a b s t r a c t

Background: procalcitonin is a new biomarker with a higher accuracy in the diagnosis of bacterial infec- tions. Utilization of PCT may reduce the number of unnecessary antibiotics prescribed to patients and conse- quently may decrease the rise in antibiotic resistance.

The aim of this systematic review is to determine if a PCT-guided algorithm can safely reduce the number of an- tibiotics prescribed to all patients with a suspected of infection in the emergency department (ED).

Methods: MEDLINE, EMBASE, Web of Science, COCHRANE central, PubMed publisher, and Google Scholar were searched. Two reviewers performed the screening independently. The QUADAS 2 tool was used to assess quality. Results: In total, 1621 articles were screened. Nine articles were included in the analysis. In the 6 studies on adult patients, only patients with respiratory tract infections were investigated. In these studies, a cutoff value of 0.25 ug/L was used, and PCT-guided therapy reduced the number of prescribed antibiotics significantly. Three studies were on pediatric patients, 2 on fever without source and 1 on respiratory complaints. Procalcitonin-guided ther- apy did not reduce antibiotic prescription in children. Procalcitonin-guided therapy did not result in an increase in adverse events in any of the studies.

Discussion: Procalcitonin-guided therapy in the ED is only studied in subpopulations, where it was effective and safe in adult patients with respiratory tract infections and not effective but safe nonetheless in specific Pediatric populations. Nonadherence is a significant problem in prospective PCT-guided therapy studies. There is not enough evidence to use PCT-guided therapy in a general ED population.

(C) 2016

1. Introduction

In the emergency department (ED), immediate treatment of bacterial infections is vital. Delay of administration of antibiotics is associated with morbidity and mortality in patients with severe sepsis and septic shock [1]. On the other side, the use of antibiotics in the ED, without laboratory confirmation of the definitive diagnosis, may result in an overuse of

? PROSPERO Systematic review registration: CRD42015023534.

?? Funding: This study is supported through the Erasmus MC efficiency research grant (2013).

This is a non-commercial grant for researchers in the Erasmus University Medical center.

? Presentation at meetings: The results of this study have not been presented at a scien-

tific meeting.

* Corresponding author at: Erasmus University Medical Center, Spoedeisende hulp, Nc019, Postbus 2040, 3000CA Rotterdam, The Netherlands. Tel.: +31 10 7030557.

E-mail addresses: [email protected] (Y. van der Does), [email protected] (P.P.M. Rood), [email protected] (J.A. Haagsma), [email protected] (P. Patka), [email protected] (E.C.M. van Gorp), [email protected] (M. Limper).

antimicrobial therapy. Consequently, adverse drug events and health care costs may rise, and antibiotic resistance may increase [2-4].

Antibiotic resistance is a growing global problem [2,3]. Governments worldwide promote the implementation of Antimicrobial stewardship programs [5]. antimicrobial stewardship programs encourage to initiate optimal Antimicrobial treatment [6]. Ideally, patients without bacterial infections would not receive antibiotics. However, in the emergency sit- uation, it is difficult to distinguish bacterial infections from viral infec- tions and other febrile conditions.

When a febrile patient presents at the ED, the standard Diagnostic approach-besides thorough medical History taking and physical examination-consists of laboratory tests such as C-reactive protein (CRP) and leukocyte count, and different image modalities. Cultures and polymerase chain reaction technology can be obtained in the ED, but results are not directly available and therefore not useful for ED decision-making.

procalcitonin can be used as a biomarker for bacterial infec- tions. Elevated levels indicate the probable presence of bacterial infec- tions. As levels of PCT rise within approximately 6 hours after the start

http://dx.doi.org/10.1016/j.ajem.2016.03.065

0735-6757/(C) 2016

Figure. Flow diagram.

of bacterial infection and remain relatively stable, its properties are suit- able for ED application [7,8]. Compared to CRP, PCT has been shown to be more accurate in different age groups, ranging from young children with fever without source (FWS) [9,10], to Geriatric patients [11]. In ad- dition, for different sites of infection, such as respiratory tract and uro- genital tract [12-14], and in multiple clinical settings, including primary care, intensive care units, and the ED, PCT is more accurate [15-17]. Although the characteristics of PCT are promising, there may be other factors that can influence the initiation of antibiotic therapy. Pro- spective studies give more insight in these factors because intention-to- treat analyses can be compared with per-protocol analyses. In addition, safety can be addressed in prospective studies because unwanted undertreatment and consequent adverse events can be quantified.

Procalcitonin-guided therapy is defined as initiation of antibiotic treatment using PCT measurements, usually using a suggested treat- ment algorithm based on the height of the PCT measurement [15]. The

overall clinical value of PCT-guided therapy in the general ED popula- tion of all ages and full spectrum of febrile complaints remains to be investigated.

The aim of this systematic review is to determine if a PCT-guided al- gorithm can safely reduce the number of antibiotics prescribed to all pa- tients suspected of infection in the ED.

