RCT indicates randomized controlled trial.

Figure 1. Flowchart depicting the process of article selection. RCT indicates randomized controlled trial.

Figure 2. Mortality rate in the procalcitonin (PCT) and control groups. The forest plot depicts the Peto odds ratios (ORs) with 95% confidence intervals (CIs), comparing patients treated in the PCT and control groups in different trials and clinical settings.

Figure 2. Mortality rate in the procalcitonin (PCT) and control groups. The forest plot depicts the Peto odds ratios (ORs) with 95% confidence intervals (CIs), comparing patients treated in the PCT and control groups in different trials and clinical settings.

Figure 3. Proposed algorithms for use of procalcitonin (PCT) values to determine antibiotic treatment of infections. A, An algorithm for low-acuity nonpneumonic infections (ie, low risk) in primary care and emergency department (ED) settings. B, Algorithm for moderate-acuity pneumonic infections (ie, moderate risk) in hospital and ED settings; C, Algorithm for high-acuity infections (ie, high risk; sepsis) in intensive care unit settings. Abx indicates antibiotics; COPD, chronic obstructive pulmonary disease; GOLD, Global Initiative for Chronic Obstructive Lung Disease; PSI, pneumonia severity index.

Figure 3. Proposed algorithms for use of procalcitonin (PCT) values to determine antibiotic treatment of infections. A, An algorithm for low-acuity nonpneumonic infections (ie, low risk) in primary care and emergency department (ED) settings. B, Algorithm for moderate-acuity pneumonic infections (ie, moderate risk) in hospital and ED settings; C, Algorithm for high-acuity infections (ie, high risk; sepsis) in intensive care unit settings. Abx indicates antibiotics; COPD, chronic obstructive pulmonary disease; GOLD, Global Initiative for Chronic Obstructive Lung Disease; PSI, pneumonia severity index.

Table 1. Quality of Studiesa
Table 1. Quality of Studiesa
Table 2. Overview of Design and Content of the RCTs Grouped by Study Setting
Table 2. Overview of Design and Content of the RCTs Grouped by Study Setting
Table 3. Primary and Secondary Outcomes of the Different RCTs, Grouped by Study Setting
Table 3. Primary and Secondary Outcomes of the Different RCTs, Grouped by Study Setting
Aug 8 2011

Procalcitonin Algorithms for Antibiotic Therapy Decisions: A Systematic Review of Randomized Controlled Trials and Recommendations for Clinical Algorithms

Author Affiliations

Author Affiliations: Department of Emergency Medicine, Harvard School of Public Health (Dr Schuetz), and Department of Medicine, Inpatient Clinician Educator Service, Massachusetts General Hospital (Drs Chiappa and Greenwald), Boston, Massachusetts; and Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada (Dr Briel).

Arch Intern Med. 2011;171(15):1322-1331. doi:10.1001/archinternmed.2011.318

Previous randomized controlled trials suggest that using clinical algorithms based on procalcitonin levels, a marker of bacterial infections, results in reduced antibiotic use without a deleterious effect on clinical outcomes. However, algorithms differed among trials and were embedded primarily within the European health care setting. Herein, we summarize the design, efficacy, and safety of previous randomized controlled trials and propose adapted algorithms for US settings. We performed a systematic search and included all 14 randomized controlled trials (N = 4467 patients) that investigated procalcitonin algorithms for antibiotic treatment decisions in adult patients with respiratory tract infections and sepsis from primary care, emergency department (ED), and intensive care unit settings. We found no significant difference in mortality between procalcitonin-treated and control patients overall (odds ratio, 0.91; 95% confidence interval, 0.73-1.14) or in primary care (0.13; 0-6.64), ED (0.95; 0.67-1.36), and intensive care unit (0.89; 0.66-1.20) settings individually. A consistent reduction was observed in antibiotic prescription and/or duration of therapy, mainly owing to lower prescribing rates in low-acuity primary care and ED patients, and shorter duration of therapy in moderate- and high-acuity ED and intensive care unit patients. Measurement of procalcitonin levels for antibiotic decisions in patients with respiratory tract infections and sepsis appears to reduce antibiotic exposure without worsening the mortality rate. We propose specific procalcitonin algorithms for low-, moderate-, and high-acuity patients as a basis for future trials aiming at reducing antibiotic overconsumption.

