Archives of Academic Emergency Medicine. 2021; 9(1): e69

REV I EW ART I C L E

Accuracy of CREST Guideline in Management of Cellulitis
in Emergency Department; a Systematic Review and Meta-
analysis
Hossein Akhavan1, Seyed Reza Habibzadeh2, Fatemeh Maleki3, Mahdi Foroughian2, Sayyed Reza Ahmadi2,
Reza Akhavan2, Bita Abbasi4, Behzad Shahi5∗, Navid Kalani6, Naser Hatami7, Amir Mangouri8, Sheida
Jamalnia9

1.. Department of Pediatrics, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran.

2.. Department of Emergency Medicine, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran.

3.. Department of Emergency Medicine, Faculty of Medicine, Birjand University of Medical Sciences, Birjand, Iran.

4.. Department of Radiology, Faculty of Medicine, Mashhad University of Medical sciences, Mashhad, Iran.

5.. Department of Emergency Medicine, Faculty of Medicine, Zahedan University of Medical Sciences, Zahedan, Iran.

6.. Research Center for Social Determinants of Health, Jahrom University of Medical Sciences, Jahrom, Iran.

7.. Student Research Committee, Jahrom University of Medical Sciences, Jahrom, Iran.

8.. Division of Vascular Surgery and Endovascular Therapy, Department of General Surgery, Sina Hospital, Tehran University of Medical
Sciences, Tehran, Iran.

9.. Medical Journalism Department, Shiraz University of Medical Sciences, Shiraz, Iran.

Received: September 2020; Accepted: October 2021; Published online: 3 November 2021

Abstract: Introduction: Skin and soft tissue infections are important causes of outpatient visits to medical clinics or hos-
pitals. This study aimed to review the literature for the accuracy of Clinical Resource Efficiency Support Team
(CREST) guideline in management of cellulitis in emergency department. Methods: Studies that had evaluated
cellulitis patients using the CREST guideline were quarried in Scopus, Web of Science, and PubMed database,
from 2005 to the end of 2020. The quality of the studies was evaluated using Scottish Intercollegiate Guideline
Network (SIGN) checklist for cohort studies. Pooled area under the receiver operating characteristic curve (AU-
ROC) of CREST guideline regarding the rate of hospital stay more than 24 hours, rate of revisit, and appropriate-
ness of antimicrobial treatment in management of cellulitis in emergency department was evaluated. Results:
Seven studies evaluating a total of 1640 adult cellulitis patients were finally entered to the study. In evaluation of
the rate of the appropriate treatment versus over-treatment, the pooled AUROC was estimated to be 0.38 (95%
confidence interval (CI): 0.06 – 0.82), indicating low accuracy (AUROC lower than 0.5) of guideline for antimicro-
bial choice. CREST II patients had a significantly lower odds ratio (OR) of revisiting the Emergency Department,
OR=0.21 (95% CI: 0.009 – 0.47). Pooled AUROC value of 0.86 (CI95%: 0.84 – 0.89) showed accuracy of the CREST
classification in prediction of being hospitalized more or less than 24 hours. Conclusion: CREST classification
shows good accuracy in determining the duration of hospitalization or observation in ED but it could lead to
inevitable over/under treatment with empirical antimicrobial agents.

Keywords: Cellulitis; emergency service, hospital; systematic review; skin diseases, bacterial; anti-bacterial agents

Cite this article as: Akhavan H, Habibzadeh S R, Maleki F, Foroughian M, Ahmadi S R, Akhavan R, Abbasi B, Shahi B, Kalani N, Hatami N,

Mangouri A S, Jamalnia S. Accuracy of CREST Guideline in Management of Cellulitis in Emergency Department; a Systematic Review and

Meta-analysis. Arch Acad Emerg Med. 2021; 9(1): e69. https://doi.org/10.22037/aaem.v9i1.1422.