1. Methods
   1. Study design

A systematic review of literature was performed according to the PRISMA guidelines [18]. The design of this systematic review is regis- tered in the PROSPERO database [19], with registration number CRD42015023534 (http://www.crd.york.ac.uk/PROSPERO/). The

Table 1

QUADAS 2

First author, year, country	Risk of bias					Concerns on applicability
	Patient selection	Index test	Reference standard	Flow and timing		Patient selection Index test	Reference standard
Baer 2013, Switzerland	Low	Unclear	Unclear	Low		Low Unclear	Low
Christ-Crain 2004, Switzerland	Unclear	Low	Low	Low		Low Low	Low
Christ-Crain 2006, Switzerland	Unclear	Low	Low	Low		Low Low	Low
Drozdov 2015, Switzerland	Unclear	Low	Low	Low		High High	Low
Lacroix 2014, Switzerland	Unclear	Unclear	Low	High		Low Unclear	Low
Long 2011, China	Unclear	Low	Low	Low		Low Low	Low
Manzano 2010, Canada	Unclear	Unclear	Low	Low		Unclear Low	Low
Schuetz 2009, Switzerland	Low	Low	Low	Low		Low Low	Low
Stolz 2007, Switzerland	High	Low	Low	High		Low Low	Low
Tang 2013, China	Unclear	Low	Low	Low		Low Low	Low

Table 2

Study characteristics

First author, Study year, country populationa	age distribution in years	Sex distribution (male)	Inclusion criteria	Exclusion criteria	PCT cutoff value used	Overruling of algorithm
Baer 2013, 337	PCT group:	PCT group:	Pediatric patients,	Unwillingness or unable	Antibiotics definitely	In patients with
Switzerland patients	median (IQR),	98 (58%).	age 1 mo-18 y,	to provide written	(N 0.5 ug/L), probably	life-threatening
	2.7 (1.1-5.2).	Control group:	presenting with LRTI,	informed consent by	(0.26-0.5 ug/L),	infections, defined
	Control group:	98 (58%).	defined as T >=38?C	patients and caretakers,	probably not	as severe comorbidity,
	2.9 (1.2-5.7).		(measured at home	severe immune	(0.1-0.25 ug/L), and	emerging ICU need
			or hospital), with 1	suppression (HIV with	definitely not	during initial
			symptom (cough,	b15% CD4 count,	(0.1 ug/L).	follow-up, or
			sputum production,	immunosuppressive		hemodynamic or
			pleuritic pain, poor feeding) and	treatment, neutropenia (b1000 x 109/L),		respiratory instability.
			1 clinical sign	cystic fibrosis, acute
			(tachypnea, dyspnea,	croup, hospital stay
			wheezing, late	within previous 14 d,
			inspiratory crackles,	other severe infection.
			bronchial breathing,
			pleural rub).
Christ-Crain 243	PCT group	PCT group:	Adult patients with	Severely	Antibiotics	Not reported.
2004, patients	(mean +- SD):	67 (54%).	suspected LRTI,	immunocompromised	discouraged
Switzerland	62.8 +- 19.8.	Control group	defined as community-	patients, ie, HIV	(0.1-0.25 ug/L),
	Control group:	61 (51%).	acquired pneumonia,	infection with CD4	suggested
	65.3 +- 17.3.		COPD, asthma, acute	count b200cells/mL,	(0.25-0.5 ug/L),
			bronchitis.	neutropenic patients,	strongly
				stem cell transplant	recommended
				recipients, cystic	(N 0.5 ug/L).
				fibrosis, active
				tuberculosis, hospital-
				acquired pneumonia
				on ED presentation.
Christ-Crain 302	PCT group	PCT group:	Adult patients with	Cystic fibrosis, active	Antibiotics strongly	Not reported.
2006, patients	(mean +- SD):	94 (62%).	suspected CAP,	pulmonary tuberculosis,	discouraged
Switzerland	70 +- 17.	Control group	defined as infiltrate	hospital-acquired	(b0.1 ug/L),
	Control group:	93 (62%).	on chest x-ray, and	pneumonia on ED	discouraged
	70 +- 17.		one of more symptoms	presentation, severely	(0.1-0.25 ug/L),
			or signs: cough, sputum	immunocompromised	encouraged
			production, dyspnea, temp >=38.0?C and up,	patients (not defined).	(0.25-0.5 ug/L), strongly encouraged
			abnormal breath		(N 0.5 ug/L).
			sounds, rales on
			auscultation, leukocytosis 10 x 109/L
			and up, or less than 4x 109/L.
Lacroix 2014, 271	Age in months.	PCT group:	Pediatric patients	Congenital or acquired	Lab score: PCT	No algorithm of
Switzerland patientsa	PCT group	65 (50%).	between 7 days and	immunodeficiency	b0.5 ng/mL (0),	antibiotic treatment
	(median [IQR])	Control group	3 years old with FWS:	syndromes, antibiotic	0.5-1.99 ng/mL (2),	advice reported.
	4.8 (1.7-10.4).	71 (51%).	Temperature >=38?C,	administration b 48 h	>=2 ng/mL (4). In	Therefore, overruling
	Control group		with no identified	of presentation,	combination with CRP	is also not reported.
	3.4 (1.5-10.4).		source of infection after	fever N 7 d.	value-based score,
			thorough history and		and urinary dipstick
			physical examination.		based score. A Lab
			Parental informed		score of >=3 was used
			consent.		as marker for severe
					bacterial illness.
Long 2011, 162	PCT group	PCT group:	Adult patients with	Pregnancy,	Antibiotics strongly	Not reported.
China patientsa	(mean +- SD):	46 (60%).	suspected CAP, defined	commencement of	discouraged
	44 +- 16.	Control group	as infiltrate on chest	antibiotic therapy	(b0.1 ug/L),
	Control group:	49 (62%).	x-ray, and one of more	N 48 h before enrollment,	discouraged
	47 +- 19.		symptoms or signs:	systemic immune	(0.1-0.25 ug/L),
			fever, cough, purulent	deficiency, withholding	encouraged
			sputum, focal chest signs,	life support and active	(0.25-0.5 ug/L),
			dyspnea or pleuritic pain.	tuberculosis.	strongly encouraged
			outpatient treatment.		(N 0.5 ug/L).
Manzano 384	Age in months.	Not reported	Pediatric patients, age	Acquired or congenital	For intention-to-treat	No algorithm of
2010, patientsa	PCT group		between 1 and 36 mo,	immunodeficiency,	analysis no cutoff defined.	antibiotic treatment
Canada	(mean +- SD):		a history of rectal	current antibiotics use.	SemiQuantitative test:	advice reported.
	12 +- 8.		temperature >= 38.0?C,		b0.5 ng/mL, 0.5 ng/mL or	Therefore, overruling
	Control group:		no identified source of		higher, 2 ng/mL or higher,	is also not reported.
	12 +- 8.		infection, indication		10 ng/mL or higher.
			for blood and urine		Per-protocol analysis with
			analysis.		Prophylactic antibiotics
					with PCT level N 0.5 ng/mL
					or higher.
Schuetz 2009, 1359	PCT group:	PCT group:	Adult patients with	Inability to give informed	Antibiotics strongly	Patients in need of ICU
Switzerland patientsa	median (IQR),	402 (60%).	suspected LRTI, defined	consent, active	discouraged (b0.1 ug/L),	admission, respiratory
	74 (59-82).	Control group	as at least 1 respiratory	intravenous drug	discouraged (0.1-0.25 ug/L),	or hemodynamic

