Early adequate empiric antibiotics is associated with improved survival for patients with sepsis; however, the failure to de-escalate or continuing antibiotics beyond recommended durations contribute to 50% of antibiotics prescribed in hospital settings being unnecessary or inappropriate (1). This may contribute to increasing rates of resistant organisms and unwarranted adverse effects associated with antimicrobials. The Surviving Sepsis Campaign (SSC) guidelines suggest that biomarkers such as procalcitonin may be used to shorten the duration of antimicrobial use in patients with sepsis (2). Numerous trials have evaluated the effects of procalcitonin-guided strategies on antimicrobial use and clinical outcomes in critically ill patients admitted to the ICU (3–18).
Several systematic reviews and meta-analyses have attempted to summarize the available literature and generate an overall effect size of procalcitonin-guided strategies on outcomes such as antibiotic exposure, mortality, and hospital and ICU length of stay (LOS) (19–26). In general, the previous meta-analyses demonstrated a statistically significant decrease in antibiotics exposure (weighted mean difference in antibiotics days ranging from 2.05 to 4.19 d) with procalcitonin-guided therapy compared with standard care, but did not detect a difference in mortality, ICU LOS, or hospital LOS between strategies. However, many of these studies suffer from the limitation of combining studies that evaluated procalcitonin-guided strategies in different phases of antimicrobial use (initiation, cessation, or mixed) to provide an average effect size (27). The combination of studies from different phases of antimicrobial use with meta-analytic methods may not be appropriate due to substantial clinical heterogeneity. This is also evident when evaluating the magnitude of statistical heterogeneity observed in available meta-analyses (22–25). Furthermore, since the publication of the previous systematic reviews, two additional large randomized control studies evaluating procalcitonin-guided strategies have been published (14 , 15). Therefore, we performed a systematic review and meta-analysis to 1) evaluate the true effect size of procalcitonin-guided strategies in different phases of antimicrobial use; and 2) incorporate the findings from the most up-to-date clinical studies.
MATERIALS AND METHODS
This systematic review and meta-analysis was performed and reported in accordance with the current guidelines as described by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (28).
Literature Search and Data Extraction
A systematic review of the medical literature was performed from MEDLINE and EMBASE databases from inception to November 1, 2017. See Supplemental Table 1 (Supplemental Digital Content 1, http://links.lww.com/CCM/D211) for Medical Subject Headings search terms and search strategy. Each abstract was reviewed by at least two authors independently and assessed for inclusion based on study criteria. Inclusion criteria encompassed randomized controlled study, procalcitonin-guided antibiotics decision compared with standard of care, critical care adult population, English language, and reported at least one of the following outcomes: mortality, LOS, or antibiotics use. Abstracts were excluded for full text review if the study was performed in children or neonates; specifically evaluated patients with neutropenia, osteomyelitis, endocarditis, septic arthritis, or abscesses; or evaluated procalcitonin guidance for strategies outside of antibiotics management. Each full text article was reviewed independently by at least two authors. All meta-analyses of procalcitonin use in critically ill patients were reviewed to identify any additional studies that may have been missed with the search strategy. Discrepancies were settled by an independent review from a third reviewer.
Data were extracted from each included study by at least two independent reviewers using a structured data collection sheet to include the following items: publication details, country of origin, design, setting, procalcitonin test and algorithm, patients’ characteristics, interventions, compliance with algorithm, and outcomes. For all studies, the intention-to-treat outcomes were utilized. Trials were classified based on whether the procalcitonin strategy guided the initiation, cessation, or both (mixed) phases of antimicrobial management (27). Briefly, the classification of the procalcitonin strategy was based on whether there was a protocolized-driven approach to the antimicrobial decisions surrounding the specific phase of antibiotics management. Specifically, initiation is specified the period when clinicians were deciding whether antibiotics should be initiated, and cessation as the period when clinicians were deciding how to manage antibiotics that patients are currently prescribed. Discrepancies in data extraction were discussed and solved by consensus.