∗Corresponding Author: Behzad Shahi; Department of Emergency Medicine,
Faculty of Medicine, Zahedan University of Medical Sciences, Zahedan,

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



H. Akhavan et al. 2

1. Introduction

Skin and soft tissue infections are important causes of out-

patient visits to medical clinics or hospitals. These infections

have a wide range of symptomatology and etiology that could

even be life threatening in some cases (1). Cellulitis is an

acute infection of the skin and soft tissues. Subcutaneous tis-

sues show redness, pain, and swelling in the affected area.

The most common etiology is Staphylococcus aureus bac-

teria, followed by Streptococcus pyogenes and many other

gram-positive cocci and rarely some gram-negative germs

(2). Clinical evaluation of the severity of the infection is very

important and decisive; however, the need for different di-

agnostic and therapeutic algorithms to guide physicians in

reaching the right and appropriate decision has not been

fully addressed (3). Koerner et al. (2011) reviewed the recent

attempts in classification of cellulitis cases and stated that

primary classifications were assorting cases based on the site

of the infection; while further studies suggested more com-

prehensive guidelines as well as the Eron criteria (4). Later,

Clinical Resource Efficiency Support Team (CREST) was de-

veloped, based on the Eron recommendations, addressed as

Eron/CREST classification by some authors (5), with an easy

method of classification for clinical application (6). But since

systemic sepsis has not been fully considered in this guide-

line, some researchers have doubted its application in clini-

cal practice (4). Since no study has pooled the clinical out-

comes of the CREST application, this study aimed to review

the literature on the accuracy of CREST guideline in manage-

ment of cellulitis in emergency department.

2. Methods

2.1. Study design and setting

This study was performed in adherence to the guidelines of

the Preferential Cases of The Report for Systematic Review

and Meta-Analysis (PRISMA) and is a systematic review and

meta-analysis of the existing literature on the accuracy of

CREST guideline in treatment of cellulitis, published in peer-

reviewed journals. English language studies were quarried

among all articles published from 2005 to the end of 2020,

on the topic of cellulitis, since CREST guideline was first es-

tablished in 2005. Studies which evaluated their study pop-

ulation using the CREST guideline were searched in Scopus,

Web of Science, and PubMed databases by two researchers

using the keywords of "CREST ", "cellulitis" and "bacterial

skin infection". Only articles about cellulitis that contained

Iran. Email: mr.shahi87@yahoo.com. Tel: 00989151913501, ORCID:
http://orcid.org/0000-0001-9884-1542.

the keywords of CREST were included in initial search.

2.2. Search strategy

In PubMed, using the search strategy of [(CREST OR Eron)

and (Cellulitis OR bacterial skin infection)], 44 results were

found. Along with multiple Scopus and Web of Science

search results found using the same strategy, in the initial

search, 135 potentially relevant articles were retrieved, 127

of which remained after removing the duplicate items. Then,

articles were selected based on the title and abstract, a list of

abstracts was prepared. After hiding the details of the arti-

cles such as the author’s name, the name of the journal, etc.,

the full texts of the articles were given to 2 trained researchers

to review the text of articles. Each article was reviewed by 2

independent researchers and in case of rejection of the arti-

cles by both researchers, the reason was mentioned and in

case of disagreement between them, the article was judged

by a third person. 115 articles were excluded from the study

due to irrelevance or not containing CREST guideline in their

methodology. Finally, 12 articles met the inclusion criteria

and entered the quality assessment process.

2.3. Quality assessment

A checklist of Scottish intercollegiate guidelines network

(SIGN) was used to evaluate the quality of these articles (7).

Seven articles had acceptable quality for being included in

study.

2.4. Statistical analysis

A web-based calculator was used to determine Area under

the Receiver operating characteristic (ROC) Curve (AUROC)

of the overall accuracy of CREST in predicting the rate of hos-

pital stay more than 24 hours (8). To combine the event rates

or values of AUROC with respect to the percentage or stan-

dard error of the values, the weighted average and the ran-

dom effects were used in the meta-analysis due to the het-

erogeneity of the studies. The I2 index and the Cochran test

were used to examine the heterogeneity between the results.