Table 2 (continued)

First author, year, country	Study populationa	Age distribution in years	Sex distribution (male)	Inclusion criteria	Exclusion criteria	PCT cutoff value used	Overruling of algorithm
		Control group:	380 (55%).	symptom (cough,	use, severe	encouraged (0.25-0.5 ug/L),	instability, positive
		72 (59-82).		sputum, dyspnea,	immunosuppression	strongly encouraged	Ag test for L
				tachypnea, pleuritic	other than corticosteroid	(N 0.5 ug/L).	pneumophila or
				pain), plus either	use, life-threatening		after consulting with
				rales or crepitation	comorbidity,		the study center.
				on auscultation,	community-acquired
				or temperature	pneumonia.
				N 38.0?C, shivering,
				leukocytosis.
Stolz 2007,	208	PCT group:	PCT group:	Adult patients with	Patients with other	Antibiotics strongly	Not reported.
Switzerland	patientsa	median (IQR),	50 (49%).	exacerbation of COPD,	explanations for	discouraged (b0.1 ug/L),
		69.5 (65-77).	Control group	who met post	presenting symptoms	discouraged (0.1-0.25 ug/L),
		Control	44 (42%).	bronchodilator therapy	other than COPD, and	encouraged (0.25-0.5 ug/L),
		group: 69.5		spirometric criteria	“vulnerable patients“	strongly encouraged
		(64.8-79).		according to GOLD	ie patients with	(N 0.5 ug/L).
				guidelines, within	Psychiatric diagnoses
				48 h of ED admission.	(not specified).
					Immunosuppression,
					asthma, cystic fibrosism
					presence of infiltrates
					on chest x-ray on
					hospital admission.
Tang 2013,	156	PCT group	PCT group:	Adult patients with	Treatment with	Antibiotics strongly	Not reported.
China	patientsa	(mean +- SD):	64 (50%).	suspected exacerbation	antibiotics within	discouraged (b0.1 ug/L),
		54 +- 14.	Control group	of asthma, with any	2 wk of recruitment,	discouraged (0.1-0.25 ug/L),
		Control group:	59 (47%).	or all GINA asthma	nonrespiratory bacterial	encouraged (0.25-0.5 ug/L),
		55 +- 15.		guidelines criteria:	infection, chest x-ray	strongly encouraged
				dyspnea, wheeze,	confirmed pneumonia,	(N 0.5 ug/L).
				acute cough, increased	other chronic respiratory
				Work of breathing,	disease, severe organ
				increased beta2	dysfunction.
				agonist use, O2 saturation b95%,
				peak flow b80%
				of known best.

Abbreviations: CAP, community-acquired pneumonia; GINA: Global Initiative for Asthma; GOLD, Global Initiative for Chronic obstructive lung disease; IQR, interquartile range; LRTI, lower respiratory tract infection.

^a Studies reported both a number of randomized and number of analyzed patients. Number of analyzed patients is reported. For number of omitted patients in analysis, see Table 1.

primary outcome measure was the effectiveness of PCT-guided therapy in the ED, defined as reduction in the initiation of antibiotic therapy.

Search strategy

A comprehensive search, supported by a professional librarian of the Erasmus University Medical Center Rotterdam, was performed. The MEDLINE, EMBASE, Web-of-science, COCHRANE central, PubMed pub- lisher, and Google scholar, containing all articles up to July 1, 2015, were searched. The results were limited to the English language. Search terms are listed in Supplement A.1.

This review was restricted to articles that prospectively reported on an intervention of PCT-guided therapy in an ED setting. Outcome mea- sures were reduction of antibiotics (defined as number or percentage of antibiotics prescriptions) and safety of PCT-guided therapy (defined as hospital mortality, hospital, or intensive care unit [ICU] admission and return visits to the ED).

Studies that were not performed in the ED, that is, in the ICU, medical or surgical wards, or primary care facilities, were excluded. Furthermore, studies performed in specific departments such as burns units were ex- cluded as well as studies where there was no comparison between a PCT-guided therapy group and a control group of Standard care. There was no limit on age distribution or subpopulation of patients.

Two authors (YD and ML) screened titles and abstracts of the search results and the full text of the selected articles. In case of disagreement, a third reviewer (PR) acted as a referee.

The QUADAS 2 tool [20] was used for assessing quality and bias in the selected full text studies. The QUADAS 2 tool is the recommended

quality assessment tool by the Cochrane Library. After positive quality assessment, data were extracted from the remaining articles.