Outcomes and Analyses
The primary outcome of this meta-analysis was short term, all-cause mortality, defined as hospital mortality or mortality within 30 days. Secondary outcomes included antibiotics duration, long-term mortality (60–100 d), hospital and ICU LOS, and recurrent infections. If recurrent infections were evaluated at multiple time points, the one reported that is closest to day 14 was utilized. For each of the outcomes, meta-analysis was only performed if there were sufficient data with at least three studies in each of the procalcitonin strategy subgroups (initiation, cessation, mixed). If any of the subgroups lacked sufficient data, then the studies from that subgroup were excluded from the meta-analysis.
Risk of Bias
The included articles were reviewed, and a quality assessment was performed using the Cochrane Collaboration tool for assessing risk of bias (29). Overall quality was independently determined by each reviewer with discrepancies solved by consensus. Funnel plots were utilized to allow for a visual assessment of publication bias based on the primary outcome analysis.
Pooled risk ratios (RRs) and 95% CIs were calculated for dichotomous variables. Continuous variables were analyzed using weighted mean differences and 95% CIs. When means and variances were not reported, they were estimated from the medians, ranges, and size of the samples (30). Between-study statistical heterogeneity was assessed by the Cochran’s Q test, and I.2 (31) For all analyses performed, if no significant heterogeneity was noted (p < 0.10 and I2 < 50%), fixed-effect model analyses using the Mantel-Haenszel method were presented; otherwise, results of the random-effects model analysis using the DerSimonian-Laird method were presented. Statistical calculations were performed using Revman 5.3.5 (The Cochrane Collaboration, London, United Kingdom).
Using the search strategy, we found 686 records through MEDLINE and 938 through EMBASE totaling 1,624 unique records. From 1,624 abstracts, 53 articles were selected for full text review. Thirty eight of the articles were excluded, leaving 15 trials that fulfilled all inclusion and no exclusion criteria (Fig. 1). Nine previously published meta-analyses were identified, which did not identify any additional trials to be included.
Supplemental Table 2 (Supplemental Digital Content 1, http://links.lww.com/CCM/D211) summarizes the main characteristics of the included trials, classified based on the phase of antimicrobial use in which a procalcitonin-guided strategy was evaluated. All 15 trials were published after 2008, with three evaluating procalcitonin for the initiation of antibiotics, nine for cessation, and three for mixed strategies. The majority of the trials (12 of 15) were performed in mixed ICU settings, with two focused on surgical ICU and one on medical ICU. The number of included patients from each trial ranged from 27 to 1,546.
The algorithms used for procalcitonin guidance varied widely. For trials that utilized procalcitonin for initiation of antibiotics, the cutoffs were generally a procalcitonin value of greater than 0.5–1 μg/L leading to initiation or escalation of antibiotics. Trials that utilized procalcitonin for cessation or de-escalation of antibiotics usually had both a relative (greater than 50–90% drop from previous procalcitonin) and an absolute value (less than 0.1–0.5 μg/mL) to determine when antibiotics should be stopped or de-escalated. All trials utilized an ultra-sensitive procalcitonin assay with a low limit of detection (0.06 μg/L). Noncompliance to the trial protocol was documented in the majority of trials and ranged from 0% (16) to 61% (14).
Risk of Bias
A summary of the risk of biases of included trials based on the Cochrane Collaboration tool can be found in Supplemental Figure 1 (Supplemental Digital Content 1, http://links.lww.com/CCM/D211). Two trials had high overall risks of bias (16 , 17), five with moderate risks (4 , 6–8 , 11), and the remainder with low risks. Supplemental Figure 2 (Supplemental Digital Content 1, http://links.lww.com/CCM/D211) illustrates a visual representation of the potential for publication bias based on the primary outcome (short-term mortality). No gross asymmetry was observed, suggesting that publication bias was not likely to influence the overall effect size.