The advent of antibiotic therapy led to dramatic reductions in mortality and morbidity rates due to bacterial infections and sepsis.1 However, overuse of antibiotics to fight infections may cause considerable harm by exposing individual patients to adverse events resulting from antibiotic use and by increasing the development of bacterial resistance. Combating the emergence of bacterial resistance to antimicrobial agents requires more effective efforts to reduce the inappropriate or unnecessarily prolonged use of antibiotics.2

Patients and physicians share the common goals of improving the patient's health and resolving infections as quickly as possible; they often believe use of antibiotics to be the most expeditious intervention to address these goals. This one-size-fits-all approach fails to consider the following basic questions: (1) Who truly benefits from antibiotic therapy? and (2) If treated, what is the optimal treatment duration?1

Considerable interest has been expressed in antibiotic stewardship programs aiming at reducing antibiotic overuse and the associated emergence of multiresistant pathogens.3,4 A previous cohort study3 of hospitalized adult patients illustrated the effect of antimicrobial-resistant infections on hospitalization duration, mortality rate, and cost. In this study, the 188 patients with antimicrobial-resistant infections stayed 6.4 to 12.7 days longer in the hospital, had attributable mortality of 6.5%, and incurred societal costs ranging from $18 588 to $29 069 (2008 dollars) per patient.

A novel approach for determining the necessity and optimal duration of antibiotic therapy is the use of biomarkers, such as procalcitonin (PCT) levels, which become upregulated during bacterial infections and appear to mirror the severity of infections.5,6Quiz Ref IDA growing body of literature supports the measurement of PCT levels to improve the ability to differentiate bacterial from nonbacterial infections because nonbacterial infections and nonspecific inflammatory reactions do not result in elevated PCT levels.7 Also, an initial study8 has shown that a decrease in PCT levels indicates a favorable patient response to antimicrobial therapy. Thus, PCT is a promising candidate marker to help physicians more rationally decide on prescription and duration of antibiotic therapy in patients with infections.

Procalcitonin, the precursor peptide of the hormone calcitonin, is released ubiquitously in response to primarily bacterial toxins and bacteria-specific proinflammatory mediators, particularly interleukin 1b, tumor necrosis factor, and interleukin 6.5 In previous studies,9,10 a strong correlation was observed between the concentration of PCT and the extent and severity of bacterial infections. Of particular interest, PCT levels are attenuated by the cytokines typically released in response to viral infections, namely, interferon-γ.10 Levels of PCT have been shown to increase within 6 to 12 hours of the initial bacterial infection, and circulating PCT levels are expected to decrease by half daily when the infection is controlled by the host immune system and antibiotics (Abx).11 A previous study12 has shown that the production of PCT, in contrast to other blood markers, is not attenuated by nonsteroidal and steroidal anti-inflammatory drugs. Based on these promising preclinical data, many studies have investigated the clinical usefulness of measuring PCT levels for different clinical settings and infections.7 However, owing to the diagnostic uncertainty associated with sepsis and other infectious diseases and the lack of a diagnostic criterion standard, the results of many observational studies have been inconclusive.7

To circumvent the limitations of observational studies, including observer bias, selection bias, sample availability, coinfection, colonization, and difficulty of pathogen identification owing to time constraints, several randomized controlled trials (RCTs) have been conducted focusing primarily on the outcomes of patients with or without the use of PCT-guided algorithms for antibiotic therapy. The clinical harm and benefit of using PCT thereby were measured by clinical outcomes, assuming that if the patient recovers without antibiotics, no relevant bacterial illness had existed, and if the patient recovers with fewer days taking antibiotics, the bacterial illness was adequately controlled with shorter antibiotic exposure.

The aim of this systemic review is to summarize the evidence based on previous RCTs for using PCT measurement in respiratory infections and sepsis from the clinical settings for which the most RCT data are available, namely, primary care, the emergency department (ED), the medical intensive care unit (MICU), and the surgical intensive care unit (SICU). Because most published studies were conducted in Europe, we also aim to propose clinical algorithms for use in future United States trials.

Literature search

We searched EMBASE (from 1974 to the present), MEDLINE via PubMED and Ovid (from 1948 to the present), and the Cochrane Central Register of Controlled Trials (from 1991 to 2011) for articles regarding PCT levels taken into account when making decisions regarding Abx. The following search terms were included: procalcitonin (mp [multiple posting, in which the term appears in the title, abstract, or subject heading]) and calcitonin AND pneumonia (exp [explode, a search term that automatically includes closely related MeSH terms]), sepsis (exp), chronic obstructive pulmonary disease (exp), or respiratory tract infections (exp). We also included procalcitonin (mp) and calcitonin AND intensive care units, emergency services, hospital, or ambulatory care facilities. We restricted the search to RCTs performed only in adults. We also identified relevant systematic reviews, meta-analyses, and controlled clinical trials and reviewed their references; in addition, we searched trial registries ( and and contacted experts in the field for additional eligible studies. We included articles in any language and did not exclude articles based on other comorbidities.

Eligibility criteria and study selection

Eligible trials had to be RCTs including adults with a diagnosis of respiratory tract infections (ie, pneumonia, acute exacerbations of chronic obstructive pulmonary disease [AECOPD] or other respiratory tract infections) or sepsis. Clinical settings included primary care, the ED, or the ICU. Interventions included measurement of PCT levels to inform decisions regarding antibiotic therapy (ie, regarding its initiation and/or duration). Abstracts or full-text articles for which no abstract was available were reviewed by 2 of us (P.S. and J.L.G.) to ensure topic appropriateness and adherence to inclusion and exclusion criteria. Disagreements were resolved by consensus.