The Egger’s test and funnel plot were used to examine the

publication bias in the Revman software version 5.4.1. In

this meta-analysis, AUROC of CREST in predicting the stud-

ied variables was estimated and entered in a random effects

model, due to high heterogeneity (I2=99%).

3. Results

3.1. Characteristics of included studies

Seven studies were finally entered into the quantitative syn-

thesis in this meta-analysis, as shown in figure 1. Six stud-

ies were prospective cohort studies and one was retrospec-

tive. Two studies were conducted in Iran, 3 studies in USA,

and 2 in Scotland. Study setting was Emergency department

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



3 Archives of Academic Emergency Medicine. 2021; 9(1): e69

(ED) in 4 studies; while 3 of the studies evaluated hospitalized

patients (9-11). A total of 1640 patients were studied in our

meta-analysis. All studies evaluated adult subjects (over 18

years old). We were not able to classify studies into subgroups

for meta-analysis to see the possible differences between old

and young adults. All studies had the same inclusion crite-

rion, which was diagnosis of cellulitis. Although evaluating

distinct areas of cellulitis involvement, none of the studies

included facial cellulitis. Various outcomes were measured

in different studies, so we could not include all studies ina

quantitative synthesis for a particular outcome (Table 1).

3.2. Accuracy of CREST guideline

For hospitalization length
Results showed pooled AUROC value of 0.86 (95%CI: 0.84 –

0.89) for CREST guideline regarding hospitalization length

based on the combination of results from 4 studies ref-

erenced in figure 2. But there was a high heterogeneity

(I2=98%). No further subgroup analysis was possible to de-

termine the source of heterogeneity.

For empirical antimicrobial choices
Appropriateness of empirical antimicrobial choices was as-

sessed in 3 studies (9-11) through evaluating biological cul-

tures, as shown in Table 3. Results were summarized as

undertreatment, overtreatment, or appropriate treatment in

studies. As shown in figure 3, the pooled AUROC of CREST

guideline for empirical antimicrobial choices was estimated

to be 0.38, (95%CI: 0.06 – 0.82).

For revisiting after being discharged
As shown in figure 4, revisiting after being discharged from

the emergency department was evaluated in studies by

Claeys (2014) and Abiri. Comparisons were only available for

CREST II. CREST II patients had a significantly low odds ratio

(OR) of revisiting the ED, OR=0.21 (95%CI: 0.009 – 0.47).

Publication bias
To assess the publication bias, the funnel plot was visually in-

spected for asymmetry, as shown in figure 5.

4. Discussion

Our study revealed that the pooled AUROC for evaluating the

rate of appropriate treatment versus overtreatment was 0.38

(95% CI: 0.06–0.82), indicating low accuracy (AUROC less

than 0.5) of CREST guideline for antimicrobial choice. The

odds of revisiting the emergency department were consider-

ably lower in CREST II patients, with an OR of 0.21 (95%CI:

0.009–0.47). The CREST classification was shown to be ac-

curate for being hospitalized for more than 24 hours with a

pooled AUROC of 0.86 (95%CI: 0.84 – 0.89). Soft tissue infec-

tions are a common group of infections that are often mild

to moderate in severity and are easily treatable. Their etio-

logical diagnosis is often difficult and unnecessary in most

cases of cellulitis, where the patient has mild symptoms. Our

study was a systematic review of the studies that used CREST

guideline in management of cellulitis. There were few stud-

ies conducted in this area and only 7 studies were included.

in our qualitative review. Further assessment of study out-

comes showed interesting findings in evaluation of empirical

antimicrobial choices and hospitalization outcomes. Labo-

ratory investigations are suggested for CREST II-IV classes

and most Class I CREST classified patients get outpatient

care with first line antibiotic choice of oral Flucloxacillin 500

mg per day. Patients classified in class II or higher classes

may receive IV therapy (5, 6). Although pooling the data of IV

versus oral treatment was not possible in our study, duration

of hospital stay in ED or observation units were evaluated in

some of our included studies, which revealed that patients

with higher CREST classification had higher rates of staying

in the hospital for more than 24 hours based on the AUROC.