1. Results
   1. Literature search

The search results are depicted in Figure. The search strategy identi- fied 1621 individual studies. Of these studies, 635 were ED-based stud- ies that investigated PCT. A total of 198 studies investigated the accuracy of PCT on various outcomes in the ED; 188 studies did not use a prospec- tive PCT-guided therapy algorithm and were therefore not included for further analysis. After full-text screening, 10 articles remained that ad- dressed PCT-guided therapy in a prospective setting. The overall quality of the studies was assessed using the QUADAS 2 guidelines [20].

Quality assessment

The quality assessment is described in Supplement C.1 and summa- rized in Table 1. Although the study of Drozdov et al [21] was eligible for inclusion in the review based on the selection criteria, it was excluded in the quality assessment. Drozdov et al [21] did not address the initiation of antibiotics but instead reported on a PCT-guided stopping algorithm for patients who already received antibiotic treatment for a urinary tract infection. This was not in line with the review question, and there- fore, the results of this study were not applicable.

Stolz et al [22] excluded patients with another explanation of dys- pnea than an acute exacerbation of chronic obstructive pulmonary dis- ease (AECOPD) and patients with psychiatric comorbidity from the

Table 3

Study outcomes
First author,	Antibiotics	Hospital	Length of	ICU	ICU length	Return visits	Mortality	Combined safety	Physician
year, country	reduction	admission	hospital	admission	of stay	to ED		end point	nonadherence
			stay						with PCT advice
Baer 2013,	PCT group	PCT group	PCT group	Not reported	Not	Not reported.	None	Defined as hospital	Not reported.
Switzerland	104 (62%),	104 (62%)	(days, median,	separately.	reported		reported.	readmission, ICU
	control group	Control	[IQR]) 2.6					admission,
	93 (56%).	group 100	(2 [0-4])					complications or
	Reduction 6%	(60%)	Control group					death,
	(95% CI, -5%	Reduction	2.7 (2 [0-5])					complications
	to 16%) in all	2% (95% CI,	Reduction:					of LRTI, disease
	patients. 28% (95% CI, 12%-	-8% to 12%).	-0.1 (95% CI, -0.8 to 0.5).					specific failure: PCT group 38 (23%)
	43%)							Control group
	in non-CAP							33 (20%) Reduction
	patients, -8%							2% (95% CI,
	(95% CI, -19%							-5% to 11%)
	to 4%) in CAP
	patients.
Christ-Crain	Antibiotics in	PCT group	PCT group	PCT group	Not	Not reported.	PCT group	Not reported.	In 9 patients (7%)
2004,	154 (63%)	101 (81%).	(days, mean	6 (5%).	reported		4 (3%).		antibiotics when
Switzerland	patients. PCT	Control	+- SD) 13.7	Control			Control		PCT was b0.1.
	group 55 (44%).	group 88	+- 7.3. Control	group 5			group 4		In 13 patients
	Control group	(74%).	group 10.8	(4%). P= .71.			(3%).		(10%)
	99 (83%).	P= .16.	+- 7.0. P= .25.				P= .95		antibiotics when
	Pb .001.								PCT was b0.25.
									After re-evaluation
									after 6-24 h: of a
									total of 29 patients
									in PCT group,
									10 received
									antibiotics, 5
									because of
									elevated PCT
									levels, 5 because
									of physician
									decision.
Christ-Crain	PCT group:	PCT group	PCT group	PCT group	Not	treatment failure	PCT group	Not reported.	In 1 patient (0%)
2006,	128 (85%).	146 (97%).	(days, mean	20 (13%).	reported	(after 6 weeks):	18 (12%).		antibiotics when
Switzerland	Control group	Control	+- SD) 12.0	Control		51 patients (17%)	Control		PCT was b0.1.
	149 (99%).	group 146	+- 9.1. Control	group 21		PCT group 24	group		(end-stage
	Pb .001. HR	(97%).	group 13.0	(14%).		(16%). Control	20 (13%).		pulmonary
	3.2 (95% CI,	P= 1.0.	+- 9.0. P= .35.	P= .87.		group 27 (18%).	P= .73.		fibrosis).
	2.5-4.2).					P= .65.			In 19 patients (6%)
									antibiotics when
									PCT was b0.25.
									(6 severe COPD,
									2 end-stage
									pulmonary
									fibrosis, 11
									other severe
									comorbidities).
Lacroix 2014,	PCT group 54	PCT group	Not reported.	Not	Not	Not reported.	Not	Not reported.	In 14 (11%) cases,
Switzerland	(41%), control	44 (34%).		reported.	reported		reported.		patients received
	group 42 (42%).	Control							antibiotics despite
	P= 1.000.	group							the low Lab score.
		50 (36%).							No patients with
		P= .810.							high Lab scores
									were withheld
									antibiotics.
Long 2011,	PCT group: 69	None.	Not	Not	Not	Treatment failure	None	Not reported.	Not reported.
China	(85%). Control		applicable.	applicable.	applicable	(after 4 wk):	reported.
	group 79 (98%).					21 patients (13%)
	P= .004. HR					PCT group
	3.2 (95% CI,					12 (15%). Control
	2.5-4.2).					group 9 (11%).
						No significant
						difference.
Manzano	PCT group:	PCT group:	Not reported.	Not	Not	Not reported	Not	Not reported.	No antibiotic
2010,	48 (25%).	50 (26%).		reported.	reported		reported.		treatment
Canada	Control group:	Control							advice was
	54 (28%). Risk	group: 48							given.
	difference -	(25%). Risk
	3 (95% CI, -	difference 1
	2 to 6).	(95% CI,
		-8 to 10).
Schuetz 2009,	PCT group	PCT group:	PCT group in	PCT group	Not	Recurrence	PCT group	Death, ICU	In 132 (20%)