Data extraction and quality assessment

Working in teams of 2, 3 investigators (P.S., V.C., and J.L.G.) independently extracted data from the included trials. Disagreements were resolved by consensus. A standardized data abstraction tool was used that included clinical setting, study design, number of study subject individuals, clinical outcomes, and study protocols. Also, an assessment of the methodological study quality was peformed, based on the following criteria: adequate sequence generation (eg, computer-generated random numbers, compared with inadequate approaches, which included the use of alternation, case record numbers, or days of the week); adequate allocation concealment (deemed adequate if a central randomization procedure [ie, telephone or Web-based] or the use of sequentially numbered, opaque, sealed envelopes was reported); adequate masking of physicians, patients, and outcome assessors; low risk of attrition bias (ie, minimal loss to follow-up and performance of adequate sensitivity analyses); and freedom from selective outcome reporting (eg, if all stated outcomes were reported).13 Two of us (P.S. and M.B.) who were coauthors of eligible studies14,15 were excluded from reviewing our own work.

Data synthesis

For all included RCTs, we report summary data, including antibiotic prescription rate, duration of antibiotic treatment, mortality rate, and adverse event rate, as defined in the individual studies. We pooled results and calculated odds ratios (ORs) with 95% confidence intervals (CIs) for overall mortality rate using the Peto method.16 The Peto ORs are appropriate when intervention effects are small (ie, when ORs are close to 1), events are not particularly common, and studies have balanced numbers in intervention and control groups.17 We tested for heterogeneity with the Cochran Q test and measured the inconsistency (I2 [the percentage of total variance across studies that is due to heterogeneity rather than chance]) of intervention effects across trials.18 Also, we performed sensitivity analyses comparing trials with low risk of bias (ie, adequate sequence generation, allocation concealment, and unlikely attrition bias) vs trials at higher risk. We investigated the presence of publication bias by means of funnel plots.19 Concerning antibiotic treatment, the different studies were too few in number and lacked adequate consistency in reporting across study design and population for a pooled analysis of antibiotic treatment data for different diagnoses and settings. Instead, we summarized antibiotic exposure and outcome data from the different studies grouped by clinical setting. This report adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.20 All calculations were performed using commercially available software (STATA, version 9.2 [StataCorp LP, College Station, Texas] and RevMan, version 5.1 [Cochrane Collaboration]).

Literature search and quality of studies

The literature searches of MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials and mining of trial registries and references yielded 120 potential results (Figure 1). A total of 58 articles were then excluded owing to nonrandomization and/or not focusing on our population of interest. Also, we excluded 4 articles focusing on pediatric populations and 14 ongoing trials, leaving 44 articles that we further assessed for eligibility. After exclusion of 2 duplicate publications in different journals and redundantly identified articles from the different databases, 14 articles remained. Review of the reference lists from the 14 articles and those from selected meta-analyses and reviews regarding the topic yielded no additional trials. Thus, the final quantitative analysis includes 14 RCTs.

Table 1 shows the quality assessment of included trials with regard to sequence generation, allocation concealment, masking, incomplete outcome data, selective outcome reporting, and other biases. The methodological quality varied across included studies; only 3 trials could be considered at low risk for bias, namely, 1 in the primary care,14 1 in the ED,15 and 1 in the ICU setting.32 The other studies lacked several quality criteria and need to be considered at high risk for bias. None of the studies managed to mask physicians with respect to treatment allocation of patients.

Study setting, design, and pct algorithms

Table 2 summarizes clinical design and study setting. It also displays the underlying diagnoses of patients, the study question, the PCT algorithm used, and the outcomes investigated of all included RCTs.

Primary Care

Two studies in the primary care setting were found.14,21 Both were noninferiority trials with regard to clinical outcomes, focused on upper and lower respiratory tract infections, and used a similar PCT algorithm (ie, no antibiotics in patients with PCT levels of <0.25 μg/L). However, although the study by Briel et al14 used repeated PCT measurements and addressed subjects' activity limitations and antibiotic exposure as outcomes, the study by Burkhardt et al21 only measured PCT levels once on admission but also measured health impairment and antibiotic exposure.

Emergency Department

A total of 6 studies in the ED setting were identified.15,22-26 All studies used a similar PCT algorithm (ie, no initiation of Abx or, if already initiated, discontinuation of Abx in patients with PCT levels of <0.25 μg/L). Two studies22,26 measured PCT levels only on admission, and the remaining 4 studies15,23-25 used repeated measurements. All studies reported antibiotic exposure and patient outcomes, but only the trial by Schuetz et al15 was powered for noninferiority with regard to clinical outcomes.