Short-term, all-cause mortality was reported in all trials, with at least three in each procalcitonin strategy subgroup. The combined estimate of the pooled RR for all studies based on the fixed effects model for short-term mortality is 0.93 (95% CI, 0.85–1.01; p = 0.08), with no significant differences observed between procalcitonin-guidance and standard of care (Fig. 2). No significant heterogeneity was observed (Q = 8.62; df = 14; p = 0.85; I2 = 0%). Examining each procalcitonin strategy subgroup, the pooled RR for the initiation, cessation, and mixed procalcitonin strategies for short-term mortality were 1.00 (95% CI, 0.86–1.15; p = 0.96), 0.87 (95% CI, 0.77–0.98; p = 0.02), and 1.01 (95% CI, 0.80–1.29; p = 0.93), respectively. The test for subgroup heterogeneity was not significant (p = 0.27).
Antibiotic duration was reported in 13 trials, with 12 able to be included for meta-analyses (nine trials from the cessation subgroup and three trials from the mixed subgroup). The mean difference in antibiotic duration, using a random effects model, was –1.26 days (95% CI, –1.98 to –0.54 d; p < 0.001) in the cessation subgroup and –3.10 days (95% CI, –6.09 to –0.11 d; p = 0.04) in the mixed subgroup (Fig. 3). Significant heterogeneity was observed (Q = 66.8; df = 11; p < 0.001; I2 = 84%).
Five trials reported long-term mortality, but there was only sufficient evidence to evaluate the cessation subgroup, which had three studies reporting results (14 , 15 , 18). The pooled RR for the cessation studies, calculated using the random-effects model since substantial heterogeneity was observed (Q = 5.54; df = 2; p = 0.06; I2 = 64%) was 0.91 (95% CI, 0.75–1.11; p = 0.36) (Supplemental Fig. 3, Supplemental Digital Content 1, http://links.lww.com/CCM/D211).
Hospital and ICU LOS were reported in 10 and 13 studies, respectively. Due to the insufficient number of trials in each subgroup, only seven trials were included in the analyses for cessation and mixed subgroups for hospital LOS and 11 trials were included the initiation and cessation subgroups for ICU LOS. No differences in hospital and ICU LOS were observed (Supplemental Figs. 4 and 5, Supplemental Digital Content 1, http://links.lww.com/CCM/D211).
The occurrence of recurrent infections was also evaluated as a secondary outcome. Eight trials reported recurrent infections, with six eligible to be combined for meta-analysis in the cessation subgroup. No difference in recurrent infections was observed with pooled risk ratios = 1.19 (95% CI, 0.86–1.65; p = 0.29) using a fixed effects model (Supplemental Fig. 6, Supplemental Digital Content 1, http://links.lww.com/CCM/D211).
This systematic review and meta-analysis sought to evaluate the impact of procalcitonin-guided antimicrobial management compared with standard antibiotics initiation and cessation practices in critically ill patients. Recognizing the possibility that the true effect size of the impact of procalcitonin guidance may be altered if procalcitonin use in different antibiotic phases yields different outcomes, this meta-analysis specifically evaluated the different strategy subgroups. Overall, a procalcitonin-guided strategy for antibiotic use did not lead to a decrease in short-term mortality. However, procalcitonin guidance for the cessation of antibiotics was associated with a decrease in mortality. This is in contrast to the use of procalcitonin for the initiation of antibiotics or mixed strategies, where no mortality benefit was observed. The SSC guidelines suggest procalcitonin can be used to support the discontinuation of empiric antibiotics, but are silent on the use of procalcitonin for antibiotics initiation (2). These recommendations emphasize the distinction of procalcitonin use between phases of antibiotics management and are supported by this meta-analysis.