Staying in the hospital and observation units for more than

24 hours may also be showing the need for IV antibiotic ther-

apy based on the findings of Claeys et al(12, 13) and Abiri et

al. (14). But further laboratory investigations were not pre-

sented in all studies, as some of studies were evaluating out-

patient cases, which had CREST I and II classes.

Our study showed that the pooled AUROC for appropriate-

ness of treatment was 0.38 (95%CI: -0.06 – 0.82), indicating

no significant difference in appropriate or under/over treat-

ment; thus, reconsiderations are needed for treatment of

middle classes. The closeness of the definitions provided for

classes I to IV may be the reason for uncertainty in the de-

cisions for treatment of these cases. However, in our study,

there was a large spatial heterogeneity in the included stud-

ies, which could suggest the role of differences in the pattern

of antibiotic resistance of gram-positive bacteria as a hetero-

geneity factor. Various studies conducted in Iran show a high

rate of antibiotic resistance in Iran (15, 16), the pattern of

which might be different from the United States and Europe

(17).

5. Limitations

There was a high amount of heterogeneity in some of our

syntheses, but the small number of the included studies that

had evaluated distinct types of study outcomes did not allow

us to have a comprehensive review of the cause of the het-

erogeneity. Since all included studies had similar inclusion

criteria for their patient recruitment, various factors could

have affected the results, causing heterogeneity. One of these

factors that our study shows to modify the heterogeneity be-

tween the studies is the different study settings in the in-

cluded articles. Study setting was emergency department

(ED) in 4 studies; while 3 studies evaluated hospitalized pa-

tients (9-11).

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



H. Akhavan et al. 4

6. Conclusion

CREST classification demonstrates good precision in decid-

ing the duration of hospitalization or observation at the ED;

however, the reliability of this guideline in antimicrobial

agent choice or route of antibiotic administration remains

unclear; and using these classifications had not been able to

prevent over/undertreatment with antibiotics, which might

be due to inadequate and vague description and potentially

overlapping definition of each class.

7. Declarations

7.1. Acknowledgments

We would like to thank the Clinical Research Development

Unit of Peymanieh Educational and Research and Therapeu-

tic Center of Jahrom University of Medical Sciences for pro-

viding facilities for this work.

7.2. Conflict of interest statement

The authors have declared that no competing interests exist.

7.3. Funding/Support

This research did not receive any grant from funding agencies

in the public, commercial, or non-profit sectors.

7.4. Authors’ contribution

HA, SH and SRH conceptualized the study questions and

performed revisions. NK, NH, AM and SJ performed the

searches. FM, MF and RA, BA, BSH conducted the statistical

analyses. Other authors provided the draft of the manuscript.

7.5. Ethical Considerations

All ethical principles are considered in this article.

References

1. Bystritsky RJ. Cellulitis. Infectious Disease Clinics.

2021;35(1):49-60.

2. Norimatsu Y, Ohno Y. Predictors for readmission due to

cellulitis among Japanese patients. The Journal of Der-

matology. 2021;48(5):681-4.

3. Cranendonk D, Lavrijsen A, Prins J, Wiersinga W. Cel-

lulitis: current insights into pathophysiology and clini-

cal management. The Netherlands Journal of Medicine.

2017;75(9):366-78.

4. Koerner R, Johnson AP. Changes in the classification and

management of skin and soft tissue infections. Journal of

Antimicrobial Chemotherapy. 2011;66(2):232-4.

5. Fulton R, Doherty L, Gill D, Harney A, Harper C, Jenk-

inson H. Guidelines on the management of cellulitis in

adults. Northern Ireland: CREST. 2005.

6. Vijayalakshmi B, Ganapathy D. Medical management of

cellulitis. Research Journal of Pharmacy and Technology.

2016;9(11):2067-70.

7. Miller J. The Scottish intercollegiate guidelines network

(SIGN). The British Journal of Diabetes & Vascular Dis-

ease. 2002;2(1):47-9.