Intensive Care Unit

In the ICU setting, 6 RCTs were identified27-32 that differed substantially in terms of underlying diagnoses (ie, postoperative infections in the SICU, severe sepsis and/or septic shock, and ventilator-associated pneumonia) and the PCT cutoffs used. All studies based their recommendation primarily on repeated PCT measurements and specified discontinuing Abx when PCT levels dropped to a range of less than 0.25 to less than 1.00 μg/L or by at least 80% to 90%. One study27 did not focus on de-escalation of antibiotics but specified to change antibiotics or to perform a diagnostic workup if PCT levels did not decrease by an adequate amount. Two studies were multicenter trials,29,32 and the trial by Bouadma et al32 was a noninferiority trial in terms of mortality rate and adverse outcomes. Antibiotic exposure, length of stay, and mortality rate were the most commonly reported outcomes.

Efficacy and clinical outcomes

Overall, a total of 4467 patients were included in the 14 RCTs (ie, 2240 in the control group and 2227 in the PCT group). Table 3 displays the detailed results of all RCTs with regard to the number of included patients and underlying diagnoses, mortality rate, absolute and relative antibiotic prescription rates, and duration between the 2 groups. Overall, no significant difference was observed in mortality in patients in the PCT groups (172 of 2227 [7.7%]) compared with the control group (186 of 2240 [8.3%]), with a summary OR of 0.91 (95% CI, 0.73-1.14) (Figure 2). We found no evidence of relevant heterogeneity among trials. The analysis for publication bias indicated no evidence of such bias for mortality rate (eFigure 1).

Primary Care

The 2 primary care trials included a total of 1008 patients with upper and lower respiratory tract infections. The study by Briel et al,14 which used repeated PCT measurements, found a 74% reduction in antibiotic prescription and a 13% reduction in mean antibiotic duration. However, the study by Burkhardt et al21 only used a single PCT measurement at admission and found a 42% reduction in antibiotic exposure but not a reduction in treatment duration. In both trials, no differences were observed in the primary safety end points (ie, days with restricted activities and days with health impairment within a 14-day follow-up) between the controls and the PCT-group patients. The mortality rate in both trials was similarly low in both groups (OR, 0.13; 95% CI, 0-6.64) (Figure 2).

Emergency Department

The 6 ED trials15,22-26 included a total of 2449 patients, mostly with community-acquired pneumonia and AECOPD. For community-acquired pneumonia, the main effect of using PCT was to shorten the duration of antibiotic treatment. Trials using only a single initial PCT measurement had less effect on duration of antibiotic therapy (ie, a 15% reduction in the first study by Christ-Crain et al22) compared with those requiring repeated PCT measurements (ie, a 55% reduction in the second study by Christ-Crain et al,23 a 40% reduction in the study by Long et al,25 and a 34% reduction in the trial by Schuetz et al15). The study by Kristoffersen et al26 had similar antibiotic prescription rates but a 25% reduction in the duration of antibiotic therapy. Those authors noted, however, that physicians were not asked to wait for PCT results before initiating antimicrobial therapy. Therefore, PCT values were, in most cases, used to motivate cessation or continuation of already-initiated treatments.

For patients with AECOPD, the trial by Stolz et al24 found a 44% reduction in prescription rates. The patients with COPD guided by a PCT-based algorithm included by Christ-Crain et al22 and Schuetz et al15 also had lower prescription rates compared with controls (ie, 87% vs 38% and 70% vs 49%, respectively). In none of the ED trials was an increase in adverse outcomes noted, although only the trial by Schuetz et al was powered for this end point. Within all ED trials, mortality in PCT groups was 5.2% (63 of 1214) compared with 5.4% (67 of 1235) in control groups (OR, 0.95; 95% CI, 0.67-1.36) (Figure 2).

Intensive Care Unit

The 2 ICU trials that focused on patients with severe sepsis/septic shock (ie, Nobre et al28 and Bouadma et al32) found reductions in antibiotic exposure of 37% and 33%, respectively. The trial by Svoboda et al27 did not report antibiotic exposure because it only focused on patient outcomes. The trial reported by Stolz et al29 included only patients with ventilator-associated pneumonia and found a 27% increase in antibiotic-free days alive. Two trials30,31 focused on patients with postoperative infection and sepsis; they reported reductions of antibiotic exposure by 20% and 25%, respectively. None of the studies reported a difference in mortality rate or adverse outcome, but only the study by Bouadma et al32 was powered for noninferiority. Within all ICU trials, mortality in the PCT group patients was 21.5% (109 of 506), compared with 23.4% (118 of 504) in standard group patients (OR, 0.89; 95% CI, 0.66-1.20) (Figure 2). A sensitivity analysis contrasting the 3 trials at low risk of bias14,15,32 with the remaining higher-risk trials yielded similar results (eFigure 2A and B).