Previous meta-analyses attempted to evaluate the effect of procalcitonin guidance on mortality in critically ill patients (20–25); however, none of them detected a significant difference between procalcitonin and standard of care. With the exception of the evaluation by Soni et al (24), none of these previous meta-analyses attempted to separate procalcitonin studies based on the different phase of antibiotics management. Soni et al (24) performed meta-analyses that evaluated procalcitonin for the discontinuation of antibiotics and, separately, procalcitonin for antibiotics initiation. However, the meta-analysis included studies that utilized procalcitonin for both the initiation and discontinuation of antibiotics (a “mixed” strategy). As demonstrated by this meta-analysis, mixed strategy procalcitonin use was not associated with an improvement of mortality. Therefore, combining mixed studies with discontinuation studies may have also lessened the true overall effect of procalcitonin for the discontinuation of antibiotics.
The secondary outcome findings in this meta-analysis demonstrated no difference in ICU LOS, hospital LOS, or recurrent infections. However, a significant decrease in antibiotic duration was observed with procalcitonin guidance. These findings were similar to other meta-analyses published on the topic (19–25). Significant heterogeneity was observed in the antimicrobial duration meta-analysis. This may be due to the differences in trial follow-up period, where antimicrobial duration was evaluated anywhere from 5 days after enrollment (13) to the entire hospital LOS (up to 547 d) (11). Lastly, noncompliance to study protocol was quite variable among included trials, which may further increase heterogeneity observed in the meta-analysis of antimicrobial duration. To accurately capture the true effect of procalcitonin guidance on antibiotic duration, future trials should standardize when and how antibiotic duration is captured and have provisions within its protocol to account for effects of protocol noncompliance.
The current meta-analysis clearly demonstrates that there is considerable clinical heterogeneity associated with trials that evaluate procalcitonin in different phases of antimicrobial management. As such, it is suboptimal and likely inappropriate to combine procalcitonin studies from different antibiotic phases for meta-analyses. Furthermore, because improved mortality with procalcitonin use was only observed in the cessation phase of antimicrobial management, future investigations should focus on isolating this strategy in their trial protocols. Inclusion of initiation or mixed strategies within the same study protocol as cessation may lead to confounding and dilution of true effects. The potential detrimental effects of prolonged or unnecessary antibiotics may be crucial. In the study by Jensen et al (12), which evaluated procalcitonin for the initiation of antibiotics, those randomized to procalcitonin guidance had significantly higher rates of mechanical ventilation, duration of ICU stay, and incidence of renal dysfunction.
Our systematic review and meta-analysis has several limitations. First, due to the low number of trials in each strategy subgroup and the differences in reported outcomes of the included trials, it was not feasible to clearly evaluate each separate procalcitonin strategy for all evaluated outcomes. Second, the procalcitonin algorithms evaluated varied widely, which likely led to differences in antibiotics management and patient outcomes. Third, with some of the analyses, there was a considerable degree of statistical heterogeneity. This may partially be due to the inconsistent procalcitonin algorithms evaluated and the effects of combining critically ill patients from different types of ICUs from different countries. The statistical heterogeneity was managed by using random-effects models when appropriate. Fourth, due to the variance in mortality reporting, the primary outcome combined reported hospital mortality or mortality within 30 days. Although combining measures such as 30-day mortality and hospital mortality may confound the final finding, it allows for comparison across all the studies. In our comparison, eight of the 15 studies utilized 28- or 30-day mortality, which accounted for about 80% of all patients in the meta-analysis. Given that the average hospital LOS among all the studies was about 23 days, we believe that the merits of combining the mortality results outweighs the potential for confounding the outcome. Furthermore, a sensitivity analysis evaluating only studies that reported 28-day mortality demonstrated similar decrease mortality finding in the cessation group (Supplementary Fig. 7, Supplemental Digital Content 1, http://links.lww.com/CCM/D211). Fifth, the meta-analysis included studies that may have high or moderate risk of bias. However, in a sensitivity analysis, when those studies were removed, similar decreases in mortality was seen in the cessation subgroup (Supplementary Fig. 8, Supplemental Digital Content 1, http://links.lww.com/CCM/D211). Lastly, many of the included trials reported high rates of noncompliance to procalcitonin algorithm, with the exception of one trial where all patients with protocol violations were excluded (16). Noncompliance with the procalcitonin algorithm likely reflects usual care practice, which would bias the trials toward finding no difference between approaches. This may further dilute the effects of procalcitonin guidance. Of note, for trials that provided more granular information regarding compliance rates (3 , 5 , 13), compliance depended on the recommendation; compliance was lower when a low procalcitonin level suggested antibiotics should be withheld or stopped and higher when a high procalcitonin level suggested continuation or initiation of antibiotics. For example, in the trial by Layios et al (5), when the baseline procalcitonin level was less than 0.5 μg/L, only 36% of clinicians were compliant with the recommendation to withhold antimicrobials (5). This is in contrast to those who had a baseline procalcitonin level greater than 0.5 μg/L, where 86% of clinicians were compliant with the recommendation to initiate antimicrobials. This preferential compliance to initiate antibiotics and noncompliance toward withholding antimicrobials would further bias finding no difference between approaches in an intention-to-treat trial design.