Table 3 (continued)

First author, year, country	Antibiotics reduction	Hospital admission	Length of hospital stay	ICU admission	ICU length of stay	Return visits to ED	Mortality	Combined safety end point	Physician nonadherence with PCT advice
Switzerland	506 (75%),	628 (93%).	days: mean	43 (6%).	reported	of LRTI/	34 (5%).	admission,	patients, the
	control group	Control	(median	Control		rehospitalization	Control	recurrence of LRTI/	PCT algorithm
	603 (88%).	group:	[IQR])	group 60		PCT group	group 33	rehospitalization	was overruled,
	Relative rate	629 (91%).	9.4 (8 [4-12])	(9%). Risk		25 (4%). Control	(5%). Risk	b30 d PCT group	of which
	difference -12.2		Control group	difference		group 45 (7%).	difference	103 (15%). Control	62 (9%) were
	(95% CI, -16.3		9.2 (8 [4-12])	-2.3 (95%		Risk difference	0.3 (95% CI,	group 130 (19%).	in violation of
	to -8.1).		Reduction: 1.8 (95% CI,	CI, -5.2 to 0.4).		-2.8 (95% CI, -5.1 to -0.4).	-2.1 to 2.5).	Risk difference -3.5 (95% CI,	predefined protocol.
			-6.9 to 11.0).					-7.6 to 0.4)
Stolz 2007,	PCT group	Hospital	PCT group in	PCT group	PCT group	Recurrence of	Any cause	Not reported.	Not reported.
Switzerland	41 (40%),	admission	days: (median	8 (8%).	in days:	ECOPD within	mortality
	control group	24 h or	[IQR]) 9 [1-	Control	(mean +-	6 mo: PCT	within 6 mo:
	76 (72%).	longer. PCT	15].	group	SD)	group 44 (43%).	PCT group 5
	Pb .0001.	group: 80	Control group	11 (10%).	3.3 +- 2.7.	Control group	(5%).
		(78%).	10 [1-15].	P= .526.	Control	43 (40%). P=	Control
		Control	P= .960.		group 3.7 +-	.607.	group 9
		group: 82			2.1. P= .351.		(9%).
		(77%).					P= .409.
		P= .852.
Tang 2013,	PCT group	Not reported.	Not reported.	Mechanical	Not	Secondary ED	PCT group 1.	Not reported.	Not reported.
China	59 (46%),			ventilation	reported	visit within	Control
	control group			treatment:		6 weeks. PCT	group 2.
	95 (75%).			PCT group		group 8 (6%).	Excluded
	Pb .01.			8 (6%).		Control group	from further
				Control		13 (10%) Pb .05.	analysis.
				group
				9 (7%).
				P= .821.

Abbreviation: CI, confidence interval; ECOPD, exacerbation of chronic obstructive pulmonary disease; HR, hazard ratio.

study population. The exclusion of a selected part of the total population resulted in a high risk of population selection and possible effect exagger- ation because patients with medical comorbidities were excluded, and patients with possible lower therapy adherence may have been excluded. There was an unclear risk of bias in patient selection in 6 studies.

Lacroix et al [10] used temperature greater than or equal to 38.0?C as an inclusion criterion. Patients were also included if parents had mea- sured a temperature of greater than or equal to 38.0?C at home. Lacroix et al [10] reported a low adherence to the combined Lab-score, a predic- tion model containing a PCT value. No reasons for nonadherence were reported. This raised concerns on the applicability of the results of this study. Furthermore, the authors reported a missed inclusion rate of 75% but gave no description of individual reasons. This may have result- ed in a selection bias. Three studies [9,13,23] used an envelope as ran- domization method. This method is associated with an increased risk of selection bias because allocation concealment can be deciphered by holding envelopes against a lightsource [24]. Christ-Crain et al [13] exclud- ed 47 of a total of 597 eligible patients because of “other reasons.” Long et al [25] excluded 115 patients without specifying the reason of exclusion. The index test description had an unclear risk of bias in 2 studies.

Baer et al [26] did not report a final diagnosis of the febrile episode. It is not possible to check if the antibiotics were indicated retrospectively. Manzano et al [9] did not give an antibiotic treatment advice. A concrete cutoff value for PCT with a treatment suggestion could have influenced the results of this study.

Main study results

The selected articles are shown in Table 2. Nine randomized con- trolled trials met the selection criteria listed in Figure and the QUADAS 2 criteria in Table 1. These studies consisted of 2 multicenter trials [17,26] and 7 single-center studies [9,10,13,14,22,23,25]. Six studies [10,13,14,17,22,26] were conducted in Switzerland, 5 of these [13,14,17,22,26] in the university hospital of Basel. The remaining stud- ies were performed in China [23,25] and Canada [9].

Study populations

Sample sizes of the studies varied widely, ranging from 156 [23] to 1359 [17] patients, with 6 studies [9,10,13,14,22,26] having sample sizes between 200 and 400 patients.

Three studies reported on pediatric patients [9,10,26], of which 2 [9,10] reported on newborns and infants with FWS and 1 on pediatric patients with respiratory tract infections [26].

Six studies reported on adult patients with subcategories of respiratory complaints: community-acquired pneumonia [14,25], acute lower respira- tory tract infections [13,17], AECOPD [22], and exacerbation of asthma [23]. The age of patients ranged from newborn children between 7 days and 3 months of age [10] to septuagenarians [17]. Most of the participants were males (N 50% men in 6 [10, 13, 14, 17, 25, 26] of the 9 studies). One

study [9] did not report sex. None of the studies reported ethnicity.