Within this systematic review, we address the question of the safety and efficacy of using a PCT-based algorithm for antibiotic therapy decisions in patients with respiratory tract infections and sepsis using data derived from previous RCTs. Quiz Ref IDWe found a marked reduction in antibiotic exposure in all settings, levels of disease acuity, and patient populations. This reduction occurred because of lower prescription rates in low-acuity infections such as bronchitis, exacerbation of AECOPD in primary care and ED settings,Quiz Ref IDand shorter duration of antibiotic courses in moderate- and high-acuity patients, such as those with pneumonia and sepsis in the hospital and ICU settings. Critically, none of the trials reported an increase in adverse outcomes, including mortality rate, although only a subset of the included trials were powered to detect changes in clinical outcomes.

Given the limited number of studies included across a breadth of diagnoses and settings in this analysis and the limited methodological quality of many of the included studies (ie, their being at high risk for bias), caution must be used when generalizing these findings. Although it is reassuring to note the uniform absence of adverse events paralleling the similarly uniform presence of antibiotic use reduction across the studies, further data are needed before PCT-based algorithms should be considered the criterion standard of care, especially for diagnoses such as ventilator-associated pneumonia or postoperative infection, in which randomized trial data are limited. Clearly, the most robust data are for pneumonia; in this diagnosis, as with the others, the findings are promising.

Previous meta-analyses have investigated the use of PCT levels for detection of sepsis in adult patients33-37 and children38 from observational studies. Also, 3 previous meta-analyses focused on randomized trials only to investigate the measurement of PCT levels for antibiotic decisions in the critical care setting39,40 and in patients with suspected bacterial infections.41 Our systematic review included all published RCTs that investigated PCT algorithms for antibiotic treatment decisions concerning escalation and de-escalation of dosage in adult patients with respiratory tract infections and sepsis from primary care, ED, and ICU settings. We focus our analysis on the differences in PCT algorithms used for low-, moderate-, and high-acuity patients; this is a basis for future trials aiming at reducing antibiotic overconsumption and it may be of particular importance for practicing physicians who want to include PCT findings in their hospital protocols. Owing to differences of PCT levels in different clinical settings and patient populations, the correct and safe practical use of this marker will likely vary with the level of disease acuity in patients.

The algorithms in the included studies varied moderately. Nevertheless, a few commonalities can be identified. Quiz Ref IDFirst, regarding the most severe diseases (eg, sepsis and pneumonia) or the highest-acuity care settings (ie, the ICU and the hospital), PCT levels were not used to determine whether antibiotic therapy should be initiated but when to discontinue it. Second, the decrease of elevated PCT levels appears to correlate adequately with sufficient resolution of bacterial infections to allow for the safe discontinuation of antibiotic therapy, even if this discontinuation occurs before the traditional length of a typical course of antibiotics has elapsed. Third, in lower-acuity settings (eg, the primary care setting) or with clinical entities that are generally less imminently dangerous (eg, bronchitis), PCT values may be used to assist in the initial determination of whether antibiotics should be prescribed at all. However, all patients should undergo reassessment in cases in which Abx are withheld to ensure that the clinical condition improves spontaneously in a clinically appropriate period.

Most of the studies of PCT-based algorithmic approaches to antibiotic management of respiratory tract infections have been conducted in Europe. Despite the recognized importance of antibiotic stewardship and health care cost containment in the United States,3,4 further studies in a US population may be needed to assuage concerns regarding practice and population differences. Therefore, we recommend the PCT algorithms described herein for use in future US studies. However, algorithms for PCT use, much like those for other biomarkers, should supplement and not supplant clinical impressions.

Low risk/low acuity

For patients with a low pretest probability of contracting a bacterial infection (eg, patients with nonpneumonic upper and lower respiratory tract infection treated in the primary care setting), a single measurement of PCT level and a cutoff ranging from less than 0.10 to less than 0.25 μg/L appears to be an appropriate, safe, and simple approach in this setting to determine the need for antibiotics. Clinical follow-up with remeasurement of PCT should be performed in all patients in whom Abx were withheld and who show no clinical improvement (Figure 3A).

Moderate risk/ moderate acuity

For patients who are clinically stable and are treated at the ED or are hospitalized with pneumonia, the initiation of antibiotic therapy should be based on clinical grounds and a PCT threshold of at least 0.25 μg/L, assuming the PCT results can be obtained expeditiously. In patients with an initial PCT level of no higher than 0.25 μg/L, alternative diagnoses (eg, viral infection and pulmonary embolism) should be considered. Thereafter, repeated measurement of PCT levels every other day should occur, with directions to discontinue antibiotic therapy when PCT levels drop to less than 0.25 μg/L or by at least 80% to 90% of the peak value and when the patient has improved clinically. If Abx are withheld initially, algorithms should suggest retesting PCT levels 6 to 12 hours after the initial measurement (Figure 3B).