In conclusion, when evaluating all studies of procalcitonin-guided antimicrobial management in critically ill patients, no difference in short-term mortality was observed. However, when only examining procalcitonin-guided cessation of antibiotics, mortality was lower with procalcitonin-guided approach. As such, future studies should focus specifically on procalcitonin for the cessation of antibiotics in critically ill patients.
1. Dellit TH, Owens RC, McGowan JE Jr, et alInfectious Diseases Society of America; Society for Healthcare Epidemiology of America: Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis 2007; 44:159–177
2. Rhodes A, Evans LE, Alhazzani W, et alSurviving sepsis campaign: International guidelines for management of sepsis and septic shock: 2016. Crit Care Med 2017; 45:486–552
3. Bouadma L, Luyt CE, Tubach F, et alPRORATA trial group: Use of procalcitonin
to reduce patients’ exposure to antibiotics in intensive care units (PRORATA trial): A multicentre randomised controlled trial. Lancet 2010; 375:463–474
4. Hochreiter M, Köhler T, Schweiger AM, et alProcalcitonin to guide duration of antibiotic therapy in intensive care patients: A randomized prospective controlled trial. Crit Care 2009; 13:R83
5. Layios N, Lambermont B, Canivet JL, et alProcalcitonin usefulness for the initiation of antibiotic treatment in intensive care unit patients. Crit Care Med 2012; 40:2304–2309
6. Nobre V, Harbarth S, Graf JD, et alUse of procalcitonin
to shorten antibiotic treatment duration in septic patients: A randomized trial. Am J Respir Crit Care Med 2008; 177:498–505
7. Oliveira CF, Botoni FA, Oliveira CR, et alProcalcitonin versus C-reactive protein for guiding antibiotic therapy in sepsis: A randomized trial. Crit Care Med 2013; 41:2336–2343
8. Schroeder S, Hochreiter M, Koehler T, et alProcalcitonin (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; 394:221–226
9. Stolz D, Smyrnios N, Eggimann P, et alProcalcitonin for reduced antibiotic exposure in ventilator-associated pneumonia: A randomised study. Eur Respir J 2009; 34:1364–1375
10. Svoboda P, Kantorová I, Scheer P, et alCan procalcitonin
help us in timing of re-intervention in septic patients after multiple trauma or major surgery? Hepatogastroenterology 2007; 54:359–363
11. Deliberato RO, Marra AR, Sanches PR, et alClinical and economic impact of procalcitonin
to shorten antimicrobial therapy in septic patients with proven bacterial infection in an intensive care setting. Diagn Microbiol Infect Dis 2013; 76:266–271
12. Jensen JU, Hein L, Lundgren B, et alProcalcitonin And Survival Study (PASS) Group: Procalcitonin
-guided interventions against infections to increase early appropriate antibiotics and improve survival in the intensive care unit: A randomized trial. Crit Care Med 2011; 39:2048–2058
13. Annane D, Maxime V, Faller JP, et alProcalcitonin levels to guide antibiotic therapy in adults with non-microbiologically proven apparent severe sepsis: A randomised controlled trial. BMJ Open 2013; 3
14. Bloos F, Trips E, Nierhaus A, et alfor SepNet Critical Care Trials Group: Effect of sodium selenite administration and procalcitonin
-guided therapy on mortality
in patients with severe sepsis or septic shock: A randomized clinical trial. JAMA Intern Med 2016; 176:1266–1276