Selection criteria studies

Inclusion criteria of the studies on lower respiratory tract infections [13,14,17,25,26] included body temperature of greater than or equal to 38?C (100.4?F), combined with at least 1 symptom of infection, that is, cough, sputum production, or dyspnea, and 1 clinical sign, that is, abnor- mal breath sounds or leukocytosis. The criterion for suspected commu- nity acquired pneumonia was an infiltrate on a chest x-ray. Inclusion criteria on asthma and chronic obstructive pulmonary disease (COPD) were based on reaction to ?-2-agonist use [22,23]. One study used tem- perature measured at home as inclusion criterion [26]. Two pediatric studies on FWS [9,10] included a measured body temperature of greater than or equal to 38?C, without the presence of a suspected cause of fever after history and physical examination [10], and the need for blood and urinary analysis [9].

Eight studies [9,10,13,14,17,22,25,26] reported immunosuppression as an exclusion criterion. This criterion was not uniformly defined. Some studies gave examples of specific conditions, that is, HIV infection with low CD4 + count [13,26], neutropenic patients [13,26], active

tuberculosis [13,14,25], and cystic fibrosis [13,14,22,25,26]. Four studies [9,10,23,25] excluded patients with current antibiotics use or within 14 days of ED presentation. Schuetz et al [17] excluded intravenous Drug users. Stolz et al [22] excluded “vulnerable patients”: patients with psy- chiatric diagnoses, which were not defined.

Procalcitonin cutoff values

Seven studies [13,14,17,22,23,25,26] used a cutoff of 0.25 ug/L to suggest or encourage the initiation of antibiotics. Two pediatric studies did not use a continuous cutoff scale. Lacroix et al [10] used PCT as part of the Lab-score, a decision rule that combined a semiquantitative PCT result with a semiquantitative CRP value and urinary dipstick out- come. The Lab-score is a severity index scale, and the outcome had no directly suggested treatment consequences. Manzano et al [9] studied PCT prospectively without treatment algorithm; only the PCT result was available, without treatment advice.

Overruling of PCT-guided therapy protocol and physician nonadherence

Two studies [17,26] described predefined criteria for overruling PCT- guided therapy. These criteria composed of life-threatening illness, de- fined as respiratory or hemodynamic instability. Schuetz et al [17] also in- cluded a positive screening test for Legionella pneumophilia as criterion.

Four [10,13,14,17] studies reported physician nonadherence, rang- ing from 6% to 20%. In the studies with predefined criteria, Schuetz et al [17] reported that 9% of the patients were excluded without meet- ing the nonadherence criteria. Baer et al [26] only mentioned predefined criteria and did not report the number of physician nonadherence or protocol violations.

Antibiotics reduction

Procalcitonin-guided therapy resulted in a significant reduction of antibiotics in all studies in adult patients [13,14,17,22,23,25] (Table 3). In the 3 pediatric studies, no significant reduction in antibiotics was noted [9,10,26]. No study reported an increase in initiation of antibi- otics. All results were from the intention-to-treat analyses.

Patient-related outcomes

Five studies [13,14,17,22,23] reported mortality and ICU admission. Death rates varied between 3% [13] and 13% [14]. Intensive care unit ad- mittance or mechanical ventilation was required for 5% [13] to 14% [14] of patients. No studies reported a significant difference between groups for death rates or ICU admission. Seven studies [9,10,13,14,17,22,26] re- ported hospital admissions. The admission rate ranged from 26% [9] to 97% [14]. No study found a significant difference between hospital ad- missions. Five studies [13,14,17,22,26] reported length of hospital stay; none found significant differences between groups. One study

[25] included patients who were sent home from the ED, and 1 study

[23] did not report a hospital admission number.

1. Discussion
   1. Findings

The results of our study show that PCT-guided therapy is only stud- ied prospectively in distinct ED patient populations, adults with respira- tory complaints [13,14,17,22,23,25], and young infants with respiratory complaints [26] and FWS [9,10]. In the studies on adult patients with re- spiratory complaints, PCT-guided therapy reduced antibiotic prescrip- tions. In the pediatric subgroups, there was no reduction.

In all included studies, there was no undertreatment, and there was no increase in adverse events in the intervention group. This suggests that PCT-guided therapy is safe in the patients of these distinct ED populations. Procalcitonin-guided therapy has been shown to reduce Antibiotic prescriptions in adult patients with respiratory complaints in various clinical settings. In primary care, reduction of antibiotics was 72% [16]. One hospital-based study on patients with lower respiratory tract infec- tions did not find a significant reduction in antibiotic prescriptions, due to a reported protocol nonadherence of 41%. The per-protocol did result in a 25% reduction of antibiotics based on a single PCT value [27]. Procalcitonin studies in the ICU mainly focus on stopping antibiotics in- stead of starting. Several ICU studies show a reduction in duration of an-

tibiotic treatment using PCT-guided therapy [15,28,29].

The most interesting finding was that nonadherence to PCT-guided algo- rithms was present in several included studies. Lacroix et al [10] reported that the use of a PCT-guided algorithm, included in the Lab-score, did not re- sult in a reduction in antibiotics in practice. However, per-protocol analysis showed that the algorithm would result in reduction, had it been followed. This illustrates the point that physicians do not always follow the advice of a PCT-guided therapy. This is confirmed by the nonadherence to the PCT- guided therapy algorithms in several of the other included studies [10,13,14,17]. Nonadherence is only visible in prospective studies because, in contrary to observational studies, randomized controlled trials report an intention-to-treat analysis, which includes the physician factor in the results. Prospective PCT-guided therapy studies in other clinical settings also show nonadherence. Briel et al [16] reported a nonadherence rate of 15% in primary care. Kristoffersen et al [27] reported a 41% nonadherence in a hospital-based setting. In the ICU setting, the PRORATA trial [15] had a pro- tocol nonadherence for stopping antibiotic therapy based on a PCT-guided algorithm of 53%; a recent ICU study [29] reported a 56% nonadherence rate when physicians were asked to stop antibiotics within 24 hours after initiation. Procalcitonin-guided therapy is accompanied by protocol nonadherence, and this finding is consistent in multiple clinical settings. We speculate that individual clinical experience is the cause of the lower reduction of antibiotics in intention-to-treat results. This may be caused

by the lack of understanding of the factors that influence PCT levels [30].