High risk/high acuity

In high-risk or ICU patients with suspected sepsis, algorithms should dictate that empirical antibiotic therapy not be delayed for PCT measurement. Periodic monitoring of PCT levels after initiation of antibiotic therapy may be the preferred strategy, and a drop of PCT levels to less than 0.50 μg/L or by at least 80% to 90% from baseline in patients who show a clinical improvement after therapy are reasonable thresholds for cessation of antibiotic therapy in this fragile population (Figure 3C). For postoperative patients in the SICU, a decrease in PCT level to less than 1.0 μg/L may be sufficient to discontinue Abx. As with moderate risk/moderate acuity algorithms, if Abx is withheld initially based on a low PCT level, a second measurement should be obtained within 6 to 12 hours.


Our review has a number of limitations. First, because it is limited to RCTs in an attempt to minimize bias, it excluded a reasonably large body of literature pertaining to PCT that did not derive from RCTs. This exclusion may have led to the inadvertant omission of relevant findings. Second, although we conducted an extensive literature search for RCTs on the topic and a funnel plot did not suggest the presence of publication bias, such a bias cannot definitely be ruled out. Quiz Ref IDThird, the methodological quality of most included trials was low to moderate. However, a sensitivity analysis focusing on 3 trials at low risk for bias yielded similar results, and no evidence was observed of heterogeneity among effects on mortality rate.

In conclusion, this systematic review of mostly moderate-quality RCTs suggests that the use of PCT-guided algorithms for antibiotic therapy decisions in patients with respiratory tract infections, including bronchitis, AECOPD, and pneumonia and in patients with sepsis appears to be effective at reducing use of Abx without sacrificing patient safety. These types of algorithms also appear to be useful in different clinical settings. Based on the available evidence, we have proposed PCT algorithms based on clinical acuity levels, which should be used in future large multicenter trials within the United States, powered for patient outcomes and aimed at reducing antibiotic overconsumption.

Back to top
Article Information

Correspondence: Philipp Schuetz, MD, MPH, Department of Emergency Medicine, Harvard School of Public Health, 667 Huntington Ave, Boston, MA 02115 (

Accepted for Publication: May 10, 2011.

Author Contributions:Study concept and design: Schuetz, Greenwald, and Chiappa. Acquisition of data: Schuetz, Greenwald, Chiappa, and Briel. Analysis and interpretation of data: Schuetz, Greenwald, Chiappa, and Briel. Drafting of the manuscript: Schuetz and Greenwald. Critical revision of the manuscript for important intellectual content: Schuetz, Greenwald, Chiappa, and Briel. Statistical analysis: Schuetz and Briel.

Financial Disclosure: Dr Schuetz was supported by research grant PASMP3-127684/1 from the Swiss Foundation for Grants in Biology and Medicine and received support from BRAHMS USA Inc and bioMérieux to attend meetings and to fulfill speaking engagements. Dr Briel is supported by grants from santésuisse and the Gottfried and Julia Bangerter-Rhyner Foundation.

Additional Contributions: Qing Wang, PhD, of the Basel Institute for Clinical Epidemiology, University Hospital Basel, Switzerland, helped with the translation of an article published in Chinese. Carole Foxman, MA, MS, Coordinator for Education and Database Services/Research Liaison at the Treadwell Library, Massachusetts General Hospital, assisted with the EMBASE and Cochrane Central Register of Controlled Trials database searches.

This article was corrected for typographical errors on August 18, 2011.