15. de Jong E, van Oers JA, Beishuizen A, et alEfficacy and safety of procalcitonin
guidance in reducing the duration of antibiotic treatment in critically ill patients: A randomised, controlled, open-label trial. Lancet Infect Dis 2016; 16:819–827
16. Ding J, Chen Z, Feng KProcalcitonin-guided antibiotic use in acute exacerbations of idiopathic pulmonary fibrosis. Int J Med Sci 2013; 10:903–907
17. Najafi A, Khodadadian A, Sanatkar M, et alThe comparison of procalcitonin
guidance administer antibiotics with empiric antibiotic therapy in critically ill patients admitted in intensive care unit. Acta Med Iran 2015; 53:562–567
18. Shehabi Y, Sterba M, Garrett PM, et alProGUARD Study Investigators; ANZICS Clinical Trials Group: Procalcitonin
algorithm in critically ill adults with undifferentiated infection or suspected sepsis. A randomized controlled trial. Am J Respir Crit Care Med 2014; 190:1102–1110
19. Agarwal R, Schwartz DNProcalcitonin to guide duration of antimicrobial therapy in intensive care units: A systematic review. Clin Infect Dis 2011; 53:379–387
20. Heyland DK, Johnson AP, Reynolds SC, et alProcalcitonin for reduced antibiotic exposure in the critical care setting: A systematic review and an economic evaluation. Crit Care Med 2011; 39:1792–1799
21. Kopterides P, Siempos II, Tsangaris I, et alProcalcitonin-guided algorithms of antibiotic therapy in the intensive care unit: A systematic review and meta-analysis
of randomized controlled trials. Crit Care Med 2010; 38:2229–2241
22. Matthaiou DK, Ntani G, Kontogiorgi M, et alAn ESICM systematic review and meta-analysis
-guided antibiotic therapy algorithms in adult critically ill patients. Intensive Care Med 2012; 38:940–949
23. Prkno A, Wacker C, Brunkhorst FM, et alProcalcitonin-guided therapy in intensive care unit patients with severe sepsis and septic shock—a systematic review and meta-analysis
. Crit Care 2013; 17:R291
24. Soni NJ, Samson DJ, Galaydick JL, et alProcalcitonin-guided antibiotic therapy: A systematic review and meta-analysis
. J Hosp Med 2013; 8:530–540
25. Andriolo BN, Andriolo RB, Salomão R, et alEffectiveness and safety of procalcitonin
evaluation for reducing mortality
in adults with sepsis, severe sepsis or septic shock. Cochrane Database Syst Rev 2017; 1:CD010959
26. Schuetz P, Muller B, Christ-Crain M, et alProcalcitonin to initiate or discontinue antibiotics in acute respiratory tract infections. Cochrane Database Syst Rev 2012; (10):CD007498
27. Lam SW, Bauer SR, Duggal AProcalcitonin-based algorithms to initiate or stop antibiotic therapy in critically ill patients: Is it time to rethink our strategy? Int J Antimicrob Agents 2016; 47:20–27
28. Moher D, Liberati A, Tetzlaff J, et alPRISMA Group: Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLoS Med 2009; 6:e1000097
29. Higgins JPT, Green SCochrane Collaboration: Cochrane Handbook for Systematic Reviews of Interventions. 2008Chichester, England; Hoboken, NJ, Wiley-Blackwell,
30. Wan X, Wang W, Liu J, et alEstimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med Res Methodol 2014; 14:135
31. Higgins JP, Thompson SG, Deeks JJ, et alMeasuring inconsistency in meta-analyses. BMJ 2003; 327:557–560