The results of this systematic review cannot be extrapolated to a general ED population because the studies included in this systematic review focused on specific subpopulations of patients with respiratory complaints. The study aim was to investigate the value of PCT-guided therapy for all patients in the ED. For this reason, we did not limit the re- sults on specific populations but included all ED studies from young children to elderly patients. However, our search results only yielded specific subpopulations. We can conclude that PCT-guided therapy is not studied in a wide enough population to use PCT as a standard bio- marker for bacterial infections in the ED.

The overall quality assessment indicated a low risk on bias in the se- lected articles. The study by Drozdov et al [21] did not give information on antibiotic initiation and was therefore excluded. Two studies had a high risk on bias. Stolz et al [22] excluded patients with possible other ex- planations for dyspnea than AECOPD. In addition, patients with psychiat- ric comorbidities were excluded. This may have resulted in an exaggeration of the effect of PCT-guided therapy because merely a part of the total population of patients with AECOPD was analyzed. In the study by Lacroix et al [10], patient selection issues were noted as well. These studies were included because the studies both used a PCT- guided algorithm and investigated reduction of antibiotics and therefore give insight in the effectiveness of PCT-guided therapy in the ED. Because of the high risk of selection bias in these studies, the results cannot be gen- eralized to either the general population of adult ED patients with AECOPD or to the general population of pediatric ED patients with FWS.

Limitations

It was not possible to pool the data of the 9 included studies because we found insufficient studies with comparable study populations. A

pooling of the results of PCT-guided therapy in adults may have resulted in a reduction of antibiotics in adult patients and to no effect in pediatric patients. However, because of the highly selective populations of the se- lected studies, these outcomes would not have had added value.

This review was performed in 2015. There are several studies being performed at the time of writing, which study PCT-guided therapy, for instance the NeoPInS trial [31]. These results are not available at this moment but may further clarify the value of PCT-guided therapy.

The review is primarily intended for emergency physicians. There- fore, only investigated ED-based studies were included. The ED is a unique clinical setting, which has specific problems such as the diagnos- tic uncertainty at a time when emergency treatment has to be initiated. Hence, the choice for this setting reduces the generalizability of the re- sults to other settings.

Five studies [10,13,14,17,23,25] reported on funding. Investigators of 3 of these studies received funding from the manufacturer of the PCT assay [10,13,17]. The authors of 3 studies [13,14,17] reported receiving payments for speaking engagements, lecture fees, and consultancy work for the manufacturer of the PCT assay. Conflicts of interest might raise concern in the appreciation of the results [32].

Conclusions

Procalcitonin-guided therapy is a valuable strategy in antimicrobial stewardship and can theoretically reduce the number of unnecessary antibiotics prescribed to ED patients. However, protocol nonadherence is a significant problem in the prospective PCT-guided therapy studies. In adult patients with suspected Respiratory infections, PCT-guided therapy may reduce antibiotic prescriptions, without increasing adverse events. However, physician judgment is still crucial and cannot be re- placed by biomarkers in these patient populations based on the avail- able evidence. In pediatric patients, PCT-guided therapy was ineffective because nonadherence to the PCT-guided algorithm reverses

the theoretical reduction in antibiotics.

Procalcitonin-guided therapy can only become Standard therapy in the ED when it is validated in a Representative sample. In addition, addi- tional evidence on the physiologic properties of PCT may result in more confidence in PCT-guided algorithms.

Supplementary data to this article can be found online at http://dx.

doi.org/10.1016/j.ajem.2016.03.065.

Acknowledgments

The authors thank W.M. Bramer, BSc, librarian of the Erasmus Uni- versity Medical Center, for the synthesis of the literature search.