Yealy DM, Fine MJ. Measurement of serum procalcitonin: a step closer to tailored care for respiratory infections?  JAMA. 2009;302(10):1115-111619738100PubMedGoogle ScholarCrossref
Whitney CG, Farley MM, Hadler J,  et al; Active Bacterial Core Surveillance Program of the Emerging Infections Program Network.  Increasing prevalence of multidrug-resistant Streptococcus pneumoniae in the United States.  N Engl J Med. 2000;343(26):1917-192411136262PubMedGoogle ScholarCrossref
Roberts RR, Hota B, Ahmad I,  et al.  Hospital and societal costs of antimicrobial-resistant infections in a Chicago teaching hospital: implications for antibiotic stewardship.  Clin Infect Dis. 2009;49(8):1175-118419739972PubMedGoogle ScholarCrossref
Ohl CA, Luther VP. Antimicrobial stewardship for inpatient facilities.  J Hosp Med. 2011;6:(suppl 1)  S4-S1521225949PubMedGoogle ScholarCrossref
Becker KL, Nylén ES, White JC, Müller B, Snider RH Jr. Procalcitonin and the calcitonin gene family of peptides in inflammation, infection, and sepsis: a journey from calcitonin back to its precursors.  J Clin Endocrinol Metab. 2004;89(4):1512-152515070906PubMedGoogle ScholarCrossref
Müller F, Christ-Crain M, Bregenzer T,  et al; ProHOSP Study Group.  Procalcitonin levels predict bacteremia in patients with community-acquired pneumonia: a prospective cohort trial.  Chest. 2010;138(1):121-12920299634PubMedGoogle Scholar
Schuetz P, Christ-Crain M, Müller B. Procalcitonin and other biomarkers to improve assessment and antibiotic stewardship in infections: hope for hype?  Swiss Med Wkly. 2009;139(23-24):318-32619529989PubMedGoogle Scholar
Karlsson S, Heikkinen M, Pettilä V,  et al; Finnsepsis Study Group.  Predictive value of procalcitonin decrease in patients with severe sepsis: a prospective observational study.  Crit Care. 2010;14(6):R20521078153PubMedGoogle ScholarCrossref
Gogos CA, Drosou E, Bassaris HP, Skoutelis A. Pro- versus anti-inflammatory cytokine profile in patients with severe sepsis: a marker for prognosis and future therapeutic options.  J Infect Dis. 2000;181(1):176-18010608764PubMedGoogle ScholarCrossref
Linscheid P, Seboek D, Nylen ES,  et al.  In vitro and in vivo calcitonin I gene expression in parenchymal cells: a novel product of human adipose tissue.  Endocrinology. 2003;144(12):5578-558412960010PubMedGoogle ScholarCrossref
Linscheid P, Seboek D, Zulewski H, Keller U, Müller B. Autocrine/paracrine role of inflammation-mediated calcitonin gene-related peptide and adrenomedullin expression in human adipose tissue.  Endocrinology. 2005;146(6):2699-270815761041PubMedGoogle ScholarCrossref
Müller B, Peri G, Doni A,  et al.  High circulating levels of the IL-1 type II decoy receptor in critically ill patients with sepsis: association of high decoy receptor levels with glucocorticoid administration.  J Leukoc Biol. 2002;72(4):643-64912377932PubMedGoogle Scholar
Higgins JPT, ed, Green S, ed. Assessing risk of bias in included studies. In: Cochrane Handbook for Systematic Reviews of Interventions, Version 5.1.0. March 2011. Accessed April 21, 2011
Briel M, Schuetz P, Mueller B,  et al.  Procalcitonin-guided antibiotic use vs a standard approach for acute respiratory tract infections in primary care.  Arch Intern Med. 2008;168(18):2000-2007; discussion, 2007-200818852401PubMedGoogle ScholarCrossref
Schuetz P, Christ-Crain M, Thomann R,  et al; ProHOSP Study Group.  Effect of procalcitonin-based guidelines vs standard guidelines on antibiotic use in lower respiratory tract infections: the ProHOSP randomized controlled trial.  JAMA. 2009;302(10):1059-106619738090PubMedGoogle ScholarCrossref
Yusuf S, Peto R, Lewis J, Collins R, Sleight P. Beta blockade during and after myocardial infarction: an overview of the randomized trials.  Prog Cardiovasc Dis. 1985;27(5):335-3712858114PubMedGoogle ScholarCrossref
Higgins JPT, ed, Green S, ed. Peto odds ratio method [chapter]. In: Cochrane Handbook for Systematic Reviews on Interventions, Version 5.1.0. March 2011. Accessed April 21, 2011
Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses.  BMJ. 2003;327(7414):557-56012958120PubMedGoogle ScholarCrossref
Sterne JAC, Egger M, Smith GD. Systematic reviews in health care: investigating and dealing with publication and other biases in meta-analysis.  BMJ. 2001;323(7304):101-10511451790PubMedGoogle ScholarCrossref
Moher D, Liberati A, Tetzlaff J, Altman DG.PRISMA Group.  Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement.  Ann Intern Med. 2009;151(4):264-269, W264Google Scholar
Burkhardt O, Ewig S, Haagen U,  et al.  Procalcitonin guidance and reductions of antibiotic use in acute respiratory tract infection.  Eur Respir J. 2010;36(3):601-60720185423PubMedGoogle ScholarCrossref