8. Eng J. ROC analysis: web-based calculator for

ROC curves. Baltimore: Johns Hopkins Univer-

sity. 2017 [updated 2014 March 19. Available from:

http://www.jrocfit.org.

9. Marwick C, Broomhall J, McCowan C, Phillips G,

Gonzalez-McQuire S, Akhras K, et al. Severity assessment

of skin and soft tissue infections: cohort study of man-

agement and outcomes for hospitalized patients. Journal

of Antimicrobial Chemotherapy. 2011;66(2):387-97.

10. Marwick C, Rae N, Irvine N, Davey P. Prospective study of

severity assessment and management of acute medical

admissions with skin and soft tissue infection. Journal of

antimicrobial chemotherapy. 2012;67(4):1016-9.

11. Hashem NG, Hidayat L, Berkowitz L, Venugopalan

V. Management of skin and soft-tissue infections at

a community teaching hospital using a severity-of-

illness tool. Journal of Antimicrobial Chemotherapy.

2016;71(11):3268-75.

12. Claeys KC, Lagnf AM, Patel TB, Jacob MG, Davis SL, Ry-

bak MJ. Acute bacterial skin and skin structure infec-

tions treated with intravenous antibiotics in the emer-

gency department or observational unit: experience at

the Detroit Medical Center. Infectious diseases and ther-

apy. 2015;4(2):173-86.

13. Claeys KC, Zasowski EJ, Lagnf AM, Sabagha N, Levine DP,

Davis SL, et al. Development of a risk-scoring tool to de-

termine appropriate level of care in acute bacterial skin

and skin structure infections in an acute healthcare set-

ting. Infectious diseases and therapy. 2018;7(4):495-507.

14. Abiri S, Foroughian M, Akbar H, Mehramiz NJ, Hatami N,

Ameri AA, et al. The investigation of valid criteria for hos-

pitalization and discharge in patients with limb cellulitis:

a prospective cohort study. Journal of Emergency Prac-

tice and Trauma. 2020;6(2):55-8.

15. Parastan R, Kargar M, Solhjoo K, Kafilzadeh F. Staphylo-

coccus aureus biofilms: Structures, antibiotic resistance,

inhibition, and vaccines. Gene Reports. 2020;20:100739.

16. Moravvej Z, Estaji F, Askari E, Solhjou K, Naderi Nasab M,

Saadat S. Update on the global number of vancomycin-

resistant Staphylococcus aureus (VRSA) strains. Interna-

tional Journal of Antimicrobial Agents. 2013;42(4):370-1.

17. Prestinaci F, Pezzotti P, Pantosti A. Antimicrobial resis-

tance: a global multifaceted phenomenon. Pathogens

and global health. 2015;109(7):309-18.

18. Abbasi M, Abiri S, Ghanbari M-J, Orimi MH. Evaluation

of Crest Guideline Validity For Diagnosis of Non-facial

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



5 Archives of Academic Emergency Medicine. 2021; 9(1): e69

Cellulites. Journal of Dermatologic Research and Ther-

apy. 2016;1(2):05.

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



H. Akhavan et al. 6

Table 1: Characteristic of studies included in the meta-analysis

Study Design Setting Outcomes N Inclusion criteria
Abbasi, 2016 (18) Iran Prospective cohort ED Hospitalization less or more

than 24 h revisit in 1 week
89 Non-facial cellulitis

Abiri, 2020(14) Iran Prospective cohort ED Hospitalization less or more
than 24 h revisit in 1 week

100 Limb cellulitis

Hashem, 2015(11) USA Retrospective Hospitalized Appropriateness of empirical
antimicrobial choices clinical

response

369 Patients admitted with
cellulitis

Claeys, 2015 (12) USA Retrospective cohort ED/or observation
units

96-h ED
revisit/hospitalization

308 Adult cellulitis patients
with less than 24 h of IV

antibiotics without
hospital admission

Claeys, 2018(13) USA Observational cohort
study

ED or observation
units

Area-under-the-
receiver-operating-

characteristic-curve (AUROC)
analysis of ED/OU versus

inpatient

506 Diagnosis of acute
bacterial skin and skin

structure infections

Marwick, 2011(9)
Scotland

Retrospective cohort Hospitalized Appropriateness of empirical
antimicrobial choices

189 Received antibiotic
treatment for cellulitis in

hospital
Marwick, 2012 (10)

Scotland
Cohort Acute Appropriateness of empirical

antimicrobial choices
79 Adult patients with

cellulitis
NA: Not Addressed; ED: Emergency Department; OU: Outpatient.