References

1. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, et al. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and sep- tic shock, 2012. Intensive Care Med 2013;39(2):165-228.
2. Cosgrove SE, Carmeli Y. The impact of Antimicrobial resistance on health and eco- nomic outcomes. Clin Infect Dis 2003;36(11):1433-7.
3. Laxminarayan R, Duse A, Wattal C, Zaidi AK, Wertheim HF, Sumpradit N, et al. Anti- biotic resistance-the need for global solutions. Lancet Infect Dis 2013;13(12): 1057-98.
4. Van Boeckel TP, Gandra S, Ashok A, Caudron Q, Grenfell BT, Levin SA, et al. Global an- tibiotic consumption 2000 to 2010: an analysis of national pharmaceutical sales data. Lancet Infect Dis 2014;14(8):742-50.
5. Spellberg B, Bartlett J, Wunderink R, Gilbert DN. Novel approaches are needed to de- velop tomorrow’s antibacterial therapies. Am J Respir Crit Care Med 2015;191(2): 135-40.
6. Dellit TH, Owens RC, McGowan Jr JE, Gerding DN, Weinstein RA, Burke JP, et al. In- fectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicro- bial stewardship. Clin Infect Dis 2007;44(2):159-77.
7. Assicot M, Gendrel D, Carsin H, Raymond J, Guilbaud J, Bohuon C. High serum procalcitonin concentrations in patients with sepsis and infection. Lancet 1993; 341(8844):515-8.
8. Soni NJ, Samson DJ, Galaydick JL, Vats V, Pitrak DL, Aronson N. Procalcitonin-guided antibiotic therapy. Rockville (MD): Agency for Healthcare Research and Quality (US); Report No.: 12(13)-EHC124-EF; 2012 Oct.
9. Manzano S, Bailey B, Girodias JB, Galetto-Lacour A, Cousineau J, Delvin E. Impact of procalcitonin on the management of children aged 1 to 36 months presenting with fever without source: a randomized controlled trial. Am J Emerg Med 2010; 28(6):647-53.
10. Lacroix L, Manzano S, Vandertuin L, Hugon F, Galetto-Lacour A, Gervaix A. Impact of the lab-score on antibiotic prescription rate in children with fever without source: a randomized controlled trial. PLoS One 2014;9(12).
11. Lai CC, Chen SY, Wang CY, Wang JY, Su CP, Liao CH, et al. Diagnostic value of procalcitonin for bacterial infection in elderly patients in the emergency depart- ment. J Am Geriatr Soc 2010;58(3):518-22.
12. Bressan S, Andreola B, Zucchetta P, Montini G, Burei M, Perilongo G, et al. Procalcitonin as a predictor of renal scarring in infants and young children. Pediatr Nephrol 2009;24(6):1199-204.
13. Christ-Crain M, Jaccard-Stolz D, Bingisser R, Gencay MM, Huber PR, Tamm M, et al. Effect of procalcitonin-guided treatment on antibiotic use and outcome in lower re- spiratory tract infections: cluster-randomised, single-blinded intervention trial. Lan- cet 2004;363(9409):600-7.
14. Christ-Crain M, Stolz D, Bingisser R. Procalcitonin guidance of antibiotic therapy in community-acquired pneumonia: a randomized trial. Am J Respir Crit Care Med 2006;174(1):84-93.2006.
15. Bouadma L, Luyt CE, Tubach F, Cracco C, Alvarez A, Schwebel C, et al. Use of procalcitonin to reduce patients’ exposure to antibiotics in intensive care units (PRORATA trial): a multicentre randomised controlled trial. Lancet 2010; 375(9713):463-74.
16. Briel M, Schuetz P, Mueller B, Young J, Schild U, Nusbaumer C, et al. Procalcitonin- guided antibiotic use vs a standard approach for acute respiratory tract infections in primary care. Arch Intern Med 2008;168(18):2000-7 [discussion 7-8].
17. Schuetz P, Christ-Crain M, Thomann R, Falconnier C, Wolbers M, Widmer I, et al. Ef- fect of procalcitonin-based guidelines vs standard guidelines on antibiotic use in lower respiratory tract infections: the ProHOSP randomized controlled trial. JAMA 2009;302(10):1059-66.
18. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for sys- tematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 2009; 151(4):264-9 [W64].
19. Booth A, Clarke M, Ghersi D, Moher D, Petticrew M, Stewart L. An international reg- istry of systematic-review protocols. Lancet 2011;377(9760):108-9.
20. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155(8):529-36.
21. Drozdov D, Schwarz S, Kutz A, Grolimund E, Rast AC, Steiner D, et al. Procalcitonin and pyuria-based algorithm reduces antibiotic use in urinary tract infections: a ran- domized controlled trial. BMC Med 2015;13(1).
22. Stolz D, Christ-Grain M, Bingisser R, Leuppi J, Miedinger D, Muller C, et al. Antibiotic treatment of Exacerbations of COPD: a randomized, controlled trial comparing procalcitonin-guidance with standard therapy. Chest 2007;131(1):9-19.
23. Tang J, Long W, Yan L, Zhang Y, Xie J, Lu G, et al. Procalcitonin guided antibiotic therapy of acute exacerbations of asthma: A randomized controlled trial. BMC Infect Dis 2013;13(1).
24. Schulz KF. Subverting randomization in controlled trials. JAMA 1995;274(18): 1456-8.
25. Long W, Deng X, Zhang Y, Lu G, Xie J, Tang J. Procalcitonin guidance for reduction of antibiotic use in low-risk outpatients with community-acquired pneumonia. Respirology 2011;16(5):819-24.
26. Baer G, Baumann P, Buettcher M, Heininger U, Berthet G, Schafer J, et al. Procalcitonin Guidance to Reduce Antibiotic Treatment of Lower Respiratory Tract Infection in Children and Adolescents (ProPAED): a randomized controlled trial. PLoS One 2013;8(8).
27. Kristoffersen KB, Sogaard OS, Wejse C, Black FT, Greve T, Tarp B, 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):481-7.
28. Nobre V, Harbarth S, Graf JD, Rohner P, Pugin J. Use of procalcitonin to shorten anti- biotic treatment duration in septic patients: a randomized trial. Am J Respir Crit Care Med 2008;177(5):498-505.
29. de Jong E, van Oers JA, Beishuizen A, Vos P, Vermeijden WJ, Haas LE, et al. Efficacy 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.
30. Gilbert D. serum procalcitonin levels: comment on “effectiveness and safety of procalcitonin-guided antibiotic therapy in lower respiratory tract infections in ‘real life'”. Arch Intern Med 2012;172(9):722-3.
31. Stocker M, Hop WC, van Rossum AM. Neonatal procalcitonin intervention study (NeoPInS): effect of procalcitonin-guidED decision making on duration of antibiotic therapy in suspected neonatal early-onset sepsis: a multi-centre randomized supe- riority and non-inferiority intervention study. BMC Pediatr 2010;10:89.
32. Friedman LS, Richter ED. Relationship between conflicts of interest and research re- sults. J Gen Intern Med 2004;19(1):51-6.