Christ-Crain M, Jaccard-Stolz D, Bingisser R,  et al.  Effect of procalcitonin-guided treatment on antibiotic use and outcome in lower respiratory tract infections: cluster-randomised, single-blinded intervention trial.  Lancet. 2004;363(9409):600-60714987884PubMedGoogle ScholarCrossref
Christ-Crain M, Stolz D, Bingisser R,  et al.  Procalcitonin guidance of antibiotic therapy in community-acquired pneumonia: a randomized trial.  Am J Respir Crit Care Med. 2006;174(1):84-9316603606PubMedGoogle ScholarCrossref
Stolz D, Christ-Crain M, Bingisser R,  et al.  Antibiotic treatment of exacerbations of COPD: a randomized, controlled trial comparing procalcitonin-guidance with standard therapy.  Chest. 2007;131(1):9-1917218551PubMedGoogle ScholarCrossref
Long W, Deng XQ, Tang JG,  et al.  The value of serum procalcitonin in treatment of community acquired pneumonia in outpatient [in Chinese].  Zhonghua Nei Ke Za Zhi. 2009;48(3):216-21919576090PubMedGoogle Scholar
Kristoffersen KB, Søgaard OS, Wejse C,  et al.  Antibiotic treatment interruption of suspected lower respiratory tract infections based on a single procalcitonin measurement at hospital admission: a randomized trial.  Clin Microbiol Infect. 2009;15(5):481-48719416298PubMedGoogle ScholarCrossref
Svoboda P, Kantorová I, Scheer P, Radvanova J, Radvan M. Can procalcitonin help us in timing of re-intervention in septic patients after multiple trauma or major surgery?  Hepatogastroenterology. 2007;54(74):359-36317523274PubMedGoogle Scholar
Nobre V, Harbarth S, Graf J-D, Rohner P, Pugin J. Use of procalcitonin to shorten antibiotic treatment duration in septic patients: a randomized trial.  Am J Respir Crit Care Med. 2008;177(5):498-50518096708PubMedGoogle ScholarCrossref
Stolz D, Smyrnios N, Eggimann P,  et al.  Procalcitonin for reduced antibiotic exposure in ventilator-associated pneumonia: a randomised study.  Eur Respir J. 2009;34(6):1364-137519797133PubMedGoogle ScholarCrossref
Hochreiter M, Köhler T, Schweiger AM,  et al.  Procalcitonin to guide duration of antibiotic therapy in intensive care patients: a randomized prospective controlled trial.  Crit Care. 2009;13(3):R8319493352PubMedGoogle ScholarCrossref
Schroeder S, Hochreiter M, Koehler T,  et al.  Procalcitonin (PCT)-guided algorithm reduces length of antibiotic treatment in surgical intensive care patients with severe sepsis: results of a prospective randomized study.  Langenbecks Arch Surg. 2009;394(2):221-22619034493PubMedGoogle ScholarCrossref
Bouadma L, Luyt C-E, Tubach F,  et al; PRORATA trial group.  Use of procalcitonin to reduce patients' exposure to antibiotics in intensive care units (PRORATA trial): a multicentre randomised controlled trial.  Lancet. 2010;375(9713):463-47420097417PubMedGoogle ScholarCrossref
Simmonds MC, Higgins JP, Stewart LA, Tierney JF, Clarke MJ, Thompson SG. Meta-analysis of individual patient data from randomized trials: a review of methods used in practice.  Clin Trials. 2005;2(3):209-21716279144PubMedGoogle ScholarCrossref
Uzzan B, Cohen R, Nicolas P, Cucherat M, Perret G-Y. Procalcitonin as a diagnostic test for sepsis in critically ill adults and after surgery or trauma: a systematic review and meta-analysis.  Crit Care Med. 2006;34(7):1996-200316715031PubMedGoogle ScholarCrossref
Tang BM, Eslick GD, Craig JC, McLean AS. Accuracy of procalcitonin for sepsis diagnosis in critically ill patients: systematic review and meta-analysis.  Lancet Infect Dis. 2007;7(3):210-21717317602PubMedGoogle ScholarCrossref
Simon L, Gauvin F, Amre DK, Saint-Louis P, Lacroix J. Serum procalcitonin and C-reactive protein levels as markers of bacterial infection: a systematic review and meta-analysis.  Clin Infect Dis. 2004;39(2):206-21715307030PubMedGoogle ScholarCrossref
Jones AE, Fiechtl JF, Brown MD, Ballew JJ, Kline JA. Procalcitonin test in the diagnosis of bacteremia: a meta-analysis.  Ann Emerg Med. 2007;50(1):34-4117161501PubMedGoogle ScholarCrossref
Yu Z, Liu J, Sun Q, Qiu Y, Han S, Guo X. The accuracy of the procalcitonin test for the diagnosis of neonatal sepsis: a meta-analysis.  Scand J Infect Dis. 2010;42(10):723-73320840003PubMedGoogle ScholarCrossref
Kopterides P, Siempos II, Tsangaris I, Tsantes A, Armaganidis A. Procalcitonin-guided algorithms of antibiotic therapy in the intensive care unit: a systematic review and meta-analysis of randomized controlled trials.  Crit Care Med. 2010;38(11):2229-224120729729PubMedGoogle ScholarCrossref
Heyland DK, Johnson AP, Reynolds SC, Muscedere J. Procalcitonin for reduced antibiotic exposure in the critical care setting: a systematic review and an economic evaluation [published online February 24, 2011].  Crit Care MedGoogle Scholar
Tang H, Huang T, Jing J, Shen H, Cui W. Effect of procalcitonin-guided treatment in patients with infections: a systematic review and meta-analysis.  Infection. 2009;37(6):497-50719826761PubMedGoogle ScholarCrossref