Table 2: Quality of studies included in the meta-analysis based on the Scottish intercollegiate guidelines network (SIGN) checklist

Study Appropriate
and clearly

focused
question

Predicting
the

outcome at
the time of
enrolment

Lost to
follow up

status
addressed

Clearly
defined

outcomes

A reliable
method of
exposure

assess-
ment

Evidence
of

outcome
assess-
ment

Exposure
level or

prognostic
factor is
assessed

more than
once

The main
potential

confounders are
identified and

taken into
account in the

design and
analysis

Have con-
fidence

intervals
been

provided?

Minimization
of the risk
of bias or

confound-
ing?

Clear
evidence

of an asso-
ciation

between
exposure

and
outcome?

Are the results
of this study

directly
applicable to the

patient group
targeted in this

guideline?

Abbasi, 2016 Y Y NA Y Y N N N Y N Y Y
Abiri, 2020 Y Y NA Y Y N N N N Y Y Y

Hashem, 2015 Y Y Y Y Y Y Y Y N Y Y Y
Claeys, 2015 Y Y Y Y Y Y Y Y Y Y Y Y
Claeys, 2018 Y Y Y Y Y Y Y Y Y Y Y Y

Marwick, 2011 Y Y Y Y Y N Y N Y Y Y Y
Marwick, 2012 Y Y Y Y Y Y Y Y Y Y Y Y

N: no; NA: not addressed; Y: yes.

Table 3: Quality of antimicrobial treatment based on the Clinical Resource Efficiency Support Team (CREST) guideline

Treatment CREST I CREST II CREST III CREST IV
Marwick, 2012 Appropriate 0(0) 4(4.6) 1(1.15) 18(20.69)

Under/Over 19(21.84) 33(37.93) 3(3.45) 9(10.34)
Marwick, 2011 Appropriate 57(33.53) 20(11.76) 10(5.88) 1(0.59)

Under/Over 12(7.06) 36(21.18) 23(13.53) 11(6.47)
Hashem, 2015 Appropriate 8(4) 65(32.5) 14(7) 3(1.5)

Under/Over 60(30) 37(18.5) 10(5) 3(1.5)
Total Appropriate 122(17.09) 113(15.83) 36(5.04) 41(5.74)

Under/Over 122(17.09) 175(24.51) 62(8.68) 43(6.02)
Data are presented as number (%) if they were available in the studies.

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



7 Archives of Academic Emergency Medicine. 2021; 9(1): e69

Figure 1: PRISMA flow chart of study.

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



H. Akhavan et al. 8

Figure 2: Forest plot of Clinical Resource Efficiency Support Team (CREST) guideline’s accuracy for hospitalization length; less versus more

than 24 hours (based on pooled area under the Receiver Operating Characteristic (ROC) curve).

Figure 3: Forest plot of Clinical Resource Efficiency Support Team (CREST) guideline’s accuracy for appropriate antimicrobial treatment

(based on pooled area under the Receiver Operating Characteristic (ROC) curve).

Figure 4: Forest plot of Clinical Resource Efficiency Support Team (CREST) guideline’s accuracy for rate of revisit after being discharged (based

on pooled area under the Receiver Operating Characteristic (ROC) curve).

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem



9 Archives of Academic Emergency Medicine. 2021; 9(1): e69

Figure 5: Funnel plot of the study to assess publication bias.

This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0).
Downloaded from: http://journals.sbmu.ac.ir/aaem


	Introduction
	Methods
	Results
	Discussion
	Limitations
	Conclusion
	Declarations
	References